Author contributions: Y.F.: data collection, analysis and interpretation, and manuscript writing; K.-S.C.: financial support, data collection, analysis and interpretation, and revising the manuscript; K.T.: data collection and analysis; S.W.L.: electrophysiology experimental design, data collection, analysis and interpretation, and manuscript writing; C.G. and H.K.: data collection, analysis and interpretation; S.F. electrophysiology experimental design, data analysis and interpretation, manuscript writing, and financial support; X.S.: financial support, administrative support, and manuscript writing; D.F.C.: financial support, conception and design, data analysis and interpretation, manuscript writing and revision, and final approval of manuscript. Y.F. and K-S.C. contributed equally to this article.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS November 16, 2012.
The ciliary epithelium (CE) of adult mammals has been reported to provide a source of retinal stem cells (RSCs) that can give rise to all retinal cell types in vitro. A recent study, however, suggests that CE-derived cells possess properties of pigmented ciliary epithelial cells and display little neurogenic potential. Here we show that the neurogenic potential of CE-derived cells is negatively regulated by ephrin-A3, which is upregulated in the CE of postnatal mice and presents a strong prohibitory niche for adult RSCs. Addition of ephrin-A3 inhibits proliferation of CE-derived RSCs and increases pigment 349 cell 359. In contrast, absence of ephrin-A3 promotes proliferation and increases expression of neural progenitor cell markers and photoreceptor progeny. The negative effects of ephrin-A3 on CE-derived RSCs are mediated through activation of an EphA4 receptor and suppression of Wnt3a/β-catenin signaling. Together, our data suggest that CE-derived RSCs contain the intrinsic machinery to generate photoreceptors and other retinal neurons, while the CE of adult mice expresses negative regulators that prohibit the proliferation and neural differentiation of RSCs. Manipulating ephrin and Wnt/β-catenin signaling may, thus, represent a viable approach in activating the endogenous neurogenic potential of CE-derived RSCs for treating photoreceptor damage and retinal degenerative disorders. STEM CELLS2013;31:349–359
Retinal degenerative diseases characterized by death of photoreceptor cells, including age-related macular degeneration, retinitis pigmentosa, and cone dystrophy, are leading causes of irreversible blindness. Photoreceptor cells, like most neurons in the mammalian central nervous system (CNS), do not spontaneously regenerate once lost . Stem cell therapy, thus, holds a great promise for reversing vision loss and treating these disorders. To date, efforts have been made to identify an endogenous source of retinal stem cells (RSCs) of adult mammals, and cells derived from the ciliary epithelium (CE) of the eye are thought to contain such a group of cells .
The ciliary marginal zone (CMZ) of lower vertebrates, such as teleosts and amphibians, is known to harbor a pool of RSCs capable of generating new retinal neurons throughout life . The regenerative process usually starts with the dedifferentiation of pigmented cells; they then proceed to depigmentation, reentrance into the cell cycle, and expression of neural retinal progenitor genes [4, 5]. Although the retinas of adult mammals do not regenerate, a population of multipotent RSCs has been isolated from the CE of adult rodents and humans that can give rise to all retinal cell types in vitro [2, 6–8]. However, their ability to proliferate and generate new retinal neurons, such as photoreceptors, appears to be limited in vivo [9–11]. Mitogens, including basic fibroblast growth factor (bFGF), insulin, Wnt3a, and pigment epithelium-derived factor, are found to promote the proliferative potential of CE-derived RSCs [12–16]. Transcription factors, such as OTX2, Crx, and Chx10, increase the photoreceptor progeny of CE-derived RSCs . Recent reports, however, suggest that the CE-derived cells of adult mammals possess pigmented ciliary epithelial cell properties and are incapable of differentiating into retinal neurons or photoreceptors [10, 11]. This study argues against the validity of the existing theory that CE-derived cells present an endogenous source of RSCs for cell replacement therapy to treat retinal neurodegeneration without prerequisite reprogramming.
We have previously reported that the neurogenic potential of CNS stem cells is critically regulated by positional cues in the local environment and is prohibited by expression of high levels of growth inhibitors, such as ephrin-A2 and -A3, in adult mice . Ephrins and Eph receptors are pivotal regulators of pattern formation, cell migration, axonal guidance, and synaptogenesis during CNS development . Accumulating evidence indicates that ephrin-As, acting through receptor EphA7, negatively regulate neural stem cell survival and proliferation in the developing telencephalon  and subventricular zone , or in the non-neurogenic CNS regions of adult mice [18, 22]. Absence of ephrin-A2 and -A3 in adult mice results in active ongoing neurogenesis in diverse CNS regions . These findings led to our hypothesis that ephrins also participate in the control of the neurogenic program of CE-derived cells in adult mice by functioning as a negative regulator. To address this hypothesis, we set out to test whether ephrin-As suppress the neurogenic potential of CE-derived cells and whether manipulating ephrin signaling is sufficient to promote the neurogenic program and photoreceptor potential of these RSCs.
MATERIALS AND METHODS
Ephrin-A3−/− mice were generated as previously described , and C57BL/6J wild-type mice were purchased from Jackson Laboratory (Bar Harbor, Maine). Mice older than 2 months were referred to as adult. All experimental procedures and use and care of animals followed a protocol approved by the Animal Care and Use Committee at the Schepens Eye Research Institute, Harvard Medical School.
Preparation of CE-Derived Cell and Neurosphere Cultures
As previously described , briefly, mouse eyeballs were hemisected; the CE was carefully dissected free from all surrounding tissues and dissociated using a papain-based dissociation system (Worthington Biochemical, Lakewood, NJ) according to the manufacturer's instructions. Dissociated cells were resuspended in Dulbecco's modified Eagle's medium (DMEM)/Ham's F-12 medium (F12) (1:1; Life Technologies, Grand Island, NY) containing N2 supplement (1:100; Life Technologies ), epidermal growth factor (20 ng/ml; Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com), bFGF (20 ng/ml; Sigma-Aldrich), heparin (2 μg/ml; Sigma-Aldrich), and 1% penicillin-streptomycin (Sigma-Aldrich). Cells were plated at a density of 2 × 104 cells per well in a 24-well culture plate. Fresh growth factors were added every other day, and the medium was changed every 5–7 days. Neurospheres (NSs), which were defined as free-floating spheres with a defined outer boundary and a diameter larger than 40 μm, were counted after 7 days of incubation . A minimum of five wells were counted per group, and each experiment was repeated at least three times. For self-renewal studies, individual NSs were collected, dissociated, and seeded at a clonal density of 2 × 104 cells per well in a 24-well plate. After 14 days of incubation, secondary NSs were counted [8, 23].
Cell Proliferation Assay
To study cell proliferation, primary NS obtained after 7 days in culture were dissociated to a single cell suspension and seeded in 96-well culture plates at a density of 1,000 cells per well, in DMEM/F12 medium containing 10% fetal bovine serum (FBS; Life Technologies) for 48 hours. 5-Bromo-2-deoxyuridine (BrdU, 0.5 μM; Sigma-Aldrich, St. Louis, MO) was added to the culture media 4 hours before fixation. Cell nuclei were stained with the nuclear marker 4,6-diamidino-2-phenylindole dihydrochloride (DAPI; Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com) and counted, and the percentage of BrdU+ cells was recorded. The results presented were obtained from a minimum of three independent experiments. To investigate the effect of ephrin-A3 on the proliferation and differentiation of CE-derived cells in vitro, antibody-clustered ephrin-A3 (300 ng/ml; R&D systems; Minneapolis, MN, http://www.rndsystems.com) or control Fc (300 ng/ml; R&D Systems) was added to dissociated NS cell cultures. Antibody-clustered forms of proteins were generated by incubating the protein with anti-human IgG (Jackson Immuno Research Laboratories, West Grove, PA, http://www.jacksonimmuno.com) for 1 hour at room temperature before application [18, 21]. In some experiments, recombinant mouse Wnt3a (20–50 ng/ml; R&D Systems) were added to the culture medium. To investigate the role of EphA4 receptor in the proliferation of CE-derived RSCs, the EphA4 soluble protein (2.5 μg/ml; R&D Systems) or control Fc (2.5 μg/ml) was added to the culture medium.
Cell Differentiation Assay
To study the differentiation potential of CE-derived RSCs, 7-day-old CE-derived NSs were dissociated in a 0.25% Trypsin-EDTA (Sigma-Aldrich) solution. Cells were plated at a density of 1 × 104 cells per well in a 24-well plate precoated with poly-L-ornithine (10 μg/ml) and laminin (100 ng/ml, both from Sigma-Aldrich). The cultures were maintained in Neurobasal medium, Life Technologies, containing B27 supplement, Life Technologies, penicillin/streptomycin, 2 mM L-glutamine, Life Technologies, and 1% FBS for 10–21 days, and the medium was changed every 3–4 days.
Immunofluorescence staining was performed using a standard protocol. Primary antibodies used included rat anti-BrdU (1:100; Abcam, Cambridge, MA, http://www.abcam.com), mouse monoclonal antibodies specific for β-III-tubulin (Tuj1; 1:200; BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) and microtubule-associated protein 2 (MAP2; 1:100; Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com), rabbit anti-recoverin antibody (1:1,000; Chemicon, Bellerica, MA, http://www.chemicon.com), sheep anti-Chx10 antibody (1:500, Abcam, Cambridge, MA), rabbit anti-glial fibrillary acidic protein (GFAP) antibody (1:1,000; Sigma-Aldrich), rabbit anti-ephrin-A3 antibody (1:100; Santa Cruz Biotechnology), rabbit anti-EphA4 antibody (1:100; Abcam), and secondary antibodies conjugated with Cy2, or Cy3 (Jackson Immunoresearch Laboratories, West Grove, PA). The slides were mounted with VECTASHIELD mounting media (H-1200; Vector Labs, Burlingame, CA).
Quantification of Cell Proliferation In Vivo
To label proliferating cells in vivo, adult C57BL/6J wild-type and ephrin-A3−/− mice received intraperitoneal injections (twice daily) of BrdU (50 mg/kg; Sigma-Aldrich) dissolved in 0.9% sterile saline for 7 consecutive days. Seven days after the first BrdU injection, mice were killed and transcardially perfused with 4% paraformaldehyde (Sigma-Aldrich). Transverse cryosections of the mouse eyes (12 μm) were pretreated with HCl (2.0 N; Sigma-Aldrich) for 1 hour at 37°C and subjected to BrdU immunolabeling. Eyeball sections were counterstained with DAPI (1 mg/ml) in phosphate buffered saline (Sigma-Aldrich) to reveal cell lamina structures. Cells stained positive for BrdU were counted, and a minimum of 20 sections per mouse, nine mice per group, were analyzed.
Reverse Transcription-Polymerase Chain Reaction and Real-Time Polymerase Chain Reaction
Total RNAs were extracted from 7-day-old CE-derived primary NS or from NS-derived cells at 10 days after induction of differentiation, using TRIzol reagent (Life Technologies) in accordance with the manufacturer's instructions. Reverse transcription was performed using TaqMan RT reagents (P/N N808-0234; Applied Biosystems, Carlsbad, CA, http://www.appliedbiosystems.com). Quantitative detection of specific mRNA transcripts was carried out by real-time reverse transcription polymerase chain reaction (RT-PCR) using an ABI7900 HT Sequence Detection System (Applied Biosystems, Carlsbad, CA). Primers were designed using OligoPerfect Designer software provided on the Invitrogen website (http://tools.invitrogen.com/content.cfm?pageid = 9716) or as previously described . The sequences of all primers are listed in supporting information Table 1. Samples were analyzed in triplicate, and relative amounts of mRNAs were determined by normalizing to β-actin and glyceraldehyde 3-phosphate dehydrogenase expression levels. The same primers were used for conventional RT-PCR, which was carried out at a 60°C annealing temperature for 32 cycles.
Immunoprecipitation and Immunoblotting
Tissues or cultured stem cells were lysed with lysis buffer (100 mM HEPES/KOH, 1 mM EDTA, 200 mM KCl, 20% glycerol, 0.1% octylphenoxy-polyethylene glycol, 1 mM DL-Dithiothreitol, 0.1 mM phenylmethyl sulfonyl fluoride, 0.2 mM sodium orthovanadate) and subjected to centrifugation at 12,000 rpm for 10 minutes. Coimmunoprecipitation assays were performed using Protein G Magnetic Beads, following the manufacturer's protocol (S1430S, New England Biolabs Inc., Ipswich, MA). Cell or retinal tissue lysates (20–50 μg per lane) were loaded onto 4%–20% Precise Protein Gels (Thermo Scientific, Rockford, IL) followed by Western blotting with an anti-mouse phosphotyrosine antibody (4G10; 1:1000, Millipore; Upstate, NY), rabbit anti-EphA4 (1:100; Abcam), rat anti-EphA7 (1:100; Abcam), rabbit anti-ephrin-A3 (1:100; Santa Cruz Biotechnology), rabbit anti-β-catenin (1:4,000; Abcam). Mouse monoclonal β-actin antibody (1:5,000; Abcam) or mouse monoclonal GAPDH antibody conjugated to chemiluminescent horseradish peroxidase (HRP) substrate (1:5,000; Abcam) was used as loading controls. Membranes were incubated with appropriate HRP-conjugated secondary antibodies (1:5,000–1:10,000; GE Healthcare UK, Ltd.) and analyzed quantitatively using ImageJ software (NIH, Bethesda, MD). Quantitative data were presented as mean ± SD from triplicate experiments.
CE-derived RSCs were induced to differentiate as describe above. At 1, 3, and 5 weeks post-differentiation, cells with round somas and neurites were selected for electrophysiological recording. Spiking was recorded with a patch electrode (4–8 MΩ, filled with Ames medium) in the cell-attached mode (voltage-clamp). By convention, depolarizing responses in voltage clamp are depicted as downward deflections, and hyperpolarizing responses are upward. Thus, action potentials consist of a downward then upward deflection. Slides were perfused at 4 ml/minute. Electric stimulation was used to deliver 3 milliseconds cathodal (depolarizing) pulses to a 100 kΩ Pt-Ir stimulating electrode (Microprobes) that was centered 25 μm above the soma of targeted cells; such pulses are known to effectively initiate action potentials in a wide range of neurons. Identical anodic pulses were delivered 100 millliseconds after the cathodal pulse to prevent charge build up on the cells or electrode; the large temporal separation between the cathodal and anodal phases allowed the response to the cathodal phase to be studied in isolation. Pulse amplitudes ranged from 0 to 20 μA; five repeats were delivered at each amplitude level. Pulse stimuli were controlled by Multi-Channel Systems STG2004 hardware and software using custom software written in LabView (National Instruments) and Matlab (MathWorks) by Maesoon Im and Daniel Freeman. Two silver chloride-coated silver wires served as the return; each was positioned approximately 8 mm from the targeted cell and approximately 6 mm from the other. Because the residual artifact(s) limited the ability to precisely determine the onset timing of the action potential, we used the peak of the depolarizing phase to compare latencies of different spikes. Threshold was determined by fitting a sigmoidal curve to the raw data points and then determining the amplitude for which there was a 50% probability of eliciting a spike. Large electrical artifacts occurred at the onset, and offset of the stimulus were removed via digital filtering. Electrical stimulus artifacts arising from the cathodal pulses were removed by subtracting the inverted artifact arising from the corresponding anodic pulse. In most cases, this completely eliminated the artifact (Fig. 6D). In some cases, however, there were slight temporal differences in the kinetics between the cathodal and anodal artifacts resulting in some remaining artifact (Fig. 6E).
Increased Expression of Ephrin-A3 with Retinal Maturation
To address whether ephrin-As are negative regulators of the neurogenic potential of CE-derived cells, we examined the temporal and spatial patterns of ephrin-A3 expression in the CE and retina of the mouse eye. Quantitative analysis of Western blots revealed an upregulation of ephrin-A3 beginning at postnatal day 4 (P4) in the mouse retina, reaching a peak in the adult (Fig. 1A). This temporal pattern of ephrin-A3 expression was confirmed by immunolabeling of ephrin-A3 in retinal and CE sections (supporting information Figs. 1, 2), which further demonstrated that ephrin-A3 was highly expressed in the retina and CE of adult mice. The increase of ephrin-A3 expression coincides with the reduction of RSC proliferation and cessation of retinogenesis in the developing mouse retina, supporting that ephrin-A3 may send a prohibitory signal for the neurogenic program of RSCs.
Increased Proliferation of CE-Derived RSCs in the Absence of Ephrin-A3
The initial step toward the regenerative response by CE-derived cells in lower vertebrates is to re-enter the cell cycle. To investigate whether ephrin-A3 suppresses the potential of CE-derived cells to display stem cell properties and turn into RSCs, we first examined their ability to proliferate or re-enter the cell cycle in vivo in the absence of ephrin-A3, using ephrin-A3 deficient mice. Following a 7-day administration of BrdU, few proliferating or BrdU+ cells were detected in either the CE or retinas of wild-type mice (Fig. 1B, 1C). In contrast, in adult ephrin-A3−/− mice, BrdU+ cells were consistently detected in both the CE (Fig. 1D) and peripheral retina (Fig. 1E). There was a greater than 20-fold increase in the number of BrdU+ cells in the CE and peripheral retinal sections (Fig. 1B) of adult ephrin-A3−/− mice (n = 9) as compared to that of wild-type mice (n = 9). Among them, approximately 95% of BrdU+ cells were colocalized with retinal progenitor cell marker Chx10  (Fig. 1F–1I), suggesting their origin as proliferating RSCs. Most BrdU+ cells were nonpigmented, while a small percentage of them were seen with pigments; however, no BrdU+ cells were detected from the central retina and retinal pigment epithelium (RPE) of ephrin-A3−/− mice, implicating distinctive mechanisms that govern the proliferation of CE-derived cells and the remaining RPE.
To corroborate this finding, we compared the ability of CE-derived cells of wild-type and ephrin-A3−/− mice to form NS in vitro (Fig. 2A, 2B). Cells taken from the CE of wild-type and ephrin-A3−/− mice were dissociated and cultured at a single cell density for 14 days. After 7 days of incubation, we observed a significantly greater number of primary NS in cultures derived from ephrin-A3−/− mice than those from wild-type mice (Fig. 2). To examine secondary NS formation, primary NS were dissociated and seeded in a 24-well plate at a density of single cell colony. After 14 days of incubation, cultures of ephrin-A3−/− CE-derived cells displayed a twofold increase in number of secondary NS when compared with cultures of wild-type mice (Fig. 2F). To confirm that the increased number of NS in cultures of ephrin-A3−/− mice was a result of the enhanced proliferative potential of RSCs, we performed a BrdU incorporation assay (Fig. 2C, 2D). A 4-hour BrdU pulse stimulation yielded a significantly higher ratio of BrdU incorporation in dissociated cell cultures of NS derived from ephrin-A3−/− mice than those from wild-type mice (Fig. 2G), further indicating enhanced RSC proliferation in the absence of ephrin-A3. In another series of experiments, antibody-clustered ephrin-A3, which is known to activate receptor signaling , was added to dissociated NS cell cultures derived from wild-type mice. Administration of antibody-clustered ephrin-A3 significantly inhibited BrdU incorporation in culture as compared to untreated controls or cultures treated with control Fc (Fig. 2H). Together, these data demonstrate that ephrin-A3 negatively regulates RSC proliferation in vivo and in vitro.
Increased Neural and Photoreceptor Cell Potency in the Absence of Ephrin-A3
To further address the role of ephrin-A3 in the neurogenic potential of CE-derived RSCs, we compared the levels of expression of neural progenitor cell markers in CE-derived NS taken from wild-type and ephrin-A3−/− mice, using real-time RT-PCR. In the absence of ephrin-A3 signal, the expression of retinal progenitor cell marker Chx10  was significantly upregulated as compared with that of wild-type mice (Fig. 3A). There was also an overall increase in the expression of neural progenitor cell markers Ngn2, Sox2, Math5, and Math1, and photoreceptor precursor cell markers Crx, Nrl, and Nr2e3 , in CE-derived NS cells of ephrin-A3−/− mice than in wild-type mice (Fig. 3B–3D). These data were supported by the drastic upregulation of neurogenic signals, such as the Wnt family members (e.g., Wnt1, Wnt3a, and Wnt7a), in NS cells derived from ephrin-A3−/− mice as compared to those from wild-type mice (Fig. 3C, 3D). In contrast, expression of the pigment epithelial progenitor cell marker Mift  was significantly downregulated in NS taken from ephrin-A3−/− mice, implicating a decreased pigment epithelial cell potency when the cells exhibited enhanced retinal progenitor cell fate. The data suggest that expression of ephrin-A3 in CE-derived cells inversely correlates with their neural progenitor cell potency.
To investigate if CE-derived RSCs of ephrin-A3−/− mice display increased potency to generate retinal neurons and photoreceptors, NS cells of wild-type and ephrin-A3−/− mice were induced to differentiate in culture. Following 7 days of incubation, CE-derived NS cells gradually lost pigmentation and exhibited a morphology that differs from the CE (Fig. 4A, 4B). By 21 days, CE-derived NS cells taken from ephrin-A3−/− mice differentiated into various types of retinal cells. These cells expressed retina-specific cell markers, such as GFAP (glial cell marker, 73.21% ± 10.22%), recoverin (rod photoreceptor cell marker, 24.66% ± 4.32%), β-III-tubulin (Tuj; neuronal marker, 1.51% ± 0.28%), and MAP2 (mature neuron marker, 1.01% ± 0.21%), and exhibited corresponding neuronal and glial cell morphologies (Fig. 4C–4I). In contrast, very few NS cells of wild-type mice were found to express the retinal neuron markers mentioned above after 21 days of incubation, suggesting their limited ability to differentiate into retina-specific cells. Quantitative analysis with real-time RT-PCR confirmed that following incubation RSCs of both wild-type and ephrin-A3−/− mice downregulated retinal progenitor cell markers Pax6 and Chx10 (Fig. 5A). However, after 10 days of incubation, cultures derived from ephrin-A3−/− mice exhibited significantly higher levels of expression of retinal neuron and glial markers, such as rhodopsin, recoverin, protein kinase C alpha (PKCα), and GFAP, indicating increased photoreceptor and retinal cell potency than in cultures from wild-type mice (Fig. 5B).
To determine if ephrin-A3 promotes an RPE cell fate while suppressing the neuronal fate, NS cells of wild-type mice were cultured in the presence of control Fc or antibody clustered form of ephrin-A3. Following 7-day induction of differentiation, CE-derived NS cells treated with antibody clustered form of ephrin-A3 exhibited significantly higher number of pigmented cells as compared to control Fc treated cultures. The percentage of pigmented cells doubled in ephrin-A3 treated group as compared to the control group (Fig. 5C), suggesting that ephrin-A3 promotes the RPE progeny of CE-derived RSCs while suppresses RSC proliferation.
To verify the functional capacity of neurons differentiated from CE-derived RSCs of ephrin-A3−/− mice, next we measured the response of postdifferentiated neurons to extracellular electric stimulation, which method effectively elicits action potentials in a wide range of neurons [25, 26]. At 1, 3, and 5 weeks postdifferentiation, RSCs with round cell bodies and axon-like neurites were recorded, and a patch-clamp electrode in cell-attached mode was established on the soma in order to record the cell's spiking response to stimulation. A small stimulating electrode was positioned approximately 25 μm above the soma of the targeted cells. Cathodal pulses, with duration of 3 millisecond and amplitudes ranging from 0 to 20 μA were delivered at a rate of 1 Hz. Responses recorded from a typical cell at 3 week postdifferentiation are shown in Figure 5D: the positive responses consisted of biphasic waveforms that are characteristic of action potential propagation; they exhibited an excitatory (downward) current with approximately 0.5 millisecond duration followed immediately by an inhibitory (upward) current that persisted for approximately 1 millisecond. Similar responses were found in 13/13 cells recorded, suggesting their maturation into functional neurons. As expected, the likelihood of eliciting a spike increased with the amplitude of stimulation (Fig. 5E, 5F). Response threshold (defined as the amplitude for which 50% of pulses elicited action potentials) was measured in 11 cells (2/13 cells displayed only a single amplitude). Among these 11 cells, we noted two types of cell populations: type A with a mean value of 2.99 μA (n = 5/11) and type B with a mean value of 9.07 μA (n = 6/11; p = .00597). Both types of responses had been detected in mature retinal neurons, and the lower thresholds in type A neurons suggest a higher concentration of sodium channels. Consistent with previous studies of spiking neurons [25, 26], the latency of spike onset decreased with increasing stimulus amplitude (Fig. 5E, 5G), further supporting the voltage-dependent ion channel activities in these neurons. Like early differentiated neurons, all of the 13 RSCs examined at 3 weeks postdifferentiation also exhibited rhythmic spontaneous discharges (Fig. 5D, bottom row) at an average frequency of approximately 0.3 Hz, with inward currents that were smaller and slightly shorter in duration than the depolarizing phase of action potentials. Spontaneous currents disappeared in RSCs that were differentiated for 5 weeks (n = 5), similar to that is seen in mature retinal neurons. Whereas, spontaneous inward currents could be detected in all RSCs at 1 week postdifferentiation, when no action potentials could be elicited from these younger cells (n = 16). Thus, neuron-like cells differentiated from CE-derived RSCs of ephrin-A3−/− mice showed major electrophysiological characteristics of typical retinal neurons in terms of spontaneous spiking activity [27, 28], temporal parameters of mature action potentials and voltage-dependent currents, which gradually developed in a time-dependent manner. In contrast, no action potential-related transient current were ever detected from astroglia-like cells, which exhibited large flat somas with no visible elongated processes (not shown). Together, these data indicate that ephrin-A3 is a negative regulator of the neurogenic potential for CE-derived RSCs which shift the cells toward a pigment epithelial cell fate.
Ephrin-A3 Functioning Through EphA4 and the Suppression of Wnt Signaling to Regulate the Neurogenic Potential of CE-derived RSCs
Our previous research showed that ephrin-A3 acts, at least in part, through forward signaling to cause neural stem cell growth inhibition . EphA4 and EphA7 were reported to be the primary receptors of ephrin-A3; therefore, in this study we examined the expression of EphA4 and EphA7 in CE-derived NS cells, using immunohistochemistry and RT-PCR. Immunodetection revealed EphA4 (Fig. 6A, 6B), but not EphA7 (data not shown), in the CE. Results of RT-PCR further confirmed that EphA4 is prominently expressed by CE-derived RSCs (Fig. 6C). Immunoprecipitation of cell lysates taken from the CE with the ephrin-A3 antibody also pulled down EphA4, instead of EphA7, supporting the notion that EphA4 is the predominant ephrin-A3 receptor in CE-derived cells (Fig. 6B). In agreement with those findings, Western blot analysis detected EphA4 activation and phosphorylation in cell lysates taken from the ciliary bodies of wild-type, but not ephrin-A3−/− mice (Fig. 6E). To further demonstrate that ephrin-A3 functions through the EphA4 receptor, we performed a BrdU incorporation assay using CE-derived NS cell cultures (Fig. 6F). Following a 4-hour BrdU pulse stimulation, 36% of NS cells derived from the CE of wild-type mice became BrdU+. Addition of soluble EphA4 protein, which competes and inhibits the activity of intrinsic EphA4 receptors, but not its control Fc, significantly increased BrdU incorporation into NS cells in culture, supporting a role for EphA4 as the functional receptor of ephrin-A3 to mediate CE-derived NS cell proliferation (Fig. 6F).
Several lines of evidence suggest that Wnt3a/β-catenin signaling plays pivotal roles in the neurogenic potential of neural stem cells [29–32]. The fact that CE-derived RSCs of ephrin-A3−/− mice had increased levels of Wnts (Fig. 3D) prompted us to ask whether the Wnt pathway is critically involved as a downstream target of ephrin-A3 to mediate the proliferative potential of RSCs. To test this, we examined expression of the canonical downstream pathway of Wnt3a signaling, β-catenin. We noted an average fourfold increase in β-catenin level in primary CE-derived NS cells of ephrin-A3−/− mice as compared to that of wild-type mice (Fig. 7A, B), strongly suggesting increased activation of the canonical Wnt pathway. In the BrdU incorporation assay, addition of Wnt3a into the culture resulted in greater than twofold increase in BrdU incorporation into CE-derived NS cells of wild-type mice as compared to the untreated control group (Fig. 7C–7E). Together, the results suggest that ephrin-A3 functions through EphA4 to present a negative niche for the proliferation and neurogenic potential of CE-derived RSCs by suppressing Wnt/β-catenin signaling.
This study identifies ephrin-A3 as a key inhibitor of RSC potency and neuronal fate of CE-derived RSCs in the adult mouse. Absence of ephrin-A3 enhances the proliferation of CE-derived RSCs in vivo. However, the number of BrdU+ cells migrating into the retina of ephrin-A3−/− mice remained low, and even fewer was found to differentiate and express a retinal neuron marker (e.g., rhodopsin or β-III-tubulin), suggesting an extremely low neurogenic potential in these mice in vivo. In part, this may be caused by the presence of other negative niche signals, such as ephrin-A2 , in the microenvironments of the adult CE and retina. In the future, it would be interesting to determine if absence of both ephrin-A2 and -A3 further enhances the neurogenic potential of CE-derived RSCs in vivo.
CE-derived cells are of neuroepithelial origin and thought to possess both epithelial and neural progenitor cell properties. It has been previously reported that these cells express neuronal progenitor markers Chx10 and nestin and are intrinsically capable of generating retina-specific neurons ex vivo. However, it was recently reported that CE-derived cells exhibit primarily pigmented epithelial cell properties, with little neurogenic potential  and limited ability to differentiate into photoreceptors . On the basis of our data, we propose a working model for how the neurogenic potential of CE-derived RSCs is regulated. We suggest that CE-derived cells maintain a balance between epithelial and neurogenic potential in the adult. This balance is controlled by two intrinsically linked but opposing signaling events—the inhibitory and stimulatory neurogenic signals the cell receives. Ephrin-A3/EphA4 pathways comprise the inhibitory signals that suppress retinal progenitor maintenance, while the Wnt pathways promote the proliferation and neuronal fate of CE-derived cells. Interactions between ephrin and Wnt pathways can also be found in several other progenitor/stem cell systems; these include small intestine stem cells [33–35] and the developing xenopus optic tectum . The list of negative and positive regulators is likely to grow as the mechanisms underlying the actions and interactions of these pathways are unveiled. Crosstalk between these pathways imposes both positive and negative feedback regulation, enabling better response to environmental cues and providing a level of plasticity in regulation in the adult mouse.
While the CE is composed of two layers of cuboidal cells—pigmented and nonpigmented cells, ephrin-A3 and EphA4 were detected in almost all cells of both layers. It has been shown that both ephrins and Eph family receptor tyrosine kinases are expressed by neural progenitors [18, 20, 37]. While ephrins can mediate a bidirectional signaling , our experiments using soluble EphA4 protein, which competes and inhibits the activity of intrinsic EphA4 receptors, strongly suggest that ephrin-A3 evokes an EphA4 forward signal to suppress the proliferation of CE-derived RSCs. We propose that the niche cells in the CE constantly interact with RSCs in vivo by making direct contact to inhibit their neurogenic potential.
Ephrin family proteins act in combination with an Eph receptor and are pivotal regulators of CNS development, including axon guidance, neural migration and patterning, synapse formation, and vascular morphogenesis [39, 40]. Recent studies also point to a role for ephrin families functioning as inhibitors of neurogenesis in the subgranular zone and other brain areas of adult mice . Results in this study support the premise that ephrin-As are also involved in the regulation of the neurogenic potential of RSCs. Expression of ephrin-A3 increases as the retina matures, along with the cessation of RSC proliferation and retinogenesis . In adult retina and CE, expression of ephrin-A3 is high, and its absence results in increased proliferation of RSCs in the CE and peripheral retina—a margin reminiscent of the CMZ of fish and amphibians [3, 41–43]. These data suggest that ephrin-A3 is a component of an inhibitory niche for adult RSCs in mice. In agreement with previous reports , we found that CE-derived RSCs of ephrin-A3−/− mice are initially pigmented but then undergo a course of depigmentation. Moreover, expression of pigmented epithelial progenitor marker Mitf was downregulated before RSCs transformed into retinal neural progenitors in culture. These progenitor cells subsequently downregulated RSC markers Chx10 and Pax6 during differentiation. Also consistent with previous reports [2, 9–11], we show that the potential of CE-derived cells of wild-type mice to differentiate into photoreceptors or other retina-specific cells is negligible. However, RSCs derived from ephrin-A3−/− mice consistently exhibited drastic upregulation of the expression of photoreceptor progenitor cell markers Crx, Nr2e3, and Nrl, as compared to that of wild-type mice, and enhanced potency to differentiate into retinal neurons and photoreceptor cells. These data suggest that adult CE-derived cells possess the machinery or properties of retinal progenitor cells; however, their retinogenic potential is negatively regulated by ephrin-A3 signaling under normal adult conditions. Thus, manipulating the ephrin-A3 signaling pathway may present a viable approach to redirecting CE-derived cells toward a retinal progenitor cell fate.
This work also contributes to resolving an issue regarding the function of and signaling events associated with ephrin-A3 in adult stem cell regulation. While EphA7 and EphA4 are reported to serve as the receptors of ephrin-A3 [18, 19, 21, 44, 45], our data suggest that ephrin-A3 primarily binds EphA4 in the CE, to suppress the activity and expression of Wnt3a/β-catenin pathways and, in turn, mediate the neurogenic potential of adult RSCs. At this point, we cannot rule out the possibility that ephrin-A3 and EphA4 also act bidirectionally to regulate the behavior of adult RSCs. Interconnection between the ephrin and Wnt pathways has been previously reported, where ephrin-B1 was shown to inhibit the Wnt3a/β-catenin pathway through binding to EphB receptors . Here we report that ephrin-A3 activates EphA4 to suppress Wnt3a/β-catenin signaling, and that addition of Wnt3a alone is sufficient to promote cell proliferation. RSCs derived from ephrin-A3−/− mice showed significantly increased levels of expression of Wnt family members, including Wnt1, Wnt3a, and Wnt7 (Fig. 4D), and their downstream signal β-catenin. While ideally, activation of Wnt/β-catenin signal should be examined by comparing the free β-catenin levels in relation to total β-catenin levels in the cells, this experiment is technically difficult as each mouse contains only limited number of CE-derived RSCs. Alternatively, we were able to show that the total β-catenin levels in RSCs of ephrin-A3−/− mice were drastically upregulated as compared to wild-type mice, suggesting increased β-catenin stability and activation of Wnt signaling. These observations are in agreement with a previous report demonstrating increased number and size of primary and secondary NS in CE-derived cell cultures stimulated by Wnt3a . Together our data support a negative role for ephrin-A3 in the Wnt pathways. Among the neurogenic signals examined, we also noted drastic upregulation of Ngn1, Sox2, Math1, and Math5 in CE-derived RSCs of ephrin-A3−/− mice as compared to wild-type mice, suggesting involvement and complex interactions among multiple signaling pathways. While further experiments are needed to unravel the molecular details of these interactions, together, our results suggest that the interactions and balance between ephrin-A3 and Wnt/β-catenin pathways mediate retinal progenitor cell potency in adult CE.
Our study demonstrates ephrin-A3 as a key component of the adult RSC niche, which acts through EphA4 to suppress Wnt signaling and inhibit the neurogenic potential of CE-derived RSCs. Deletion of ephrin-A3 increases the proliferation and expression of neural retinal progenitor cell genes, as well as photoreceptor potency for CE-derived RSCs while presence of ephrin-A3 suppresses the neurogenic potential and shift CE-derived cells toward a pigment epithelial cell fate. Despite the fact that the CE is easily obtained in humans by a minimally invasive procedure, to date, the clinical promise of CE-derived RSCs serving as an endogenous source of retinal neural progenitors for the treatment of photoreceptor damage or degenerative disorders is yet to be realized. However, this potential avenue for treatment can only be fully explored with a comprehensive understanding of the mechanisms and molecular signals that control the cellular properties and differentiation of CE-derived cells.
The project was supported by grants from NIH/NEI (R01EY017641), NIDA (R21DA024803), Department of Veterans Affairs (1I01RX000110), Department of Defense (W81XWH-09-2-0091) to D.F.C., Lion's Foundation Grant to K.-S.C., the National Basic Research Program of China (No. 2007CB512204) to X.S., and National Natural Science Foundation (NSFC81100667) to Y.F.