The irreversible loss of retinal neurons is a common final pathway of many ocular diseases, including age-related macular degeneration, glaucoma and retinitis pigmentosa. Currently there are no therapies available to repair or replace damaged retinal cells, prompting investigations to develop methods for stem cell transplantation to treat these diseases. This review describes the different cell populations that have the potential to be used for cell based therapies to regenerate the retina, and discusses various factors that need to be addressed before cell transplantation can be implemented and developed. Cells that could be used as a source of retinal neurons for human therapies have been isolated from a variety of sources including the embryo and the adult eye. Most studies have focussed on the differentiation of stem/progenitor cells into photoreceptors, as the development of these retinal neurons has been extensively characterised. However, little work has been so far undertaken on the differentiation of these cells into other major type of retinal neurons and this merits further investigation.

Embryonic stem cells arise from the eight-cell stage morula and are defined as cells that have an unlimited capacity for self-renewal and an ability to differentiate into any of over 200 cell types in the human body [1]. With their plasticity and migratory capacity, ESCs have been proposed as ideal candidates for cell based therapies to treat human retinal diseases. However, much research is still required in order to differentiate these cells into retinal specific precursors. ESCs have been differentiated in vitro into neural cell types [2] as well as retinal pigmented epithelium (RPE) [3], but controlling their differentiation has proved challenging. In the absence of appropriate intracellular signals, ESCs appear to differentiate towards a neuronal fate by default [4], and although evidence for the production of fully functional retinal neurons is lacking, several studies suggest that it may be possible to derive photoreceptors from ESC cultures.

A few groups have reported the induction of a retinal fate in mouse embryonic stem cells either using growth factors [5, 6], co-culture with explants of embryonic retina [7], or by cell insertion of retinal progenitor genes [8]. The most efficient induction so far achieved involves the use of growth factors that are normally involved in retinal development, such as lefty-A, Dkk-1 and Activin-A [5]. This treatment caused 25-30% of cells to express the eye-field transcription factors Rx and Pax6 and upon co-culture with explants of adult mouse retina, they formed rhodopsin and recoverin photoreceptors. Other studies have also directed human ESCs towards a retinal fate by using a combination of a BMP antagonist, a Wnt pathway inhibitor and IGF-1 [9]. This protocol resulted in 80% of cells expressing eye-field transcription factors and when differentiated up to 10% of cells expressed early markers of photoreceptors including Crx, Nrl and recoverin. These ESC-derived photoreceptors have been shown to restore some visual function in Crx deficient mice (a model of Leber’s Congenital Amaurosis) as detected by electroretinography [10].

In 2008, a study by Osakada et al. showed expression of late photoreceptor markers in differentiated human ESCs, although this differentiation involved a 5-6 month protocol [11]. Using Wnt and Nodal antagonists to differentiate pluripotent cells towards a neural fate and later a combination of retinoic acid and taurine for retinal differentiation, cells expressed functional components for light responses including transducin-alpha, phosducin, phosphodiesterase and rhodopsin kinase. The differentiation of ESCs into other types of retinal neurons has not been extensively studied, but emerging new research has shown that human ESCs can also be directed towards a retinal ganglion cell fate [12]. To achieve this differentiation, ESCs were induced to form embryoid bodies and then differentiated towards a neuronal fate by culturing cells on Poly-L lysine and laminin in the presence of FGF2 and serum. Cells were then differentiated for 6 days in the presence of FGF2 and Shh. These cells expressed genes coding for RGC markers such as ATH5, BRN3B, RPF-1, THY1 and ISL-1. Unfortunately, only 2% of cells possessed these RGC-like properties, and when transplanted intravitreally into rats, very few cells integrated into the ganglion cell layer [12].

Although studies of ESC differentiation into retinal neurons show great promise, there are significant limitations to using these cells, including reproducibility of differentiation amongst different ESC lines and the risk of teratoma formation [13]. In addition, elucidating developmental cues that guide the reliable differentiation of ESCs towards specific retinal neurons may prove to be a very long process. It may therefore be more feasible to use a tissue-specific type of stem cell. In this context, transplantation of neural stem cells from the adult mammalian brain have been shown to be able to integrate into the host retina, but have failed to express retinal-specific markers [14, 15]. Recently, various populations of stem/progenitor cells have been isolated from the adult mammalian (including human) neural retina [16] and ciliary body [17]. Compared to ESCs, these adult stem cells may be easier to differentiate into retinal neurons as they have already undergone several processes of differentiation within the eye which are unknown to us. Therefore these cells would have already undergone crucial stages of development to make them committed towards a retinal fate. Their use would bypass the long and extensive protocols that may be needed to differentiate ESCs into the type of retinal progenitors required for efficient retinal transplantation.

The ciliary body is part of the uveal tract that is continuous with the iris anteriorly and the choroid posteriorly. The ciliary body is lined by an inner non-pigmented and outer pigmented epithelium, that are continuous with the neural retina and retinal pigment epithelium respectively. Developmentally, both the ciliary epithelium and the retina share a common origin and arise from the inner layer of the optic cup [18]. However, they follow different fates as demonstrated by their distinct roles in the adult eye. The neural retina translates light stimuli into neural signals and conveys them to the brain whereas the CE secretes aqueous humour to maintain the anterior chamber of the eye and thus intraocular pressure. Since the epithelium of the ciliary body is derived from the neuroepithelium of the optic vesicle during embryonic development, it is thought that this region of the eye may contain cells that retain their neurogenic potential. This has been suggested by observations that the non-pigmented epithelium of the chicken eye is capable of generating neurons following intraocular injection of growth factors such as insulin, EGF and FGF2, which stimulate the proliferation and neuronal differentiation of non-pigmented cells within the ciliary body of this species [19]. These cells express the transcription factors Pax6 and Chx10 and are found around 3mm anterior to the peripheral edge of the retina. This region of Pax6/Chx10-expressing cells coincides with the region where newly generated amacrine cells, Müller glia and ganglion cells, but not of bipolar cells or photoreceptors, appear in response to growth factor treatment [19].

In situ studies in the mouse have shown that during retinal histogenesis, many rod photoreceptors and their precursors can be seen in various stages of morphological differentiation in the pars plana (see Fig. 1 for anatomical localization). However, once development is complete these precursors are rarely seen [20]. In mice with photoreceptor degeneration (induced by NMU injections), cells within the non-pigmented pars plana were induced to proliferate and incorporated BrdU. In addition, some cells also expressed recoverin [21].

Fig. (1). Schematic diagram showing the anatomy of the most anterior neural retina and ciliary body of the adult human eye. A non-laminated region, similar to that seen in fish and amphibians and known as the ciliary marginal zone (CMZ) is present in fish and amphibians throughout life [33-36]. A similar region has been identified in the early postnatal life of avians [38] and rodents [55], as well as in young primates [22] and human adults [25].

Fig. (2). Expression of neural progenitor markers in the non-laminated region (ciliary margin-like zone) of the adult human retina. (A) Confocal image of adult human retina showing intense expression of Nestin (green) in the most peripheral region adjacent to the ciliary body (asterisk). Pigmented (solid arrow) and non-pigmented (segmented arrow) ciliary epithelium do not express Nestin in situ [25]. (B) Confocal image of non-laminated anterior retina showing expression of the Müller cell marker CRALBP (green) and the progenitor marker CHX10 (red).

Fig. (3). A population of Müller glia from the adult human retina express neural stem cell characteristics in vitro. (A) Cells cultured in the absence of extracellular matrix and growth factors display a characteristic glial morphology. In contrast, cells cultured in the presence of extracellular matrix and FGF2 for 5 days acquire a neural morphology with characteristic axon-like processes (red arrows). (B) Müller stem cells in culture express the retinal progenitor markers CHX10, SOX2 and intermediate filament nestin.

Neuron-like cells have also been identified in the non-pigmented epithelium of non-human primates [22, 23]. In the monkey ciliary body, cells with a rounded shape showing short processes stain for the cone photoreceptor marker arrestin. There is also evidence that the human non-pigmented CE expresses components involved in phototransduction including rhodopsin, rhodopsin-kinase and arrestin. The presence of photosensitive proteins within the CE has led to the suggestion that it plays a role in circadian processing [24]. In explants of human retina, a few cells within the non-pigmented CE have been shown to proliferate in response to EGF [25]. In situ staining of the adult human ciliary body has also revealed that non-pigmented CE cells express SOX2, CHX10 and NOTCH1, with characteristic nuclear staining of CHX10 only visible in the non-pigmented layer [25].

Cells from the CE expressing stem/progenitor characteristics have been isolated and cultured in vitro from many species including mice [17], rats [26], pigs [27] and humans [28]. Stem cells from the adult mouse ciliary epithelium were first reported by Tropepe et al. in 2000. The report stated that single “pigmented ciliary margin cells” proliferated in vitro to form sphere colonies of cells expressing markers of retinal neurons including rod photoreceptors, bipolar cells and Müller glia [17]. This inaccurate anatomical description became repeatedly cited in the literature, with later reports describing the isolation of these cells from the pars plana and pars plicata of the ‘retinal ciliary margin’ [27]. In this context, it is important to clarify that anatomically both the pars plana and pars plicata are part of the ciliary body and not part of the neural retina (illustrated in Fig.1). The pars plana refers to the thin flat region of the ciliary body in close proximity to the retinal margin whereas the pars plicata refers to the folded region of the ciliary body situated between the pars plana and the iris [18, 25].

In 2004, the same group that described the presence of stem cells in the mouse CE reported that stem cells from the human pars plana and pars plicata could be isolated and grown as both spheres and monolayers in culture [28]. These spheres purportedly consisted of a mixture of both pigmented and non-pigmented cells, but no attempt was made to dissociate the two cell types. It was then suggested that the human eye contains a small population of about 10,000 retinal stem cells in the CE [28], and that they appeared to exist throughout life and in vitro form spheres of pigmented and non-pigmented cells within 7 days of isolation. When cultured as monolayers and differentiated in the presence of FGF2, these cells were shown to express markers of progenitor cells such as Nestin, PAX6 and CHX10, as well as markers of differentiated cells such as rhodopsin and MAP2. Interestingly, some cells also stained positive for the RPE marker, RPE 65 [28] and both mouse and human CE progenitor cells were found to proliferate very rapidly compared to data published for neural stem cells [17, 28]. However, when cells dissociated from spheres were transplanted into the mouse eye, they showed limited migration into the neural retina and RPE [17, 28].

Since the first report of the presence of progenitor/stem cells in the mammalian CE, many papers have been published characterising these cells and investigating their ability to differentiate into retinal neurons. Several groups have reported that extended culture of both pigmented and non-pigmented cells of the CE causes their differentiation into progenitor-like cells [19, 27, 29]. As a result of these characteristics, these cells were incorrectly termed ‘retinal stem cells’. However, no evidence was presented showing that these cells were able to differentiate into functional retinal neurons. The relationship between these ‘epithelial’ sphere-forming cells and the retinal stem cells in the peripheral retina of fish and amphibians is not clear, since the latter are not pigmented [30]. However, it may be possible that functional variations of CE cells exist amongst different species.

More recent studies have strengthened the argument that CE cells from the mouse and human eye do not possess retinal stem cell properties. These investigations have demonstrated that CE spheres contain cells with epithelial properties and limited expression of neuronal markers compared to neurospheres formed by neural stem cells [31, 32]. Furthermore, investigations have shown that although cells isolated from the CE could be induced to express low levels of neuronal markers, they retained their epithelial morphology and failed to differentiate into retinal neurons [31]. These results are supported by insitu observations in the human eye that the pigmented ciliary epithelium does not express markers of stem/progenitor cells [25]. In summary these findings suggest that cells of the CE are not true retinal stem cells and therefore may not be suitable for retinal regeneration therapies.

The recent advent of iPS cells has opened a new potential avenue in the pursuit of cells for transplantation in regenerative medicine. Takahashi & Yamanaka [58] demonstrated that by retroviral induction of Oct3/4, Sox2, c-Myc and Klf4 pluripotent stem cells could be obtained from fibroblast cultures. Further study of these “reprogrammed” iPS cells revealed that their biological behaviour was indistinguishable from that of ESCs [59]. Recent refinements to the technique have enabled iPS cell generation without the use of viral vectors [60] and without induction of c-Myc [61] which is associated with an increase in tumorigenesis. However before such cells can be taken through to clinical trials, concerns regarding the effect of the reactivation of pluripotency upon a cell, alterations in target cells and characterisation of the cells need to be addressed [62].

The study of early human retinal development has been limited to comparative studies in other species and more recently to ESCs. It has been shown that iPS cells can also undergo a targeted differentiation process towards retinogenesis and have the ability to provide a further in vitro model of the developing human retina [63]. Both mouse and human IPS cells have been induced to differentiate into retinal progenitor cells in vitro [64]. Subsequent treatment of these iPS derived progenitors with retinoic acid and taurine resulted in cells expressing markers of photoreceptors, thus raising the possibility of an alternative source of photoreceptor progenitors for transplantation.

Although the field of iPS cell research is rapidly evolving, it needs further refinement before development of autologous cell based therapies for retinal disease can be safely developed.

Transplantation of stem and progenitor cells to regenerate the CNS is a subject of continuous investigations. Proposed mechanisms of therapeutic efficacy include neuroprotection of surviving cells [65] and/or retinal replacement [66]. However, various studies have identified a series of natural barriers that need to be overcome before these therapies can be brought to the clinic. Two of the major problems facing the development of stem cell therapies in the CNS is the presence of extracellular matrix molecules that prevent the outgrowth of axons and the low efficiency with which stem cells migrate and integrate into the retina. This has been shown by observations that when neural stem cells are transplanted into an injured spinal cord, most of the cells remain in the cavity wall and do not migrate or integrate with the host spinal cord [67]. In the eye, photoreceptors have been transplanted into the degenerating retina, but these attempts have resulted in little evidence of host-graft synaptic formation and functional recovery [68-70]. Studies involving retinal transplantation have shown that when a small number of cells integrate, they produce a short term improvement of function but are not sufficient to regenerate the diseased retina [10]. These findings have led researchers to investigate factors that may be potentially modified to improve the migration and integration of retinal stem cells. These include studies into the modulation of the extracellular matrix and the identification of the most appropriate ontogenetic stage at which transplanted cells will optimally integrate into the retina.

Although studies in the retinal stem cell field have led to the identification of various cell sources and methods to improve transplant cell migration and integration, much work is still needed before cell based therapies can be implemented to treat human retinal disease. Present knowledge would suggest that the development of stem cell therapies may be a viable proposition in the not too distant future, and efforts should be made to design practical approaches to source the most appropriate stem cells and to develop transplantation strategies that promote the long term functional integration of these cells into the diseased retina.