Stem cells as source for retinal pigment epithelium transplantation

Inherited maculopathies, age related macular degeneration and some forms of retinitis pigmentosa are associated with impaired function or loss of the retinal pigment epithelium (RPE). Among potential treatments, transplantation approaches are particularly promising. The arrangement of RPE cells in a well-defined tissue layer makes the RPE amenable to cell or tissue sheet transplantation. Different cell sources have been suggested for RPE transplantation but the development of a clinical protocol faces several obstacles. The source should provide a sufficient number of cells to at least recover the macula area. Secondly, cells should be plastic enough to be able to integrate in the host tissue. Tissue sheets should be considered as well, but the substrate on which RPE cells are cultured needs to be carefully evaluated. Immunogenicity can also be an obstacle for effective transplantation as well as tumorigenicity of not fully differentiated cells. Finally, ethical concerns may represent drawbacks when embryo-derived cells are proposed for RPE transplantation. Here we discuss different cell sources that became available in recent years and their different properties. We also present data on a new source of human RPE. We provide a protocol for RPE differentiation of retinal stem cells derived from adult ciliary bodies of post-mortem donors. We show molecular characterization of the in vitro differentiated RPE tissue and demonstrate its functionality based on a phagocytosis assay. This new source may provide tissue for allogenic transplantation based on best matches through histocompatibility testing.


Introduction
The retinal pigment epithelium (RPE) is a highly specialized epithelium with a neuroectodermal embryonic origin like the retina. While the retina was first described by Galen in the second century A.D., discovery of the RPE required the use of the first rudimentary microscopes in the 18 th century and was described by Carlo Mondini of Bologna in his "Commentationes Bononienses" (1790) as "a real membrane formed by innumerable globules which makes an excessively delicate network" (Marmor and Wolfensberger, 1998). The histology of the RPE was then elucidated at the end of the 19 th century and further characterized in more recent times.
The eyes derive from two evaginations of the forebrain that generate the optic vesicles connected to the brain by the optic stalks. The optic vesicles then invaginate to form the optic cups with the outer layer destined to become the RPE. The outer stratum is a monolayer of cells that differentiate during embryonic/fetal development and is characterized by pigmentation, which appears during the 5 th week of human embryogenesis. RPE differentiation is induced by several factors including the signaling molecule Activin, a member of the TGF family, which is secreted by adjacent mesenchymal cells. These signals induce expression of transcription factors, such as microphthalmia-associated transcription factor (MITF), orthodenticle homolog 2 (OTX2) and paired box 6 (PAX6), that are essential for RPE specification and to drive expression of proteins necessary for the distinguishing functions of the RPE (Bharti et al., 2012;Fuhrmann et al., 2000;Housset et al., 2013). The fully differentiated RPE consists of a polarized monolayer of pigmented cells with a basal side adherent to the Bruch's membrane, which separates the RPE from the choroid, and an apical membrane facing the photoreceptor cells.
In this paper we summarize molecular and functional characteristics of the RPE tissue since the characterization of these features is required when RPE is generated in vitro. We also discuss RPE impairment and diseases that will be amenable to cell replacement strategies. We review several stem cell sources to produce RPE in vitro. Finally, we present a new protocol for the differentiation of adult human retinal stem cells into RPE sheets. cadherins, namely cadherin 2 (N-cadherin) and cadherin 3 (P-cadherin) (Burke et al., 1999;Lagunowich and Grunwald, 1989;Murphy-Erdosh et al., 1994). Formation of adherens junctions is followed by the formation of circumferential bundles of actin filaments to build zonula adherens junctions (Nabi et al., 1993;Owaribe and Masuda, 1982;Williams and Rizzolo, 1997). This remodeling affects the cytoskeleton leading to the development of distinct apical, lateral and basal membrane territories hosting specific proteins. Gap-junctions in the lateral membrane are transmembrane channels that mediate cell-cell communication between adjacent RPE cells. Connexin 43 is one of the components of the gap-junctions in the RPE tissue and was reported to play a role in differentiation of RPE cells (Kojima et al., 2008). The apical and basolateral domains are segregated by tight junctions and different proteins and lipids are specifically sorted to the two compartments. For example, the basolateral side of the RPE cell harbors the chloride channel bestrophin 1 (Marmorstein et al., 2000;Sun et al., 2002). Na + /K + ATPase can be found preferentially at the apical plasma membrane of RPE cells, rather than at the basolateral membrane as in most epithelial cells (Caldwell and McLaughlin, 1984). In fact, RPE cells exhibit a reversed polarity compared to other epithelial cells and the specific localization of the Na + /K + ATPase appears to be directed by the preferential expression of N-cadherin compared to E-cadherin . Members of the ezrin/moesin family of actin-binding proteins are localized at the apical plasma membrane in RPE cells (Höfer and Drenckhahn, 1993) and apical microvilli are characterized by the expression of v5 integrin (Finnemann et al., 1997).

RPE function in photoreceptor outer segment turnover
The RPE serves several functions that are essential for vision and survival of retinal neurons and several eye diseases are caused by impairments of these functions. One of the most important functions of the RPE is the removal and degradation of the tips of POS ( Figure 2A). Phagocytosis of POS distal tips follows a circadian rhythm and is triggered by light. The direct role of the RPE in phagocytosis was demonstrated more than 40 years ago by radioactive amino acid delivery as a pulse and detection of radioactive phagosomes inside the RPE (Young and Bok, 1969). Young and Bok also demonstrated that RPE cells actively participate in the disposal of POS distal tips because radioactive phagocytized POS were then eliminated from RPE cells. In the renewal process POS are newly built at the base of the outer segment while the tip of the outer segment contains the highest concentration of old and photo-damaged proteins and lipids that need to be removed. Through coordinated POS tip shedding and new POS production a constant length of the POS is maintained. Diurnal binding of POS to the RPE is mediated by v5 integrin through the secreted glycoprotein milk fat globule-EGF 8 (MFG-E8) that acts as a ligand mediating the binding of POS to v5 integrin via its RGD motif (Finnemann et al., 1997;Nandrot and Finnemann, 2006;Nandrot et al., 2007Nandrot et al., , 2004. The scavenger receptor CD36, a transmembrane glycoprotein in the RPE, acts in POS uptake through detection of oxidized phospholipids but not in initial binding of POS (Ryeom et al., 1996;Sun et al., 2006). CD36 is also involved in the clearance of oxidized low-density lipoprotein subretinal deposits (Picard et al., 2010). Nevertheless, v5 integrin is not necessary for POS internalization. Internalization of POS requires c-mer proto-oncogene tyrosine kinase (MERTK), a receptor activated upon tyrosine phosphorylation by focal adhesion kinase (FAK), a downstream effector of v5 integrin (D'Cruz et al., 2000;Vollrath et al., 2001).

Role of the RPE in the visual cycle
Another fundamental function of the RPE is retinoid recycling ( Figure 2B). Absorption of a photon of light by rhodopsin causes isomerisation of the chromophore from 11-cis retinal to all-trans-retinal, which is released from opsin. In order to reconstitute rhodopsin, alltrans retinal must be converted back to 11-cis retinal through a multistep pathway called visual cycle. The visual cycle begins in the photoreceptor outer segment discs where the ATP-binding cassette subfamily A member 4 (ABCA4) transfers all-trans retinal from the intradiscal membrane surface to the cytoplasmic membrane surface (Sullivan, 2009). The all-trans-retinal is then reduced to all-trans retinol by retinol dehydrogenases (RDH8 and RDH12) (Belyaeva et al., 2005;Chen et al., 2012). All-trans retinol is transported through the interphotoreceptor matrix to the RPE by the interphotoreceptor retinoid-binding protein (IRBP) (Okajima et al., 1989). All-trans retinol is also supplied to the RPE by the choroidal vasculature and enters the RPE cells in a receptor-mediated process involving recognition of a serum retinol-binding protein/transthyretin (RBP/TTR) complex by the STRA6 receptor (Kawaguchi et al., 2007;Pfeffer et al., 1986). Within the RPE, all-trans retinol is bound to the cellular retinaldehyde-binding protein (CRALBP) encoded by the RLBP1 gene (Saari et al., 1982). The conversion of all-trans retinol to 11-cis retinal requires an enzymatic cascade involving at least three enzymes associated with the RPE smooth endoplasmic reticulum. The reaction starts with esterification of all-trans retinol to all-trans retinyl esters by the enzyme lecithin:retinol acyltransferase (LRAT) (Ruiz et al., 1999;Saari et al., 1993).
Then an isomerohydrolase, called retinal pigment epithelium-specific protein 65kDa (RPE65), catalyses the concerted hydrolysis of all-trans retinyl ester and the isomerization to 11-cis retinol (Jin et al., 2005;Moiseyev et al., 2005;Redmond et al., 2005). 11-cis retinol dehydrogenase (RDH5) converts 11-cis retinol to the final product 11-cis retinal (Driessen et al., 1995;Simon et al., 1995). 11-cis retinal then exits the RPE and combines with the opsin protein in photoreceptor cells to form the visual pigment. This step is mediated by IRBP with its capacity to remove 11-cis retinal from membranes (Carlson and Bok, 1999). In many human retinal dystrophies the visual cycle is disturbed resulting in the inability to either produce an adequate supply of 11-cis retinal or a failure to remove intermediate retinoid products.

Polarization of the RPE
Photoreceptors are tightly interdigitated with RPE microvilli and this is required to maintain the retina in a fixed plane necessary for the optics, as well as for exchange of nutrients and oxygen from the choriocapillaris to the photoreceptors. This exchange is strictly regulated by the RPE ( Figure 2C). The tight nature of the subretinal space between RPE and retina necessitates fluids to be kept out. Under physiological conditions aquaporins, water channel proteins, allow water transport apical to basal through RPE cells at a rate of 2 to 18 μL/cm 2 per hour (Stamer et al., 2003) and in cases of RPE detachment a transport of fluid could be confirmed in human RPE (Chihara and Nao-i, 1985). Tight junctions among RPE cells provide a barrier against the passage of substances from the fenestrated choroidal capillaries to the retina and generate the external blood-retinal barrier. Electrical parameters can be recorded with RPE tissue mounted in an Ussing chamber and the difference between resting potentials of apical and basolateral membranes in the RPE generates a transepithelial potential that is linked to the different types and distribution of ion channels and other transporters (Gallemore et al., 1993). A specific and regulated traffic of ions either in the retina-to-choroid or choroid-to-retina directions relies on the RPE.
The RPE actively transports Cland HCO 3 from the retina to the choroid and Na + in the opposite direction (Miller and Edelman, 1990). Clenters through an apical membrane Na + -K + -2Clco-transporter and is extruded at the basolateral membrane. The basolateral membrane is characterized by the presence of several chloride channels including cystic fibrosis transmembrane conductance regulator (CFTR) and bestrophin 1 (Weng et al., 2002). Bestrophin 1 is a homo-oligomeric integral membrane protein forming anion channels (Marmorstein et al., 2000;Stanton et al., 2006). Bestrophin 1 transports Clin a Ca 2+ dependent manner but may also have intracellular functions not well characterized yet (Qu and Hartzell, 2008;Strauss et al., 2014). Mutations in the BEST1 gene encoding for bestrophin 1 are linked to Best Vitelliform Macular Dystrophy (Marquardt et al., 1998;Petrukhin et al., 1998). Ion channels in RPE cells are important to buffer ion changes in response to light stimuli because light induces a decrease of K + ion concentration in the subretinal space that causes hyperpolarization of the apical membrane of the RPE and activation of inward rectifier K + channels. The K + conductance is electrically coupled to the basolateral Clconductance and this coupling is important for buffering the chemical composition of the subretinal space (Bialek and Miller, 1994). The RPE also secrets trophic factors ( Figure 2C) such as the pigment epithelium-derived factor (PEDF), a protein implicated in the survival and normal function of the retina (Barnstable and Tombran-Tink, 2004). PEDF is found in the healthy human eye and its levels are altered in eyes affected by retinal degenerative processes (Holekamp et al., 2002;Mohan et al., 2012;Ogata et al., 2004Ogata et al., , 2002Wang et al., 2013). The basal side of the RPE secretes the vascular endothelial growth factor (VEGF-A), required to maintain the choroidal circulation (Marneros et al., 2005;Saint-Geniez et al., 2009). VEGF-A is target of therapies for wet Age-related Macular Degeneration (AMD) (Kaiser, 2013).

Pigmentation of the RPE
Melanogenesis in the RPE is fundamental for sight as reduced pigmentation causes vision impairment. Melanin is synthesized inside intracellular organelles called melanosomes.
Other proteins residing on the melanosomes are the P-protein (pink-eyed dilution homolog protein, OCA2), membrane-associated transport protein (MATP, OCA4) and the melanosome-specific G-protein coupled receptor called ocular albinism type 1 (GPR143/OA1). All these proteins are linked to different forms of albinism (Bassi et al., 1995;Brilliant, 2001;Newton et al., 2001;Oetting and King, 1994). The fully pigmented and mature melanosomes are stored within the RPE and contain black eumelanin. Two shapes can be observed in melanosomes: ellipsoid granules (1 m in diameter and 2-3 m in length) are primarily located at the apical portion and spherical granules reside in the mid-portion of the cell. Melanosomes redistribute to the apical processes of the RPE after light onset and this movement is mediated by Rab27a, melanophilin, and myosin Va (Futter et al., 2004). Melanin contributes to vision by preventing light reflection in the fundus and by protecting photoreceptors from an excess of dispersed light. Pigment granules are important not only for absorption of light but they also protect from oxidative stress and are involved in the binding of zinc and drugs (Schraermeyer and Heimann, 1999;Ugarte and Osborne, 2014). Pigmentation in the RPE also influences ganglion cell projections and fovea development but the molecular mechanisms have not been well characterized yet.

RPE in retinal diseases
The RPE plays important roles in vision: absorption of stray light, formation of the bloodretinal barrier, regeneration of visual pigments and phagocytosis of the tip of POS (Strauss, 2005). Several retinal diseases are caused by loss or failing of one of these functions.
Albinism represents a group of genetic abnormalities due to defects in melanogenesis in the RPE. All albino patients have decreased visual acuity due to fovea hypoplasia and loss of stereoscopic vision due to the misrouting of the optic tracts (Harvey et al., 2006;Summers, 2009). During development, embryonic ganglion cell axons must find their way out of the retina into the optic nerve and extend along the optic tract to reach targets in the central nervous system. In the optic chiasm retinal axons segregate to pass into the optic tracts on either the same side (ipsilateral) or the opposite side of the brain (contralateral).
The albino visual pathway is abnormal because of a reduction in the number of ipsilateral projecting ganglion cells. The molecular mechanisms regulating optic nerve formation and fovea development have not been completely elucidated and it is still not well understood why RPE pigmentation is necessary for these developmental processes (McAllister et al., 2010;Rachel et al., 2002;Rebsam et al., 2012). Albinism is not always associated to reduced melanin synthesis but can also be due to reduced number of melanosomes that likewise results in reduced pigmentation. In fact, mutations in the OA1 gene, causing with recent evidences of early photoreceptor anomalies (Gomes et al., 2009;Ritter et al., 2013) adding to the consolidated hypothesis that RPE damage occurs initially, with secondary photoreceptor degeneration (Eagle et al., 1980;Steinmetz et al., 1991).
Nevertheless, it is interesting to note that mutations in a gene expressed in photoreceptors may lead to substantial impairment of RPE cells. Hence, we should be cautious in defining a priori the therapeutic potential of a given cell therapy based solely on the gene defect location, and improving our knowledge on the details of cell degeneration in different stages of a disease may become a critical element in deciding whether and when to initiate cell therapy, i.e. in defining the indications for a novel treatment.
The majority of the known mutations causing RP are associated with genes expressed in rod photoreceptors, however also genes specifically expressed in RPE, such as genes encoding for visual cycle enzymes (LRAT, RPE65, RDH5), can cause RP. Progressive degeneration in RP first leads to rod photoreceptor loss but subsequently affects also cones as well as the RPE with the formation of pigment depositions in the retina visible with fundus examination, hence the term "pigmentosa" for this family of retinal dystrophies (Koenekoop, 2009;Sancho-Pelluz et al., 2008;Wright et al., 2010).
No cure has been found yet for treating retinal degeneration, but several strategies are achieving initial results. These approaches can be divided into two broad groups: (1) those aiming at restoring health and function of surviving cells through gene supplementation (Bainbridge et al., 2008;Cideciyan et al., 2008;Cremers et al., 2002;Hauswirth et al., 2008;Maguire et al., 2008;Smith et al., 2009) or by treatments with neuroprotective molecules to slow the degeneration process (Frasson et al., 1999;Hamel, 2006;Léveillard and Sahel, 2010;Nakazawa et al., 2011;Punzo et al., 2009;Sieving et al., 2006), and (2) those attempting to restore retinal photosensitivity. The latter includes electronic retinal implants (Chader et al., 2009;Zrenner, 2002;Zrenner et al., 2011), optogenetic approaches (Busskamp et al., 2012), and cell implantation therapies (Gust and Reh, 2011;MacLaren et al., 2006;West et al., 2010). An effective cell therapy can act through different mechanisms: transplanted cells may secrete trophic/neuroprotective factors to slow degeneration or provide new photoreceptors to replace lost cells. In the latter case not only implanted cells need to survive and function as phototransducers, but they also have to form functional synapses with second order retinal neurons. Moreover, implanted photoreceptor cells cannot survive and be active without a functional RPE that almost invariably is affected at some stage of the disease process. Developing a combined RPE/photoreceptor sheet for transplantation might be a solution in the future, but at present cell therapy approaches are focusing on the implantation of one cell type (either photoreceptor or RPE cells) selecting conditions where the other cell type is still present.
This approach appears more feasible in diseases where RPE cells degenerate first, subsequently leading to photoreceptor loss and thus to blindness, as it occurs in some common macular diseases, like dry AMD and STGD. As functional recovery depends on photoreceptor rescue by implanted healthy RPE cells, it will be critical to treat patients early in the disease process and before extensive damage occurs. Although success rates are not high, case numbers low and no long-term follow-up is available, the positive There are several elements that researchers need to account for when developing such therapeutic tools. Therapeutic potentials and transplantation procedures have to face the need for 40,000-60,000 RPE cells to replace the macula area (Lu et al., 2009). Current Good Manufacturing Practices (cGMPs) are required to monitor the facility, processes, testing, and practices to produce a consistently safe and effective product for human use.
In the context of clinical manufacturing of a cell therapy product Current Good Tissue Practices (cGTP) are also required to oversee donor consent, traceability, and infectious disease screening.
Transplantation of allogenic or autologous RPE aims at replacing pathological RPE with healthy tissue. Several cell sources have been evaluated for RPE replacement therapies including immortalized cell lines (i.e. ARPE19) as well as other non-RPE cell lines or sheets of fetal and adult RPE (Pinilla et al., 2007;Algvere et al., 1997;Chen et al., 2009;Tezel et al., 2007). Autologous peripheral RPE have also been transplanted in AMD patients (Binder et al., 2002;van Meurs and Van Den Biesen, 2003). Only partial rescue of vision could be shown in these studies and translation into the clinic was hindered by the limited accessibility to these cells or by ethical concerns. Furthermore, prenatal tissue may be very variable with regards to quality. Lately, studies focused on the search of new cell sources to overcome these limitations and to offer in vitro culture techniques that allow the expansion of homogeneous cells to gain numbers suitable for cell replacement therapies. ESCs are based on either spontaneous or defined culture procedures (Ramsden et al., 2013). In order to induce spontaneous differentiation, ESCs were grown to confluence on a feeder layer of PA6 stromal cells and when basic fibroblast growth factor (bFGF) was withdrawn from the culture medium pigmented clones spontaneously formed and were subsequently isolated and expanded (Kawasaki et al., 2002). A second protocol required that ESC were first grown as embryoid bodies and then plated on dishes coated either with laminin and fibronectin or with gelatin until they formed visible pigmented colonies (Lu et al., 2009;Osakada et al., 2008). The RPE tissue generated by these methods expressed many RPE-specific proteins, formed tight junctions and showed apical-basal polarity. Molecularly directed methods to differentiate RPE from ESCs were based on blocking the Wnt and Nodal signaling pathways or on exposure to Nicotinamide and Activin A or on treatment with bFGF, Retinoic acid and Sonic hedgehog (Idelson et al., 2009;Osakada et al., 2009;Zahabi et al., 2012). ESC-derived pigmented cells could form a cell monolayer with polygonal actin bundles, as defined by the staining with F-actin, and with tight-junctions as shown by localization of the tight junction protein zona occludens 1 (ZO-1). These cells were shown to express PAX6, at levels similar to fetal RPE, and MITF, a key regulator of RPE differentiation (Bharti et al., 2012). The presence of mature RPE specific markers, such as the visual cycle enzyme RPE65, could also be demonstrated. showing their equivalence to ESCs Osakada et al., 2009).

Embryonic stem cells and induced pluripotent stem cells as sources of RPE
Provided that many ethical obstacles can be overcome by the use of iPSC derived RPE, the in vitro generated tissue can be envisaged as a promising source for cell replacement therapies because it displays many of the features of native RPE such as pigmentation, apical-basal polarity, tight-junctions, phagocytosis activity and expression of enzymes for the visual cycle. The eye is also an immune-privileged location thanks to the blood-retina barrier making this organ a favorable site for tissue transplantation therapies.
Pre-clinical evaluation of integration competence of in vitro differentiated cells were undertaken by subretinal injections of human ESC-derived RPE into retinal degeneration rodent models or into the immune-deficient NIH III mouse model (Haruta et al., 2004;Lu et al., 2009 injected into the submacular space of one patient affected by STGD and one patient affected by dry AMD. The follow-up at four months did not show any evidence of teratoma formation and no loss of vision in treated eyes. Some vision improvement was reported (Schwartz et al., 2012).

Adult RPE as a source of RPE
Mature RPE cells are mitotically quiescent under physiological conditions in the eye.
Several studies on amphibians showed that, upon injury, the RPE could proliferate and regenerate both RPE and neural retina. These findings demonstrated intrinsic plasticity of RPE cells and ability to transdifferentiate. These observations also suggested that the RPE has some capacity for self-repair upon proper stimuli. Plasticity of the human RPE cells was also uncovered in pathological conditions, for example in proliferative vitreoretinopathy, in which epithelial-mesenchymal transition and proliferation was reported (Casaroli-Marano et al., 1999). RPE cultures derived from adult human donors could be propagated in vitro, however the identification of two phenotypically distinct cell populations suggested some heterogeneity in the culture (McKay and Burke, 1994). Long post-confluent periods were required for adult RPE cells to form tissue sheets and mature adherens junctions (Kaida et al., 2000). The prospective use of adult RPE in regenerative therapy was recently reconsidered when a subpopulation of RPE cells was discovered to exhibit self-renewal ability in vitro and multipotentiality (Salero et al., 2012). After mature RPE monolayer prior to transplantation was recently shown to limit these harmful outcomes (Stanzel et al., 2014).

Retinal neurospheres (RNS) as source of RPE
Retinal stem cells (RSCs) have been identified in the adult ciliary body and these cells can be cultured in vitro as pigmented neurospheres. The procedure requires dissection of the ciliary epithelium followed by dissociation of the cells and culture at very low density to generate floating clones of pigmented cells called retinal neurospheres (RNS) (Ahmad et al., 2000;Tropepe et al., 2000). Molecular analysis of these cells showed that they derive from large pigmented ciliary epithelial cells expressing low levels of P-cadherin (Ballios et al., 2012). Pigmented RNS can be obtained from different species such as man, rat, mouse, pig and rabbit (Coles et al., 2004;Gu et al., 2007;Inoue et al., 2005;Ahmad et al. 2000) and Published studies on the murine tissue suggested that the ciliary body can be a promising source of cells to generate RPE for transplantation purposes (Aruta et al., 2011). While ESCs are already used in clinical trials and are paving the way for subretinal transplantation procedures, there is still need for accessible sources of cells without the ethical burdens of ESC. The ciliary body appears to be a fairly accessible tissue that can be easily collected from post-mortem donors. Nevertheless, in case efficient isolation techniques become available, partial ciliectomy surgical procedures may be considered on patients to allow autologous transplantation. Here we will discuss improvement of the procedures to generate a highly pigmented and differentiated human RPE tissue that may provide a source for cell therapy approaches in the future.

Tissue collection and dissociation
The procedures of the present study involving human participants were approved by the local ethical committee, and were conducted in accordance with the principles of the Declaration of Helsinki (http://www.wma.net/en/60about/70history/01declarationHelsinki/).
Human ciliary bodies were collected from deceased donors after their legal representatives had expressed a written informed consent for the inclusion in the study.
Donors were not affected by retinal degeneration and had an age between 70 and 85 years. After disinfection with povidone iodine, the conjunctiva and Tenon's capsule were dissected for 12 clock hours and a concentric scleral cut was performed 12 clock hours with a blade and curved scissors preserving the integrity of the uveal tissue. The corneoscleral graft was then explanted and a circular cut was done posteriorly to the pars plicata and containing the ciliary bodies. The ciliary bodies, the lens and the iris were finally explanted en-bloc and preserved at 4°C in a synthetic media for corneal storage (EUSOL-C ® ).
For cell preparation, the iris was separated from the pars plicata, and the lens and the capsule were removed. The cleaned pars plicata was cut into 4 portions and treated with 2mg/ml of dispase in DMEM-F12 medium for 30 min at 37°C. The pieces of ciliary body were then moved to a solution containing 1.3mg/ml of trypsin and 0.67mg/ml of hyaluronidase and incubated for 20 minutes at 37°C. The ciliary body epithelium was scraped from the ciliary muscle and dissociated to a single cell suspension (for a detailed protocol see Giordano et al., 2007). We obtained an average of 6x10 5 cells per eye.

RNS culture
Cells dissociated from the ciliary body were seeded at the density of 20,000 cells/ml and cultured with DMEM-F12 medium added with N2 supplement (Gibco), 20ng/ml of EGF and 20ng/ml of bFGF. Floating clones of cells were visible 2-3 days after plating. Clones consisted of pigmented cells that did not attach to the plastic substrate of the Petri dish ( Figure 3A), as previously published for RNS (Aruta et al., 2011;Demontis et al., 2012;Giordano et al., 2007). After 7 days in culture RNS reached a diameter of about 100m ( Figure 3A-C). At different time points cell proliferation in RNS was assessed by exposure to 10 M BrdU (5-bromo-2'-deoxyuridine) for 3 hours followed by immunofluorescence with an anti-BrdU antibody to detect the newly incorporated nucleotide in the genomic DNA.
Cells in the RNS were actively proliferating during the first days of culture when most nuclei were positive to BrdU staining ( Figure 3D). The number of proliferating cells decreased with time in culture and no proliferation could be detected after 9 days ( Figure   3E-F). Based on the proliferation assay we chose to start differentiation of RNS at the 7 th day of culture.

Differentiation of RPE sheets from RNS
We modified the published protocol for murine RNS and we seeded human RNS cells after dissociation for 30 min with Accutase TM (Millipore). We obtained about 1.2x10 5 cells per eye that underwent differentiation into RPE. Dissociated cells were seeded on an ECM substrate at the concentration of 23,000 cells/cm 2 and cultured for 2 days with the RNS growth medium containing bFGF and EGF. On the 3 rd day of culture growth factors were withdrawn and cells were exposed to differentiation medium as previously published (Aruta et al., 2011). Cells actively proliferated during the first days of differentiation and proliferation was still high after 7 days of culture (75%, Figure 3G Figure 3K).  Figure 4M-N). RPE65 was clearly present in RPE cells after 21 days of differentiation in vitro but the amount of protein was much lower compared to adult human RPE ( Figure 5A-B). Similarly, CRALBP, necessary for the binding of retinoids inside the RPE, was observed at all times of differentiation but at lower levels compared to expression in mature human RPE ( Figure 5A-B).

Molecular characteristics of the RPE cells derived from RNS in vitro
A key transcription factor necessary for RPE differentiation is MITF. MITF is characterized by different isoforms and H-MITF is one of the isoforms specific for RPE cells and is not expressed by other pigmented cells of the eye (Bharti et al., 2008). Primers for real-time qPCR were designed such as they could discriminate at the mRNA level among several MITF isoforms (Aruta et al., 2011). Expression of the RPE specific H-MITF isoform showed highest expression levels at 21 days of differentiation ( Figure 5C). Interestingly, expression increased with differentiation and the levels at 21 days were higher than in the adult human RPE. To confirm that the MITF transcription factor localized, as expected, in the nuclei of RPE cells we used an anti-MITF antibody and found very low MITF in RPE cells differentiated for 7 days. Strong and nuclear specific immunostaining was observed in RPE cells after 14 days of differentiation ( Figure 5D-E).
Finally, we wanted to define if in vitro differentiated RPE expressed PEDF, a trophic factor with neuroprotective properties (Barnstable and Tombran-Tink, 2004). RPE cells differentiated for 21 days showed a detectable expression of PEDF mRNA although expression levels were much lower compared to adult RPE ( Figure 5F). Progressive pigmentation of cells with differentiation time inferred that melanin synthesis enzymes were active in the RPE cultures ( Figure 3G, I and K). We thus also assessed expression of one of the genes important for melanosome biogenesis, OA1. OA1 transcripts could be detected after 21 days of differentiation and displayed higher levels than in the adult RPE ( Figure 5G). It is known that melanosomes are not actively regenerated in the fully mature RPE and therefore some genes involved in melanogenesis are down-regulated in adult RPE. The observation of a higher level of OA1 mRNA in RPE cells in vitro compared to the RPE from adult donors may be explained by the fact that RPE cultures at 21 days have not reached full maturation, as also suggested by the levels of expression of other RPE markers ( Figure 5A-C). Consistent with this hypothesis a previous report showed that the OA1 gene is expressed at low levels in adult RPE (Surace et al., 2000).

Phagocytosis activity of RPE cells in the newly formed sheets of tissue
A phagocytosis assay was performed to test one of the functions of RPE cells in the tissue differentiated in vitro. To define the phagocytic ability of in vitro differentiated RPE we used fluorescein labeled bovine POS, as previously published (Aruta et al., 2011). RPE cells differentiated for 21 days were fed for 1 or 3 hours with POS. We assessed whether RPE cells were able to internalize fluorescein labeled POS by confocal microscopy. Intracellular localization of POS in RPE cells confirmed the phagocytic ability of the RPE tissue differentiated in vitro ( Figure 5H). Increased internalization of POS was observed over time of exposure ( Figure 5I), in agreement with studies of phagocytosis in primary RPE cell cultures (Kennedy et al., 1994).

Future directions
In vitro differentiated human RPE derived from ESCs have been transplanted subretinally as cell suspension for the treatment of AMD and STGD in phase I and II clinical trials Immunosuppressive therapy may be used to prevent and counteract the immune reaction, but its side effects can be severe. The development of technologies capable of reducing the immunogenicity of in vitro expanded RPE cells, or even to ensure immunological compatibility with the patient would definitely facilitate the application of cell therapies.
iPSCs express low levels of MHC class one and 2-microglobin proteins at their surface but expression of these proteins increases upon differentiation into RPE . Hence grafts of in vitro differentiated RPE derived from allogenic iPSC may evoke immune response with subsequent rejection. Furthermore, the surgical procedure for transplantation can induce some degree of inflammation to which the transplanted tissue will be exposed. Another complication of the surgery can be the possible reflux of  .
Several studies are evaluating the best substrate for the culture of a continuous RPE monolayer (Sheridan et al., 2004).  (Kanemura et al., 2014). RNS-derived RPE tissue has not been evaluated in tumorigenicity tests yet, but data shown in this study provided evidence that RPE derived from RNS stopped proliferation upon differentiation. Further analyses will be required to fully characterize RNS-derived RPE and its characteristics when cultured on matrixes suitable for tissue transplantation.
Another issue discussed more recently in transplantation studies is the differentiation stage that RPE cells need to reach before transplantation. Photoreceptors were shown to be able to integrate in the host tissue if they were postmitotic but not fully differentiated (MacLaren et al., 2006). Similarly, not heavily pigmented cells attached better to the Bruch's membrane suggesting that the control of in vitro differentiation before transplantation may allow optimization of cell survival and functionality (Schwartz et al., 2012). RPE cells derived from RNS did not reach a fully differentiated stage, based on the gene expression studies presented here. A full characterization of the differentiation level of this tissue will require microarray analysis followed by comparison with fetal and adult human RPE.
The ability of the recipient eye to accept the transplanted cells may also be affected by factors present in a fully differentiated tissue and by the unhealthy environment. The modulation of the recipient environment by ectopic expression of IGF1 helped the integration of transplanted cells in a photoreceptor transplantation murine model (West et al., 2012). It is possible that integration of RPE may also be facilitated by supplementation of growth factors during transplantation.
Finally, the success in restoring vision via cell therapy with RPE cells will ultimately depend on the presence of residual photoreceptors in the retina. Hence, it is critical that adequate diagnostic tools are available to accurately visualize and quantify photoreceptor cells, and to improve the characterization of the Bruch's membrane and choriocapillaris.
The confocal near-infrared reflectance (NIR) imaging and the Spectral-domain optical coherence tomography (SD-OCT) were recently shown in preclinical studies to be able to detect transplanted human ESC derived RPE cells in Royal College of Surgeons (RCS) rat in vivo (Hu et al., 2012;Ribeiro et al., 2013). Most recent OCT technologies already provide high resolution images of the RPE-photoreceptor complex, but more advanced features would be desirable, such as extending high density 3D scanning to larger retinal areas, and automatizing the extraction of measurements from such high volume data.
Such improvements in diagnostics tools will allow a more accurate classification and selection of the cases to be treated, a better identification of the preferred retinal location for cells transplantation, as well as more quantitative data in follow-up studies. In