|Home | About | Journals | Submit | Contact Us | Français|
Some of the most common causes of blindness involve the degeneration of photoreceptors in the neural retina; photoreceptor replacement therapy might restore some vision in these individuals. Embryonic stem (ES) cells could in principle provide a source of photoreceptors to repair the retina. We have previously shown that retinal progenitors can be efficiently derived from human ES cells. We now show that retinal cells derived from human ES cells will migrate into mouse retinas following intra-ocular injection, settle into the appropriate layers and express markers for differentiated cells, including both rod and cone photoreceptor cells. After transplantation of the cells into the subretinal space of adult Crx -/- mice (a model of Leber's Congenital Amaurosis), the hES cell derived retinal cells differentiate into functional photoreceptors and restore light responses to the animals. These results demonstrate that hES cells can, in principle, be used for photoreceptor replacement therapies.
Degeneration of photoreceptors in the retina is a common cause of blindness. Age related macular degeneration, in which cone photoreceptors in the central retina degenerate, is the leading cause of blindness in the elderly, while many inherited diseases of the retina result in rod photoreceptor degeneration. There are currently no effective treatments available to prevent the loss of photoreceptors in most of these disorders. Photoreceptor replacement has been shown to be feasible in animal models of these diseases, but any cell replacement strategy will require a source of new retinal cells.
Several sources of cells have been tested for their ability to replace photoreceptors. Fetal or embryonic retinal progenitors can be grown in vitro (Anchan et al., 1991) and used for transplantation (Klassen et al., 2004). Neurospheres can be grown from the adult pigmented ciliary epithelium, and these cells can also be transplanted to the retina (Coles et al, 2004). Neural stem cells derived from the hippocampus show a remarkable ability to integrate into the retinal layers and form morphologically normal appearing retinal neurons (Takahashi et al., 1998). The best evidence for functional photoreceptor replacement comes from the study of MacLaren et al (2006), in which freshly dissociated, postmitotic rod photoreceptors were transplanted to the subretinal space; however, the number of cells cannot be increased in vitro due to their postmitotic state.
Embryonic stem (ES) cells might also be a source for replacement of photoreceptors. Their indefinite self-renewal and pluripotentiality make them an ideal source. We have previously shown that human ES cells can be directed to a retinal cell fate using a combination of a BMP inhibitor, a Wnt inhibitor and IGF-1 (Lamba et al., 2006). Up to 10% of the cells differentiated with this protocol also expressed early markers of photoreceptor differentiation like Crx, Nrl, and recoverin at the end of 3 weeks. A previous study transplanted hES cells into the subretinal space of wild-type mice (Banin et al., 2006); however, as they did not use a directed retinal differentiation protocol, they failed to see significant photoreceptor differentiation. In this study, we have used our retinally directed hES cells for transplantation into both wild type and Crx deficient mice. When transplanted to newborn wild type mice, we find that the cells integrate into all the retinal layers and express markers appropriate for the lamina in which they have settled. When transplanted into the sub-retinal space of adult mice, the cells integrate into the outer nuclear layer, express photoreceptor markers and differentiate outer morphologically identifiable outer segments. Lastly, when transplanted into the subretinal space of adult Crx -/- mice, the cells integrate into the retina, and restore a light response to otherwise unresponsive animals. Together these results show that human ES cells can be used as a source of functional photoreceptors for potential cell replacement therapy.
Human ES cells were passed through a three-week “retinal determination” protocol as previously described (Lamba et al., 2006). This protocol typically results in cultures with 80% retinal cells. In addition, the cells also expressed markers of inner retinal neurons and photoreceptors, including Pax6, Hu C/D (30.5%), PKCα, NRL (15.42% +/- 2.19) and Crx (32.83% +/- 3.42). The expression of these genes was confirmed by PCR for Crx, S-opsin, recoverin, arrestin, IRBP, Rhodopsin and TRβ2 (Supplementary Figure 1). To identify the cells following transplantation, the retinal cultures were infected with a replication incompetent lentivirus expressing eGFP under a ubiquitous promoter (hEF1a) or an adenovirus driving eGFP under CMV promoter. This resulted in expression of GFP by 50-70% of all cells in the plate. After an additional 5-7 days of culture, the cells were transplanted using two approaches: (i) into the intra-vitreal space of the newborn mice and (ii) into the subretinal space of the adult mice. In both approaches, 50,000-80,000 cells were transplanted in 1 μl volume. The mice were allowed to survive for a period of one to six weeks following transplantation. To prevent immune-rejection of the transplanted human cells, the immunosuppressant cyclosporine A was injected daily.
When the cells were injected intra-vitreally either at birth or post-natal day one (P0 or P1), cells migrate into all three cell layers: ganglion cell layer (GCL), inner nuclear layer (INL) and the outer nuclear layer (ONL) (31.67+/-2% in ONL, 52.43+/-2.8% in INL, 15.9+/- 1.98% in GCL, total GFP cells in retina = 9657+/- 2595, n=4) (Figure 1A, Supplementary Figure 2). Interestingly, migration was restricted to these early time points as intra-vitreal transplants from P2 onwards resulted in absence of migration of cells into retina. Many of the cells that settled in the inner nuclear layer expressed markers of inner retinal neurons such as Hu C/D-a marker of ganglion and amacrine cells (Hu C/D cells=19.8%, Figure 1 B,B′) and PKCα - a marker of rod bipolar cells (not shown). The cells that migrated to the outer nuclear layer, where photoreceptors normally reside, express recoverin (a photoreceptor-expressed protein; Figure 1D,D′, recoverin = 25.79%). To determine whether the transplanted cells had integrated into the mouse retinal circuit, we labeled sections with a synaptic marker synaptophysin. Synaptophysin is a pre-synaptic protein expressed in the axon terminal of both rod and cone photoreceptors (Yang et al., 2002a). The GFP expressing cells in the outer nuclear layer expressed synaptophysin at their terminals in the outer plexiform layer (Figure 1C, C′). Further evidence for cone photoreceptor differentiation of the transplanted human ES derived retinal cells was obtained by labeling sections for the cone pigment, S-opsin (the pigment in the cone photoreceptors responsible for short-wavelength/blue cone vision; Figure 1E-E′″). The morphology of the transplanted cells that were present in the outer nuclear layer was very similar to the host photoreceptors, though the human cells expressed slightly lower levels of S-opsin immunoreactivity than the host mouse cones. Although the hES cell derived cells migrate throughout the retinal layers, we only detected S-opsin expressing cells in the ONL. To rule out any GFP expression by host retinal cells due to potential carry over of virus, we transplanted lentivirus infected fibroblasts and did not see any GFP expression in the host retina (data not shown). We also examined all eyes for evidence of teratomas, since these can arise from undifferentiated ES cells. However, we did not find teratomas in any of the mice that received intra-vitreal transplanted cells (n=36, n>4weeks=10). Teratomas likely arise from undifferentiated ES cells, and it is unlikely that any of the cells remain undifferentiated after the 4 weeks of retina-directed cell cultures prior to transplantation. These data show that the transplanted human ES derived retinal cells migrate into the appropriate layers in the retina following intra-vitreal transplantation in newborn mice, express markers consistent with retinal neuronal and rod and cone photoreceptor differentiation, and express a synaptic marker in the outer plexiform layer.
Previous results have shown that mouse photoreceptors incorporate effectively in adult retinas when transplanted to the sub-retinal space (MacLaren et al., 2006). To determine whether hES cell derived retinal cells would incorporate into the retinas of adult mice, we injected the cells into the subretinal space in wild type adult mice, from 4-6 weeks of age. The mice were analyzed two-three weeks later. In these animals, we found that GFP-expressing cells migrated into the ONL of the retina in the region of the transplant; many expressed recoverin (78.2+/-5.1% of the cells in the ONL, n=4) (Figure 2C-C″″) and had morphology highly reminiscent of photoreceptors. Some of the cells that migrated into the ONL also expressed rhodopsin (63.24+/-2.25% of the cells in the ONL, n=4), consistent with some of the human ES cells differentiating as rod photoreceptors (Figure 2D, E-E′, F-F″). We compared the morphology of the transplanted human cells to those from similar mouse to mouse transplantation from a GFP expressing donor mouse. We found that the human cells (Figure 2A, A′) looked remarkably similar to integrated mouse cells (Figure 2B, B′). Thus, the human ES derived retinal cells behave very similarly to postnatal mouse retinal cells following transplantation (MacLaren et al., 2006). We also tested whether the human ES derived retinal cells could be transplanted to animal models of retinal degenerations. We transplanted cells to the vitreous of newborn albino mice; the rods and cones of these mice are sensitive to normal levels of illumination and so undergo a progressive degeneration (LaVail et al., 1987). We examined their retinas for presence of transplanted photoreceptors after sacrifice, and found that the transplanted rods survived in the outer nuclear layer as clusters of cells, that expressed rhodopsin and recoverin, while the host rods underwent extensive degeneration (Supplemental Figure 3). We found an average of 9728 (+/-1992, n=3) GFP expressing cells in the retina. Of these, 27.7% were recoverin expressing +/-0.47) and 21.3% were rhodopsin positive (+/- 5.57%). The transplanted cells failed to extend inner and outer segments either due to the lack of supporting host photoreceptors or healthy interaction with host RPE cells which are important in their maintenance. These results suggest that the transplanted human photoreceptors may be more resistant to degeneration than host mouse photoreceptors, to light damage.
To determine whether the hES cell derived retinal cells were able to form functional photoreceptors, we transplanted the cells into Crx -/- mice. Crx is one of the earliest genes expressed by differentiating photoreceptors and acts as a key regulator of expression of photoreceptor-specific genes. Knocking out Crx gene results in severe reduction or complete loss of components of the photo-transduction cycle, like rhodopsin, cone opsin, rod transducin, cone arrestin, and recoverin (Furukawa et al., 1999). These mice have been analyzed for visual function and Crx-/- completely lack any rod or cone ERG response even though the photoreceptor layer is intact, at least early. The hES cells were subjected to the retinal differentiation protocol labeled with the GFP-expressing virus and injected into the subretinal space of mature (4-6 week old) mice. After two to three weeks, the mice were analyzed for the restoration of visual responses with ERG analysis. The animals were dark-adapted and progressively more intense light flashes were presented. Figure 4 shows the results of this analysis. The traces shown are responses to the most intense flashes, likely a combination of both rod and cone responses. The un-injected eye of the Crx -/- mouse has essentially no light response (Figure 4C), as described previously (Furukawa et al., 1999). However, the eye that received the transplanted cells had a clear response to a flash of light (Figure 4D). 15 animals had some ERG b-wave response in the injected eye (b-wave amplitude of 22+/-4μV, n=15, p value <0.0001 compared to control eyes, b-wave amplitude of wild-type mouse under similar excitation conditions was 203+/-28.43μV, n=3, Figure 4C). Histological analysis following the ERG testing confirmed that the animals that showed a response to light had a successful transplant. We found that 36(+/-1.1%) cells were recoverin positive (Figure 3A, B, C-C″, D-D″), though, like the transplant into light damage retinas they failed to elaborate inner and outer segments in the absence of supportive host outer segments. The transplanted cells also expressed rhodopsin and Nrl (Figure 3E-E″, 3F-F″). Even though the transplanted human cells do not extend clearly detectable outer segments, it has been previously shown that a diminished ERG response occurs in retinal degeneration models lacking outer segments (Machida et al., 2000). Rds mice have detectable ERGs as late as 6-7 months despite the absence of outer segments (Reuter and Sanyal, 1984). None of the eyes that did not receive injections of cells had a response to light (b-wave amplitude of 5.13+/-0.5μV n=23). In injected eyes where no cells were found in subretinal space or inside the retina at time of analysis, we also did not see a detectable ERG (b-wave amplitude of 4.5+/-1.3μV n=9). We also found a strong correlation between the size of the transplanted area and the corresponding b-wave response (r=0.8139, p<0.0001, Figure 4F). For a more accurate correlation between the ERG response and number of rod photoreceptors, we used the Nrl antibody, which marks human rod photoreceptors but not those of the mouse (Suppl. Fig. 6). We found an average of 2985 Nrl+ cells in the retina (+/- 722, n=7). Also, we were able to detect a number of human rods that no longer expressed GFP. Upon correlating either total number of Nrl cells per eye or per section analyzed, we found a strong correlation with the corresponding ERG b-wave (for per eye r=0.8977 and p=0.0025, and for per section, r=0.8537, p=0.0070, Figure 4G, H). To confirm the integration of these cells into the retinal circuitry, we analyzed their expression of synaptic markers synaptophysin (Figure 4A-A″) and PSD95 (Figure 4B-B″), and we found that both markers were present in their terminals in the outer plexiform layer. Additionally, we tested the integration of GFP expressing mouse retinal cells into Crx deficient mice. We found a similar integration (48.7+/-6.3% recoverin expressing cells, Supplementary Figure 4) and b-wave restoration (b-wave amplitude of 38+/- 10μV, n=3, Figure 4E) in these transplanted mice. The data thus show that at least some of the transplanted human ES derived photoreceptors can respond to light, and likely can transmit this response to the bipolar cells (b-wave). The small ERG response is consistent to what would be expected with a few surviving photoreceptors in the retina (Takahashi et al., 2005).
Taken together our results show that human embryonic stem cells can be directed to a retinal fate, and following transplantation will differentiate into light responsive photoreceptors. The long-term goal of these studies is to develop a procedure for photoreceptor replacement to treat retinal degenerations, such as Macular Degeneration and Retinitis Pigmentosa. Recent success with photoreceptor transplantation from neonatal mouse retina highlights the fact that the mature retina can incorporate new photoreceptors into preexisting circuitry (MacLaren et al., 2006). The use of human embryonic stem cells thus provides a potentially inexhaustible source for human photoreceptors. Other potential sources for human photoreceptors for transplantation include sphere-producing cells isolated from the ciliary epithelium (called retinal stem cells) or retinal progenitors isolated from embryonic retina and expanded in vitro (Coles et al., 2004; Humayun et al., 2000; Kelley et al., 1995; Yang et al., 2002b). Although both sources have shown potential for in vitro expression of photoreceptor-specific proteins, neither of these sources displays potential for photoreceptor differentiation following transplantation. Attempts have been made to stimulate photoreceptor differentiation from these other sources by expressing potent photoreceptor-specific transcription factors (see Lamba et al, 2008 for review); however, the present studies indicate that this is not necessary for photoreceptor development from human ES cells.
The restoration of an ERG response, albeit small, in the Crx-/- mice correlates with the efficiency of transplantation. As seen in Figure 3, there is extensive migration into the outer nuclear layer of transplanted human cells (see Supplemental Figure 4 for mouse to mouse transplants for comparison). The extent of the transplant (area covered by the transplanted cells), as well as number of Nrl expressing cells, correlates well with the amplitude of the B-wave (Figure 4). The transplanted mouse cells also restore a small, but significant B-wave, whereas sham injections, or injections of fibroblasts do not. In our best transplant to date, we found coverage of approximately 25% of the retinal surface with between one and four rows of rods. We find an average of 3000 Nrl+ human cells (presumed rod photoreceptors) in the Crx-/- mouse retina after transplantation. Since there are at least twice as many Crx-expressing photoreceptors as Nrl expressing rod photoreceptors in the cultures prior to transplantation, there may be as many as 6000 human photoreceptors in these mice. It is surprising that we see even a small ERG with so few functional rods, given that the normal mouse retina has several million rods. However, the relationship between rod number and B-wave amplitude is not a linear one, particularly when there are low numbers of rods. Robson et al, in 2004 have noted that there is a significant non-linearity in the rod-to-bipolar synapse and it is likely that only a few rods are sufficient to drive the rod bipolar at super-threshold intensities (Robson et al., 2004). Moreover, in their analysis of structure-function relationships in retinal degeneration caused by light damage, Takahashi et al found that the photopic B-wave ERG drops off much more rapidly with cell number than does the rod mediated scotopic B-wave (Takahashi et al., 2005). Similar to that reported by the Machida et al (2000) report, there appears to be rod “buffering” at low numbers of rods. The authors in that paper suggest that the wide receptive field of the bipolar cell provides buffering against loss of photoreceptors. Bipolar cells receive input from as many as 45 rods, but a maximal bipolar response results from signaling by only a small subset.
In a relatively short period of time, human ES cells have been shown to be useful for deriving a variety of different differentiated cells and tissues (D′Amour et al., 2006; Kang et al., 2007; Laflamme et al., 2007; Lamba et al., 2006; Osakada et al., 2008; Yang et al., 2008). The use of these cells for repair of the central nervous system has received a great deal of attention, due to the fact that there are few alternatives available for neurological repair. Early attempts to treat animal models of neurological disease, including retinal degenerations, with human ES derived TH-positive neurons failed to show a significant functional contribution for the transplanted neurons, and in some cases reported frequent teratomas as a complication(Li et al., 2008). We have found that when the human ES cells are subjected to an extensive differentiation program prior to transplantation, they do not form teratomas when transplanted to either the vitreous or the sub-retinal space. In addition, sufficient numbers of cells migrate into the retina and differentiate and survive long enough to provide a small, but significant, functional restoration to mice without a light response. Recent studies in other areas of the CNS are consistent with these conclusions(Yang et al., 2008), and together with our results in retina indicate that human ES cells may provide the best source for replacement neurons throughout the nervous system.
Neonatal (P0 to P3) mice were anesthetized on ice while adult mice were anesthetized with ketamine/xylazine (ketamine 0.13 mg/g of body weight, xylazine 8.8 mg/kg). In neonatal mice, the eyelids were prematurely opened, and the cells were injected into the vitreous through a glass micropipette through a small hole made with a 26G needle. In adult mice, the subretinal space was reached by an incision in the sclera, choroid and RPE of the dorsal eye, near the ora serrata. GFP expressing retinal cells derived from human embryonic stem cells or the retinas of P2-P12 GFP mice were injected in one microliter of DMEM using a pulled fine glass micropipette (Drummond wiretrol II). A small piece of gel foam with a dab of bacitracin was placed over the incision. Injected animals or mothers of neonatal pups were injected with Cyclosporine A (10 mg/kg/day) from the day before surgery until tissue preparation.
Mice were dark adapted for at least 15 hr and anaesthetized with Isoflurane inhalation (1.5L/min). Pupils were dilated by topical application of tropicamide and phenylephrine 10 mins before ERG recordings. Mice were placed on a heating pad held at 37°C. All manipulations were done under infrared il lumination. A gold ring electrode embedded in a contact lens was placed on a drop of 2%–3% methyl cellulose on the cornea and a copper reference electrode was placed in the forehead. Responses to three-four different test flash intensities were collected from each mouse. All ERGs were carried out under scotopic conditions. ERG signals were amplified 10,000×, filtered between 1 Hz to 3 kHz, and sampled at 5 kHz. ERG data was analyzed using EMWin software (LKC Technologies, Inc).
All counts are presented as average and standard error of mean (SEM). Statistical analysis for correlation and significance was carried out using Graphpad Prism software.
Supplementary Fig.1. (A) PCR analysis for expression of various photoreceptor markers in the cultures. (B) Staining of the cultures shows expression various markers of various retinal neurons including Crx, Nrl, recoverin, Pax6, Hu C/D and PKCα.
Supplementary Fig.2. Transplantation into newborn wild-type mice. A shows a low-power view of the retina with the transplanted human GFP+ cells (green) migrating into all layers of the retina. B is the same section stained with DAPI (blue) to mark all nuclei.
Supplementary Fig.3. Transplantation into a light-damage model, albino mice. (A) Low-power view of the retina 4 weeks after transplantation of human ES-derived retinal cells stained with anti-recoverin (red). Arrows indicate region of surviving photoreceptors which are also GFP+ indicating their human origin. Arrowheads show lack of photoreceptors in regions where human cells did not migrate. (B-J) Higher magnification images confirm that a number of surviving GFP expressing human cells are rhodopsin+ (B-G, red) and recoverin+ (H-J, red).
Supplementary Fig.4. Subretinal transplantation of GFP mouse retinal cells in adult Crx-/- retina. A is a lower power view of the transplanted retina showing extent of transplantation after 2 weeks. B, B′ and C, C′ are higher power views of the boxed regions in A showing co-expression of recoverin (red). Arrow in B marks the axon of a transplanted cell extending into the inner retina and arrowhead indicated GFP and recoverin co-expressing cell. In C, C′, arrow marks a recoverin expressing transplanted GFP cell while arrowhead marks an integrated GFP cell not expressing recoverin anymore.
Supplementary Fig.5. Anti-human NCAM-FITC antibody analysis. A, B show a section of 115 day human fetal retina stained with the antibody (A, Green) and a brightfield view. The antibody marks all human fetal retinal cells. The antibody does not cross-react with mouse retina (Crx-/- retina shown in C, D). E, F shows a section of the human retinal cell transplant into Crx-/- mice. E shows that the anti-NCAM-FITC (green) marks all human cells including those (marked by arrow) which did not get labeled with the GFP expressing lentivirus (stained in red in F).
Supplementary Fig.6. Nrl staining of 115 day human fetal retina (top) and post-natal day 6 wild-type mouse retina (bottom) showing absence of staining of the mouse rods by the antibody in mice.
The authors gratefully acknowledge the technical support of Ms. Paige Etter, the gift of Crx deficient mice from Dr. C. Cepko, the help with the ERG analysis from Dr. Jim Hurley, the gift of the viral constructs from Drs. C. Murry and Dr. A. Lieber, the helpful suggestions of the members of the Reh lab on various aspects of the work, the critical comments on the manuscript by Dr. Rachel Wong, and criticisms of a more general nature by Dr. Olivia Bermingham-McDonogh. This work was funded by grants from the NIH (PO1 GM081619), a gift from Peter C. Kovner and support from the Foundation Fighting Blindness to T.A.R.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.