We acquire information from the outside world through our eyes which contain the retina, the photosensitive component of the central nervous system. Once the adult mammalian retina is damaged, the retinal neuronal death causes a severe loss of visual function. It has been believed that the adult mammalian retina had no regenerative capacity. However, the identification of neuronal progenitor cells in the retina sheds some light on cellular therapies for damaged retinal regeneration. In this review, we highlight three potential stem/progenitor cells in the eye, the ciliary body epithelium cells, the iris pigmented epithelium cells, and Müller glia. In order to make them prime candidates for the possible treatment of retinal diseases, it is important to understand their basic characters. In addition, we discuss the key signaling molecules that function extracellularly and determine whether neuronal progenitors remain quiescent, proliferate, or differentiate. Finally, we introduce a secreted protein, Tsukushi, which is a possible candidate as a niche molecule for retinal stem/progenitor cells.
Although it has long been believed that the central nervous system (CNS) is incapable of regeneration, the possibility of neural regeneration has gained credence with the identification of neural stem cells seeded within different regions of the adult CNS. The vertebrate retina is a well-characterized structure, consisting of seven major cell types, which are arranged in a stereotypical laminar organization (Fig. 1A,B; Harada et al. 2007). As the retina is the photosensitive component of the CNS located in the eye, retinal neuronal death causes a severe loss of visual function. The accessibility of the retina, together with an extensive knowledge base relating to the organization and function of this structure, make it a prime candidate for developing cellular therapies for CNS disorders. In mammals and other warm-blooded vertebrates, several studies have identified a number of different cellular sources of neural stem cells within the adult eye (Perron & Harris 2000). Neuronal progenitors with retinal potential are present in the ciliary body epithelium (Ahmad et al. 2000; Tropepe et al. 2000), the iris pigment epithelium (Haruta et al. 2001; Sun et al. 2006; Asami et al. 2007), and the peripheral margin of the postnatal retina (Moshiri & Reh 2004; Zhao et al. 2005). Furthermore, Müller glial cells have been demonstrated to possess retinal progenitor properties in the adult retina (Fischer & Reh 2001; Ooto et al. 2004; Osakada et al. 2007). These progenitor cells are capable of differentiating along neuronal- and glial-pathways and generating specific retinal cell types when exposed to a conductive environment. The location of these progenitor cells in the eye is summarized in Figure 1C. Despite the demonstration of multipotentiality, they are not defined as stem cells because of the requirement for complex cellular interactions and the lack of self-renewal potential in vitro. Therefore, in this review, we refer to the multipotent neural progenitors in the optic vesicle derivatives as retinal stem/progenitor cells.
The retinal stem/progenitor cells can be categorized based on their differential responsiveness to extrinsic factors. Depending on the type of progenitors, some proliferate extensively when removed from their niche and differentiate along neuronal and glial lineages, while others do not. Since these cells respond to growth factor treatment by increasing their level of proliferation in vivo, they are hypothesized to be a potential source of stem cells to support retinal regeneration and may contribute to retinal regeneration therapy. This review mainly focuses on recent reports for three kinds of vertebrate retinal stem/progenitor cells that are located in the ciliary body (CB), the iris epithelium, and the central retina, and does not include the potential of retinal pigmented epithelium (RPE) and the embryonic stem cells; the reader is referred to an excellent review on this possibility by Fischer (2005) and Vugler et al. (2007), respectively. We also describe the responsiveness of the retinal stem/progenitor cells to signaling molecules, which regulate whether they remain quiescent, proliferate, or differentiate. Finally, we introduce a soluble factor Tsukushi (TSK), identified as a BMP antagonist (Ohta et al. 2004), and discuss the possibility that TSK may play a role as a niche molecule for retinal stem/progenitor cells.
Ciliary body epithelium cells
In cold-blooded vertebrates, such as fish and amphibians, the growth of the retina is coordinated with the overall growth of the eye and the retinal neurogenesis continues throughout the life of the organism. This is supported by stem/progenitor cells found in the peripheral margin region of the retina, called the ciliary marginal zone (CMZ; Straznicky & Gaze 1971; Johns 1977). Fischer and Reh (2000) found a zone of cells at the retinal margin of postnatal chick that resembles the CMZs of fish and amphibians, and cells comprising this zone contribute to the growth of the retina that occurs postnataly. However, a report that searched for a comparable region in the mammalian retina indicated that there is no CMZ in the mouse retina (Kubota et al. 2002). Moreover, the CMZ has progressively decreased in size during vertebrate evolution (Kubota et al. 2002). Notably, the CB in the adult mammalian eye contains a quiescent population of cells that can be expanded in vitro (Ahmad et al. 2000; Tropepe et al. 2000). The CB is a neuroepithelial derivative and anatomically located, together with the iris, at the most distal tip of the neural retina. The CB is composed of two distinct epitheliums opposed at their basal surfaces, which form a complex tri-dimensional network with capillaries (Morrison & Freddo 1996). The inner layer of the ciliary body epithelium (CBE) cells consists of non-pigmented cells, whereas the outer layer contains pigmented cells. Cell labeling experiments by chronic injection of BrdU revealed that a rare population of mitotically quiescent cells is located in the pigmented CE cells of adult rats (Ahmad et al. 2000). When the CB from adult mice was dissociated into single cells by enzymatic treatment and cultured in serum-free medium, cells proliferated to form pigmented sphere colonies (PSCs) regardless of growth factor condition (Tropepe et al. 2000). Furthermore, PSCs can be subcloned for at least six generations, which might depend on endogenous FGF2 partially. On the other hand, exogenous FGF2 can influence the formation and the proliferation of PSCs (Tropepe et al. 2000). Non-pigmented sphere colonies can be isolated from adult albino CD1 mice, suggesting that the biochemical components of pigment formation are not required for progenitor cell proliferation (Tropepe et al. 2000). In situ hybridization and reverse transcription–polymerase chain reaction (RT–PCR) analyses showed that PSCs express retinal progenitor markers Chx10 and nestin and markers that are involved in early eye morphogenesis such as Pax6, Six3, and Rx (Ahmad et al. 2000; Tropepe et al. 2000; Lord-Grignon et al. 2006). Transcription profiling of PSCs indicated that PSCs also express genes that are involved in the control of embryonic development, retinal identity, redox metabolism, and cellular proliferation (Ahmad et al. 2004; Lord-Grignon et al. 2006). Taga and colleagues demonstrated that the number of PSCs was increased in the presence of Wnt3a in vitro (Inoue et al. 2006). Moreover, Wnt3a-treated PSC cells retained multipotency and formed a greater number of secondary spheres than non-treated cells (Inoue et al. 2006). They also reported that the canonical Wnt signaling pathway contributes to the proliferative effect of Wnt3a on PSCs and the combination of FGF2 and Wnt3a has a strong additive effect on their proliferation (Inoue et al. 2006). When PSCs were cultured under conditions known to promote retinal cell differentiation (Ahmad et al. 2000; Tropepe et al. 2000), some cells that migrated away from PSCs expressed the pan-neuronal marker microtubule-associated protein 2 (MAP2) and neurofilament, while other cells expressed the astrocytic marker glial fibrillary acidic protein (GFAP). Later, differentiated PSCs contained a small number of nestin-positive cells that remained confined to the centers of the colonies (Tropepe et al. 2000), and they expressed several markers that are seen in different differentiated retinal cells; rod photoreceptors (RetP1, ROM-1, 309L, Rho1D4, D2P4), bipolar cells (protein kinase C, Chx10), retinal ganglion cells (β-tubulin), and Müller glia (10E4). Thus, the cells comprising PSCs possess the potential to differentiate into different retinal cell types in vitro. Interestingly, stem/progenitor cells in the CBE respond to injury and growth factor treatment by extensive proliferation in vivo, suggesting that they are capable of endogenous activation (Ahmad et al. 2004). Further work will be required to elucidate the precise identification of retinal stem cells in the CB although PSC cells are proliferative, multipotent, and self-renewable. Considering the extra-retinal location of cells comprising the CB, an important question is whether these multipotential cells can migrate into and differentiate in the retina.
Iris pigment epithelium cells
Iris pigment epithelium (IPE) cells have the same developmental origin as RPE, CE, and retinal cells, which are located in the most peripheral region of ocular tissue. IPE has long been shown to exhibit a remarkable and inherent capacity for lens regeneration in newt, chick, and human in vitro (Eguchi 1986; Tsonis et al. 2001; Kosaka et al. 2004). In the newt, if the lens is surgically removed from the adult eye, a structurally and functionally complete lens always regenerates from the dorsal margin of IPE (Eguchi 1971; Eguchi 1986). This phenomenon of lens regeneration from the IPE suggests the possible function of IPE cells as stem/progenitor cells in ocular tissue.
When the iris tissue of adult rat was cultured in the presence of FGF2, many cells migrated out of the iris, and some of them expressed neurofilament but not rhodopsin, a specific marker for rod photoreceptor (Haruta et al. 2001). However, the adult rat iris-derived cells could acquire photoreceptor-specific phenotype as a result of the ectopic expression of Crx, which is the homeobox gene specifically expressed in the photoreceptors of the mature retina and is crucial for photoreceptor differentiation (Haruta et al. 2001; Akagi et al. 2005). In the case of primates, a combination of Crx and NeuroD, a basic helix-loop-helix gene expressed in developing and matured rod photoreceptors, were needed to induce monkey iris-derived cells to adopt photoreceptor-specific phenotypes (Akagi et al. 2005). Takahashi et al. showed that NeuroD expression was inducible by transferring the Crx gene into rat-iris-derived cells but not monkey-iris-derived cells (Akagi et al. 2005). Furthermore, these photoreceptor-like cells derived in the above manner have a photoresponse to light stimuli and the potential to integrate and survive within the developing retina, suggesting the possible application of these cells to the treatment of human retinal diseases (Akagi et al. 2005).
Recently, Kosaka and colleagues succeeded in isolating postnatal chicken and mammalian IPE cells and maintaining them for long periods in vitro (Sun et al. 2006; Asami et al. 2007). When isolated primary IPE cells from 2-day-old chickens were cultured in the presence of FGF2 and/or epidermal growth factor (EGF) on a gyratory shaker, they formed the neuroshpere-like structures and gradually lost their melanin. RT–PCR and immunostaining analyses showed that IPE-derived spheres expressed Pax6 and vimentin, which are retinal progenitor markers, but not pan-neural differentiation markers such as Tuj1 and GFAP (Sun et al. 2006). When primary IPE cells were plated onto laminin-coated dishes in the presence of FGF2, they extended thin cell processes and expressed retinal specific neuronal markers, rhodopsin, iodopsin (cone photoreceptor), PKC, and HPC-1 (amacrine), indicating that IPE cells have the potential to generate retinal-specific neurons in vitro (Sun et al. 2006). By transplanting into the space between the RPE and the photoreceptor layer, IPE cells were incorporated into the subretinal space and differentiated into the photoreceptor cells (Sun et al. 2006). Using isolated IPE cells from postnatal mammalian eye, Kosaka et al. also demonstrated their multipotent capacity and plasticity for cell differentiation (Asami et al. 2007). In the mammalian IPE, the inner and outer IPE layers differentially expressed nestin in a manner suggesting that they shared origins with the neural retina and the pigmented epithelial layers, respectively. The nestin-expressing cells are located only in the IPE inner layer. They proliferate and exhibit the multipotent ability to differentiate into retina-specific cells in response to FGF2, whereas the nestin-nonexpressing cells in IPE are non-proliferative and exhibit restricted neuronal potency; hence, they develop into pan-neural marker expressing cells (Asami et al. 2007). Their data collectively suggested that discrete populations of highly pigmented cells with heterogeneous developmental potencies exist postnatally within the IPE, and that some of them are able to differentiate into multiple neuronal cell types, including retinal specific neurons (under certain conditions) without gene transfer. Taken together, the relative ease with which the IPE can be surgically isolated as a recognizable structure to yield multipotent cells in vitro raises the possibility that these cells constitute a potential source of retinal transplantation in patients with retinal degenerative diseases or damaged retinae.
The neurons and Müller glia that make up the mature retina are generated in a precise sequence during histogenesis and specifically, Müller glia are generated last (Marquardt & Gruss 2002; Cayouette et al. 2006). Ficher and colleagues have tested the potential of the avian retina to regenerate after they are neuro-toxic damaged (Fischer & Reh 2001; Fischer & Reh 2003). When extensive cell death of inner retinal neurons, such as amacrine cells, was induced by making intraocular injections of toxic levels of N-methyl-D-asparate (NMDA), large numbers of mitotically active cells, shown by BrdU incorporation and immunoreactivity for Phospho-Histone H3, were found in the retina (Fischer & Reh 2001). By double labeling for BrdU and glutamine synthetase (GS), a marker for Müller glia, the mitotically active cells were identified as Müller glia after 2 days of toxin treatment. They followed the fate of the double positive cells over the next few weeks and it was found that the BrdU-labeled cells can adapt one of three different fates; a small percentage (less than 4%) of the Müller glia-derived cells differentiate into retinal neurons that express Hu or calretinin, a greater percentage (about 20%) of the newly generated cells differentiate into Müller glia that express GS, and the rest of the cells remained undifferentiated with continued expression of pax6 and Chx10 (Fischer & Reh 2003).
Ahmad and colleagues enriched Müller glia from rat retina and cultured them in the presence of FGF2 (Das et al. 2006). These cells generated neurospheres and transdifferentiated into the retinal neurons, astrocytes, and oligodendrocytes in vitro (Das et al. 2006). They demonstrated that Müller glia display cardinal features of neural stem cells in vitro, for example, they self-renew, and generate both neurons and glia (Das et al. 2006). When enriched Müller glial cells were transplanted into injured retina, they integrated and generated lamina-specific retinal neurons (Das et al. 2006). The upregulation of Notch and Wnt pathway genes in neurotoxin-damaged retina suggests the involvement of these pathways in the stem cell potential of Müller glia.
Using NMDA as a neurotoxicity model for adult mammalian retina, Takahashi and colleagues observed that some Müller glial cells were stimulated to proliferate and produce bipolar cells and rod photoreceptors after retinal damage (Ooto et al. 2004). They also showed that retinoic acid (RA) treatment increased bipolar cell genesis, while misexpression of homeobox and bHLH genes promoted the regeneration of various retinal neural cells (Ooto et al. 2004). Considering the number of newly generated retinal neurons, the risks associated with invasive surgery, and the rejection of grafted cells in cell transplantation, they sought an alternative by examining the effects of Wnt signaling in the regeneration process after retinal injury with regard to intrinsic progenitor cells. Retinal explants isolated from adult rats were used as an in vitro injury model (Osakada et al. 2007). In the damaged retinal explants, Müller glia proliferated and re-entered the cell cycle to a limited degree in response to injury. When these explants were cultured in the presence of Wnt3a protein, the proliferation of Müller glia-derived retinal progenitors was increased after damage. In both the non-injured retina and the damaged retina, β-catenin was observed in the cytoplasm and at the plasma membrane. However, Wnt3a treatment only promoted nuclear translocation of β-catenin in the damaged retina (Osakada et al. 2007). They also showed that Wnt signaling activation by applying an inhibitor of glycogen syntase kinase (GSK)-3β, which phosphorylates and destabilizes β-catenin, promoted the neural regeneration of the retina, whereas Wnt signaling inhibition by Dkk-1, a soluble negative modulator of Wnt signaling, attenuated the retinal regeneration (Osakada et al. 2007). When the damaged retinal explants received Wnt3a treatment, followed by incubation in regular medium without Wnt3a, the Müller glia-derived retinal progenitors migrated into the outer nuclear layer (ONL), in which they differentiated into rod photoreceptor cells (Osakada et al. 2007). They also showed that the differentiation of progenitor-like cells into photoreceptor cells was promoted in the presence of RA, which is crucial for photoreceptor genesis (Hyatt & Dowling 1997). Taking these findings together, they concluded that the Wnt/β-catenin signaling pathway regulates the regeneration of retinal neurons after injury in the adult retina. Thus, the low-molecule compounds that activate Wnt/β-catenin signaling may have therapeutic potential for promoting the regeneration of neurons in the adult mammalian CNS.
Expression of Tsukushi in the eye
We have introduced three types of cells that might function as neuronal stem/progenitor cells in the adult warm-blooded vertebrate retina. However, the molecular mechanism of the stem/progenitor cells that remains dormant under the constraint of the non-neurogenic environment of the adult normal retina is still unclear. The failure of neuronal stem/progenitor cells in the mammalian and avian retina to renew retinal cells in the postnatal period suggests that the progenitor cells find themselves in an inhibitory environment. Nevertheless, no studies to date have identified definitive stem cells in the eye or examined the nature of the niche that regulates the self-renewal and differentiation of retinal stem/progenitor cells.
We have identified a secreted factor, Tsukushi (TSK), which belongs to the small leucine rich repeat proteoglycan (SLAP) family (Iozzo 1997; Ohta et al. 2004). During chick gastrulation, TSK interacts with and modulates activities of the TGF-β family members, BMP and Vg1, and is involved in the primitive streak and Hensen's node formation (Ohta et al. 2004; Ohta et al. 2006). In addition, TSK interacts with the Notch ligand, Delta-1, and controls ectodermal patterning and neural crest specification in Xenopus embryos (Kuriyama et al. 2006). Recently, we identified two further signaling pathways that are regulated by TSK. During Xenopus germ layer formation, TSK binds to and inhibits FGF8 signaling and binds to and enhances Xnr2 signaling (Morris et al. 2007). Our accumulating data suggest that TSK is a key coordinator of these multiple pathways outside the cell through regulation of an extracellular signaling network.
To examine TSK expression in mice, TSK exon was replaced by the LacZ gene and β-gal staining was carried out. In adult mice eyes, TSK is expressed in the inner nuclear layer (INL), the CB, and lens epithelium. This expression pattern is conserved among Xenopus, chick, and mouse (Fig. 2). The expression of TSK in the CB suggested that TSK has some biological roles not only on CBE cells but also on the growth factor-responsive progenitors at the peripheral edge of the postnatal mammalian retina (Zhao et al. 2005). In addition, TSK is also expressed in the INL where the cell bodies of Müller glia are located, suggesting that TSK regulates the proliferation and maintenance of Müller glia. As it is difficult to observe the expression of TSK in the iris epithelium because of its high level of pigmentations, we will generate a heterozygote between the TSK KO mice and the albino CD1 mice to examine TSK expression. Since Wnt signaling promotes the proliferation and regeneration of CBE cells and Müller glia (Das et al. 2006; Inoue et al. 2006; Osakada et al. 2007) and because TSK inhibits the role of Wnt2b, which supports prolonged proliferation of chick retinal progenitor cells (Kubo et al. 2005; Ohta et al. 2007), it is likely that TSK functions as a niche molecule through the inhibition of Wnt signaling.
The identification of retinal stem/progenitor cells in the adult mammalian retina opens the possibility that these cells may have potential for transplantation in the treatment of retinal degeneration disorders such as retinitis pigmentosa and age-related macular degeneration (Ali & Sowden 2003). Experimental studies using rodents confirm the potential for retinal stem/progenitor cells transplantation and emphasize the importance of inductive signals for the proper differentiation and integration of donor cells into the host retina. However, recovery from retinal degeneration disorders is highly dependent on the restoration of functional neuronal retinal circuits after such transplantation procedures. While we have learned a workable understanding of retinal stem/progenitor cell properties, our next step must overcome several difficulties concerning how to control neuronal circuitry, not only among retinal cells, but also between the eye and the brain.
We thank Douglas S Campbell, Sandy Chen, Mitsuko Kosaka, Fumitaka Osakada, and Yohei Shinmyo for comments and help with writing the manuscript, Kumiko Hori for help with mouse embryos, and all members of our labs for their valuable help. This work was supported by PRESTO of the Japan Science and Technology Agency (K.O.) and Grants-in-Aid from the Ministry of Education, Science, Sports, and Culture of Japan (K.O., H.T.), by 21st Century COE Research (H.T.), and by Global COE Research (A.I., H.T.).