Molecular Vision 2008; 14:2309-2320
Received 21 June 2008 | Accepted 25 November 2008 | Published 12 December 2008
Weiming Mao, Run-Tao Yan, Shu-Zhen Wang
Department of Ophthalmology, University of Alabama at Birmingham, Birmingham, AL
Correspondence to: Shu-Zhen Wang, DB104, Callahan Eye Foundation Hospital, 700 South 18th Street, Birmingham, AL, 35294-0009; Phone: (205) 325-8628; FAX: (205) 325-8679; email: firstname.lastname@example.org
Purpose: Harnessing a cell culture of retinal pigment epithelium (RPE) to give rise to retinal neurons may offer a source of developing neurons for cell-replacement studies. This study explores the possibility of reprogramming RPE progeny cells to differentiate toward retinal neurons with achaete-scute homolog 1 (ash1), a proneural gene that is expressed in progenitor cells in the developing retina and promotes amacrine cell production when overexpressed in the chick retina.
Methods: Replication Competent Avian Splice (RCAS) retrovirus was used to drive the ectopic expression of ash1 in cell cultures of dissociated RPE isolated from day 6 chick embryos. RCAS expressing green fluorescent protein (RCAS-GFP) was used as control. The cultures were examined for de novo generation of neuron-like cells by molecular, cellular, and physiologic criteria.
Results: In control cultures infected with RCAS-GFP, RPE cells appeared cobblestone-like and often darkly pigmented. In cultures infected with RCAS-ash1, however, cells remained de-pigmented and frequently formed clusters. Further examination at the morphological and molecular levels showed the development of elaborate processes characteristic of neurons and the expression of genes/markers that identify different types of retinal neurons. The most prevalently expressed neural marker was calretinin, which in the chick retina identifies amacrine, ganglion, and horizontal cells. As an assay for functional maturation, the reprogrammed cells were analyzed for the presence of functional, ionotropic glutamate receptors that lead to a rise in the cytosolic free calcium (Ca2+) concentration. Calcium imaging showed that reprogrammed cells responded to glutamate and N-methyl-D-aspartate (NMDA) by increasing their Ca2+ concentrations, which, after reaching a peak level, returned to the basal level. The response curves of reprogrammed cells resembled those of cultured retinal neurons.
Conclusions: These results suggest that RPE progeny cells can be reprogrammed by ash1 to develop molecular, morphological, and physiologic properties that are characteristic of retinal neurons.
The vertebrate retina contains five major types of neurons: photoreceptor, horizontal, bipolar, amacrine, and ganglion. Visual signals are initiated by photoreceptors, modulated by horizontal, bipolar, and amacrine cells, and transmitted to the brain through the axons of ganglion cells. Retinal neurons, like other neurons in the central nervous system, are terminally differentiated and do not reenter the cell cycle for proliferation. Thus, degenerated retinal neurons are not automatically replaced. In the teleost retina, both progenitor/stem cells at the ciliary marginal zone and Müller glia in the retina can give rise to new retinal cells in response to injury [1-5]. Chick Müller glia have also been shown to exhibit some properties of retinal stem cells . Nonetheless, such an inborn stem-cell-based regeneration mechanism seems elusive in the retina of higher vertebrates. As a result, attention in recent years has been directed toward inducing retinal neurogenesis through programming or reprogramming the differentiation of cells that can be propagated in large numbers in vitro, including embryonic stem cells and adult stem cells of the brain, retina, or bone marrow [7-11].
Unlike retinal neurons, retinal pigment epithelial (RPE) cells from many species, including human, can reenter the cell cycle. More importantly, their progeny are able to differentiate into cell types other than RPE , suggesting the possibility of RPE serving as retinal stem cells . The idea of using RPE cell cultures as a novel source of developing retinal neurons has been experimentally explored. Most of the published reports examined genes/factors that are involved in the production of photoreceptors [14-17] or ganglion cells [18,19].
A large number of studies have shown that ash1, a vertebrate homolog of the Drosophila proneural gene achaete-scute, exhibits proneural activity during the development of both the central and peripheral nervous systems [20-28], where ash1 promotes neurogenesis against gliogenesis [29-32]. Ash1 encodes a transcription factor of the basic helix–loop–helix family. In the retina, expression of ash1 is restricted to progenitor cells [29,33]. In the mouse, ash1 is required in the production of late-born neurons, including rod photoreceptors and bipolar cells . In ash1;ath3 double knockouts, the production of bipolar cells is virtually abolished . However, misexpression of ash1 or ath3 alone does not promote bipolar cell genesis, but co-misexpression of ash1, ath3, and chx10 increases bipolar cell number . Transgenic expression of ash1 in mouse RPE initiates retinal neurogenesis in the RPE cell layer . In the chick, the temporal and spatial pattern of ash1 expression coincides with amacrine cell genesis , and overexpression of ash1 expands the amacrine population .
Making use of RPE cells’ plasticity and the suggested roles of ash1 in the production of retinal progenitor cells and amacrine cells, we studied the possibility of coaxing cultured RPE cells with ash1 into differentiating along retinal neural pathways. Our experiments demonstrate neural differentiation at the molecular, morphological, and physiologic levels in the otherwise non-neural RPE cell cultures infected with a retrovirus expressing ash1.
Fertilized, pathogen-free White Leghorn chicken eggs were purchased from Spafas (Preston, CT) and incubated in a Petersime egg incubator (Gettysburg, OH). The care and use of animals adhered to the procedures and policies comparable to those published by the USA Public Health Service (Public Health Service Police on Humane Care and Use of Animals) and set by the Institutional Animal Use and Care Committee at the University of Alabama at Birmingham.
Replication Competent Avian Splice (RCAS) retrovirus  expressing ash1 (RCAS-ash1) was generated as described . During cloning process, a recombinant (rb) form of Ash1, Ash1ΔCrb, was generated. It lacked the C-terminal 33 amino acids of the wild-type Ash1 and contained (at the C-terminus) 19 extra residues derived from the shuttle vector Cla12Nco. RCAS-ash1ΔCrb was included in the study to determine whether the C-terminus of Ash1 was dispensable for its function. All recombinant RCAS constructs were sequenced for verification and for ruling out possible changes including frameshift or nonsense mutations. Sequencing was performed at the DNA Sequencing Service of the Genomics Core Facility at the University of Alabama at Birmingham. RCAS-green fluorescent protein (GFP) was produced as previously described .
E6 chick eyes were enucleated, and tissues outside the PRE were removed. A circular incision in the ciliary region was made to get rid of the anterior portion of the eye, producing eye cup consisting of the RPE, the retina, and the vitreous. The RPE was then peeled off from the neural retina. RPE tissues were dissected from day 6 chick embryos (E6) as previously described . Cells in RPE tissues were dissociated by trypsin/EDTA and gentle trituration. Fully dissociated RPE cells were resuspended in Gibco medium 199 (Invitrogen, Carlsbad, CA) plus 10% fetal bovine serum. The cells were then seeded at low density (approximately 1×105 cells) in a 35 mm dish and cultured at 37 °C under 5% CO2 with Knockout Dulbecco's Modified Eagle Medium (D-MEM) supplemented with 20% serum replacement (Invitrogen). After 3–4 days in culture, when 50% confluency was reached, 10–15 μl of retrovirus stock (2–5×108 pfu/ml) was added to each 35 mm dish. After an additional 8–12 days in culture, cells were directly subjected to Ca2+ imaging or fixed with ice-cold 4% paraformaldehyde in phosphate-buffered saline (PBS) for analyses with immunocytochemistry and in situ hybridization.
To facilitate visualizing the morphologies of reprogrammed cells, we transfected RPE cell cultures 6 days after the administration of RCAS-ash1ΔCrb. Transfection was done with recombinant DNA of adeno-associated virus expressing GFP (AAV-GFP) driven by a chimerical promoter of Cytomegalovirus (CMV)-actin-β globin. Cellular morphologies were imaged with a blue filter under a 20X or 40X objective attached to a Nikon TE300 (Melville, NY) inverted microscope 2–4 days after AAV-GFP DNA transfection.
The calcium indicator fluo-4 AM  (Molecular Probes, Eugene, OR) was also used to visualize the morphologies of reprogrammed cells. Cells in RPE cell cultures infected with RCAS-ash1 or RCAS were incubated with fluo-4 AM as described in detail late under “Calcium imaging” . Cellular morphologies were imaged under a 40X objective attached to a Nikon TE300 inverted microscope with a blue filter.
Retinas were dissected from day 16 chick embryos. Retinal cells were dissociated with 0.05% trypsin and 0.53 mM EDTA (Invitrogen) and mechanical trituration. Dissociated retinal cells were cultured in polyornithine-treated 35 mm dishes with medium 199 (Invitrogen) plus 10% fetal bovine serum at 37 °C under 5% CO2 for 5 days. Culture medium was changed every other day, and the cells were subjected to Ca2+ imaging in the same way as those in RPE cell cultures.
The following monoclonal antibodies were obtained from the Developmental Studies Hybridoma Bank (Iowa University, Iowa City, IA): anti-AP2 (3B5; 1:50; developed by Dr. T. Williams, Yale University, New Haven, CT); anti-islet-1 (1:100; developed by Dr. T. Jessell, Columbia University, New York, NY), anti-LIM (4F2; 1:50; developed by Dr. T. Jessell, Columbia University, New York, NY), anti-vimentin (H5; 1:500; developed by Dr. J. Sanes, Harvard University, Cambridge, MA), and anti-visinin (7G4; 1:500; developed by Dr. C. Cepko, Harvard University, Cambridge, MA). Antibodies obtained from commercial sources included antiviral protein p27 (1:500; Spafas, Preston, CT), polyclonal antibody against calretinin (1:500; Chemicon, Temecula, CA), polyclonal antibody against red opsin (1:200; Chemicon), polyclonal antibody against AP2α (1:200; Santa Cruz Biotechnology, Santa Cruz, CA), and monoclonal antibody against MAP2 (1:200; Sigma, St. Louis, MO). Monoclonal antibody RA4 (1:1,000 dilution) was a gift from Dr. Steven McLoon (University of Minnesota, Minneapolis, MN). Standard procedures for immunocytochemistry were followed.
The number of immunopositive cells was counted under a Nikon TE300 microscope (Melville, NY). In each experiment, 3 dishes were assigned for each marker. From each dish, the number of immunopositive cells was scored from 10 view areas under a 20X objective, and the total number of immunopositive cells per 35 mm dish was calculated by multiplying the average per view area by 200 (the total number of view areas per dish).
Briefly, cells in the culture were fixed with ice-cold 4% paraformaldehyde for 30 min before being permealized with 0.25% triton X-100 for 30 min. Hybridization was carried with 40–80 ng of RNA probes overnight at 48–55 °C. In situ hybridization was used to detect the expression of interphotoreceptor retinal binding protein (IRBP), GAD67 (GenBank 4103977), ath3, and chx10, as previously described [15,40].
Calcium imaging was used to study the physiologic responses of reprogrammed cells to neurotransmitters, as described by Lamba et al.  with minor modifications. Cells in the 35 mm culture dish (either of RPE cell culture or of retinal cell culture) were washed and incubated with medium 199 for 1 h at 37 °C. Then cells were washed with a magnesium-free bath solution that contained 5.3 mM KCl, 115.0 mM NaCl, 3.1 mM CaCl2, 5.6 mM glucose, and 3.0 mM HEPES, pH 7.2. The bath solution was adapted from that given in Sosa and Gleason . Next, cells were incubated with fluo-4 AM as described  in 1 ml of bath solution for 15 min at room temperature. After a brief rinsing and then incubation with the bath solution for 30 min at room temperature, the culture dish was placed under an Olympus BX60 microscope with constant perfusion with the bath solution at a rate of 1 ml per minute at room temperature. Fluorescence images were captured sequentially without time interval with 400 ms-exposure for each image. Captured images were analyzed with WCFI ImageJ software. Briefly, a region of interest covering the soma of a cell was selected in stacked images. A region of the same shape and area was also selected from the background for background subtraction. Changes in fluorescence intensity were expressed as percentage of ΔF/Fo, where F is the averaged integrated optic density (IOD) of an image, Fo is the average of IOD the images before glutamate application, and ΔF=F-Fo. Approximately 3–5 μl of 1 mM glutamate in the bath solution, or 1 mM N-methyl-D-aspartate (NMDA) and 1 mM glycine, was puffed onto the area of interest within 500 ms through a micropipette driven by a PV800 Pneumatic Picopump (World Precision Instruments, Sarasota, FL). The recording chamber was under constant perfusion (around 1 ml/minute) at room temperature. ΔF/Fo was plotted against time using Origin Pro7 (OriginLab, Northampton, MA).
Dissociated RPE cells (seeded at a density of approximately 1×105 cells in a 35 mm dish) lose their dark pigment during the initial days in culture, but they become repigmented as the culture approaches confluency . Infection with RCAS-GFP has little or no effect on the repigmentation of the culture (Figure 1A,B; ). Infection with RCAS-ash1, however, prevented the RPE culture from repigmentation, except at places where patches of pigmented cells were present (Figure 1A).
In addition, cultures infected with RCAS-ash1 also contained cell clusters noticeable under a microscope with phase contrast or Hoffman modulation optics (Figure 1C). The clusters of cells were irregularly shaped and contained cells with a round cell body and often with processes, reminiscent of neural clusters. No such cell clusters were present in the control culture, in which hexagonal cells were organized in a cobblestone fashion (Figure 1B).
To determine whether ash1 reprogrammed RPE progeny cells to differentiate toward retinal neurons, RPE cell cultures infected with RCAS-ash1 were examined for the expression of genes/markers that are normally expressed in various types of retinal neurons. These markers included calretinin, expressed in horizontal and ganglion cells and some amacrine cells [43,44]; general neural marker MAP2 present in all retinal neurons except bipolar cells ; ganglion cell markers RA4  and Islet-1 , which also labels bipolar cells at late developmental stages, amacrine markers AP2α  and GAD67 ; bipolar cell markers ath3 and chx10 ; horizontal cell marker LIM ; photoreceptor markers red opsin, IRBP [52,53], and visinin , whose immunoreactivity is also detected weakly in developing amacrine cells in the chick retina [55,56].
Overall, calretinin was the most expressed among the markers tested. While infection by RCAS-GFP did not induce calretinin expression (Figure 2A), numerous calretinin+ cells were detected in cultures infected with RCAS-ash1 (Figure 2B,C). Because calretinin+ cells were frequently localized in clusters (Figure 2B,C, arrows), calculating their exact number was impractical; nonetheless, we estimate that as many as 20% of the cells present in the culture were calretinin+ (>30,000 in a 35 mm dish; Table 1). Long, thin processes typical of neurons emanated from and webbed among the calretinin+ cells. Neuron-like processes were also observed on MAP2+ cells from RPE cultures infected with RCAS-ash1ΔCrb (Figure 2D,E).
Immunocytochemistry and in situ hybridization of RPE cell cultures infected with RCAS-ash1 showed the following: a substantial numbers of cells immunopositive for RA4 (>4,000 in a 35 mm dish; Table 1) or visinin (>11,000; Figure 2G); a small number of cells expressing IRBP (>1,000); a few expressing GAD67 (>100) or red opsin (>90); and none positive for AP2α, ath3, or chx10 (Table 1). None of the genes/markers were expressed in the control infected with RCAS-GFP (Figure 2A) [14,16], with the exception of RA4, which typically is detected in hundreds of cells in the control . We also examined the expression of neural markers in RPE cell cultures infected with RCAS-ash1ΔCrb, which expressed a truncated Ash1 lacking the C-terminal sequence and with a short sequence acquired from the cloning vector (see Methods). Chick embryos infected with RCAS-ash1ΔCrb exhibited all, but more pronounced, phenotypes observed with RCAS-ash1 . In RPE cell culture, infection with RCAS-ash1ΔCrb produced similar results as infection with RCAS-ash1, providing further support to the C-terminal sequence being nonessential to Ash1’s function .
The prevalence of calretinin+ cells along with ash1 promoting amacrine overexpression in the retina  raised a question of whether some of the visinin+ cells were amacrine-like, as differentiating amacrine cells in the retina transiently express of photoreceptor markers [55,56]. Double-labeling showed the presence of visinin+/calretinin+ cells (Figure 2F-H, arrows). Double-labeled cells accounted for an estimated 75% of visinin+ cells (in 5 view areas under a 40X objective) and 5%–10% of calretinin+ cells (in the same areas).
The calcium indicator fluo-4 AM selectively identifies reprogrammed, neuron-like cells in an RPE cell culture transduced with proneural gene neuroD . Thus, we used fluo-4 AM coupled with fluorescence photomicrography to directly visualize the cellular morphologies of reprogrammed cells in RPE cultures infected with RCAS-ash1. Cells in the control cultures infected with RCAS were “invisible” under our experimental conditions (Figure 3A,B), as their intracellular Ca2+ concentrations were below the level of detection with a 2 s exposure in image-capture. Under the same experimental conditions, reprogrammed cultures (i.e., infected with RCAS-ash1) contained a large number of fluoresced cells (~15%), with the rest of the cells (presumably not being reprogrammed) remaining “invisible.” Like individual, calretinin+ cells with adornment by elaborate processes (Figure 3C), most of the fluoresced cells emanated multiple processes from their soma (Figure 3D), a morphological characteristic of horizontal, amacrine, or ganglion cells of the retina.
The morphology of reprogrammed cells was also examined with AAV-GFP DNA transduction. As shown in Figure 3E, reprogrammed cells in an RPE culture infected with RCAS-ash1ΔCrb displayed cellular processes remarkably complex and closely resembling those possessed by retinal horizontal, amacrine, or ganglion cells.
Because Műller glia exhibit spindle-like processes that may appear similar to neural process and its Ca2+ concentration may be sufficiently high to be visible with fluo-4 AM, the RPE cell cultures were examined for the presence of Müller glia-like cells by immunostaining for vimentin, which is detected in Müller glial processes . Because cultured RPE cells also express vimentin and other commonly used Műller glial markers [58,59], vimentin+ cells with spindle-like processes were the observation target. In an RPE culture infected with RCAS-GFP, antivimentin staining was weak and observed in essentially all cells with morphologies of culture RPE cells (Figure 4A,B). Similar to the control, cultures infected with RCAS-ash1 contained vimentin+ cells that exhibited morphologies typical of culture RPE cells, but lacked vimentin+ cells that displayed spindle-like processes (Figure 4C,D). No particular antivimentin immunostaining was observed at places occupied by cells with compacted cell bodies (Figure 4C,D, arrows). Double-labeling for vimentin and calretinin showed the absence of Műller glia-like, vimentin+ cells (Figure 4E,F,H) in a culture containing reprogrammed cells that expressed calretinin (Figure 4G,I). The absence of Müller glia-like, vimentin+ cells is consistent with published reports that ash1 favors neurogenesis against gliogenesis [29-32].
The molecular and morphological resemblance of the reprogrammed cells to retinal neurons prompted a further analysis of whether the reprogrammed cells developed physiologic properties typical of retinal neurons. The cells were assayed for functional glutamate receptors. Glutamate is a common neurotransmitter in the retina, and its receptors are either ionotropic or metabotropic [60-62]. Ionotropic receptors include NMDA and non-NMDA receptors, varying in their selective agonists. NMDA receptors are permeable to Ca2+, and their activation results in an increase in intracellular Ca2+ concentration . By comparison, most of the metabotropic receptors cause a decrease in intracellular Ca2+ concentration when they are activated .
The development of functional glutamate receptors was assayed by examining Ca2+ concentration before and after the administration of glutamate or NDMA along with glycine, which acts as a co-agonist of the NMDA receptor and potentiates the NMDA-triggered response [11,63]. Ca2+ concentration was monitored using fluo-4 AM coupled with fluorescence microscopy. We found that reprogrammed cells, particularly those with neural morphologies, responded to the administration of NMDA and glycine by increasing their Ca2+ concentrations (Figure 5A). Increases in fluorescence intensity were detected a few seconds after the application of the neurotransmitters, indicative of a Ca2+ influx through calcium channels upon receptor activation. Fluorescence intensity peaked usually within15 s, quickly declined, and eventually reached pre-activation level (Figure 5). The response was similar to that observed with retinal cells in culture. When the changes in fluorescence intensity (as ΔF/Fo) were plotted over time, the responses to the neurotransmitters by reprogrammed cells (Figure 5B,C) resembling the responses by retinal neurons (Figure 5D) was evident. Notably, the degree of response by different cells varied (Figure 5B,C), likely due to varied extents of neural differentiation and varied physiologic conditions. Additionally, approximately 20% of the 86 cells subjected to the analysis showed responses. Cells in the control culture (infected with RCAS) showed no response (Figure 5E).
Reprogrammed cells also responded to 1 mM exogenous glutamate (Figure 6A,B). The response curves of the reprogrammed cells were similar to those of retinal cells. Like responses to NDMA+glycine, the extent of responses to glutamate varied among the cells. Varied responses were also observed with retinal neurons (Figure 6C,D). No similar responses were detected with cells in the control culture infected with RCAS (Figure 6E,F).
In this study, we tested ash1 for its ability to induce de novo genesis of retinal neuron-like cells from RPE cell cultures. The first sign of ash1-induced changes in RPE progeny cells was the altered appearance of the culture, in which the presumptive RPE cells were unable to regain pigmentation, and clusters of neuron-like cells emerged in the otherwise cobblestone-like arrangement. Ash1-induced reprogramming was evidenced by the development of elaborate cellular processes. Fluorescence microscopy after labeling the cells with fluo-4 AM or with AAV-GFP DNA transfection illustrated that the reprogrammed cells formed multiple processes, which are typical of horizontal, amacrine, or ganglion cells, but not photoreceptors or bipolar cells.
Immunostaining and in situ hybridization analyses showed the expression of genes/markers that identify different types of retinal neurons. The most prevalently expressed marker was calretinin, suggesting that a major product of ash1-induced reprogramming might be cells that resembled amacrine, horizontal, as well as ganglion cells. Ganglion-like cells, nonetheless, were unlikely to be the major product, because of the relatively moderate number of RA4+ cells. Photoreceptor-like cells were unlikely to be the major product either, because most of the reprogrammed cells exhibited multiple processes, and a relatively moderate number of cells were IRBP+ and a very small number of cells were red opsin+. There were a large number of visinin+ cells in the ash1-reprogrammed RPE cell culture. However, the majority of them coexpressed calretinin, which indicates that, like in the retina where developing amacrine cells express photoreceptor-specific genes visinin [55,56,] and RxRγ , those amacrine-like cells also transiently expressed visinin. The absence of expression of bipolar marker ath3 and chx10 does not necessarily preclude the presence of bipolar-like cells in the reprogrammed cultures, as not all of the cell type-specific markers observed with retinal neurons are expressed in reprogrammed cells under experimental conditions [19,65]. Additionally, even though there were tens of thousands of calretinin+ cells in ash1-induced reprogramming, no AP2α+ cells were observed. The partial gene expression may reflect partial cell differentiation due to microenvironment or to a lack of other factors necessary for advancing neuronal differentiation to maturation. Nonetheless, the morphologies of the reprogrammed cells were less indicative of differentiation toward bipolar-like cells than toward other secondary neurons. Thus, the results from both molecular analysis and morphological examination favor the scenario of ash1 reprogramming RPE progeny cells to differentiate toward amacrine-like cells as a major product, along with photoreceptor-like, ganglion-like, and possibly horizontal-like cells as accompanying products. This scenario is consistent with the observation that in the chick retina ash1 increases the amacrine cell population (data not shown).
Reprogrammed cells, especially those with multiple processes, responded to exogenous glutamate and the administration of NMDA+glycine by transiently increasing their intracellular Ca2+ concentrations. Their fluorescence intensity increased in the cell body and in processes seconds after administration of the neurotransmitters, suggesting a fast calcium influx through calcium channels upon activation of their receptors. After reaching its peak, the fluorescence intensity decreased quickly and eventually returned to the basal level, a profile closely resembling the response profiles of retinal neurons. This suggests the presence of functional, ionotropic glutamate receptors in the reprogrammed cells. Together, the evidence of neural differentiation at the molecular, morphological, and physiologic levels indicates that ash1 can reprogram RPE progeny cells to differentiate as cells that resemble certain neurons of the retina.
We thank Dr. Stephen Hughes for retroviral vector RCAS (B/P) and shuttle vector Cla12Nco, and Dr. Steven McLoon for RA4 monoclonal antibody. This work was supported in part by NIH/NEI grants R01 EY11640 and P30 EY03039, EyeSight Foundation FY2005–06–21, and a research grant from the International Retinal Research Foundation, and an unrestricted grant to the Department of Ophthalmology, University of Alabama at Birmingham from Research to Prevent Blindness.