Induced pluripotent stem cells as custom therapeutics for retinal repair: Progress and rationale

Abstract

Human pluripotent stem cells have made a remarkable impact on science, technology and medicine by providing a potentially unlimited source of human cells for basic research and clinical applications. In recent years, knowledge gained from the study of human embryonic stem cells and mammalian somatic cell reprogramming has led to the routine production of human induced pluripotent stem cells (hiPSCs) in laboratories worldwide. hiPSCs show promise for use in transplantation, high throughput drug screening, “disease-in-a-dish” modeling, disease gene discovery, and gene therapy testing. This review will focus on the first application, beginning with a discussion of methods for producing retinal lineage cells that are lost in inherited and acquired forms of retinal degenerative disease. The selection of appropriate hiPSC-derived donor cell type(s) for transplantation will be discussed, as will the caveats and prerequisite steps to formulating a clinical Good Manufacturing Practice (cGMP) product for clinical trials.

1. Introduction

Efforts to provide personalized cell-based treatments for degenerative diseases were greatly aided by the advent of somatic cell reprogramming technology, which holds potential to generate patient-specific cells of any type for research and clinical purposes. The retina is well-suited to be the first site for this therapeutic application, as all of the major retinal cell types, including photoreceptors and retinal pigmented epithelium (RPE), have been successfully differentiated from human induced pluripotent cells (hiPSCs). In addition, insights into the donor cell types and differentiation stages needed to treat particular retinal diseases are available from transplantation studies using embryonic stem cells (ESCs) and primary retinal tissues, as well as iPSCs. As a result, strategies have been developed to usher hiPSCs into clinical trials for retinal disease, at least one of which will be realized in the near future using autologous donor cell sources. The capability for gene repair in hiPSCs further enhances their potential to be employed in autologous cell transplantation. However, challenges and concerns remain regarding transplantation of at least some hiPSC-derived retinal cell types. In particular, inherent difficulties in achieving clinical-grade status for hiPSC-derived photoreceptors will need to be addressed before these critical cell types can be used to treat outer retinal degenerative diseases such as age-related macular degeneration (AMD) and retinitis pimentosa (RP).

2. Background

Loss of vision exacts a tremendous societal burden in terms of quality of life, decrease in productivity and health care expenditure. Retinal degenerative diseases such as AMD and RP contribute to a large proportion of such costs, with the certainty of a dramatic increase in the prevalence of AMD given the growing population of senior citizens. As such, there is an urgent need to develop strategies for retinal cell survival, repair and replacement for these and other retinal disorders.

The relative simplicity of the retina compared to other regions of the central nervous system, its surgical accessibility, and the existence of sophisticated, noninvasive anatomical and functional assessment tools make the retina particularly amenable to the application of novel, cutting edge treatment modalities. To date, these inherent properties have been exploited for the development and delivery of gene- and drug-based therapeutics. For example, gene therapy has been used in clinical trials to treat one form of Leber congenital amaurosis (Jacobson et al., 2012; Maguire et al., 2009), while pharmacological agents have been introduced via intraocular injection or encapsulated cell technology in an attempt to mitigate diseases such as wet AMD and RP (Birch et al., 2013; Haller, 2013; Kauper et al., 2012). However, for these types of treatments to succeed, retinal cell types (e.g., photoreceptors and RPE) and structural connections must still exist in vivo.

In disease states where there is significant destruction of retinal architecture and/or cell loss, a cell bypass or replacement approach is required. The prospect of cell bypass has been explored using epiretinal prostheses (such as the FDA-approved Argus II) or subretinal micro-photodiode arrays, which have succeeded in restoring some visual function to blind individuals (da Cruz et al., 2013; Zrenner et al., 2011). While such results constitute a major accomplishment, retinal prosthetics and other potential photoreceptor bypass methods, including optogenetics, may not be the optimal or most convenient choice for many patients suffering from vision loss. A potential alternative is the use of cell-based therapy to replace host cells within the neural retina (NR) and/or RPE, provided a suitable supply of donor cells can be identified. Human prenatal retinal tissue was one of the first donor sources to be examined in patients. In one report, subretinal injection of a suspension of prenatal NR cells in RP patients resulted in a transient improvement in visual acuity (Humayun et al., 2000). Improvements in visual acuity were also documented in a subset of AMD and RP patients who underwent submacular transplantation of sheets of prenatal NR with attached RPE (Radtke et al., 2008). However, the use of human fetal tissue is problematic due to ethical issues surrounding its procurement, as well as limitations in the amount of donor material that can be obtained (Gamm et al., 2008b; Kelley et al., 1995).

Human pluripotent stem cells provide another potential donor source for retinal cell transplantation. Embryonic stem cells (ESCs) are obtained from the inner cell mass of the blastocyst and can be maintained and expanded indefinitely in culture as undifferentiated cells. When prompted to differentiate, ESCs have the theoretical capacity to produce any cell type, with the caveat that the cell-specific developmental program can be adequately recapitulated in vitro. In particular, cells of a retinal lineage have been derived from mouse, nonhuman primate and human ESCs (hESCs) (Ikeda et al., 2005; Lamba et al., 2006; Meyer et al., 2009; Osakada et al., 2008), and are the subject of a separate review in this issue (Reynolds and Lamba). In contrast to ESCs, induced pluripotent stem cells (iPSCs) can be derived from somatic cells of adult individuals, and therefore constitute a unique, powerful and patient-specific tool for modeling disease and developing cell-based treatments for retinal degenerative diseases. The present review will focus on the latter indication, with special attention given to data accumulated thus far with human iPSCs (hiPSCs). In addition, the challenge of generating clinical grade cells from hiPSCs and the indications for using them for retinal cell replacement will be examined.

3. Generation of RPE and neural retinal cell types from iPSCs

3.1 Production of iPSCs

Landmark studies describing the reprogramming of vertebrate somatic cells began more than 50 years ago (Gurdon, 1962) and culminated in the discovery of the essential transcription factors that were necessary and sufficient to convert a somatic cell to a state highly similar to that of a pluripotent embryonic stem cell in mice (Takahashi and Yamanaka, 2006). Shortly afterward, both the Yamanaka and Thomson laboratories succeeded in using integrating retroviruses to deliver a combination of either OCT4, SOX2, KLF4 and cMYC (Takahashi et al., 2007) or OCT4, SOX2, NANOG and LIN28 (Yu et al., 2007) to reprogram human fibroblasts to a pluripotent state with the requisite capacity to yield progeny indicative of the three germ layers. Since those initial reports, the production of normal and disease-specific hiPSC lines has escalated rapidly [for review, see Egashira et al., 2013; Grskovic et al., 2011]. The ability to recapitulate a pathological phenotype with hiPSCs in vitro is particularly noteworthy, as it has important applications for disease modeling and drug discovery.

A more ambitious goal is to generate pluripotent lines from an individual patient, repair any underlying genetic defect(s) ex vivo, and then transplant appropriate differentiated cell type(s) in an effort to advance autologous cell replacement therapies. Paramount in each of these applications of iPSC technology is the ability to target their differentiation toward cell types of interest. A potential concern is the finding that iPSCs can exhibit varying degrees of epigenetic, transcriptional and genomic changes when compared to hESCs [for review see Bilic and Izpisua Belmonte, 2012]. These differences appear to be of a stochastic nature and may stem from low programming efficiency, culture conditions and/or epigenetic memory, among other possible causes. As these questions regarding reprogramming technology continue to be investigated and satisfactorily addressed, it is likely that hiPSCs will be increasingly employed in patient therapies.

3.2 Production of RPE from hiPSCs

Following the successful generation of retinal cells from mouse and human ESCs, researchers applied analogous differentiation protocols to iPSCs derived from prenatal and adult fibroblasts, RPE and T lymphocytes and also succeeded in converting them to retinal cells (Buchholz et al., 2009; Buchholz et al., 2013; Carr et al., 2009; Hirami et al., 2009; Meyer et al., 2009; Phillips et al., 2012). A variety of methods have been used to generate RPE from iPSCs, most of which are based on ESC protocols [for review, see (Rowland et al., 2012)]. Most commonly, patches of RPE are allowed to spontaneously differentiate after mitogen withdrawal from confluent pluripotent stem cells in adherent cultures, which are subsequently isolated by manual dissection. A second method involves formation of suspended embryoid cell spheroids (“embryoid bodies”) from hiPSCs as a first step in the neuroectodermal differentiation process, followed by plating of the differentiating spheroids to a coated surface. Thereafter, adherent pigmented RPE patches are visualized over time and dissected away from unwanted cell types. Mixed cultures can be treated with exogenous factors to accelerate the production of large populations of spontaneously differentiated RPE. When dissociated en masse and re-plated, these highly proliferative monolayers of RPE show a tendency to outcompete contaminating cells over a series of passages (Buchholz et al., 2013). ESC- and iPSC-derived RPE will also grow in aggregate suspension as pigmented spheroids, similar to RPE spheroid cultures derived from human donor retinal tissue (Gamm et al., 2008a; Meyer et al., 2009).

Regardless of the method used to derive them, there are common criteria for evaluating stem cell-derived RPE populations, which include 1) formation of characteristic hexagonal cell morphology, 2) appearance of pigmentation, 3) establishment of apical/basal polarity, and 4) evidence of RPE functions such as phagocytosis of photoreceptor outer segments, tight junction formation, growth factor secretion, and/or vectorial fluid flow, among others [for review, see Bharti et al., 2011]. Lastly, iPSC-derived RPE should express signature genes and proteins consistent with bona fide prenatal and adult human RPE (Strunnikova et al., 2010), such as those involved in melanogenesis and retinoid recycling.

Using the embryoid body method, Singh et al. derived hiPSC-RPE from two patients bearing distinct mutations in BEST1, the gene responsible for Best vitelliform macular dystrophy (BVMD) (Singh et al., 2013b). When compared to RPE cultured from human prenatal eye, hiPSC-RPE from the BVMD patients and their unaffected siblings displayed identical cellular morphology, capacity for pigmentation and tight junction formation, and RPE gene and protein expression. Disease-specific functional differences were also demonstrated, with mutant hiPSC-RPE showing increased accumulation of autofluorescent material following chronic rod outer segment feeding, as well as altered fluid flux and delayed RHODOPSIN degradation when compared to control hiPSC-RPE cultures. These findings further demonstrate the ability of iPSC-derived cell types to approximate the morphological and functional standards set by prenatal human RPE cultures.

3.3 Neural retina differentiation from hiPSCs

In contrast to protocols aimed at generating RPE, methods for the production of human neural retinal (NR) cell types from pluripotent stem cells tend to require greater manipulation of the culture environment (Hirami et al., 2009; Lamba et al., 2010; Mellough et al., 2012; Meyer et al., 2011). Techniques differ in their reliance on endogenous versus exogenous factors to modulate key signaling pathways (e.g., Wnt, BMP and Nodal), with the purpose of guiding iPSCs in a stepwise fashion towards a retinal fate (Rowland et al., 2012; Meyer et al., 2011). Both adherent and 3-D aggregate methods of retinal differentiation (and combinations thereof) have been employed, each possessing distinct advantages and disadvantages with regard to complexity and expense, scalability, adaptability and capacity to yield specific retinal cell types. However, most culture systems support an orderly, highly conserved sequence of events that approximates normal retinal development.

In a seminal report published in 2011, the Sasai laboratory generated self-organizing 3-D structures from mouse ESC aggregates that mimicked in vivo optic cups to a remarkable degree (Eiraku et al., 2011).These structures displayed interkinetic nuclear migration, self-patterning into NR and RPE domains, and retinal stratification. 3-D optic vesicle-like structures (OVs) have also been reported using human iPSCs (Meyer et al., 2009; Phillips et al., 2012) and ESCs (Boucherie et al., 2013; Meyer et al., 2011; Meyer et al., 2009; Nakano et al., 2012). Building on an earlier study (Meyer et al., 2009), Meyer et al. (2011) showed in 2011 that human iPSCs and ESCs could generate neuroepithelial-like clusters of retinal progenitors with numerous characteristics of developing optic vesicles. Based on their distinct light microscopic appearances, these human pluripotent stem cell-derived OVs could be manually separated from coexistent populations of early forebrain neurospheres and cultured in isolation. Upon further differentiation, hiPSC- and hESC-OVs produced all major NR cell types in a time frame and sequence that resembled retinal development in vivo. Furthermore, RECOVERIN+ cells demonstrated a characteristic photoreceptor electrophysiological response after stimulation with membrane-permeable 8-Br-cGMP (Meyer et al., 2011). The potential for human pluripotent stem cells to produce OV structures was also shown by the Sasai laboratory in 2012 following adaptation of their mouse ESC optic cup protocol to yield OV structures from hESCs (Nakano et al., 2012). The 3-D hESC-OVs thus produced displayed a precise apical-basal orientation with retinal progenitors located in a region approximating the prenatal neuroblastic layer. Subsequently, retinal cell types appeared in a sequential fashion that self-stratified into organized tissues similar to the developing human retina.

Like OVs derived from hESCs, hiPSC-OVs have also been shown to self-assemble into rudimentary, multi-layered retinal tissues (Phillips et al., 2012). The hiPSC-derived tissue-like structures initially possessed a neuroblastic layer comprised of proliferating VSX2+ cells (Fig. 1A), which gives rise to an inner layer of BRN3+/HUc/d+/ TUJ1+ ganglion cells (Fig. 1B, C), an intermediate layer of retinal interneurons (e.g., CALRETININ+/BRN3+ amacrine cells and post-mitotic VSX2+ bipolar cells), and an outer layer of RECOVERIN+ photoreceptor-like cells (Fig. 1D,1E) that express synaptophysin (Fig. 1F). RECOVERIN+ cells begin to predominate in hiPSC-OV cultures by day 90, although by this time the discrete laminar structure of the OVs often dissipates (Fig. 2A). Gentle dissociation of hiPSC-OVs at this stage of differentiation facilitates the preparation of hiPSC-derived photoreceptors (Fig. 2B) and other neuroretinal cell types for further study and/or transplantation. Conversely, few glia are present in hiPSC-OVs at this time point (Fig. 2C). Consistent with their early birth during retinogenesis, cones represented the earliest photoreceptor cell type in these cultures, whereas rods were much less prevalent until later differentiation time points. In contrast, rods are abundant in mouse pluripotent stem cell cultures, likely due to the shorter maturation time needed for mouse vs. human retina (Eiraku et al., 2011).

Figure 1.

Optic vesicle-like structures (OVs) derived from human induced pluripotent stem cells (hiPSCs) can form layered retinal structures containing photoreceptor-like cells with potential to form synapses. (A) After 20 days of differentiation, proliferating Ki67+/VSX2+ retinal progenitor cells derived from hiPSCs form a structure resembling the optic vesicle in vivo. (B) An inner layer consisting of post-mitotic HuC/D+ neurons and an outer neuroblastic layer of Ki67+/VSX2+ retinal progenitors can be detected by 50 days. (C) A similar pattern is observed at day 50 with the retinal ganglion cell marker BRN3 and the retinal progenitor marker SOX2. (D) RECOVERIN+ cells with the morphology of immature photoreceptors are located in the outermost layer of the OV where they (E) send processes towards the inner neuronal layers of hiPSC-OVs by day 70. (F) Synaptic proteins such as SYNAPTOPHYSIN are expressed in RECOVERIN+ cells by 90 days, indicating the potential for synapse formation. Scale bar = 50 μm (panels A, B and C); 20 μm (panel D); 10 μm (panel E); 5 μm (panel F).

Figure 2.

(A) A large number of RECOVERIN+ cells is present in bulk cultures of isolated hiPSC-OVs at day 90. (B) Upon dissociation of day 90 hiPSC-OV cultures, an enriched population of CRX+/RECOVERIN+ photoreceptor-like cells can be obtained. (C) Compared to the number of RECOVERIN+ cells, relatively few GFAP+ glia are found in dissociated day 90 hiPSC-OVs. Scale bar =100 μm.

With regard to neuronal connectivity in differentiating hiPSC-OVs, the appearance and appropriate co-expression of pre- and post-synaptic markers such as VGLUT-1,GLUR2, SNAP-25, MGLUR6 and SYNAPSIN-1 indicated the potential for the formation of synapses (Fig. 1F). Thus, hiPSCs can serve not only as a source of individual NR cell types for transplantation, but perhaps also for retinal structures capable of more complex tissue reconstruction.

4. iPSC-based retinal transplantation strategies

4.1 Cell transplantation for the inner retina

The aforementioned studies verify that iPSCs can be differentiated into cells that are indisputably retinal and theoretically could be used as a source for transplantation. However, many challenges lie ahead in the quest to optimize hiPSC-based therapeutic strategies. These issues include the choice, characterization and preparation of the donor cell population and selection of disease model(s) for preclinical testing. Retinal ganglion cells (RGCs) are the first differentiated neuronal cell type to be detected in both hESC and hiPSC-derived retinal cultures, mirroring the cell birth order of normal mammalian retinogenesis. They are also the major cell type affected in glaucoma and other optic nerve pathologies; thus, there is considerable interest in developing methods to replace or regenerate RGCs in vivo (Johnson et al., 2011). To accomplish the lofty goal of complete RGC replacement, donor cells must extend processes out of the retina along the optic nerve and into the visual centers of the brain, where they would need to synapse with retinotopically arranged resident neurons. Although the degree of plasticity within these centers is probably underestimated, it remains to be seen if they can be “re-wired” sufficiently for the brain to make sense of information coming from donor RGCs. For these and other reasons, RGC replacement likely faces the most formidable hurdles of any donor retinal cell type. Alternative approaches aimed at axonal regeneration instead of whole cell replacement may be more fruitful, but such strategies require the viable presence of at least the cell bodies of host RGCs. Interestingly, it was recently shown that hematopoetic stem cells or mESCs can undergo fusion with damaged host ganglion and amacrine cells and transiently reprogram them to a proliferative precursor/pluripotent state (Sanges et al., 2013). These reprogrammed cells proved capable of differentiating into neurons with the phenotype of the original retinal cells, extending projections to the optic nerve, and partial restoring light-evoked responses.

4.2 RPE transplantation

Compared to diseases that cause RGC loss, degenerative diseases of the outer retina (i.e., RPE and/or photoreceptors) theoretically pose fewer challenges to cell replacement therapies, due in part to the highly localized, short-range intercellular connections they require. In particular, the outward simplicity of the RPE, with its monolayer structure and relative cell homogeneity, has made it an appealing target for repair strategies. Furthermore, the RPE layer is situated adjacent to the vestigial embryonic retinal ventricular space, and the relative ease of generating a transient retinal detachment at this site has been routinely exploited in the subretinal delivery of donor graft materials and other therapeutics. The RPE carries out a number of complex functions that are critical for maintenance of photoreceptor health and activity (Strauss, 2005) and, as mentioned previously, many of these functions have been recapitulated in vitro by RPE derived from hESCs and hiPSCs (Kokkinaki et al., 2011; Meyer et al., 2011; Osakada et al., 2009; Singh et al., 2013b).

The capacity to generate substantial amounts of highly enriched, functional RPE from pluripotent stem cells has made this cell type a prime candidate for transplantation; however, there is limited data on the efficacy of hiPSC-RPE in animal models of retinal disease. One model system that has been used to test hiPSC-RPE is the Royal College of Surgeons (RCS) rat. The dystrophic RCS rat has a mutation in a gene crucial for photoreceptor outer segment phagocytosis (Mertk), which results in primary loss of RPE and secondary photoreceptor degradation. Similar to studies using hESC-RPE and other retinal and non-retinal cell types (Gamm et al., 2007; Idelson et al., 2009; Lund et al., 2006; McGill et al., 2007; Wang et al., 2008), injection of dissociated hiPSC-RPE into the subretinal space of RCS rats resulted in long-term photoreceptor survival and retention of visual function (Carr et al., 2009). Similarly, Li et. al. treated 2 day postnatal SCID Rpe65^rd12/ Rpe65^rd12 mice with a subretinal bolus of hiPSC-RPE and showed that donor cells integrated into the host RPE. ERG recordings revealed that the b-wave response was maintained in the treated eye compared to the untreated fellow eye (Li et al., 2012). Furthermore, no evidence of tumor formation was seen up to 6 months following transplantation.

Although the RCS rat and other rodent models of RPE degeneration have value for testing RPE replacement strategies, the similarity in therapeutic response observed from a wide array of subretinal manipulations may in part stem from a general neuroprotective mechanism. Nevertheless, efficacy studies in RCS rats have been sufficient to support the initiation of clinical trials for atrophic AMD and Stargardt macular dystrophy using subretinal injections of dissociated hESC-RPE. In addition, a trial using hiPSC-RPE to treat exudative AMD has been announced (Cyranoski, 2013). Of note, the hiPSC-RPE trial utilizes monolayer sheets of cells, which provides assurance of proper RPE layer organization prior to implantation. This detail is potentially critical, since cell polarization and tight junction formation are necessary for many RPE functions (Diniz et al., 2013). Looking to the future, prefabricated RPE monolayers could also serve as a platform for combined transplantation of photoreceptors and RPE, which will likely be necessary to treat diseases where both of these cell types are lost (Hynes and Lavik, 2010; Rowland et al., 2013).

4.3 Photoreceptor transplantation

While replacement of RPE alone may be beneficial for certain disease indications, a source of photoreceptors will be needed to treat retinal degenerative diseases where there is extensive photoreceptor degeneration (e.g., most RP and RP-like disorders) with or without RPE involvement. In one of the few transplantation studies published to date using retinal hiPSCs, human photoreceptors expressing GFP under the control of the interphotoreceptor retinoid binding protein (IRBP) receptor were isolated by FACS and injected into the subretinal space of wildtype mice (Lamba et al., 2006; Lamba et al., 2010). A limited number of donor photoreceptors was shown to migrate into the outer nuclear layer (ONL) after subretinal injection, similar to FAC-sorted, hESC-derived photoreceptors. Using a similar strategy, photoreceptors derived from porcine iPSCs were labeled with a lentiviral construct (retinol-binding protein 3 (RBP3)-GFP) and detected at a 1% frequency in the ONL following chemical-induced host retinal damage (Zhou et al., 2011). As opposed to subretinal injection, intravitreal introduction of iPSCs appeared less effective at promoting outer retinal integration (Parameswaran et al., 2010). Even less data is available regarding the effect of iPSC-derived donor retinal cells on host retinal function. However, in a study that employed the Rho^−/− mouse model, a dose-dependent improvement in ERG responses was seen following subretinal transplantation of allogeneic iPSC-derived retinal donor cells (Tucker et al., 2011).

Although work is ongoing to optimize hiPSC-photoreceptor survival, integration and function post-transplantation, prior studies using other photoreceptor cell sources offer considerable guidance in this regard. As first shown by MacLaren et al. (2006), postmitotic rod precursors isolated from early postnatal Nrl-GFP mice differentiated into mature rods and established synaptic connectivity in wildtype mice. This same donor rod population also restored light responses in Rho^−/− mutant mice (MacLaren et al., 2006). Since that initial report, rod precursor cells isolated from both developing mouse retina and mouse ESCs have been transplanted into many mouse models (Barber et al., 2013; Gonzalez-Cordero et al., 2013; Homma et al., 2013; Singh et al., 2013a). In the case of the Gnat1^−/− model of congenital stationary night blindness, scotopic vision was restored when P4-P8 rod precursors were injected into the subretinal space (Pearson et al., 2012). Donor photoreceptors from more mature mice have also been shown to integrate into host retina, albeit at a lower efficiency (Gust and Reh, 2011). In the aforementioned studies, reporter line transgenic animals were used to identify and isolate the desired donor cell population; however, such manipulation is not optimal for clinical applications. At present, no human photoreceptor-specific cell surface markers suitable for FACS have been described, but a combination of CD73 and CD24 has been used to enrich for photoreceptor precursors from stage-specific mouse embryonic retina (Lakowski et al., 2011).

Producing a population of donor photoreceptors from a human source is one of many necessary steps in the development of an effective strategy for outer retinal repair. Understanding and manipulating the host retinal environment, which varies substantially based on the type and stage of disease, are critical tasks as well. The integrity of the outer limiting membrane (OLM), the presence and extent of glial scarring, and the status of the inner retinal circuitry, among other factors, have a profound influence on donor cell survival and integration (Barber et al., 2013). Although the use of aminoadipic acid or ZO-1 siRNA to disrupt the mouse retina OLM can significantly increase the number of integrated photoreceptors, untoward effects of such treatment, including glial activation and dissolution of RPE tight junctions, preclude their use in human patients (Pearson et al., 2010; West et al., 2008). Fortunately, not all retinal diseases appear to possess complex barriers to integration. Without pretreatment, late-stage rd1 mouse retina transplanted with mouse rod precursors formed an anatomically distinct donor cell stratum that matured into a putative outer nuclear layer, with concurrent restoration of light-activated pupillary responses and light-mediated behavior (Singh et al., 2013a).

The integrity of the host retina may also be important in the host immune response to donor cells. The subretinal space has been shown to be an immune privileged site in the healthy eye. However, when ocular integrity is compromised by RPE loss and/or degradation of the blood-retinal barrier, macrophage invasion, and microglial and T-lymphocyte activation can initiate an immune response, resulting in graft rejection (Streilein et al., 2002). Even when postnatal photoreceptor donor cells share partial MHC haplotype identity with the host, a chronic immune response can attenuate long-term grant survival in the absence of immune suppression (West et al., 2010). Taken together, these findings underscore the key role of the disease-specific milieu in donor cell survival and integration, and imply that an unmatched or partially matched allogenic hiPSC–based treatment scheme may not be adequate for all degenerative diseases. In contrast, the use of autologous hiPSC cell types carries a high likelihood of circumventing immune complications.

5. The potential for iPSCs to be used for autologous tissue repair

The ability to derive customized hiPSC lines has opened the door to new frontiers in personalized medicine. In a landmark study, homologous recombination was used to repair a single copy of the β^s globin gene in iPSCs derived from a humanized sickle cell mouse. When autologous transplantation was performed using hematopoietic cells bearing the corrected sickle cell allele, the anemia phenotype was corrected (Hanna et al., 2007). Since that publication, gene correction has been performed in hiPSCs for several monogenetic blood disorders and tested in vitro and in xenografts (Simara et al., 2013).

Gene correction in hiPSCs has also been used to repair a mutation responsible for gyrate atrophy (GA), a progressive blinding disease that primarily affects RPE, leading to secondary photoreceptor loss. By employing bacterial artificial chromosome-mediated homologous recombination, an A226V mutation in ornithine aminotransferase (OAT) was corrected in hiPSCs derived from a GA patient (Howden et al., 2011). RPE differentiated from the uncorrected GA hiPSCs showed very low levels of OAT enzymatic activity, whereas RPE from gene-corrected hiPSCs had OAT activity comparable to human prenatal RPE and RPE differentiated from hESC and control hiPSC lines (Meyer et al., 2011). In addition, it was shown that the process of OAT gene repair did not add to the mutational load, nor did it increase genetic instability (Howden et al., 2011). However, the cost and effort that would be required to produce and test clinical grade, gene-corrected cells makes this option unwieldy with currently available technology. With this in mind, in cases where the consequences of genetic defects or risk factors are not manifested for many years, gene repair may not be necessary for autologous transplantation. An important example is AMD, where multiple genetic risk factors are known to increase patients’ susceptibility to RPE dysfunction and death, but only after many decades. In this scenario, genetic manipulation of donor hiPSCs would not be required, assuming an adequate restoration of cellular “youth” occurs through the reprogramming process. In keeping with this thought process, the first clinical trials using autologous, uncorrected hiPSC-RPE will target AMD.

Regardless of whether gene repair is necessary, other potential safety concerns remain for hiPSC transplantation. For example, tumor formation could result if residual undifferentiated cells are present within the donor cell population. While progress has been made in the development of methods to detect and remove residual pluripotent cells (Kobayashi et al., 2012; Kuroda et al., 2012; Tsuji et al., 2010; Tucker et al., 2011), further advances in this area are anticipated.

Another prerequisite for the use of hiPSCs in retinal disease is the evaluation of immunogenicity of donor retinal progeny in animal models. Zhao et al. reported that the transplantation of undifferentiated ESCs and iPSCs into syngenic mice incited an immune response resulting in regression of teratoma growth (Zhao et al., 2011). Although the issues raised in that report deserve serious consideration, investigations into immune rejection would optimally be performed using the specific differentiated cell populations needed for tissue repair. In a separate report, ten iPSC and five ESC lines derived from C57BL/6 were injected into syngeneic recipients, whereupon negligible teratoma regression and immune responses were found (Araki et al., 2013). Furthermore, no untoward effects were detected when neurospheres derived from allogenic murine iPSCs were transplanted into a spinal cord injury model (Tsuji et al., 2010).

Xenograft studies are inherently insufficient to fully examine the issues surrounding the clinical use of human iPSCs, leading some groups to generate iPSCs from nonhuman primate species to test in autograft experiments [for review, see Wu et al., 2012]. To date, several cell types have been derived from nonhuman primate iPSCs, including RPE (Okamoto and Takahashi, 2011), but there have been only a few publications describing the use of these cells in transplantation experiments. Recently, iPSCs derived from nonhuman primates were partially differentiated toward a dopaminergic neuronal fate and transplanted into an autologous hemiparkinsonian model (Emborg et al., 2013). After six months, mature neurons and glial cells arose from the transplanted progenitor population with little evidence of microglia, macrophage or lymphocyte infiltration. Although no functional assays were performed, this study in a nonhuman primate model provides support for autologous hiPSC transplantation in humans.

6. The path to hiPSC clinical trials: manufacturing clinical grade donor cells

6.1 Autologous vs allogenic donor cells

Although hiPSC technology allows for the development of autologous cell therapies, this approach may not be universally optimal or necessary. Thus, it is important during the early stages of designing hiPSC therapeutics to decide between pursuing autologous vs. allogeneic cell therapy. Autologous therapeutics address the key issue of rejection of the cell transplant; however, this choice also comes with a number of potential challenges from a manufacturing and regulatory perspective. First of all, a manufacturing process for an autologous therapeutic must be able to handle patient-to-patient variabilities that could arise from a number of factors, including donor age, gender and health. These issues could impact the consistency, quality and safety of the final cell product. Extensive development and validation studies may be required to demonstrate the ability of the cell manufacturing process to handle such variabilities in the starting cell source. It is also worth noting that an autologous therapeutic may pose an increased risk of tumor formation and require long-term tumorigenicity studies since the patient would not be expected to launch an immune response against the graft.

The alternative choice to use allogeneic cells allows well-characterized cell banks to be established for each therapeutic. Of course, the issue of rejection remains present for allogeneic cell therapeutics, although the relative immune privilege believed to be afforded by the eye might offset this risk (Streilein et al., 2002). However, if retinal immune privilege is disrupted in certain retinal diseases, the availability of HLA diverse banks of hiPSCs may offer a compromise (Taylor et al., 2011; Taylor et al., 2012). Alternatively, disruption of the BETA-2-MICROGLOBULIN gene in hiPSCs could be used to generate lines with reduced immunogenicity, analogous to a report using hESCs (Riolobos et al., 2013). By employing one or more of these approaches, the need for immunosuppression following hiPSC transplantation may be minimized or eliminated, depending on the degree of HLA matching required.

6.2 Early-stage cell manufacturing process and assay development

As noted above, one of the biggest challenges in harnessing the potential of hiPSCs is the inherent variability in the cell production process. This variability can be influenced by the quality of the starting cell source, differences in the raw materials used for reprogramming and cell culture, and protocols employed for hiPSC growth and differentiation. Identifying and controlling these variables, and demonstrating that they do not introduce unacceptable risk, is necessary to develop a cGMP manufacturing process. The first steps are to establish standardized, robust methods for growing and differentiating cells, evaluating critical raw materials (e.g., culture medium, growth factors), and performing quality control (QC) testing, the latter of which should be done early and often. Qualification trials and QC assays are needed to demonstrate process reproducibility and the ability to consistently produce a cell therapeutic that meets key specifications (e.g., viability, purity, function, sterility, and efficacy).

6.3 Material for pre-clinical testing and human clinical trials

Once the process and assay development stage of a hiPSC project is completed, material is typically produced for pre-clinical animal studies to support federal applications for human clinical trials. It is important to ensure that the final cell product used in the animal studies is representative of the therapeutic that will be provided for the human clinical trial. If possible, identical cell banks, raw materials and QC testing should be employed. The pre-clinical animal studies must address key issues regarding the safety of the transplanted cell product, which include the impact of off target cell contaminants present in the donor cell population, the potential for teratoma formation due to residual undifferentiated cells, and the generation of karyotypically abnormal cells. As such, it is critical that any aspects of the process that could impact the quality of the therapeutic are identical between cells produced for animal studies and human clinical trials. In addition, a dialogue should be established with the appropriate federal agency (e.g., Federal Drug Administration) to ensure that any process changes that are anticipated in moving from animal studies to clinical production will not pose significant regulatory problems.

6.4 Human tissue source

Regardless of what tissue is used as the starting cell material, donors for hiPSCs will have to meet stringent eligibility requirements (e.g., HCT/P regulations 21 CFR 1271, Subpart C – Donor Eligibility) (FDA, 2007). In addition, an informed consent document should be executed at the time of collection to ensure that the donor is aware of the intended use and required testing of his or her samples, and that the tissue and data will be collected and treated in an ethical manner (Lowenthal et al., 2012). For allogeneic hiPSCs, donor eligibility requirements typically include testing of the patient for human pathogens such as HIV 1, HIV 2, Hepatitis B virus (HBV), Hepatitis C virus (HCV), and Treponema pallidum. In addition, for leukocyte-rich donor tissues, Human T-lymphotropic virus (HTLV-I/II) and CMV testing is often necessary. While tissues that are collected for generating autologous iPSCs are exempt from donor eligibility requirements, testing may still be prudent given the risk of facility cross-contamination and the potential to activate latent pathogens during reprogramming and differentiation.

6.5 Reprogramming method

Initial reprogramming methods relied on the use of retroviral vectors (Takahashi et al., 2007; Yu, 2007). However, these methods are not ideal for generating clinical-grade hiPSCs for several reasons, including the potential for insertional mutagenesis (Okano et al., 2013). To address this problem, major advancements in reprogramming methodology have been made since the original viral integrating vectors were described. Alternatives now include nonviral methods (e.g., episomal and plasmid-based vectors, recombinant protein transduction, synthetic mRNA), or higher efficiency, nonintegrating adeno-, recombinant adeno-associated and Sendai viruses or excisable lentivirus constructs [for reviews, see Hussein and Nagy, 2012; Okita and Yamanaka, 2011]. Each of these reprogramming methods has advantages and disadvantages with respect to reprogramming efficiency, procedural logistics, somatic cell source, requirement for feeder cells and cGMP compliance. For example, the use of Sendai virus is limited by the fact that cGMP compliant vectors are not currently available. Viral vectors that are produced using untested cell lines may introduce additional risks of pathogens, which constitute another major concern for clinical applications. However, EBV vectors can easily be produced in a cGMP-compliant manner using expression in standard E. coli cell lines with defined, animal-free fermentation medium. The EBV vector system has been used to reprogram a variety of cell types including human fibroblasts, mononuclear cells from bone marrow, and cord blood (Hu et al., 2011; Yu et al., 2009). More recently, a method of reprogramming CD34⁺ cells from whole blood using EBV episomal vectors was shown to support the generation of hiPSCs from a wide range of donors (Mack et al., 2011).

While hiPSC reprogramming vectors used for clinical applications should ideally be produced under cGMP guidelines, the level of cGMP compliance that is required depends in part on the system used to create the vector, and whether the vector is present in the final hiPSCs. For example, viral vectors that are used in reprogramming are typically produced in mammalian cell lines utilizing culture medium that may contain animal-derived raw materials such as fetal bovine serum. These viral vectors, and ideally the cell line used to produce the viral vector, should be subjected to adventitious agent testing to rule out introduction of pathogens from the cell line and/or animal-derived raw materials (Table 1).

Table 1.

Quality Control Testing of Reprogramming Vectors

Test	Description
Identity	Restriction digest with AGE, sequencing
Vector Concentration	DNA concentration by UV, viral vector titer
Vector Quality	HPLC assay for plasmid form, assay for viral vector activity
pH/osmolality	pH/osmolality measurement
Sterility/mycoplasma	Testing for contamination by bacteria, fungi, or mycoplasma as required by 21 CFR 610
Purity - Endotoxin	Limulus Amebocyte Lysate (LAL) assay
Host impurities	Assays for residual DNA, RNA, protein from the vector production host
Viral Pathogens	In vitro, In vivo and species-specific pathogen testing as recommended by ICH for the specific vector production host and method

6.6 hiPSC bank production

For allogeneic cell therapeutics, a bank of starting cell material is usually established in the early stages of product development. A single Master Cell Bank (MCB) or a two-tiered system with a Working Cell Bank is produced and tested to insure that the starting material for the manufacturing process is consistent. For autologous therapeutics, generating a large hiPSC bank is not necessary. However, creating a small bank of hiPSCs may help identify and minimize variability that can arise from the initial reprogramming step. The description of cell bank production and testing provided below focuses on allogeneic applications; however, smaller banks and scaled-back testing schemes may be more appropriate for some applications.

To produce a MCB, several hiPSC colonies are selected and screened to ensure that they meet specifications (e.g., bacterial/fungal/mycoplasma contaminate, growth characteristics, karyotype, hiPSC marker expression) before choosing one to expand into an MCB. The entire MCB manufacturing process should minimize the use of animal-derived reagents wherever possible and maintain complete traceability of all materials. If animal components cannot be replaced, they should be obtained from companies that tightlycontrol and rigorously test the source material.

Methods have been described for the expansion of pluripotent stem cells using completely defined, xeno-free cell culture medium and attachment matrices (Chen et al., 2011; Kim et al., 2013; Rajala et al., 2010; Wang et al., 2012). hiPSCs are expanded a predetermined amount based on the intended use of the bank, taking into consideration the disease indication and expected future need. Cells are usually stored in vapor phase liquid nitrogen using a controlled freezing method and, preferably, a cryopreservation medium formulation that is free of animal-derived components. Cell Bank testing for adventitious agents is a critical part of developing cell lines for use in clinical production, which need to meet requirements established by appropriate federal agencies (e.g., FDA and the International Conference on Harmonization) (FDA, 1993; ICH, 1998) (Table 2). In addition, special project-specific testing (e.g., directed differentiation testing) should be considered.

Table 2.

Proposed Testing for iPSC Master Cell Banks

Test	Description
Identity	Short Tandem Repeat (STR) testing
Viable Cell Count	Trypan Blue dye exclusion
Bacterial/fungal Contamination	Bacteria, fungi testing according to 21 CFR 610.12
Mycoplasma Contamination	Direct culture in broth and agar, indirect test using indicator culture/DNA stain. Assay conditions meet the FDA’s PTC requirements
Karyotype	G-band on 20 metaphase spreads
Cell Marker Expression	Flow cytometry for human ES cell marker expression including: Oct-3/4, SSEA-1/3/4, TRA-1-60, TRA-1-81
In vitro Adventitious Agent Testing	ICH cell line testing on 3 cell lines – MRC5, Vero, NIH 3T3/Hs68
Residual Reprogramming Vector	PCR or Southern Blot to detect and map residual reprogramming vector

6.7 hiPSC differentiation to the final cell product

The differentiation stage of the manufacturing process will be initiated from a cell bank or culture that has been subjected to the QC testing discussed above. From the research stage of development to cGMP production, the differentiation process should be developed to accomplish four main tasks: 1) minimize undefined or animal-derived raw materials, 2) ensure that the process is robust enough to handle variability in the starting hiPSCs, 3) confirm that the process is scalable to meet projected demands, and 4) reduce or eliminate the presence of unwanted cell types and residual undifferentiated cells. Toward this end, chemically defined and non-xenogenic methods have been developed for retinal differentiation from hiPSCs (Sridhar et al., 2013; Tucker et al., 2013).

Scalability of the manufacturing process is an important issue to consider from the onset of product development. Research-based methods that are very manual in nature are not ideal for producing larger batches of cells (scale-up) for allogeneic therapies, or for use in high-throughput culture methods (scale-out) that may be required to make an autologous therapeutic commercially feasible. Moreover, scale-up or scale-out of the manufacturing process could result in significant changes in the quality of the final product. The differentiated cells are grown to a final scale that provides adequate cell numbers for dosing and sufficient cells for QC testing and archive samples. The final cell product could take on many forms, including fresh or cryopreserved dissociated cells, 3-D structures, or cells seeded onto a scaffold. A cryopreserved final cell product or intermediate cell type is ideal as it allows QC testing to be performed on the cell therapeutic prior to release to the clinic. Of note, a method for efficient cryopreservation of stratified 3-D retinal cultures from hESCs has been described (Nakano et al., 2012) that may be applicable to similar structures derived from hiPSCs. However, cryopreserved cells will likely require some level of manipulation (e.g., thaw and wash to remove DMSO) prior to administration. Therefore, additional studies will need to be performed to evaluate the effects of post-release handling of the cell therapeutic, including shipping, storage, thawing/ washing or post-thaw culture.

QC testing should also be designed to accommodate the final cell format and preparation procedures. A summary of anticipated QC testing for a hiPSC therapeutic is provided in Table 3, although abbreviated QC testing may be permissible for products that are prepared fresh for each administration, followed by more rigorous testing. For example, when fresh cells are administered to patients, there is not sufficient time to perform the standard sterility test (14 days). In this case, federal regulatory agencies typically ask for a rapid method (e.g., gram stain) to detect gross contamination with a follow-up full sterility test. A plan must be in place to address potential failing results in the sterility test if the product is administered to a patient. Other critical assays, such as the determination of residual undifferentiated cells, will be more challenging to address in this situation and may need to be performed on an intermediate sample so that results are available prior to release of the cell therapeutic to the clinic. Once manufacturing and QC testing are complete, all documentation and test results need to be reviewed to insure that the cell therapeutic was produced in compliance with cGMP guidelines and appropriate Standard Operating Procedures, and that the final cell therapeutic meets all QC testing specifications.

Table 3.

Quality Control Testing Plan of the Final Cell Product

Attribute	Method	Specification
Pre and Post-thaw Viable cell recovery	Trypan blue or other viability assessment	> 70%
Identity test	Short Tandem Repeat	Matches iPSC MCB
Bacterial Endotoxin	Kinetic chromogenic LAL (final cell prior to seeding on substrate)	< 5 EU/mL
Mycoplasma	PCR (validated) or PTC method (direct and indirect culture)	No Contamination
Sterility Test	Gram stain for fresh product 21 CFR 610.12, bacteristasis and fungistasis	No Contamination
Cell Surface Markers	Flow cytometry assay for - Positive markers: specific for cell type Negative markers: specific for off-target cells	Above (positive) or below (negative) pre-determined limits of marker expression*
Residual Undifferentiated iPSCs	Flow cytometry or qPCR assay for expression of Oct-3/4 or SSEA4	Below limits of detection
Potency	Functional assessment of cell performance	Establish by Phase 3

^*The appropriate positive and negative markers for each specific cell type should be based on the individual target cell(s), typical cell impurities, potential safety concerns, and specifications developed through the manufacturing process for that particular cell product.

8. Conclusions

Prevention remains the most desirable strategy to address human disease. However, other approaches must be explored when a particular disease cannot be held in check or when conventional therapies are ineffective. Toward this end, cell replacement may provide a viable treatment option for some severe degeneration disorders such as AMD and RP. Key issues for a cell-based therapeutic include 1) adequate production and expansion of desired cells, 2) purification of donor cell type(s) with elimination of contaminating cells, 3) long term graft survival and immunogenic compatibility, and 4) ethical implications. With these considerations in mind, human pluripotent stem cells offer many promising treatment avenues. Figure 3 summarizes the potential role of hiPSCs in retinal repair in relation to current hESC trials, with an emphasis on minimizing the cost and complexity of the process. Unmatched hESCs are currently the most convenient and inexpensive option, but could be subject to immune rejection as well as ethical objections. In contrast, cell types derived from patient-specific hiPSCs can be used for autologous transplantation, with less concern for immune rejection. However, customized cell therapy is expensive and time-consuming, and hiPSCs may need to undergo gene correction prior to transplantation, further increasing the overall burden of the treatment approach. The use of HLA-matched allogeneic hiPSCs, although not ideal, may provide a middle ground for some cell therapies. If each of these strategies fails to eliminate the specter of immune rejection, some level of chronic immunosuppression therapy or host immune conditioning may be necessary. Ultimately, the optimal strategies will only be definitively revealed through human clinical trials. However, the groundwork for hiPSC-based retinal therapies has already been laid, offering guarded hope to individuals suffering from presently untreatable blinding diseases.

Figure 3.

Flowchart depicting the potential role of hiPSCs in retinal transplantation therapies relative to current and future hESC-based trials.

Highlights.

• We review efforts to produce retinal lineage cells from iPS cells.

• We discuss the potential roles of human iPS cells in the treatment of retinal disease.

• The challenges of using human iPS cells in transplantation are reviewed.

• We outline the steps needed to create clinical grade iPS cell therapeutics.

Abbreviations

AMD

Age-related macular degeneration

cGMP

clinical good manufacturing practices

ESCs

embryonic stem cells

hESCs

human embryonic stem cells

iPSCs

induced pluripotent stem cells

hiPSCs

human induced pluripotent stem cells

MCB

master cell bank

NR

neural retina

QC

quality control

RPE

retinal pigmented epithelium

RP

retinitis pigmentosa