Design  The focus is on the use of SCs for regeneration of retinal degenerations. Cell populations include embryonic, neural, and bone marrow–derived SCs, and engineered grafts will also be described.

Results  Experimental approaches have successfully replaced damaged photoreceptors and retinal pigment epithelium using endogenous and exogenous SCs.

Conclusions  Stem cells have the potential to significantly impact retinal regeneration. A combination with bioengineering may bear even greater promise. However, ethical and scientific issues have yet to be solved.

Data suggest that widespread oxidative damage occurs in the retina of patients with advanced geographic atrophy.⁵ Proteins and nutrient metabolites related to oxidative stress are also upregulated in retinas of patients with AMD.^6-7 Therefore, oxidative damage is thought to play a major role in advancement of RPE loss, either as a pathogenic factor or as evidence for a major deficiency in the oxidative defense system. Additional reports support a role for the immune system, including complement activation, whereby accumulation of extracellular plaques and deposits elicits a local chronic inflammatory response that exacerbates the effects of primary pathogenic stimuli, in the development of drusen and the pathogenesis of AMD.^8-10 Areas of bare Bruch's membrane are also found after submacular surgery for choroidal neovascularization performed in exudative or "wet" AMD.^11-14 Furthermore, damage to the RPE has been also observed after uncomplicated photodynamic therapy, an established treatment for choroidal neovascularization.¹⁵ Degenerative changes within the neurosensory retina and/or choriocapillaris occur simultaneously to the RPE atrophy.¹⁶ Fundus changes may predispose the eye to develop the neovascular/exudative stages of AMD and lead to irreversible severe vision loss.⁹ Although treatment options exist for wet AMD, eg, injection of ranibizumab providing visual improvement in 25% to 40% of patients with choroidal neovascularization,^17-18 none for dry AMD has been adapted.¹⁹ Thus, transplantation of RPE cells to fill in this defect before irreversible atrophy of the foveal photoreceptors occurs has been pursued. However, it has been shown that allogeneic RPE cells do not attach to senescent Bruch's membrane efficiently and do not undergo proliferation and spreading to fill in the defect.^20-21 Even under circumstances in which allogeneic RPEs do attach and proliferate on aged Bruch's membrane, their ability to survive long-term is compromised.²² Transplantation of autologous RPE cells has also been explored in man, but evidence of functional recovery has not been achieved.²³

Hereditary retinal degenerations, such as retinitis pigmentosa, are major causes of blindness in the Western world, with an incidence of 1 in 2000 individuals.²⁴ The progressive loss of vision is due to mutations in more than 100 identified genes and affects different cellular compartments in either the photoreceptor cells (PRCs) themselves or the underlying RPE.²⁵ The disease is characterized by the onset of night blindness, an early loss of the peripheral visual field, and the late loss of central vision.²⁶ Different therapeutic strategies have been developed, and in recent years, major advances were reported to rescue retinal degeneration in animal models, such as mutation-specific gene-based treatments (gene replacement/gene silencing),³ mutation-independent approaches such as the administration of neurotrophic factors (genetically expressed and subretinally injected), or administration of donor cells (PRCs, RPE cells, stem cells [SCs]).^27-29 At the same time, the search for a mutation-independent pharmacological treatment has been intensified, especially after reports of the neuroprotective potential of various growth factors (eg, glial-derived neurotrophic factor or ciliary neurotrophic factor) emerged.^30-31 The latter, a phase 1 trial in patients with retinitis pigmentosa, indicated that ciliary neurotrophic factor is safe for the human retina, even with severely compromised photoreceptors, and showed visual improvement in the study.

REGENERATIVE MEDICINE: A MEANS TO RESTORE FUNCTION

This review will focus mainly on the use of living cells for regenerative medicine, their sources in general, and the challenges associated with their use. Stem cells are of interest because of their plasticity and the capacity to self-renew as well as to give rise to specialized cell types. They remain uncommitted and self-renewable until they receive a signal(s) to develop into distinct cell types.³³ In addition, because SCs can proliferate indefinitely in their undifferentiated state, they are expected to alleviate the problem of the shortage of donor cells for cell replacement therapy.³⁴ The current challenges in SC-mediated regenerative therapy are how to maintain the stemness of the cells while promoting regeneration, identifying the optimal source of cells, controlling host vs donor alloreactivity if allogeneic cells are used, and promoting efficient function while preventing loss of control of regulation that could result in teratoma formation.³⁵

A different approach to replace degenerated cells would be the use of reprogrammed somatic cells. For this, several strategies, such as nuclear transplantation, cellular fusion, introduction of defined transcription factors, and culture-induced reprogramming have been used.³⁶ Thereby, conversion of differentiated cells into an embryonic state with an increase in potency has been induced. Nuclear reprogramming has special therapeutic potential as it can be used to create patient-specific cells.

Furthermore, combining SCs and progenitor cells with bioengineering to generate tissue equivalents in culture is another promising therapeutic strategy. Replacement of tissue lost to disease or trauma using transplants delivered on polymer scaffolds can be applied to improve survival and promote the organized differentiation of grafted cells. This might be especially important in models of retinal degeneration and injury to create the appropriate architecture for integration of transplanted retinal SCs and precursor cells. Results have shown that the scaffold may promote differentiation of retinal precursor cells and is associated with increased cell survival.³⁷

TYPES OF SCs

Embryonic SCs are not the only SC candidates for generation of differentiated cell types.⁴⁹ Other pluripotent cell types include SCs derived from the primordial germ cells of the gonadal ridge⁵⁰ and possibly cancer SCs, which recently have been identified in leukemia and several solid tumors.⁵¹ In the past decade, researchers have also defined committed SCs or progenitor cells from various tissues in both adult animals and humans.⁵² The so-called adult SC can give rise to cells of a particular tissue and include, for example, bone marrow (BM)–derived SCs (BMSCs), one of the earliest clinically used and most widely studied SC populations.⁵³ These multipotent SCs differentiate into cell types from the germ layer of their origin. However, it has been found that they are generally less effective than ESCs in generating other lineages (Figure 1A).

View larger version (52K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 1. Population of CD45– tissue-committed stem cells (TCSCs). A, Hierarchy of stem cell development and potential under physiological and culture conditions. B, Rare-events cell sorting is a useful tool to select candidate stem cell populations. The TCSCs are stem cell antigen 1 (Sca-1)+/CD45–/lineage– as shown. SSC, sideward scatter; FSC, forward scatter; lin FITC, lineage fluorescein isocyanate; PE, phycoerythrin; APC, allophycocyanin.

Today, it is widely agreed that several subsets of SCs can be found in the BM: (1) hematopoietic SCs, the source of all blood cells; (2) mesenchymal SCs, nonhematopoietic stromal cells that differentiate into mesenchymal tissues; and (3) tissue-committed SCs (TCSCs) that are already committed to neural, cardiac, and other lineages.⁵⁴ Figure 1B demonstrates the population of CD45^– TCSCs. Recent publications showing the existence of CD45^– TCSCs have led to renewed interest in adult BMSCs as a source for tissue repair in vivo.⁵⁵

Pluripotency and differentiation potential of BMSCs, even across germ layers, have been extensively examined in vitro and in vivo.^56-60 In the central nervous system, multipotential precursors for neurons, astrocytes, and oligodendrocytes have been found. These so-called neural SCs from both the fetal and adult central nervous system have proliferated in vitro and can differentiate to show morphological and electrophysiological features characteristic of neural cell types.^61-62 For cell replacement therapy in the ocular system, the isolation and characterization of retinal SCs (RSCs) in 2000 represented an important step forward. Tropepe and colleagues⁶³ reported the identification of a SC in the adult mouse eye that represents a possible substrate for retinal regeneration. Retinal SCs are localized in the pigmented ciliary margin and clonally proliferate in vitro to form spherical colonies of cells that can differentiate into retinal-specific cell types, including rod photoreceptors, bipolar neurons, and Müller glia. Other organs seem to also contain cell types with SC-like properties, including the heart⁶⁴ and islet.⁶⁵ Taken together, these results provide evidence that SCs exist within the adult mammalian organism and that there is an endogenous program to incorporate these cells into damaged structures. Although effective at maintaining existing tissues and organs, it appears that this endogenous physiological mechanism is not capable of repairing significant damage.

MECHANISMS OF REGENERATION

The mechanisms for SC-mediated differentiation events, including documented functional recovery, are still under considerable scientific debate. For adult SCs, the controversy between transdifferentiation and fusion has still to be solved. Recently, it was reported that BMSCs are able to "transdifferentiate" or change commitment into cells that express early heart, skeletal muscle, neural, or liver cell markers.^56-59,64 Similarly, SCs from the BM contributed to the regeneration of infarcted myocardium.^68-69 This was supported by the observations in humans that transplantation of SCs from mobilized peripheral blood expressing the early hematopoietic CD34⁺ antigen led to the appearance of donor-derived hepatocytes, epithelial cells, and neurons.⁷⁰ Therefore, it was initially presumed the repair seen in damaged host tissues following SC transplantation or homing was due to incorporation and transdifferentiation of the BMSCs at the sites of damage. However, a number of studies have challenged this concept, providing evidence that BMSCs may instead incorporate into host tissues via fusion with host cells.^71-72 One of the key studies demonstrating evidence for fusion was that of Alvarez-Dolado et al,⁷³ where they examined CD45⁺ BMSCs and found that essentially all integration of these cells into host tissues involved fusion. Because the CD45⁺ subset of BM-derived cells is now thought to be committed to the hematopoietic lineage (as opposed to CD45^– cells, which give rise to other nonhematopoietic lineages), one could argue that Alvarez-Dolado et al did not examine the appropriate subset of cells capable of contributing to nonhematopoietic tissues. Additionally, they only examined incorporation of CD45⁺ cells into normal tissues, as opposed to sites of damage, where the mechanism of integration may be different because of the inflammatory milieu. Nevertheless, other studies have also found evidence that SCs can fuse with host cells, forming heterokaryons.⁷¹ Such studies raise the question as to whether the transdifferentiation of BMSCs seen in culture really occurs when these cells are injected in vivo or whether their primary therapeutic contribution is instead via fusion with host cells. Other reports found no proof of host cell fusion when SCs differentiate in the host.⁷⁴ Further, there is evidence that a single BMSC can repopulate the BM, giving rise to all hematopoietic precursors, and that such cells could in turn contribute to other nonhematopoietic tissues. The question of fusion with host cells vs transdifferentiation in vivo is the topic of several comprehensive reviews and reports that conclude that definitive experiments on this topic are lacking.^75-76 Therefore, it remains unresolved as to whether BMSCs integrating into damaged tissues undergo transdifferentiation in vivo to express markers of differentiation or whether they express such markers as a result of fusion with endogenous host cells.

APPLICATIONS OF REGENERATIVE SC THERAPY IN DISORDERS OF RETINAL DEGENERATION

Transplantation of RSCs with the potential to generate new retinal cells provides an alternative approach to enable the replacement of lost PRCs or RPE. Retinal SCs may restore vision in patients who have degenerative retinal diseases by 2 possible means: (1) repopulation of the damaged retina (eg, PRCs) and/or (2) rescue of retinal neurons from further degeneration.⁸⁰ Different research groups have successfully isolated murine putative RSCs from the ciliary margin and human RSCs in the pars plana and pars plicata.^81-82 However, the transplantation of these cells in normal and degenerative rodent retina was only minimally successful because of the limited ability of the cells to invade and integrate into the host retina.²⁷ On the other hand, transplantation of immature postmitotic rod precursors from the developing retina (postnatal day 1) improves retinal integration.⁸³ MacLaren et al⁸³ showed that timely selection of the progenitors is pivotal for success. The optimal result occurs when selected cells were biochemically committed but not yet morphologically differentiated. The capability of subretinally or intravitreously injected RSCs to invade and integrate into the neural retina remains restricted to sites of retinal injury. Breakdown of physical barriers, such as the outer limiting membrane, and/or release of unknown neurotrophic factors are most likely required to stimulate RSC integration.⁸⁴ To date, only sparse data are available regarding factors that might stimulate migration, integration, and differentiation of RSCs into the neural retina. However, it is assumed that neurotrophic factors, such as transforming growth factor,⁸⁵ fibroblast growth factor,⁸⁶ or epidermal growth factor,^87-88 might play a role. Recent evidence has suggested that hepatocyte growth factor/scatter factor, a pleiotrophic factor with mitogenic and morphogenic activities, may also be involved in the development and maintenance of neurons and PRCs.⁸⁹

In AMD, the replacement of diseased RPE would be pivotal to protect or rescue the adjacent PRCs. Unfortunately, no convincing animal model for AMD exists to date. Therefore, the sodium iodate (NaIO₃) model of RPE damage, established by Korte et al⁹⁰ in 1984, has been used to study at least the repopulation of bare areas of normal Bruch's membrane.⁹¹ Briefly, the selective and patchy degeneration of the RPE monolayer after intravenous NaIO₃ injection is directly correlated with decreased visual function, decreased electrophysiological function, and anatomical cell loss in the RPE (and subsequently in the retina) (Figure 2). Furthermore, the extent of the RPE damage is time and concentration dependent, as we recently published.⁹² Interestingly, NaIO₃-damaged RPE cells express higher amounts of cytokine/growth factors involved in SC homing. After treatment with NaIO₃, murine RPE cells express higher levels of stromal-derived factor 1 (SDF-1), as well as other signaling factors (complement factor C3 and hepatocyte growth factor/scatter factor). Stromal-derived factor 1 is a chemokine whose receptor CXCR4 is expressed on BM-derived progenitor cells and SCs.⁹³ While there was no evident change in expression of vascular endothelial growth factor and RANTES (regulated on activation, normal T-cell expressed and secreted), there was increased expression of the cytokine leukocyte inhibitory factor, known to promote self-renewal in ESCs.⁹⁴ Furthermore, supernatants of NaIO₃-damaged RPE exert a priming effect on BMSC migration in vitro as they enhance their transwell migration.⁹⁴ These results provide evidence that damage to the RPE leads to production of soluble factors that can cause specific chemotaxis of BMSCs and raises the possibility of their recruitment to the site of damage. These data support the possibility of using BMSCs to replace damaged cells, especially RPE, in eyes with retinal degenerations. To investigate this further, we have undertaken endogenous as well as exogenous approaches using BMSCs using the earlier-described NaIO₃ model. Endogenous refers to existing BM cells in the host while exogenous refers to adoptively transferred cells.

View larger version (147K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 2. Sodium iodate model of retinal pigment epithelium (RPE) degeneration. A-F, Autofluorescence in flat-mount whole-eye preparations of control (D) and sodium iodate–treated mice (A-C, E, and F). The top row (A-C) compared different doses of sodium iodate at 7 days postinjection (PI): 35 mg/kg (A), 50 mg/kg (B), and 70 mg/kg (C) of body weight. E, B, and F compare different times PI at the same dose (50 mg/kg): 3 days PI (E); 7 days PI (B); and 21 days PI (F). Beginning on 3 days PI, a patchy loss of RPE can be detected by the decrease in autofluorescence (black areas). The total area bare of RPE (autofluorescent areas) is dose dependent and increased over time (original magnification x1000).

Two types of approaches can be used to promote SC-mediated regenerative repair of RPE: endogenous and exogenous. Endogenously, RPE injury combined with pharmacologically enhanced growth factor–mediated mobilization lead to migration of BM-derived cells into the subretinal space. Bone marrow–derived SCs (c-kit⁺), macrophages (F4/80), and leukocytes, such as granulocytes and monocytes (CD11b), could be identified. Thereby, the number of c-kit⁺ BMSCs in the eye after NaIO₃ injection and mobilization increased dramatically compared with the mobilized control mice who did not have RPE damage.⁹¹ The migrated BMSCs had incorporated in a monolayer along the RPE 4 weeks after transplantation and expressed the RPE markers RPE65 and microphthalmia-associated transcription factor (MITF) (Figure 3). These findings suggest that BMSCs are attracted to damaged RPE and are induced to differentiate into components of RPE. Mobilization enhances the outcome.

View larger version (70K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 3. Immunocytochemical staining of vertical sections of a green fluorescent protein (GFP) chimeric mouse eye 4 weeks after sodium iodate treatment and bone marrow cell (BMC) mobilization. The GFP+ BMCs were stained with anti-GFP (green fluorescence) and anti–microphthalmia-associated transcription factor (MITF) (red fluorescence). The merged Nomarski image shows a monolayer of MITF+ GFP+ BMCs (yellow) on Bruch's membrane. The monolayer of BMCs is adjacent to pigmented host RPE cells (arrows) outlined with white dashes.

These results demonstrated that a physiological process is in place in vivo to recruit SCs to the damaged RPE and that endogenous BM-derived cells are able to integrate into the damaged RPE and express markers of RPE differentiation. Nevertheless, the significant experimental damage to the RPE could not be repaired by this endogenous approach, nor does this endogenous program appear capable of repairing or preventing the progressive damage to the RPE that occurs in AMD and retinitis pigmentosa. Thus, it appears that such recruitment of endogenous cells may not be sufficient to physiologically repair significant damage to the RPE in the same fashion that recruitment of endogenous SCs cannot repair major damage to the spinal cord or heart.

To optimize the number and availability of circulating BMSCs, we then examined an exogenous approach for regeneration of damaged RPE. Additionally, this allowed us to define the precise cell types involved, using cell sorting as opposed to the mixture of SCs and other BM-derived cells mobilized into the periphery with the endogenous approach. We injected fluorescent-activated cell–sorted BMSCs with the phenotype lin^– (negative for all lineages of differentiated BM cells), SC antigen 1 (Sca-1)⁺ intravenously into NaIO₃–treated animals. The BMSCs could be detected in the subretinal space on Bruch's membrane in areas of RPE loss on day 4 after cell injection, whereas controls without NaIO₃ injection showed no BMSCs. The double staining for Sca-1 and green fluorescence protein confirms the BM origin of the cells systemically transferred and confirms that hematopoietic SCs home to the area of damaged RPE after NaIO₃ injection. One and 2 weeks after transfer, BMSCs could be identified in the subretinal space but they did not express RPE markers. Immunocytochemical staining showed the expression of RPE65 in BMSCs 4 and 6 weeks after transplantation. These results suggest that, as with the endogenous cells, BMSCs injected systemically into the host home to the site of damage, where they integrate and express markers of RPE differentiation in a time-dependent fashion.⁹⁵ One critical area of research is to define the optimum milieu in which to promote the potential for repair.

A third route for BMSC delivery is by direct subretinal injection. We found that subretinally injected BMSCs integrated into the RPE and expressed markers of differentiation (eg, RPE65). The optimal route for SC delivery remains to be determined. Concentrating the cells might provide a kinetic advantage for incorporation of the cells into the altered tissue. Thereby, the cells would not have to home to sites of damage from the circulation.

However, the use of SCs to replace degenerated RPE cells has not yet demonstrated the ability to rescue PRCs at risk of damage. If SC differentiation and reconstitution of the damaged RPE monolayer occur after photoreceptor degeneration, a rescue effect will not be possible. Alternatively, if the mobilization of endogenous SCs occurs continuously or over a prolonged period, photoreceptor damage and/or rescue may be possible.⁹⁶

The regenerative capability of BMSCs in the ocular system is not only restricted to RPE replacement. Chiou et al⁹⁷ showed that BMSCs have multilineage differentiation potential in vitro and differentiate into retinal cells and photoreceptor lineages after coculture with RPE cells. Other groups have followed different approaches to replace diseased RPE cells. Haruta and colleagues⁹⁸ harvested RPE-like ESCs in vitro and achieved functional improvement after subretinal transplantation into Royal College of Surgeons rats.

Because only a small percentage of total BM cells are chemoattracted to supernatants from damaged RPE in vitro, as well as into damaged RPE in vivo, the properties of this subset of BM-derived cells need to be considered. Recent data indicate that the CD45⁺ population of SCs is committed to hematopoietic lineages, while the CD45^– population is believed to remain pluripotent and thus capable of differentiation into various nonhematopoietic tissues.⁹⁹ Kucia et al⁵⁵ showed that CD45^– BMSCs are composed of subsets of cells already committed to skeletal muscle, heart muscle, liver, and neural tissues. These so-called TCSCs, more recently renamed very small embryonic-like (VSEL) cells,¹⁰⁰ express Oct4, an SC marker, in addition to markers of tissue-specific progenitors. These TCSCs are mobilized into peripheral blood during organ injury.¹⁰¹ Stromal-derived factor 1–based chemotactic isolation combined with real-time reverse transcriptase–polymerase chain reaction analysis of messenger RNA revealed that early TCSCs (1) reside in the normal human and murine BM; (2) express CXCR4 on their surface; and (3) can be highly enriched in humans and mice after chemotaxis to a SDF-1 gradient. These studies were performed on freshly isolated cells, ruling out the potential contribution of culture-related transdifferentiated hematopoietic SCs or mesenchymal cells. In our experiments, we found that Sca-1⁺ CD45^– BMSCs are highly enriched in messenger RNA for retinal/RPE progenitors (Six-3, OTX, Pax-6, MITF; data not shown) and, furthermore, that this is the subset of BMSCs that has migrated in response to supernatants from damaged RPE in transwell assays. Thus, it appears that RPE-committed VSEL cells (approximately 0.05% of the population) are present within the Sca-1⁺ CD45^– subset of BMSCs. This is supported by data from in vitro experiments using a coculture of BMSCs and RPE cells to trigger SC differentiation into the RPE lineage (Figure 4).⁹¹ The BMSCs changed their morphology from round to epithelial-like and expressed the epithelial markers cytokeratin; MITF, expressed on common progenitors of retina and RPE and persisting expression following RPE differentiation (its expression diminishes in cells that progress along a retina lineage); and the RPE-specific marker RPE65 after 2 weeks (Figure 5). The process required direct cell-cell contact between BMSCs and RPE. No staining for RPE markers was detected when a membrane separated both populations of cells. This was a specific effect, as no positive staining was detected when RPE cells were replaced with fibroblasts.⁹¹

View larger version (104K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 4. Coculture with retinal pigment epithelial (RPE) cells for 2 weeks leads to the expression of RPE-specific markers on sorted stem cell antigen 1 (Sca-1)+ bone marrow–derived stem cells (BMSCs). A and B, Green fluorescent protein (GFP)+ BMSCs are stained with anti-RPE65 (Cy3, red) (A) and with anti-GFP (green) (B). C, Cell nuclei are shown in blue (4,6-diamidino-2-phenylindole). Arrows indicate similar positions in the panels. D, A merge of the top 2 panels is shown. Overlapping expression of red and green indicates BMSCs that express RPE markers; red cells, which are not green, correspond to RPE cells in the coculture.

View larger version (62K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 5. Cross section of a mouse eye 6 weeks after sodium iodate injection and intravenous transplantation of enhanced green fluorescent protein (GFP)+ bone marrow–derived stem cells (BMSCs). The donor cells were stained with anti-GFP and anti-RPE65. The panel shows a merged image combined with Nomarski optics to visualize the location in the subretinal space (scale bar = 5 µm). There is a cluster of cells with yellow fluorescence representing transplanted BMSCs (arrow) that indicates colocalization of GFP (green fluorescence) and RPE65 (red fluorescence). The remaining host retinal pigment epithelium (asterisks) has been damaged by sodium iodate treatment, as well as the sensory retina.

Finally, degenerations in the mammalian retina, initiated by defects in photoreceptors or RPE, often leave the neural retina deafferented. It responds to this challenge by remodeling, first by subtle changes in neuronal structure and later by large-scale reorganization, and represents the invocation of mechanisms resembling developmental and central nervous system plasticity. This neuronal remodeling and the formation of a glial seal may abrogate many cellular and bionic rescue strategies. On the other hand, survivor neurons appear to be stable, healthy, active cells, and given the evidence of their reactivity to deafferentation, it may be possible to influence their emergent rewiring and migration habits.¹⁰²

THE FUTURE OF REGENERATIVE MEDICINE

Currently, a number of experimental SC treatments are being evaluated, including research in cancer, Parkinson disease, cardiac disease, Huntington disease, multiple sclerosis, and acute spinal cord injury.^108-109 Although a number of challenges must still be addressed, the potential impact of SC-based regenerative medicine holds great promise.

AUTHOR INFORMATION

Submitted for Publication: November 3, 2008; final revision received January 26, 2009; accepted February 3, 2009.

Financial Disclosure: None reported.

Funding/Support: This work was supported in part by National Institutes of Health grants R01 DK069766 and 5RO1 HL063442; Juvenile Diabetes Research Foundation grants 1-2005-1037 and 1-2006-1466; The Department of the Navy, Office of Naval Research; The Department of the Army, Office of Army Research (this publication was made possible by award W81XWH-07-1-0185 from the Office of Army Research); the National Foundation to Support Cell Transplant Research; the Commonwealth of Kentucky Research Challenge Trust Fund; the W. M. Keck Foundation; The Jewish Hospital Foundation; and Research to Prevent Blindness.

Disclaimer: Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the Office of Army Research.

Additional Contributions: Luisa M. Franco, MD, and Yang Li, MD, provided technical assistance, and Carolyn DeLautre prepared the manuscript. We thank the staff of the animal facility for outstanding animal care.