Advances in repairing the degenerate retina by rod photoreceptor transplantation.

Despite very different aetiologies, age-related macular degeneration (AMD) and most inherited retinal disorders culminate in the same final common pathway, loss of the light-sensitive photoreceptors. There are few clinical treatments and none can reverse the loss of vision. Photoreceptor replacement by transplantation is proposed as a broad treatment strategy applicable to all degenerations. The past decade has seen a number of landmark achievements in this field, which together provide strong justification for continuing investigation into photoreceptor replacement strategies. These include proof of principle for restoring vision by rod-photoreceptor transplantation in mice with congenital stationary night blindness and advances in stem cell biology, which have led to the generation of complete optic structures in vitro from embryonic stem cells. The latter represents enormous potential for generating suitable and renewable donor cells with which to achieve the former. However, there are still challenges presented by the degenerating recipient retinal environment that must be addressed as we move to translating these technologies towards clinical application.

Introduction

Retinal degenerations leading to the loss of the light sensitive photoreceptors are a major cause of untreatable blindness in the UK. Inherited retinal dystrophies affect 1 in 3000 of the population, and Age-Related Macular Degeneration (AMD) affects 1 in 10 people over 60 years, a figure that is rising with an ageing population (Owen et al., 2012). Both conditions culminate in the same final common pathway, the loss of the light-sensing photoreceptors, which causes severe or complete loss of vision. In each case, there are few effective treatments and none of those currently available is able to replace lost photoreceptor cells and restore visual function. There is thus a need for new therapeutic approaches. Photoreceptors are afferent neurons and as such require no incoming connections. Moreover, they need only to make short, single synaptic connections to the remaining inner retinal circuitry to contribute to visual function. These features, arguably, make photoreceptor transplantation one of the most feasible types of Central Nervous System (CNS) repair and an excellent candidate for exploring regenerative neural stem cell therapies. The past decade has seen enormous progress in novel ocular therapies, including the first gene therapy (Bainbridge et al., 2008; Maguire et al., 2008) and retinal implant based (Chader et al., 2009) clinical trials for retinal disease, which have set the scene for pioneering new therapies for retinal disease. The success of gene therapy relies on the delivery of new functional genes to cells that lack such genes and is therefore dependent upon endogenous cell survival. In cases where the degenerative process has already led to cell death or in those conditions that are not amenable to gene therapy approaches, cell replacement therapies may offer a complementary approach. Given its accessibility, the eye has also been the model of choice for the study of neural development. As such, there is a wealth of knowledge regarding the intrinsic and extrinsic factors that regulate retinal histogenesis, knowledge that is now being employed to great effect in attempts to generate retinal cells from stem cells for transplantation (Eiraku et al., 2011; Lamba et al., 2009; Osakada et al., 2009). In this review, I will present a brief overview of the progress in photoreceptor replacement, in our ability to generate photoreceptors from stem cells and discuss some of the challenges that must be addressed as we begin to take this strategy towards clinical application.

Critique and discussion

Transplantation strategies

Retinal sheet transplantation

A central requirement of successful photoreceptor replacement therapy is the identification of an appropriate donor cell that has the ability to both migrate into the recipient retina following transplantation and differentiate into a fully functional, synaptically connected photoreceptor. Several transplantation strategies have been attempted, including the transplantation of whole sheets and microaggregates of developing neural retina and of dissociated adult hippocampal neural stem cells.

Work by Aramant and others have demonstrated that whole retinal sheets derived from either embryonic or neonatal retinae can survive and continue to differentiate after subretinal transplantation (Ghosh et al., 2004; Turner et al., 1988). More recently, they have shown that full-thickness retinal sheets can make limited connections with the recipient retina (Seiler et al., 2008). While there were some improvements in basic visual responses, for example light–dark discrimination, some authors have attributed this to the enhanced survival and function of endogenous photoreceptors by means of trophic factors released from the healthy transplanted tissue (Gouras and Tanabe, 2003). To date, retinal sheet transplantation in patients has shown some subjective visual improvement (Humayun et al., 2000; Radtke et al., 1999). A clinical study of retinitis pigmentosa (RP) and AMD patients who received foetal retinal sheet transplants (neural retina and retinal pigment epithelium, [RPE]), reported improvements in vision for 7 out of 10 patients, although the direct beneficial effects of the foetal retinal grafts are difficult to assess as all patients also received intraocular lens implants (Radtke et al., 2008). A further complication of full thickness retinal sheets is the inclusion of the inner retinal neurons that by definition are juxtaposed between the graft photoreceptors and the inner retinal neurons of the recipient retina. It remains to be determined to what degree this affects the processing of visual signals within the retina and beyond. An interesting related approach is the use of partial thickness grafts, encompassing the photoreceptor layer but omitting the inner retinal layer of the donor retina (Ghosh et al., 1999). Such a strategy might be of interest in severely degenerate retinas where the endogenous photoreceptor layer is completely absent although results to date have indicated only limited connectivity between the graft and the recipient retina.

Transplantation of dissociated cells

The limited integration between sheets and the recipient visual circuitry has prompted many groups to look at the transplantation of dissociated cell types. Given that the brain and the retina are both derived from the neuroectoderm and that immature neurons and progenitor cells are intrinsically capable of migrating and differentiating during neural development, numerous studies have investigated the potential of brain-derived neural stem/progenitor cells transplanted to the neural retina (Klassen et al., 2007; Mellough et al., 2007; Sakaguchi et al., 2004). However, transplantation of these cells into the adult retina has demonstrated only limited integration (Sakaguchi et al., 2005; Young et al., 2000). Moreover, these non-retinal sources of donor cells frequently fail to differentiate into mature retinal phenotypes, including photoreceptors, as assessed by immunohistochemistry (Young et al., 2000). More recent studies using tissue-restricted reporter genes to demonstrate retinal cell fate have confirmed that the integrated cells derived from such sources do not exhibit intrinsic features of mature retinal neurons (Sam et al., 2006).

To address this problem, progenitor cells isolated from embryonic retinas, which by definition possess the potential to differentiate into retinal neurons, have been tried; depending upon the conditions used to expand these cells in vitro prior to transplantation, these cells survive and differentiate into glial cells (Yang et al., 2002) and/or cells expressing retinal-specific markers, including some specific for photoreceptors (Qiu et al., 2005) upon transplantation. However, they also fail to become correctly integrated within the laminar structure of the neural retina, remaining instead at the site of transplantation. Greater success has been achieved when transplanting immature retinal cells into immature recipients, suggesting that the maturation state of the recipient may play a role in determining transplantation outcome: Murine progenitor cells transplanted into the eyes of neonatal Brazilian Opossums, which provide a foetal-like environment, survived and differentiated in vivo, although integrated cells were not found within the outer nuclear layer (ONL), where photoreceptors normally reside (Sakaguchi et al., 2004). Since the same donor cells failed to migrate into the retina of more mature recipients, it was suggested that the age of the host tissue had a key role in determining the fate of transplanted precursor cells and their ability to integrate into the circuitry of a non-autologous retina (Sakaguchi et al., 2003, 2004; Van Hoffelen et al., 2003).

By using postnatal donor cells from a transgenic mouse ubiquitously expressing Green Fluorescent Protein (GFP) and taking these cells from the peak of rod photoreceptor genesis and transplanting them into recipients of exactly the same developmental stage, MacLaren & Pearson et al., found that the transplanted cells migrated into the ONL (and no other layer) of the recipient retina and matured into morphologically normal photoreceptors. Moreover, these same cells could also integrate with equivalent efficiency into the non-neurogenic, adult retinal environment (MacLaren et al., 2006), as well as a number of models of retinal degeneration (Barber et al., 2013, 2008; MacLaren et al., 2006; Pearson et al., 2010) (but see further discussion). This indicated that transplantation success depended upon the developmental stage of the donor cell, rather than that of the recipient (Fig. 1).

Fig. 1

A. Schematic adapted from MacLaren & Pearson et al. (2006), summarising how the developmental stage impacts on the ability of donor cells to migrate and integrate within the recipient retina. Integration-competent cells can be identified by the post-mitotic rod marker, Nrl. Cells taken from this stage of development are capable of migrating into and integrating within the recipient retina circuitry. B. Examples of Nrl.GFP + ve donor cells transplanted into the Gnat1−/− murine model of stationary night blindness taken from Pearson et al. (2012) and Barber et al. (2013). Integrated donor cells express the protein missing in the recipient photoreceptors, rod α-transducin.

Importantly, by using another transgenic model, in which GFP expression is under the control of the rod-specific transcription factor Nrl, which is first expressed shortly after terminal mitosis (Akimoto et al., 2006), the authors were able to demonstrate that these results are only achieved using post-mitotic photoreceptor precursor cells, cells that are specified to differentiate into rod photoreceptors, but not progenitor cells or photoreceptors at other stages of development (Lakowski et al., 2010; MacLaren et al., 2006; Pearson et al., 2012). In contrast, work by Reh and colleagues reported that fully mature photoreceptors taken from the adult retina could also integrate within the wild type retina with similar efficiency but exhibited poorer survival (Gust and Reh, 2011). However, more recent reports have consistently demonstrated that while it is possible for mature photoreceptors to integrate, they do so with markedly lower efficiency, even when accounting for reduced viability (Gonzalez-Cordero et al., 2013; Pearson et al., 2012). Thus, photoreceptor transplantation is possible, even into the mature recipient retina, provided that the donor cells are at the correct stage in development at the time of transplantation (Bartsch et al., 2008; MacLaren et al., 2006).

Function

Transplanted rod-precursor cells are able to migrate into the adult retina and differentiate to acquire the specialized morphological features of mature photoreceptor cells (Bartsch et al., 2008; Eberle et al., 2012; MacLaren et al., 2006). However, the fundamental question remained as to whether these transplanted photoreceptor cells could actually improve vision. Although there have been reports of improvement in a single measure of visual function, none had conclusively demonstrated an impact on vision. Moreover, in these earlier studies, the percentage of cells able to integrate and make connections was extremely low (< 1000 cells per transplant) (Bartsch et al., 2008; MacLaren et al., 2006) and could restore only basic light sensitivity to the retina. Thus a major challenge for the cell suspension approach was, and indeed still is, to generate sufficient numbers of transplantable cells and in parallel to achieve high integration efficiency. By optimizing the transplantation protocols sufficiently to increase integration efficiency ~ 30 fold, Pearson, Ali and colleagues were able to provide the first definitive proof of functional rod-mediated vision after photoreceptor transplantation in a mouse model of congenital stationary night blindness (Pearson et al., 2012). Three weeks after transplantation, the newly integrated donor rod cells are light sensitive, have dim-flash kinetics very similar to wild type rods and form classic triad synaptic connections with second-order bipolar and horizontal cells in the recipient retina. Moreover, the visual signals generated by transplanted rods are projected to higher visual areas, including V1 of the visual cortex, and are capable of driving visually guided behaviour. Of note, there was a robust positive correlation between the number of integrated cells and the amount of visual function restored. Interestingly, in contrast to some reports, the authors found that despite robust integration of transplanted photoreceptors, the levels achieved were still insufficient to drive an electroretinogram (ERG) response in the recipient retinae. The ERG is recorded at the corneal surface and is an average of the electrical responses of all the cells in the retina in response to a light stimulus. By using gene replacement in the same model, Pearson et al. demonstrated that approximately 120,000 functioning rod photoreceptors were required to generate a reproducible scotopic ERG response (Pearson et al., 2012).

Challenges in repairing the degenerating retina

The role of the recipient retina in transplantation outcome

Photoreceptor transplantation is proposed as a broad treatment strategy, applicable to all photoreceptor degenerations. This requires the efficient integration of donor cells within the degenerating retinae since, to date, the best transplants have been achieved in normal retinae and retinae with relatively mild degeneration. Recently, the first comprehensive assessment of rod-photoreceptor transplantation across different models of inherited photoreceptor degeneration was described (Barber et al., 2013). Importantly for clinical application, the study found that photoreceptor transplantation is feasible in all models examined, including severely degenerated retinae. However, disease type impacts significantly on both the number and the morphology of integrated rods. For example, rod precursors transplanted into the Gnat1−/− model of stationary night blindness integrated in numbers similar to healthy wild type retina and formed robust outer segments and synaptic connections. Conversely, cells transplanted into the moderately fast degenerating rhodopsin knockout, integrated poorly and rarely elaborated outer segments or synapses. In contrast, this and another recent report found that transplants into the PDE6β model, which degenerates rapidly, demonstrate surprisingly good integration and even some restoration of visual function, although the resulting morphology of the transplanted photoreceptors is poor (Barber et al., 2013; Singh et al., 2013).

Successful photoreceptor transplantation requires the donor cell to migrate from the site of transplantation – typically the subretinal space – through the inter-photoreceptor matrix (IPM), across the outer limiting membrane (OLM), a series of tight-junctions that separate the neural cell bodies from the inner/outer segments of the photoreceptors, and into the recipient outer nuclear layer (ONL) (Warre-Cornish et al., 2013). Two factors that change to variable extents in different models of degeneration, namely the integrity of the OLM and the extent of recipient retinal gliosis, have been found to play key roles in determining the success of transplanted donor cell migration and integration (Barber et al., 2013; Kinouchi et al., 2003; Pearson et al., 2010; West et al., 2008).

Photoreceptor transplantation outcome and gliosis

Gliosis is well known to be a limiting factor in the regeneration of other areas of the CNS such as the spinal cord. Reactive gliosis is thought to represent a cellular attempt to protect the surrounding tissue from further damage, to promote repair and to limit neuronal remodelling (Eng and Ghirnikar, 1994). It includes morphological, biochemical and physiological changes, which can vary with the type and severity of the insult. In the retina, gliosis primarily involves the Muller glial cells, which undergo upregulation of the intermediate filament proteins vimentin and glial fibrillary acidic protein (GFAP), hypertrophy of the Muller glial terminal processes at the edge of the ONL (Bignami and Dahl, 1979) and a concomitant increase in the deposition of inhibitory extracellular matrix (ECM) molecules, including Chondroitin Sulphate Proteoglycans (CSPGs) (Inatani et al., 2000; Landers et al., 1994). Like elsewhere in the CNS, the glial scar in the retina may represent a physical barrier to cell migration or act as a reservoir of inhibitory ECM molecules or a combination of these. Indeed, gliosis can impact on the efficacy of many proposed therapeutic approaches including the efficiency of viral transduction in gene therapy (Calame et al., 2011) and impair the ability of retinal grafts and electronic implants to contact the underlying retina (Zhang et al., 2003). Similarly, photoreceptor transplantation outcome is inversely correlated with the extent of GFAP expression and deposition of CSPGs (Barber et al., 2013). Transplantation studies into the retinae of GFAP−/vimentin− double knockout mice have also reported an increase in integration compared to wild type animals (Kinouchi et al., 2003) although the precise mechanism behind this is unclear. CSPGs bind many different ECM proteins and growth factors making them important players in a variety of regulatory processes including cell adhesion and migration. In the CNS, CSPGs are upregulated after injury and participate in the inhibition of axon regeneration mainly through their GAG side chains. Our understanding of their role in retinal degeneration is limited. Using the broad-spectrum CSPG antibody CS-56 we have seen very different patterns of expression in different models of degeneration (Barber et al., 2013). A recent study showed the CSPG, aggrecan, to be markedly upregulated in two rat models of retinal dystrophy (Chen et al., 2012), while microarray analysis of individual Muller glial cells from a mouse model of RP identified a significant increase in CSPG5 (neuroglycan) (Roesch et al., 2012). ChABC, a broad-spectrum chondroitinase that can break down a number of CSPGs has been used with good effect in promoting axonal regrowth in the damaged spinal cord. Similarly, application of ChABC can break down retinal CSPGs and improve photoreceptor transplantation outcome in some models (Barber et al., 2013; Singhal et al., 2008; Suzuki et al., 2007). Matrix Metalloproteases (MMPs) are proteolytic enzymes that degrade ECM molecules, including collagen, fibronectin, laminin, and a variety of proteoglycans. Inducers of MMP-2 activity have been shown to promote transplanted donor cell migration into the recipient retina and there have been successful developments of biodegradable polymers to deliver MMPs in conjunction with cell transplants (Suzuki et al., 2006; Yao et al., 2011; Zhang et al., 2007).

Photoreceptor transplantation outcome and the OLM

The observation that the ability of transplanted cells to integrate within the host opossum retina decline with host maturation, made by Sakaguchi et al. (Sakaguchi et al., 2003, 2004), is striking since this coincides with the formation of anatomical barriers within the host retina including the OLM, a series of adherens junctions between the terminal processes of the Muller glia and the inner segments of the photoreceptors. Similarly, there is an inverse correlation between the degree of transplanted photoreceptor integration and OLM integrity. In mice with retinal dystrophy caused by defects in Crumbs homologue-1 (Crb1), a protein associated with OLM adherens junction formation and stabilization, the OLM is disrupted and transplanted photoreceptor precursor integration is significantly higher than wild type controls with an intact OLM (Barber et al., 2013; Pearson et al., 2010). However, OLM integrity is maintained in a number of retinal degenerations caused by other gene defects and transplant outcome is markedly worse in these models (Barber et al., 2013). This suggests that the OLM remains a significant barrier to transplanted photoreceptor cell integration in many forms of retinal degeneration. Support for this notion comes from the finding that pre-treatment of the recipient retina with the glial toxin α-aminoadipic acid (AAA), which disrupts the integrity of the OLM, leads to an increase in photoreceptor precursor cell integration (West et al., 2008). AAA would not be suitable for clinical application due to its toxic effects on the supportive Muller glia (Pedersen and Karlsen, 1979). An alternative strategy is to induce transient OLM disruption using small interfering ribonucleic acid (siRNA) to promote transcriptional gene silencing of OLM-related proteins. Proof of principle for this strategy has been demonstrated using siRNA targeted against the adherens junction protein ZO-1, which significantly improves transplantation outcome both in wild type and degenerate retinae (Barber et al., 2013; Pearson et al., 2010). ZO-1 is not an ideal target however since it is also expressed throughout the RPE and silencing it can lead to RPE proliferation. A target whose expression is restricted to the OLM would therefore be ideal.

Renewable sources of donor cells

The findings described above define a strategy for photoreceptor cell replacement therapy and overcome a number of obstacles. They show that cell replacement is feasible provided the correct stage cell is transplanted. Moreover, as this cell type is post-mitotic it avoids the potential hazards associated with tumour formation due to unregulated proliferation of transplanted undifferentiated stem cells. However, the postmitotic photoreceptor precursor cells used in these studies are derived from the early postnatal mouse retina. The equivalent stage in human development occurs early in the second trimester. Putting the obvious ethical considerations aside, such tissue is in very limited supply and of variable quality. An expandable source of cells that could be cultured in vitro to the correct ontogenetic stage for transplantation may, therefore, represent a more appropriate and reproducible source of photoreceptor precursor cells. These findings have renewed interest in the potential to generate new photoreceptor precursors from various stem cell sources. A few of the more promising are briefly discussed below. While beyond the scope of the current review, there is also interest in the ability of stem cells to perform neuroprotective roles (see review by Bull and Martin, 2011).

Embryonic stem cells

Pluripotent embryonic stem (ES) cells have undoubtedly yielded the most promising results to date with regard to generating retinal cell types. ES cells are derived from the inner cell mass of the pre-implantation blastocyst. They are able to maintain an undifferentiated state or can be directed to mature along lineages deriving from all three germ layers—ectoderm, endoderm and mesoderm. Several studies have produced convincing demonstrations of the generation of retinal cells in vitro from mouse, monkey and human ES cells using growth factors, retinal co-culture and genetic modification (Lamba et al., 2009; Osakada et al., 2009). Following the pioneering work by Watanabe and colleagues on telencephalic specification (Watanabe et al., 2005), Ikeda and colleagues used antagonists of the nodal (LeftyA) and Wnt (DKK-1) pathways in addition to activin-A and serum to derive mouse retinal progenitor cells (Rx+/Pax6+) after 10 days of differentiation. These ES-derived cells were able to integrate, albeit in small numbers, into embryonic retinal explants and expressed markers of retinal cell types including the photoreceptor photopigment, rhodopsin (Ikeda et al., 2005). Mouse ES cells were further directed towards retinal differentiation by the inhibition of the Notch signaling pathway (Osakada et al., 2009), which has been demonstrated to promote photoreceptor differentiation. Using a different but related approach, Reh and colleagues exposed human ES cells to Noggin (a Bmp inhibitor), IGF-1 and DKK-1 (Lamba et al., 2006). Under these conditions, almost 80% of the cells were directed to a retinal fate (Pax6+/Chx10+) within 3 weeks and ~ 12% of these went on to express the transcription factor and cone/rod photoreceptor precursor marker, Crx. Meyer and colleagues avoided the use of recombinant protein – an important requirement for translation into clinical application – using a system that takes advantage of the endogenous secretion of DKK-1 and Noggin by ES cells in neural differentiation media. The generation of Rx+ neural rosettes and their subsequent culture as neurospheres gave rise to retinal progenitors after 40 days (26% of cells were Chx10+) and by day 80, 20% of the neurospheres contained Crx+ photoreceptor precursors (Meyer et al., 2009). Together these demonstrate the ability to direct ES cells towards a retinal fate. However, it has proved harder to show that these cells can integrate as well as those isolated from the postnatal retina. Integration has been observed following the transplantation of ES-derived retinal cell (Lamba et al., 2006, 2009), although others have found that cells derived using similar protocols integrate only poorly, if at all (Ikeda et al., 2005; West et al., 2012).

In a recent ground-breaking study, Eiraku et al. have developed a 3-D culture system that enables ES cells to self-organise and mimic the morphological development of the retina (Eiraku et al., 2011). Unlike many of the 2D methods described above, this system appears to allow a more complete development of the photoreceptors. Indeed, a recent report by Ali, Pearson and colleagues (Gonzalez-Cordero et al., 2013) shows that this 3-D system permits the generation of large numbers of stage-specific photoreceptors that are capable of integration and maturation following transplantation into both wild type and degenerating retinas at specific stages of development. Thus far, the efficiency is low and optimisation will be required in order to assess whether these cells are capable of restoring visual function, as has been demonstrated for donor cells isolated from the developing post-natal retina.

Induced pluripotential stem cells

Although mouse, monkey and human ES cells have all been used to generate retinal cells, including photoreceptors, they are not without limitations. The derivation and use of human ES cells are limited by the restricted availability of human embryos, low efficiency of isolation of human ES cells, the potential for immune rejection of non-matched tissue (West et al., 2010), and by ethical concerns surrounding the destruction of human embryos for the purpose of cell isolation. One way to avoid these issues is by reprogramming the nuclei of differentiated somatic cells to a pluripotent state. The now seminal work by Takahashi and Yamanaka (Takahashi and Yamanaka, 2006) demonstrated that just four transcription factors (Oct4, Sox2, c-Myc and Klf4) were needed to turn adult mouse fibroblasts into a pluripotential stem cell that had characteristics very similar to ES cells, including multipotentiality and self-renewal. The use of iPS cells avoids many of the limitations of ES cells noted above. Like ES cells, it offers the prospect of providing retinal cells or even entire retina for the production of donor cells for transplantation but iPS cells can additionally be derived from patients with inherited retinal disorders to study disease mechanisms and develop new targets for therapy (e.g. Jin et al., 2011; Singh et al., 2013). However, since such cells would be derived from the patient themselves, they would also require replacement of the dysfunctional gene that led to the degeneration in the first place (Howden et al., 2011).

Induced pluripotent stem cells are very similar to ES cells and should, theoretically, have a similar capacity to adopt a retinal fate. Several protocols similar to those described above for ES cells have been tested on mouse and human iPS cells. While several have resulted in cells expressing markers of retinal fate, including photoreceptors, the efficiency of retinal differentiation appears to vary, at least between human iPS and ES cells. Although the protocols described by Lamba, Osakada and colleagues can generate photoreceptor-like cells at a similar efficiency to ES cells (Lamba et al., 2006, 2010; Osakada et al., 2009), Hirami, and others reported that some iPS cell lines can be resistant to retinal differentiation (Hirami et al., 2009). Even within the lines that do permit retinal induction, there is significant variation in terms of efficiency. Nonetheless, as we begin to develop more effective protocols for ES differentiation, so too is it likely that iPS differentiation will also improve and these cells remain an exciting avenue of investigation for sources of donor cells.

Retinal stem cells

In lower vertebrates, the eye continues to grow throughout the life of the organism. New neurons are produced from a region called the ciliary marginal zone (CMZ), a site of continual retinal neurogenesis and a stem cell niche at the anterior margin of the retina (Reh and Levine, 1998). Regenerative potential in the central retina is more limited. A population of rod precursor cells exists in teleost fish that can regenerate all retinal neuronal types after damage (see Raymond et al., 2006). Similarly, recent work showed that in zebrafish Muller glial cells can function as multipotent retinal stem cells (RSCs) that respond to the loss of photoreceptors by specifically regenerating the missing neurons (Bernardos et al., 2007; Ramachandran et al., 2010, 2012). Exciting recent developments also hint at the possibility of reactivating the Muller glial cells of the mammalian retina in a similar matter, although generation of photoreceptors from these cells appears limited (Karl and Reh, 2010; Karl et al., 2008).

In the embryonic chick, the cells at the retinal margin, a zone highly reminiscent of the lower vertebrate CMZ, also display significant regenerative capacity. Dissociated cells derived from this region retain the capacity to proliferate and form organized laminar re-aggregates beyond the time of normal neurogenesis in the developing retina (Layer and Willbold, 1989; Willbold and Layer, 1992). These stem cell-like cells persist in the periphery of the adult chicken eye and can be induced to proliferate and add new neurons and glia to the retinal margin (Fischer and Reh, 2000; Fischer et al., 2002). Although the adult mammalian eye appears to lack such regenerative capabilities, Van der Kooy and Ahmad independently reported that the mammalian ciliary epithelium (CE), part of the ciliary body, a structure analogous to the lower vertebrate CMZ, contains a population of stem-like cells (Tropepe et al., 2000). While quiescent and of unknown function during adult life, these cells demonstrate characteristics typical of neural stem cells, including multipotentiality and self-renewal when cultured in vitro. Single pigmented CE cells clonally proliferate in vitro in the presence of the mitogens FGF and EGF to form sphere colonies of cells (termed neurospheres). When differentiated, they reportedly expressed markers of rod, bipolar cell and Muller glial markers, as assessed by immunocytochemistry (Ahmad et al., 2000; Giordano et al., 2007; Tropepe et al., 2000). Follow-up research with human and pig CE cells introduced techniques to expand these cells in monolayer cultures (Coles et al., 2004; MacNeil et al., 2007). However, the lack of robust criteria for defining the progeny and fate of these cells in vitro has made it difficult to evaluate the ability of these cells to differentiate into rods. Indeed other groups have recently called into question the full multipotentiality of these cells and thus their classification as true stem cells (Cicero et al., 2009; Gualdoni et al., 2010). Nonetheless, they remain a population of cells with significant proliferative capacity and appear to be evolutionarily related to the true retinal stem cells found in lower vertebrates and are thus worthy of consideration. Given recent demonstrations of the importance of the conditions in which embryonic stem (ES) cells are grown in for determining cell fate (Eiraku et al., 2011) it remains possible that these cells have the potential to generate retinal cell types under the right circumstances (Fang et al., 2013; Kiyama et al., 2012).

One eye on the future

While there are still a number of steps to be addressed before photoreceptor transplantation can be effectively applied in the clinic, the prospect is looking ever more realistic. The first clinical trials transplanting ES-derived retinal cells, including ES-derived RPE for the treatment of Stargardt's disease, are under way in the USA and the UK (Schwartz et al., 2012). These represent important landmarks in the development of cell therapies for blinding retinal diseases. Of particular importance will be the assessments to confirm a lack of tumorogenicity and whether such grafts are subject to immunological reaction.

Perhaps the greatest challenge yet to be addressed is cone photoreceptor transplantation. The mouse is nocturnal and thus has a rod dominant retina that lacks a macula, the region of high-density cones that permits high visual acuity. For clinical application, however, cones are clearly of great importance. While the same principles appear to apply for cones as for rods, in so far as they must be post-mitotic in order to integrate (Lakowski et al., 2010), they do so only at very low efficiency. However, building upon the knowledge gained in the transplantation of rod photoreceptors, it is surely only a matter of time before restoring cone-mediated vision is achieved.

Conclusions

The need for new therapeutic approaches that can not only halt but also reverse the loss of sight after degeneration has occurred has created intense interest in the potential of photoreceptor replacement by transplantation. Recent advances in transplantation and stem cell technology mean that this is now a realistic prospect. Transplanted donor photoreceptors have been shown to migrate and integrate within the neuronal circuitry of the recipient retina. Moreover, these cells are capable of restoring vision at the behavioural level. Large numbers of potentially transplantable cells can now be generated from embryonic stem cells by growing optic cup like structures in vitro. While iPS cells lag a little way behind, these too are showing great potential both as donor sources and as systems for disease modeling. Together, these advances provide hope that stem cell therapies for the treatment of retinal degeneration may soon move towards clinical application.

Conflicts of interest statement

The author declares no conflicts of interest.

age-related macular degeneration

Central Nervous System

Embryonic Stem (cell)

induced Pluripotential Stem (cell)

Retinitis Pigmentosa

Retinal Pigment Epithelium

Inter Photoreceptor Matrix

Outer Limiting Membrane

Outer Nuclear Layer

Extracellular Matrix

Glial Fibrillary Acidic Protein

Ciliary Marginal Zone

Ciliary Epithelium

Retinal Stem Cell

Matrix Metalloproteases

small interfering ribonucleic acid

Chondroitinase ABC

∝-aminoadipic acid