signalling pathways in Müller cells [45], implicating the existence of autocrine functions mediated by this NT.

Müller glia can be induced to release NGF upon inflammatory stimulation with interferon-γ and lipopolysaccharides [46]. As seen with other NTs and growth factors, NGF has autocrine activities, as illustrated by the fact that whilst Müller glial cells are able to produce this factor, they are also receptive to its activity. This is supported by observations that programmed cell death of retinal ganglion cells and Müller glia observed in diabetic rats can be prevented by treatment with NGF [47].

Evidence for the expression of NT4 by Müller glia mainly derives from in situ observations of retina from eyes with proliferative vitreoretinopathy, a common complication of retinal detachment [48]. Presence of NTs in Müller cells within degenerated retina suggests that pro-inflamma- tory events that lead to the local production of NTs may reflect the mechanisms by which Müller glia exert their neuroprotective functions.

Müller Glial Cells as a Source of Retinal Neurons in the Adult Eye

The ability of fish and amphibians to regenerate retinal tissue throughout life has been known for many years. These species harbour a population of stem cells that are located at the peripheral margin of the retina adjacent to the ciliary epithelium, known as the ciliary margin zone [49]. A similar population of retinal stem cells has been identified in birds and small mammals [50] during early postnatal life, and although a population of retinal progenitor cells has been identified in the ciliary body of adult mice [51], pigs [52] and humans [53], these appear to constitute a different progenitor cell population to that observed in the ciliary margin zone of the neural retina.

During development, Müller glia and retinal neurons share a common progenitor that is multipotent at all stages of retinal histogenesis

[54]. This evidence derives from examination of the progeny of a single mouse retinal progenitor cell transfected with a retrovirus, which generated clones containing up to three types of neurons, while others contained Müller glia alone, a combination of neurons and Müller glia, or a single neuron type. Markers of glial progenitors are therefore shared with retinal neural progenitors, and these can be used to identify Müller glia with stem cell characteristics, and to modulate the ability of these cells to differentiate into specified neurons in vitro and in vivo. Figure 2 summarizes the main developmental pathways that lead to Müller cell maturation and differentiation from the neural lineage.

One of the markers expressed by neural stem cells before they differentiate into the glial pathway is the bone morphogenetic protein-4 (BMP- 4), a member of the transforming growth factor-β (TGF-β) family. It has been implicated as a regulator of neural and glial differentiation. Work in mouse embryonic stem cells suggested that BMP- 4 acts in an inhibitory manner to prevent the neural differentiation of progenitors, directing them instead to a mesodermal fate [55]. Studies in the postnatal chick retina have demonstrated the ability of Müller glia to de-differentiate and become neurogenic in nature following chemically induced injury of the retina [56]. In the adult rat retina, they have been shown to proliferate and differentiate into bipolar cells and photoreceptors following neurotoxic injury to the adult retina [10].

Dividing Müller glia, as identified by Brd-U uptake, have been shown to express Cash-1, Chx10 and Pax-6, which are characteristic markers of multipotent retinal progenitors. During the ensuing period, the majority of these cells remained as undifferentiated progenitors expressing Chx-10 and Pax-6, whilst the others differentiated into either Müller glia or retinal neurons. Other factors such as the basic loop-helix-loop (bHLH) transcription factors also play a key role in the regulation of multipotent progenitor

Fig. 2. Ontogenesis of Müller glial cells. Summary of the major developmental pathways that lead to the development of Müller glia from a common progenitor that is shared with retinal neurons.

differentiation into either neural or glial cells. The proneural bHLH factors have been also demonstrated to possess an intrinsic activity to induce the generation of neurons, whereas the inhibitory type has been shown to stimulate the production of astrocytes [57].

Müller glial stem cells express the transcription factor Notch, whose signalling pathway involves a cascade of activation and repression of various transcription factors that lead to specific neuronal differentiation within the eye. The Hes1/Hes5 genes, known as Notch effectors, are activated by induction of the Notch pathway. These lead to downstream inhibition of further genes that in turn lead to inhibition of neural differentiation. Therefore, inhibition of the Notch pathway induces the differentiation of neural progenitors to differentiate into neurons, in particular into retinal ganglion cells. Upregulation of the Notch pathway permits the proliferation of progenitor cell populations whilst permitting the differentiation of sub-group towards a glial fate [58].

Several studies have elucidated the possibility of Müller glia being a source of progenitor cells over the past decade. Müller cells have been described as a source of retinal progenitors in the postnatal chick retina [56], and have been shown to proliferate and differentiate into bipolar cells and photoreceptors following neurotoxic injury to the adult rat retina [10]. The neurogenic capacity of Müller glia has been clearly demonstrated in the zebrafish, where they form the retinal stem cell niche, and also show an ability to regenerate retinal neurons following injury [59].

Recent studies have identified a population of Müller glia with neural stem cell characteristics in the adult human eye, independent of sex or age [9]. These cells become spontaneously immortalized in vitro, an important characteristic of stem cells, and can be frozen and thawed without losing their progenitor ability for many passages in culture. Under normal culture conditions, they express neural stem cell markers such as nestin, βIII tubulin, Sox-2, Pax-6, Chx10 and Notch-1 [9]. In vitro differentiation studies of

these Müller stem cells in the presence of growth factors provided further evidence of the pluripotency of this cell population. Under specific conditions in vitro they can be induced to express markers of post-mitotic retinal neurons, including peripherin, S-opsin and recoverin (markers of photoreceptor cells), calretinin and neurofilament protein (markers of ganglion, horizontal and amacrine cells), and HuD and Brn3 (markers of ganglion cells; fig. 3) [9].

Transplantation of Müller stem cells into the subretinal space of the dystrophic RCS rat and the neonatal and adult Lister hood rats has been

performed to investigate the in vivo potential of these cells to migrate and differentiate into the retina and to restore damaged retina function. Initial experiments showed that only a small proportion of these cells were able to migrate into the retina. However, they expressed neural markers characteristic of the retinal cell layer to which they migrate. Transplanted cells found within the outer nuclear layer of the retina, where photoreceptors are present, expressed markers of photoreceptor cells such as recoverin and rhodopsin. Those cells that migrated into the ganglion cell layer expressed calretinin and HuD,

characteristic markers of ganglion cells [9]. These findings suggest that although Müller stem cells have the ability to differentiate and repopulate damaged retinal neurons, there are barriers that prevent a successful integration of transplanted cells into the retina.

Potential Barriers for Stem Cell Transplantation to Regenerate Retinal Neurons in the Diabetic Retina

Investigations have shown that the limited migration and integration of grafted Müller stem cells into the degenerated retina of the RCS rat are partly due to the accumulation of a group of ECM proteins produced during reactive gliosis [60]. These proteins, known as chondroitin sulphate proteoglycans (CSPGs), are known to prevent axonal regeneration and synapse formation following injury [61]. Remodelling of the retina following a pathological process similar to that seen elsewhere in the central nervous system often results in the deposition of glial scar rich in CSPGs [61]. The diabetic retina exhibits all the characteristics of reactive gliosis, which in addition to neovascularization, presents with accumulation of abnormal extracellular proteins and glial cell proliferation [7]. In addition, abnormalities of retinal pericytes and endothelial cells observed in the DR have been associated with increased expression of CSPGs [62]. Based on this evidence, methodologies designed to facilitate migration and integration of transplanted Müller stem cells into other experimental models of retinal degeneration may be applied to transplantation into the diabetic retina.

In eyes with DR, microglia have been shown to be markedly increased in number and display hypertrophic features at different stages of the disease [63]. These cells accumulate around the retinal vasculature, especially the dilated veins, microaneurysms, intraretinal haemorrhages, cotton-wool spots, and areas of retinal

neovascularisation. Microglia have also been observed in the outer retina and subretinal space of retinae with cystoid macular oedema, and in epiretinal membranes of PDR [63]. These observations present with another potential barrier for retinal regeneration of the diabetic retina, due to evidence that microglia prevent a successful integration of transplanted cells [64] In addition, microglia have been shown to produce the inhibitory CSPGs described above, both in vitro and in vivo, which potentiates the deposition of abnormal ECM in the diabetic retina. In this context, the demonstration of breakdown of the ECM with chondroitinase ABC in conjunction with strong microglial suppression [60] has provided major knowledge that can be applied to facilitate the development of cell-based therapies to restore retinal function lost by diseases such as DR. Further work to optimise transplantation strategies and to identify the most appropriate cell sources for retinal therapies are at present being developed in many laboratories across the world.

Potential of Müller Stem Cells for the Development of Human Therapies to Restore Retinal Function Damaged by Disease

Death of retinal neurons is the major cause of blindness in neovascular and inflammatory retinal conditions such as PDR. Current treatments currently offer a palliative role in the prevention of visual loss, but in the long term it is difficult to prevent loss of vision in many individuals. The only realistic expectation for restoration of retinal function is the development of stem-cell based therapies to replace and integrate lost neurons into the neural network.

Müller stem cells have a potential advantage over other stem cells in that they can be derived from adult human retina and can be potentially isolated from the same individual to avoid problems arising from lack of histocompatibility and crossed infections. Although Müller glial cell