Generation of Light-Sensitive Photoreceptor Phenotypes by Genetic Modification of Human Adult Ocular Stem Cells with Crx

Abstract

Purpose.: This study compared the effect of the transcription factor Crx (cone, rod homeobox) on the differentiation of human adult corneal (hCSC) and retinal (hRSC) stem cells into functional photoreceptors.

Methods.: Stem cells isolated from postmortem human corneas and retinal ciliary bodies were maintained in serum-free culture and genetically modified by electroporation to express exogenous epitope-tagged murine Crx. Expression of stem cell markers (Pax6, Oct3/4, and proliferating cell nuclear antigen [PCNA]), neuronal markers (nestin, neuron-specific class III β-tubulin, Map2 a/b, and neurofilament), and photoreceptor-specific markers (rhodopsin, cyclic nucleotide-gated cation channel-3, blue-cone opsin, and β-6-PDE) was evaluated by immunocytochemistry. A cGMP enzyme-linked immunoassay was used to assess phototransduction cascade activity by measurement of light-induced hydrolysis of cGMP.

Results.: Expression of the stem cell markers of proliferation and pluripotency Pax6, PCNA, and Oct3/4 was decreased by exogenous Crx expression in both hCSCs and hRSCs. Correspondingly, the expression of the mature neuronal markers Map2 a/b and neurofilament was increased. Both hCSCs and hRSCs displayed photoreceptor-specific immunolabeling. However, light-activated GMP hydrolysis was observed only in hRSCs after exogenous expression of Crx.

Conclusions.: The present study extends previous findings that exogenous Crx expression can promote differentiation of human retina-derived stem cells into light-sensitive photoreceptor phenotypes. Although Crx can induce human cornea-derived stem cells to express photoreceptor-specific proteins, it does not seem to be sufficient to direct their differentiation into functional photoreceptors. Nevertheless, this study demonstrates that genetic modification of adult human retinal stem cells can cause differentiation into light-sensitive photoreceptor phenotypes.

The persistence of stem cell pools in the eyes of adult vertebrates has opened the possibility of new therapeutic avenues to repair ocular tissues and restore visual function. Retinal stem cells (RSCs), with the ability to produce new neurons throughout life, are located at the periphery of the retina in the ciliary marginal zone of fish and amphibians (for review, see Refs. 1, 2), and mammals (including humans). ^3–6 These cells can be expanded in vitro, can differentiate into different retinal neurons, and are capable of integration after transplantation (for review, see Refs. 7, 8). Therefore, adult stem cell–based therapies can now be envisaged to restore retinal function in a wide range of retinal injury and in degenerative diseases. Accessible adult stem cells from the ciliary body and the corneal limbus can be used to restore visual function (for review, see Refs. 8, 9). Retinal cells and graft transplantations, ^10–14 as well as immature primary photoreceptors, ¹⁵ have been shown to improve visual function during retinal degeneration. However, ocular stem cells transplanted into damaged or degenerating retinas have shown, in general, poor differentiation into a retinal phenotype. ^16–18 Differentiation of these cells into a compatible phenotype is needed to improve photoreceptor commitment and differentiation after transplantation. To this end, we have reported in a prior study the development of protocols that shift phenotypic expression of adult mouse retinal stem cells toward the production of functional photoreceptor phenotypes. ¹⁹ Given the potential for such cells to be used therapeutically for retinal degeneration, the donor cells should be human derived to be clinically suitable. In the present study, we extended our investigation to generate light-sensitive photoreceptor phenotypes by genetically modifying adult ocular stem cells isolated from human postmortem ciliary body and corneal limbal epithelium. 

Materials and Methods

Cell Culture

Control eyes and corneas from donors with no history of ocular disease were obtained from the U.K. Transplant Support Service, Bristol, in accordance with the principles of the Declaration of Helsinki. In all cases, written informed consent was obtained before enucleation. The postmortem interval between death and the isolation of stem cells from the tissue was between 24 and 72 hours. Stem cells were isolated as previously described from the ciliary margin ^20,21 and the corneal limbal region. ²² Briefly, the corneal limbal region was carefully dissected from the peripheral cornea, the lens and iris were removed, and the ciliary margin was dissected. After dissection, the tissues were enzymatically digested at 37°C (trypsin, DNase, collagenase) and mechanically disrupted (trituration). Cells were cultured in DMEM/F12, 1× N2 supplement (Invitrogen-Gibco, Grand Island, NY), and 2 mM l-glutamine, supplemented with 20 ng/mL EGF (Invitrogen, Carlsbad, CA). 

When necessary, the differentiation of the human retinal stem cells (hRSCs) and human corneal limbal epithelium stem cells (hCSCs) was induced by supplementing the culture medium with 1% fetal calf serum for 8 to 10 days, as described elsewhere. ²³  

Electroporation

For each electroporation, 3 to 5 μg DNA/10⁶ cell suspensions up to the 10th passage were used. Different electroporation conditions were tested by varying capacitance and voltage (25 μF: 125 and 250 V; 500 μF: 150 and 300 V; Gene Pulser II; Bio-Rad, Munich, Germany). After electroporation, the cells were plated in DMEM/F12, 1× N2 supplement (Invitrogen-Gibco), and 2 mM l-glutamine, supplemented with 20 ng/mL EGF (Invitrogen). 

The pMCrx.1 plasmid containing the full-length murine Crx (cone, rod homeobox) cDNA coding sequence (907 bp) was used to genetically modify hRSCs and hCSCs by electroporation, as previously described. ¹⁹ The efficacy of electroporation was assessed by immunocytochemistry and Western blot analysis. Cells labeled with anti-V5 antibody were quantified 48 hours after electroporation by counting three random nonoverlapping, low-power (×20) images captured with a digital camera (Hamamatsu, Hamamatsu City, Japan) mounted on a microscope (Nikon, Tokyo, Japan). The captured images were obtained at a constant camera setting and exposure time, and the image montages were assembled (Photoshop CS; Adobe, San Jose, CA). A pcDNA plasmid containing the LacZ reporter gene (Invitrogen) was used as a control plasmid. 

Immunocytochemistry

Both hRSCs and hCSCs, nonelectroporated and electroporated, were plated on poly-d-lysine-coated slides 24 hours before analysis, as described elsewhere, ²³ in the differentiation medium described herein. The cells were fixed with 4% paraformaldehyde at room temperature for 30 minutes, washed in PBS, and incubated with primary antibodies overnight. Antibodies were obtained from Sigma-Aldrich, Ltd. (Dorset, UK): a polyclonal rabbit anti-bovine neurofilament 200 (NF 200) antibody (1:1000), ²⁴ a monoclonal anti-mouse proliferating cell nuclear antigen (PCNA) clone PC10 antibody (1:6000), ²⁵ and a monoclonal anti-mouse rhodopsin (Rho) clone RET-P1 antibody (1:1000) ²⁶ ; Santa Cruz Biotechnology (Santa Cruz, CA): a polyclonal goat anti-mouse cyclic nucleotide-gated (CNG) cation channel 3 (CNG-3) antibody (0.4 μg/mL) ²⁷ and a monoclonal anti-human POU transcription factor Oct3/4 (C10) antibody (1 μg/mL) ²⁸ ; R&D Systems Europe, Ltd. (Abingdon, UK): a monoclonal anti-rat neuron-specific β-III tubulin (β III Tub; clone TUJ-1) antibody (1.25 μg/mL) ²⁹ ; Developmental Studies Hybridoma Bank (University of Iowa, IA): a monoclonal anti-rat nestin (clone RAT-401) antibody (1:90) ³⁰ and a monoclonal anti-chicken Pax6 (1:40) antibody ³¹ ; Invitrogen: a monoclonal anti-V5 epitope (5.5 μg/mL) ³² ; and Chemicon (Temecula, CA): a polyclonal rabbit anti-human blue opsin (1.25 μg/mL). ³³ A polyclonal rabbit anti-phosphodiesterase β-6 (PDE) antibody (1:500) was a gift from Yee-Kin Ho (University of Illinois, Chicago, IL). ³⁴ The antibodies were visualized as appropriate by using goat anti-rabbit IgG, goat anti-mouse IgG, or rabbit anti-goat IgG conjugated to fluorescein (Sigma-Aldrich, Ltd.) at concentrations in accordance with the manufacturer's recommendations. 

Western Blot Analysis

Nonelectroporated hCSCs and hRSCs were kept in the differentiation medium (as described earlier) before harvesting and were lysed in lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% Triton X-100, 2 mM EGTA, 2 mM EDTA, and protease inhibitor cocktail 1 μg/mL; Sigma-Aldrich, Ltd.). Total protein extracts were analyzed by Western blot. ³⁵ Subsequent stripping and detection of α-tubulin was used to assess protein loading and transfer. Antibodies were obtained from the same sources as described earlier and were used at the following concentrations: anti-rhodopsin (1:2500), ³⁶ anti-α-tubulin (0.2 μg/mL, sc-1615; Santa Cruz Biotechnology), anti-V5 epitope (0.22 μg/mL), ³⁷ and anti-blue-cone opsin (2.5 μg/mL). ³⁸ The immunoreactive proteins were visualized with horseradish-peroxidase–linked donkey anti-goat antibody, goat anti-mouse or goat anti-rabbit antibody, and enhanced chemiluminescence detection (Santa Cruz Biotechnology). 

Functional Characterization of Crx-Transduced hRSCs and hCSCs

Functional phototransduction pathways were assessed by measuring the level of cyclic (c)GMP hydrolysis with an enzyme immunoassay kit, according to the manufacturer's protocol (Biotrack EIA System; Amersham Biosciences-GE Healthcare, Buckinghamshire, UK), in differentiated hRSCs and hCSCs (in the culture conditions described herein) kept in the dark or maintained in daylight. The PDE inhibitor IBMX (3-isobutyl-1-methylxanthine; Sigma-Aldrich, Ltd.), was added (50 mM) 48 hours before determination of cGMP levels. 

Results

Phenotypic Characterization of Crx-Electroporated hCSCS

Optimal electroporation conditions were defined by varying the capacitance and voltage. The percentage of hCSCs expressing exogenous Crx was evaluated by immunodetection of the V5-tagged protein 48 hours after electroporation (Fig. 1A). The proportion of cells expressing the transgene was found to be ∼22% at a capacitance of 25 μF at both 125 and 250 mV (22.8% ± 6.5% and 22.8% ± 8.0%, respectively). Increasing the capacitance to 500 μF increased the transfection efficiency to ∼29% (29.3% ± 5.0%) at 150 mV and more than 50% (51.3% ± 6.3%, n = 3, P < 0.036, unpaired t-test) at 300 mV (Fig. 1A). No nonspecific immunofluorescence was observed in cells without primary antibody in this and in the other immunocytochemical analyses to be described later. In addition, comparisons of plasmid control electroporated hCSCs and control nonelectroporated hCSCs showed no observable differences in the immunocytological patterns of any markers analyzed in this study (data not shown). Expression of the transgene was confirmed by Western blot analysis (Fig. 1C). No expression of V5 was detected in the hCSCs and hRSCs maintained in identical culture conditions, either nonelectroporated (Fig. 1C) or electroporated with a control plasmid (data not shown). 

Figure 1.

 

Optimization of the electroporation of pMCrx.1 in hCSCs (A) and hRSCs (B). Exogenous expression of Crx was measured by detecting V5-tagged protein 48 hours after electroporation and was quantified by counting three random, nonoverlapping images. Results are expressed as the mean ± SEM (n = 3). Statistically significant differences: for hCSCs (A) *P < 0.036, and for hRSCs (B) *P < 0.004 compared with 250 mV, P < 0.007 compared with 300 mV, and P < 0.035 compared with 150 mV; unpaired t-test. (C) Exogenous Crx expression was analyzed by Western blot analysis of cell lysates from hCSCs and hRSCs electroporated with pMCrx.1 or control cells. Densitometric analysis of the ratio of V5 to α-tubulin is shown. Statistically significant differences: for hCSC-CRX •P < 0.0001 compared with hCSCs; for hRSC-CRX ◊P < 0.0001 compared with hRSCs; unpaired t-test.

The effect of exogenous Crx expression on differentiation of hCSCs was assessed by immunocytochemistry in cells electroporated with the pMCrx.1 plasmid, as previously described. ¹⁹ Exogenous Crx induced a notable decrease in the expression of the proliferation markers used in this study (Fig. 2). Immunolabeling of the homeodomain transcription factor Pax6, the proliferating cell nuclear antigen (PCNA), and the POU transcription factor Oct3/4, decreased significantly after Crx electroporation (Figs. 2B, 2D, 2F). Expression of the neuronal stem cell marker nestin was also downregulated (Figs. 3A, 3B). Concomitantly, increased immunolabeling signals for differentiated neuronal markers—namely, class III β-tubulin (β-III Tub; Fig. 3D), Map2 (Fig. 3F), and neurofilament (NF 200), were observed in Crx-electroporated hCSCs (Fig 3H). To address the question of whether these neuronal cells acquire photoreceptor phenotypes, the expression of photoreceptor-specific genes was assessed by immunocytochemistry (Figs. 4, 5). Increased immunostaining for rhodopsin (Rho), blue-cone opsin, phosphodiesterase (β-6 PDE), and cyclic nucleotide-gated cation channel-3 (CNG3) was detected in Crx-electroporated hCSCs (Figs. 4B, 4D, 4F, 4H). Double immunostaining with anti-V5-tagged protein, detecting the expression of exogenous Crx, and either anti-blue opsin or anti-β-6PDE, confirmed coexpression of the Crx transgene with photoreceptor-specific markers (Fig. 5). Quantification of double-labeled cells showed that the hCSCs expressing the transgene also expressed specific markers of photoreceptors (Figs. 5G, 5H). Although ∼25% of blue cone opsin–positive cells did not coexpress the transgene (Fig. 5G), the majority of cells expressing β-6 PDE were immunopositive for V5-tagged protein (Fig. 5H). In addition, introduction of Crx gene increased the expression of blue cone opsin by ∼35%, but rhodopsin expression was increased by ∼76% in electroporated hCSCs compared with that in nonelectroporated cells (Figs. 5I, 5J). 

Figure 2.

 

Exogenous Crx gene expression decreased expression of the proliferation and differentiation markers in hCSCs. Immunocytochemical labeling of the homeodomain transcription factor Pax6 (A, B), PCNA (C, D), and the POU transcription factor Oct3/4 (E, F) in control hCSCs (A, C, E) and Crx-electroporated hCSCs (B, D, F). Cell nuclei were counterstained with DAPI (C–F). Scale bar, 100 μm.

Figure 3.

 

Exogenous Crx gene expression increased expression of the neuronal markers in hCSCs. Immunolabeling of nestin (A, B), class III β-Tub (C, D), Map2a/b (E, F), and NF 200 (G, H) in control hCSCs (A, C, E, G) and Crx-electroporated hCSCs (B, D, F, H). Cell nuclei were counterstained with DAPI (A–H). Scale bar, 100 μm.

Figure 4.

 

Exogenous Crx gene expression increased expression of photoreceptor-specific markers in hCSCs. Immunolabeling of Rho (A, B), cGMP PDE (C, D), blue cone opsin (E, F), and CNG3 (G, H), in control hCSCs (A, C, E, G) and Crx-electroporated hCSCs (B, D, F, H). Cell nuclei were counterstained with DAPI. Scale bar, 100 μm.

Figure 5.

 

Exogenous Crx gene induced expression of photoreceptor-specific markers in hCSCs. Double immunolabeling in hCSCs showed V5-positive cells expressing exogenous Crx (A, D), blue cone opsin (B), and cGMP PDE (E). Double-labeling with anti-V5 and anti-blue cone opsin (C), and anti-V5 and anti-β-6 PDE (F). Scale bar, 100 μm. Expression of blue cone opsin, cGMP PDE, and exogenous Crx was quantified by counting three random nonoverlapping images (G, H). Western blot analysis of photoreceptor markers rhodopsin (I) and blue cone opsin (J). Densitometric analysis of the ratio to α-tubulin of (I) Rho (P < 0.009, unpaired t-test) and (J) blue cone opsin (P < 0.03, unpaired t-test). Results are expressed as the mean ± SEM (n = 3).

To examine whether these cells acquire a light-sensitive photoreceptor phenotype, we evaluated activation of the phototransduction cascade in vitro by measuring the intracellular level of cGMP. Since light activation of the phototransduction cascade results in hydrolysis of cGMP in functional photoreceptors, the levels of cGMP were measured by immunoassay in cell extracts obtained from cells subjected to ambient light or maintained in dark as previously described. ¹³ No difference in cGMP levels, between light and dark conditions, was observed in nonelectroporated hCSCs and Crx-electroporated hCSCs (Fig. 6). 

Figure 6.

 

Exogenous Crx has no effect on cGMP hydrolysis in hCSCs. Nonelectroporated cells (hCSC), Crx-electroporated hCSCs (CRX) were exposed to light or kept in the dark in control conditions (A) and in the presence of an inhibitor of phosphodiesterases (IBMX) (B). Results are expressed as the mean ± SEM (n = 3).

Differentiation of Crx-Electroporated hRSCs into Retinal Neurons

Electrical parameters were determined to optimize Crx transgene expression after electrotransfer of pMCrx.1 into hRSCs as described earlier (Fig. 1B). A capacitance of 25 μF and a voltage of 125 mV were found to give more than 45% (47.5% ± 2%, n = 3) V5-positive hRSCs (Fig. 1B). Increasing the voltage significantly decreased the electroporation efficiency, to ∼12% at 25-μF capacitance (12.4% ± 0.9%, n = 3, P < 0.004, unpaired t-test) and to ∼10% at 500-μF capacitance (10.8% ± 2.2%, n = 3, P < 0.007, unpaired t-test). In addition, only ∼35% of the cells (36.1% ± 0.9%, n = 3, P < 0.035 unpaired t-test) were found to be positive at 150 mV and 500 μF. Therefore, 125 mV and 25 μF were determined to be the optimal electroporation conditions for assessing the effect of Crx on the differentiation of hRSCs. 

Exogenous Crx expression decreased the expression of the proliferation markers Pax6, PCNA, and Oct3/4 in electroporated hRSCs (Figs. 7B, 7D, 7F). Concurrently, Crx-electroporated hRSCs displayed diffuse β-III Tub, but strong Map2 and NF immunoreactivity (Figs. 8D, 8F, 8H) and reduced nestin immunolabeling (Fig. 8B), suggesting that Crx induces the differentiation into mature neurons. To ascertain whether these neurons display photoreceptor phenotypes, we evaluated the expression of rhodopsin, blue cone-opsin, β-6 PDE, and CNG3. Crx electroporated hRSCs were found to be immunopositive for all the photoreceptor markers tested (Fig. 9). Double immunolabeling with anti-V5-tagged protein and either anti-blue opsin or anti-β-6 PDE confirmed the coexpression of the Crx-transgene with photoreceptor-specific markers in Crx-electroporated hRSCs (Fig. 10). Quantification of double-labeled cells showed that hCSCs expressing the transgene also expressed specific markers of photoreceptors (Figs. 10G, 10H). Most of the blue cone opsin– and β-6 PDE–positive cells coexpressed the transgene (Figs. 10G, 10H). In addition, introduction of the Crx gene increased by ∼90% the expression of both rhodopsin and blue cone opsin, compared with that of nonelectroporated cells (Figs. 10I, 10J). 

Figure 7.

 

Exogenous Crx gene expression decreased the expression of proliferation and differentiation markers in hRSCs. Immunocytochemical labeling of the homeodomain transcription factor Pax6 (A, B), of the proliferating nuclear antigen (PCNA) (C, D), and the POU transcription factor Oct3/4 (E, F) in hRSCs (A, C, E) and Crx-electroporated hRSCs (B, D, F). Cell nuclei were counterstained with DAPI (C–F). Scale bar, 100 μm.

Figure 8.

 

Exogenous Crx gene expression increased neuronal markers expression in hRSCs. Immunolabeling of nestin (A, B), class III β-Tub (C, D), Map2a/b (E, F), and NF 200 (G, H) in hRSCs (A, C, E, G) and Crx-electroporated hRSCs (B, D, F, H). Cell nuclei were counterstained with DAPI (A–H). Scale bar, 100 μm.

Figure 9.

 

Exogenous Crx gene expression increased photoreceptor-specific marker expression in hRSCs. Immunolabeling of Rho (A, B), cGMP PDE (C, D), blue cone opsin (E, F), and CNG3 (G, H) in hRSCs (A, C, E, G) and Crx-electroporated hRSCs (B, D, F, H). Cell nuclei were counterstained with DAPI. Scale bar, 100 μm.

Figure 10.

 

Exogenous Crx gene induced expression of photoreceptor-specific markers in hRSCs. Double immunolabeling in hRSCs showed V5-positive cells expressing exogenous Crx (A, D), blue cone opsin (B), and cGMP PDE (E). Double-labeling with anti-V5 and anti-blue cone opsin (C) and anti-V5 and anti-β-6 PDE (F). Scale bar, 100 μm. Expression of blue cone opsin, cGMP PDE, and exogenous Crx was quantified by counting three random nonoverlapping images (G, H). Western blot analysis of the photoreceptor markers rhodopsin (I) and blue cone opsin (J). Densitometric analysis of the ratio to α-tubulin of (I) Rho (*P < 0.02, unpaired t-test) and (J) blue cone opsin (P < 0.02, unpaired t-test). Results are expressed as the mean ± SEM (n = 3).

Differentiation of Crx-Electroporated hRSCs into a Photoreceptor-Specific Light-Sensitive Phenotype

We evaluated the physiological characteristics of the differentiated Crx-electroporated hRSCs. Their phototransduction function was established by measuring cGMP levels (by immunoassay) in cell extracts obtained from Crx-electroporated hRSCs subjected to either ambient light or maintained in the dark, as described earlier (Fig. 11). No difference in cGMP levels, between light and dark conditions, was observed in nonelectroporated hRSCs (Fig. 11A). Similar results were observed with plasmid control-electroporated cells (data not shown). However, an increase in cGMP level was observed when Crx-electroporated hRSCs were kept in the dark (1.7 times higher, P < 0.03; unpaired t-tests, Fig. 11A). 

Figure 11.

 

Exogenous Crx gene expression increased phototransduction function in hRSCs. Hydrolysis of cGMP levels was measured in nonelectroporated (hRSC) and Crx-electroporated hRSCs (CRX).The cells were exposed to light or kept in the dark in control conditions (A) and in the presence of an inhibitor of phosphodiesterases (IBMX) (B). Results are expressed as the mean ± SEM (n = 3). *P < 0.03, unpaired t-test.

Inhibition of phosphodiesterase by a nonselective inhibitor (IBMX), increased cGMP levels (∼3.6 times higher) in nonelectroporated cells maintained in light and dark conditions and in the light-exposed Crx-electroporated hRSCs (Fig. 11B). The cGMP level difference observed between Crx-transfected hRSCs in light and dark conditions was abolished by addition of the phosphodiesterase inhibitor (Figs. 11A, 11B). The level of cGMP hydrolyzed by light-sensitive PDE corresponded to the difference between light and dark conditions and represented ∼81 fentomoles/10⁶ cells in the Crx-electroporated hRSCs (Fig. 11A). The difference between the cGMP levels in the presence and absence of IBMX, in the light condition, corresponded to a total of ∼390 fentomoles/10⁶ hRSCs that were cGMP hydrolyzed by all the PDEs. Therefore, light-sensitive PDE represented ∼21% of the total PDE activity in Crx-electroporated hRSCs (Fig. 11). 

Discussion

Currently, generation of functional photoreceptor phenotypes from adult human–derived ocular stem cells remains a challenge. In previous work, we showed that mouse retinal stem cells, isolated from the ciliary body, could be genetically modified to efficiently produce cells with a functional photoreceptor phenotype exhibiting light-sensitive properties. The present study extends this investigation in showing that human adult retinal stem cells, unlike human corneal stem cells, can be directed to differentiate into light-sensitive photoreceptors after Crx gene transfer. 

The hCSCs and hRSCs were genetically manipulated by introducing Crx using simple and reliable electroporation protocols, ^19,39 that resulted in transfection efficiencies of up to 50%. The optimal electroporation conditions defined for hCSCs and hRSCs were found to be different. These findings suggest that, although both cell types are derived from embryonic ectoderm, ⁴⁰ there may be minor differences in their respective cell membrane properties. The higher percentage of hRSCs expressing exogenous Crx was obtained by using the same electroporation conditions as reported previously for the mouse retinal stem cells. ¹⁹ However, the percentage of cells expressing the transgene was slightly lower in human than in mouse tissue. This slight discrepancy could reflect a difference in the cell membrane sensitivity to transient permeabilization by high voltage and/or the postelectroporation viability. The prolonged postmortem times of the human tissues used to generate the hRSCs could also be responsible for the difference in electroporation efficiencies observed. Several nonhuman and human studies have shown that neurospheres can be generated from the central nervous system ^41–43 and from human retina ⁴⁴ even with prolonged postmortem times. However, variability in differentiation capacity of these human cells has not been tested in regard to extended postmortem times. Indeed, variable postmortem times before culturing hRSCs has been suggested to influence the sphere-forming frequency. ⁴⁵ Nevertheless, the present study showed that genetic modification of hRSCs was efficient, as previously shown for mouse RSCs. ¹⁹  

We elected to use the mouse Crx cDNA because the gene has been reported to be conserved throughout evolution. The mouse and human putative protein sequences display 97% homology, ^46–49 and spatial expression of the CRX gene appears to be conserved. ⁵⁰ The results presented in this study confirmed that exogenous expression of mouse Crx induces differentiation and maturation of hRSCs and hCSCs. 

Indeed, the expression of the proliferation and expansion markers Pax6 ^51–53 and PCNA ⁵⁴ and the stem cell marker Oct3/4 ^53,55–57 were found to be downregulated, suggesting that exogenous Crx expression mitigates hRSCs and hCSCs pluripotency and proliferation properties as previously reported. ¹⁹ Concomitantly, upregulation of the neuronal markers (TUJ-1, Map2a/b and NF 200) indicates that Crx shifts hRSCs and hCSCs toward neuronal phenotypes, in accordance with our previous study in mouse RSCs. ¹⁹ Expression of photoreceptor-specific genes (cone and rod: β-6 PDE; rod-specific: rhodopsin; cone specific: blue opsin, CNG3) was found to be increased in both the hRSCs and hCSCs. 

Since Crx controls the expression of a core set of photoreceptor genes, ^58,59 our results strongly suggest that its exogenous expression induced both stem cell types to differentiate into photoreceptor phenotypes. Colocalization of exogenous Crx with cone opsin and β-6 PDE in both hRSCs and hCSCs supports this hypothesis. These proteins were found to show cytoplasmic localization in the hRSCs and hCSCs, indicating that contrary to photoreceptors in the mature retina, these cells lack polarity. Akagi et al. ³⁶ have reported a similar rod opsin immunodistribution in iris-derived cells transfected with Crx. Diffuse localization of cone opsins has also been detected in the mouse cone photoreceptor cell line 661W. This finding implies that photoreceptors lose their polarity when cultured in vitro. ⁶⁰ In addition, retinal pigment epithelium (RPE) transplantation induces regeneration of photoreceptor outer segments in RCS rats. ⁶¹ This effect indicates that physical contact with the RPE is needed for photoreceptor polarity. It remains to be determined whether hRSCs and hCSCs expressing exogenous Crx acquire this property when they integrate into mature retina. This property has been demonstrated, however, for iris-derived cells expressing Crx. ⁶²  

Moreover, a significant increase in light-induced cGMP hydrolysis was observed in Crx-transduced hRSCs but not in hCSCs, suggesting that the phototransduction cascade can be activated only in hRSCs. Although hCSCs expressed photoreceptor-specific markers, hRSCs displayed a greater increase in the level of expression of these genes after Crx electroporation, suggesting that key elements of the transduction cascade are missing in hCSCs. In support of this hypothesis is the report that adult rat limbal epithelial stem cells differentiate into functional neurons expressing photoreceptor-specific markers. However, no evidence of photoreceptor functionality was substantiated. ²² Moreover, there is evidence that Crx expression alone is insufficient to induce primate iris-derived cells into cells expressing rhodopsin. Rather, Crx must combine with NeuroD to achieve photoreceptor differentiation. ⁶² Whether addition of NeuroD would induce hCSCs to differentiate into light-sensitive photoreceptors remains to be determined. Nevertheless, Crx can induce retinal stem cells from both human and rodent species to differentiate into light-sensitive photoreceptor phenotypes. 

A lower percentage of functional PDE activity versus total PDE activity was observed in adult stem cells of human origin. That the electroporation efficiency was found to be lower in human than in mouse RSCs could explain this discrepancy. Comparing human and mouse clones stably expressing the transgene will help determine the capacity of mouse Crx to differentiate RSCs into light-sensitive photoreceptor phenotypes in both species. 

Our results support the hypothesis that exogenous Crx is sufficient to induce differentiation of adult human retinal stem cells into light-sensitive photoreceptor phenotypes in vitro. Transplantation of these genetically modified cells will address integration and survival in the host retina. Their functional status and efficiency in restoring visual function should be evaluated in animal models of retinal degeneration. This approach could lead to successful cell-based therapies for human retinal degenerations and so is worth pursuing. 

Footnotes

 Supported by the British Retinitis Pigmentosa Society, the Iris Fund for Prevention of Blindness, Lord Sandberg, Foresight, and the Gift of Sight Appeal.

Footnotes

 Disclosure: C. Jomary, None; S.E. Jones, None; A.J. Lotery, None