Chapter 125 Transplantation Frontiers
Introduction
In 1959, Paris Royo and Wilbur Quay reported experiments that were among the earliest steps taken in intraocular retinal cell transplantation.1 Using a paradigm originally developed by Van Dooremaal,2 Royo and Quay transplanted rat fetal retina into the anterior chamber of maternal rat eyes. Prolonged graft survival and a normal rate of graft development and differentiation were noted. Expanding on this work, del Cerro and his colleagues3 transplanted embryonic rat retina into the anterior chamber of adult eyes from various rat strains with similar results. After Gouras and coworkers4 successfully transplanted cultured human retinal pigment epithelium (RPE) cells to monkey Bruch’s membrane from which the retina had been removed, Li and Turner, in 1988, reported prevention of photoreceptor cell degeneration by RPE transplants in the Royal College of Surgeons (RCS) rat, an animal with an inherited defect causing impaired RPE phagocytosis of outer segments and progressive RPE and photoreceptor death.5,6 Li and Turner’s work, which has been replicated by many investigators, was the first successful treatment of retinal degeneration with cellular transplantation.
In principle, successful transplantation requires graft survival and integration with the host. Graft survival depends on a number of factors, both immunological and nonimmunological, that will be discussed in this chapter. Regarding graft–host integration, RPE cell transplants integrate readily with host photoreceptors through the extension of apical villous RPE processes around photoreceptor outer segments. In contrast, synapse formation between retinal grafts and host retina is much more complex. Experiments in the central nervous system, however, have established the possibility that synapse formation between donor and host neural tissue can occur. For example, Lund and others7,8 demonstrated that embryonic retina transplanted into neonatal and adult rat tectum develops the proper layering of the normal retina and extends neuronal projections only to the normal retinorecipient structures, i.e., the superior colliculus, the pretectum, and the dorsal lateral geniculate nucleus. Similarly, intraocular retinal transplants establish with their targets functional connections that elicit light-driven visual responses9 and mediate visual behaviors.10,11 These and other studies12,13 have stimulated interest in transplantation as a treatment for retinal disease. The background and rationale for transplantation, results of transplants in experimental animals and humans, immune response to transplanted tissue, non-immunological causes of graft failure, and future directions for development of transplantation for RPE and photoreceptor grafts will be considered in this chapter.
Background and rationale for RPE transplantation in age-related macular degeneration
Late complications of age-related macular degeneration (AMD) are major causes of irreversible loss of central vision among the elderly.14,15 Severe visual loss in AMD occurs due to growth of abnormal blood vessels (choroidal new vessels, CNVs) under the RPE and retina with secondary exudative retinal detachment, subretinal hemorrhage and lipid exudation, and outer retinal degeneration.16 It can also result from atrophy of RPE with consequent atrophy of overlying photoreceptors and underlying choriocapillaris (geographic atrophy, GA). The prevalence of AMD-associated CNV or GA is approximately 1.47%, increasing dramatically to 10% in people older than 80 years.17 There is no fully effective therapy for either CNVs or GA. Blocking the effects of vascular endothelial growth factor (VEGF) by frequent intravitreal injection of antibodies is the best currently available treatment for CNVs, but only 30–40% of the patients experience a moderate visual improvement.18,19 Cell-based therapy, which involves placing cells in the subretinal space, potentially offers a better option compared to pharmacological monotherapy with anti-VEGF agents. Advantages of such an approach would include long-term therapy without frequent minor surgical injections with medications; since cells like RPE produce numerous factors that help maintain the normal retinal and choroidal anatomy and physiology,18,20–24 transplantation of such cells could provide a richer and more effective therapy that is less susceptible to treatment resistance;25 cell transplantation could potentially restore the subretinal anatomy that is altered by growth (and surgical removal) of CNVs or by atrophy of RPE. Another indication for transplantation of RPE in AMD might be in patients who develop RPE tears,26 while receiving anti-VEGF therapy,27 which have a poor visual prognosis. RPE transplantation in patients with exudative macular degeneration might require prior removal of CNVs, which would result in loss of adjacent native RPE and the subjacent RPE basement membrane,28,29 causing a more complex graft bed than normal. On the other hand, in GA, transplantation of cells would involve resurfacing of the atrophic areas with less surgical manipulation with the goal of restoring the normal structure and function of the photoreceptor–RPE–choriocapillaris complex.
A number of pre-clinical experiments in animal models have demonstrated that transplantation of cells, e.g., RPE or stem cells or cells secreting growth factors, into the subretinal space can prevent photoreceptor degeneration and preserve function.30–36 The RCS rat is used commonly for these studies. A recessive mutation in merTK tyrosine kinase receptor in RPE6 causes a failure of outer segment phagocytosis by the RPE. Consequently, debris accumulates in the subretinal space, and there is eventual loss of photoreceptors causing blindness. Transplantation of RPE into the subretinal space of RCS rats causes a decrease in the amount of subretinal debris,32 prevents loss of photoreceptors,5 maintains synaptic connections of rescued photoreceptors,37 maintains electroretinography responses in areas of rescued retina,38 preserves pupillary light reflexes,39 cortical visual functions,30,33 and visual acuity.33,40,41
Results of macular translocation surgery may mean that establishment of a healthy subfoveal RPE monolayer in AMD eyes can restore macular function. In macular translocation, the retina is moved so that the fovea is relocated to an area away from submacular disease. This maneuver can result in improved visual acuity.42,43 However, after translocation, the fovea rests on previously extrafoveal RPE, Bruch’s membrane, and choriocapillaris. Thus, the procedure is not completely analogous to an RPE transplant per se.
Results of RPE transplants in humans
The first human RPE transplant in AMD was reported by Peyman and coworkers in 1991.44 Since then, numerous reports have described the results of allogeneic and autologous RPE transplantation.45–54 These surgeries have involved removal of CNV and placement of suspended allogeneic RPE cells,44 autologous RPE cells,48,54 patches of RPE,46,55 or autografts of RPE-choroid.48,51,52,54 Thus far, RPE transplantation has not been successful in the majority of patients undergoing surgery although a few patients show improvement of two or more lines of visual acuity following autografts.48,50,51,53,54 Allogeneic RPE transplants in AMD patients that have undergone CNV excision have failed with poor visual outcome and, in patients who are not immune-suppressed, subretinal fibrosis and chronic fluid leakage in the dissection bed.45,46,56 In autologous transplants, graft failure has resulted from intra- and postoperative subretinal hemorrhage, failure of graft revascularization, fibrous encapsulation of the graft, and development of epiretinal membrane, proliferative vitreoretinopathy, and retinal detachment.50,51,53 Optical coherence tomography studies indicate the presence of outer retinal atrophy in the majority of patients undergoing autologous RPE transplants, whether as cell suspensions or as sheets of RPE-Bruch’s membrane-choroid.54,57 Immune rejection and graft failure may underlie these results. The allogeneic RPE transplants may not have survived in the subretinal space independent of immune rejection. Tezel and coworkers have shown that if RPE cells cannot adhere to their basement membrane (or comparable surface) within 24 hours, they undergo apoptosis.58 All previous demonstrations of successful RPE transplants in laboratory animals have involved transplantation onto normal Bruch’s membrane (Fig. 125.1) or onto native RPE.4,5,31,32,59–62 In AMD, Bruch’s membrane is itself abnormal.63–66 Uncultured RPE isolated from adult human donor eyes show very limited adherence to aged submacular human Bruch’s membrane in vitro.67 In organ culture experiments, even cultured fetal RPE, which exhibit robust attachment to Bruch’s membrane, cannot survive for more than 1–2 weeks on aged or AMD submacular human Bruch’s membrane,22,68–70 presumably due to age- and AMD-related changes in this substrate. Histopathology of an immune-suppressed patient that underwent CNV excision plus uncultured adult RPE transplantation indicates that the cells were not organized as a monolayer, and there was photoreceptor atrophy over the transplant.55 These results are consistent with the observations of RPE behavior on human submacular Bruch’s membrane in organ culture.
Immune response to RPE transplants
Immune privileged sites and immune privileged tissue
Immune privilege in the eye is the result of multiple anatomic and physiologic factors acting on both the innate and adaptive immune systems.71,72 Streilein et al.72 noted that among the factors and mechanisms responsible for ocular immune privilege are: (1) the blood–ocular barrier, which minimizes contact of allografts with the cells and molecules of the systemic immune system, thus blunting the immune response to alloantigens; (2) deficient lymphatic drainage of the eye, which leads to initial alloantigen presentation in the spleen (versus regional lymph nodes) and, therefore, relatively less inflammatory immune responses; (3) an unusual distribution and functional properties of bone-marrow-derived antigen presenting cells; and (4) an ocular microenvironment rich in soluble or cell membrane-associated immuno-modulatory factors. Examples of such factors include transforming growth factor (TGF)-β2, α-melanocyte stimulating hormone, vasoactive intestinal peptide, calcitonin gene-related peptide, macrophage migration inhibitory factor, interleukin (IL)-1 receptor antagonist, Fas ligand, CD46, CD59, and free cortisol.73–80 These molecules help to create an immune-suppressive microenvironment by affecting the function of immune cells that come in contact with transplanted tissue or inhibiting complement activation. Ocular sympathetic innervation is necessary for maintenance of high ocular levels of TGF-β, and denervation leads to loss of immune privilege.81
Ocular immune privilege is a dynamic state of immune regulation that results in the induction of an antigen-specific “deviant” systemic immune response characterized by: (1) active downregulation of donor-specific delayed hypersensitivity reactions; (2) activation of regulatory cytotoxic CD8+ T-cells that suppress delayed hypersensitivity as well as suppress B-cells producing complement-fixing antibodies; (3) activation of regulatory CD4+ cells that block terminal differentiation of delayed hypersensitivity effector cells; and (4) enhancement of B-cells that produce non-complement-fixing antibodies. Initial descriptions of this response were based on experiments in which alloantigens were placed in the ocular anterior chamber, which led to the acronym, ACAID, anterior chamber-associated immune deviation.73,74,82,83
Immune privilege exists in the subretinal space, but it is not absolute. Subretinal allografts resist immune rejection and induce the development of systemic tolerance in the form of a suppressed antigen-specific delayed hypersensitivity reaction.83,84 RPE cells may create a local immune suppressive microenvironment through the production of immunomodulatory factors such as TGF-β85–87 and through suppression of T-cell activation88,89 and the induction of activated-T-cell apoptosis.90,91 In addition, RPE expresses ligands for T-cell inhibitory receptor programmed cell death-1 (PD-1) receptors on T-cells. These ligands (PDL1 and PDL2), whose expression is upregulated by interferon-γ, bind to PD-1 and suppress T-cells.92,93 RPE also expresses a novel immunosuppressive factor, cathepsin L inhibitor, CTLA-2α, that converts exposed T-cells to regulatory T-cells.94
The RPE can behave as immune privileged tissue. Neonatal RPE sheets, e.g., resist immune rejection at heterotopic sites (i.e., anatomic locations in which the transplanted tissue is not found normally) when compared to non-privileged tissue.95 It is not known whether adult RPE, as a sheet or single cell suspension, is immune privileged. Constitutive expression of Fas ligand on the RPE may contribute to its status as immune privileged tissue.95,96 This status does not mean, however, that the RPE is incapable of sensitizing its host. Allogeneic neonatal RPE grafts can sensitize recipient mice if the grafts are placed at a nonimmune privileged site.95
RPE cells express transplantation antigens. Low levels of major histocompatibility (MHC) class I antigens are present on the surface of fetal RPE cells.97 Also, RPE cells may express minor histocompatibility antigens. They do not seem to express MHC class II antigens normally although they can do so if exposed to interferon (IFN)-γ.98 Culturing RPE cells also can induce MHC antigen expression.99 In addition, RPE cells can process antigen prior to antigen presentation.100 Higher levels of IFN-γ-induced MHC class II antigen expression on RPE cells leads to increased activation of antigen-specific T-cells as manifested by the production of pro-inflammatory tumor necrosis factor-α.101 This activation does not include T-cell proliferation or production of IL-2, however, as is the norm for T-cell activation.
Implantation of neonatal allogeneic RPE cells into ocular immune privileged sites (e.g., the subretinal space) suppresses typical delayed hypersensitivity in the recipients, and the capacity to do so appears to be related to the immune privileged status of the graft site rather than to the immune privileged status of the graft.95 Thus, it may be that if RPE transplantation is done in a setting in which the immune privilege of the recipient site is compromised (e.g., breakdown of the blood–ocular barrier due to removal of native RPE), then allogeneic RPE grafts will be rejected despite their immune privileged status. If native RPE cells are compromised with sodium iodate, for example, tumor cells or ovalbumin injected into the subretinal space do not induce immune deviation.83 Immune deviation appears to be impaired persistently (i.e., after restoration of the blood–retinal barrier is complete) following iodate treatment and may be due to entry of plasma proteins into the subretinal space.83
In patients receiving RPE transplants after CNV excision, the outer blood–retinal barrier will be compromised both by the presence of CNVs preoperatively and by the iatrogenic RPE defect in the transplant bed postoperatively. Experimental studies indicate that the outer blood–retinal barrier is disrupted for approximately 1 week after hydraulic debridement of the native RPE.102,103 Experiments in mice indicate that breakdown of the outer blood–retinal barrier for approximately 14 days (beginning a few days before antigen inoculation) abolished immune privilege and abrogated immune deviation to cell-associated and soluble antigens in the subretinal space.83 It should be noted that placement of alloantigen in the eye in ocular immune privilege experiments involves disruption of the blood–ocular barrier. That immune privilege is extended despite this disruption indicates the importance of the factors in the microenvironment. Persistent loss of the blood–retinal barrier (e.g., due to CNVs) may result in dilution or absence of immunomodulatory factors in the microenvironment in addition to increased access to subretinal space by systemic immune cells. Also, the role of age-related changes in the RPE vis-à-vis subretinal immune privilege is unexplored. Finally, the extent to which immunological abnormalities associated with AMD,104 even in the absence of CNVs, compromise the normal immune suppressive environment of the subretinal space is unknown.
Are RPE transplants rejected?
It is not clear that orthotopic (i.e., transplants placed in an anatomic location in which they are found normally) allogeneic RPE transplants are rejected. Some clinical and experimental evidence indicates that they are rejected.45,46,56,62,105–109 In contrast, other studies indicate that RPE transplants may not be rejected.59,110 In still other studies, it is not clear whether rejection occurred or not.111 Interpretation of most of these studies, however, is complicated by the fact that it was difficult to identify unambiguously the transplanted RPE cells due to lack of a unique histological or clinical marker.
In a pilot study, transplanted allogeneic uncultured adult human RPE did not appear to be rejected in AMD patients who had undergone CNV excision and were treated with systemic azathioprine, prednisone, and cyclosporine, which suggests that immune suppression might prevent RPE transplant rejection in this setting.55 However, these elderly patients were not able to tolerate systemic triple immune suppressive therapy for an extended time. Local immune suppression is somewhat effective in laboratory experiments involving intravitreal injection of cyclosporine,112 but it has not been reported in human RPE transplants.
Autologous transplants (e.g., transplants of uncultured peripheral RPE cells to the submacular space of the same eye) should not undergo graft rejection. Methods for harvesting RPE cells for autologous transplants exist,50,113 but in vitro data indicate that harvested aged human RPE do not adhere well to aged submacular Bruch’s membrane, even in the presence of native RPE basement membrane.67 This result may explain why patients undergoing autologous RPE transplants after CNV excision have generally shown very limited visual improvement.50,54 Also, aged autologous adult RPE may exhibit aging and AMD changes that render them unsuitable for transplantation.114,115 Use of allogeneic cultured RPE could be considered116,117 if immune suppression is either unnecessary or, if necessary, it can be achieved.118 Attempts at xenograft transplantation, even using triple immune suppression, have been unsuccessful thus far in preclinical models.119
RPE graft failure
As noted above, RPE transplants in humans may have failed independent of immune rejection of the graft. Attachment of RPE to Bruch’s membrane after transplantation is crucial as RPE undergo apoptosis otherwise.120 In vitro studies indicate that trypsin-harvested, cultured fetal human RPE can adhere to adult submacular Bruch’s membrane even on surfaces lacking native RPE basement membrane.22,68,69,121 In contrast, uncultured collagenase IV-harvested adult RPE adhere poorly to aged human submacular Bruch’s membrane even in the presence of native RPE basement membrane.67 Del Priore and coworkers have shown higher attachment rates with cultured RPE from older donors when seeded onto peripheral Bruch’s membrane of older persons.122–125 This higher attachment rate might reflect a difference in integrin expression of primary isolated RPE cells versus cultured cells.68,126 It also might reflect differences in extracellular matrix composition of peripheral versus submacular Bruch’s membrane.63,127 Histology of excised CNVs,28,29,128 histopathology of eyes following CNV excision,129,130 and postoperative clinical findings28,131 all suggest that CNV dissection exposes both the superficial and deeper layers of the inner collagenous layer of Bruch’s membrane, which will constitute much of the surface to which transplanted RPE must adhere and on which they must survive if transplants are done after CNV excision. Cultured fetal human RPE can adhere to the superficial and deep inner collagenous layers of aged human submacular Bruch’s membrane, but long-term survival is impaired.22,69,126 An additional age-related change in Bruch’s membrane that could contribute to poor graft survival and function is the increased deposition of cholesterol.132 It is not known if the lipid deposits, which may contribute to the decreased hydraulic conductivity of aged Bruch’s membrane,127 might affect RPE transplant function or survival by masking extracellular matrix ligands. In vitro studies show that fetal RPE can attach to a high degree on the surface of the inner collagenous layer and that the majority of cell death occurs after 7 days in culture, indicating that poor cell survival is a post-attachment event.22 The poor visual results associated with autologous iris pigment epithelium (IPE) transplants might be due to graft failure and, perhaps less likely, from limitations in the ability of IPE to replace RPE.133–135 Stem cell precursors of RPE are an attractive option for transplantation (see below). They are being evaluated for their ability to attach and resurface Bruch’s membrane. One study found that human embryonic stem-cell-derived RPE cells have the potential to resurface aged Bruch’s membrane to a similar extent as cultured fetal RPE, but analysis of protein secretion indicates that these two cells may not behave identically.22
RPE replacement: future directions
Immune rejection
Many different strategies can be pursued to prevent graft rejection, including systemic or local drug therapy with established (e.g., prednisone, azathioprine, cyclosporine) or new (e.g., recombinant cytokines) compounds or via other techniques, e.g., diminishing MHC class-II-positive cells in the transplant.136 Cyclosporine does not appear to interfere with the ability of allogeneic neonatal retinal grafts to induce ACAID, nor does prolonged treatment with systemic cyclosporine interfere with persistence of allospecific suppressor cells for 35 days after transplantation.137 Nonetheless, retina grafts in the anterior chamber do deteriorate in 35-day grafts suggesting that graft destruction is the result either of immune mechanisms not inhibited by cyclosporine and/or non-immunologic factors. Independent of its effectiveness, systemic immune suppression can be complicated by side-effects such as nephrotoxicity, hypertension, risk of malignancy, susceptibility to infection, hepatotoxicity, seizures, and even anaphylaxis, depending on the medications used.138 Thus, local immune suppression has been considered. In one study, local immune suppression of rabbit eyes with cyclosporine prolonged cultured human fetal RPE xenograft survival in the subretinal space, but by 25 weeks, only ~10% of the green fluorescence protein-labeled grafted cells could be identified.112 It may be that the viral vector used to introduce green fluorescence protein induced an inflammatory response. Sustained slow-release cyclosporine delivery was as effective at promoting transplant survival as repeated intravitreal injections of much higher doses. Thus, it is not clear that intraocular immune suppression with cyclosporine alone will prevent allogeneic RPE transplant rejection. High-dose dexamethasone therapy is known to be effective in preventing acute allograft rejection139,140 although lymphocyte resistance to prednisolone can develop.141 Repeated intravitreous injections of dexamethasone can be well tolerated by the retina,142 which leads us to believe that sustained intravitreal dexamethasone delivery may not be complicated by retinal damage. Use of the sustained dexamethasone delivery system has prevented allogeneic corneal transplant rejection in rats.143 Currently, there are several intravitreal sustained-release steroid formulations that could be considered in human transplants.144,145 Thus, if immune suppression of RPE grafts is necessary, this approach may also be effective in preventing RPE transplant rejection.
Transplanted RPE survival and differentiation on aged Bruch’s membrane
There is no fully validated animal model of AMD at present. In vivo and in vitro RPE wound healing and transplantation models exist, but they do not appear to be directly relevant to AMD patients undergoing CNV excision. Previously reported successful in vivo RPE transplants involve attachment to normal Bruch’s membrane or native RPE as noted above. Experiments using aged submacular human Bruch’s membrane in organ culture indicate that aged human RPE do not adhere well to this surface unless they are cultured.67,68 Cultured fetal RPE appear to adhere to aged submacular Bruch’s membrane better than cultured aged RPE, but even in this case the cells do not appear to survive and differentiate well at 7 days in organ culture. It appears that aging changes in Bruch’s membrane are responsible for this deleterious effect on the cells.70,146,147 Thus, one approach to improving RPE transplantation success is to identify and manage these Bruch’s membrane changes. Experiments in vitro have included “cleaning” Bruch’s membrane with Triton X-100 (a detergent) and/or resurfacing the Bruch’s membrane with extracellular matrix molecules, singly or in combination148,149 or resurfacing the Bruch’s membrane with a cell-secreted matrix.70 The latter has shown promise, but translation to clinical application remains a challenge. Another approach has included enhancing the expression of cell-matrix adhesion molecules, i.e., integrins, to improve RPE adhesion and survival.126,150 Finally, an artificial scaffold could be used to transplant RPE. Such a scaffold could provide an environment that would allow easy placement of an RPE graft in the right orientation and/or promote RPE survival and differentiation.151,152 The scaffold might even prevent the age-related changes in Bruch’s membrane from damaging the transplanted RPE cells. An ideal scaffold will be biocompatible, biodegradable or capable of integrating, and should not interfere with membrane transport or permeability. Various polymers like polymethylmethacrylate, polyL-lactide, polyglycolic acid, expanded polytetrafluoroethylene, and others have been used with RPE, stem cells, and Schwann cells.153–155
Native RPE resurfacing of aged Bruch’s membrane
An alternative to RPE transplantation is to stimulate RPE ingrowth from the edge of the dissection bed. Human pathological studies indicate that RPE ingrowth following CNV excision in AMD patients is incomplete and aberrant.28,129,156 Previously described cell culture RPE wound-healing models are associated with complete wound resurfacing, in contrast to the situation in AMD patients undergoing CNV excision.157–159 In vivo RPE wound-healing models (e.g., rabbit, pig) are associated with complete resurfacing of the RPE defect160,161 or involve administration of compounds (e.g., mitomycin C) that complicate further experimental study or have unclear relevance to resurfacing Bruch’s membrane in AMD patients.102,162,163 Experiments from a Bruch’s membrane organ culture model indicate that resurfacing of ~65–85% of a ~3-mm-diameter circular RPE defect in aged submacular human Bruch’s membrane occurs by approximately 10 days if RPE basement membrane or the superficial inner collagenous layer is present. If deeper defects are created, exposing deeper regions of the inner collagenous layer, significantly less resurfacing is observed (~54%) at day 10 indicating that one should expect less resurfacing with greater damage to Bruch’s membrane.164,165 Additional experiments are needed to determine whether one can alter the denuded Bruch’s membrane surface (or the RPE cells at the edge of the defect) so that resurfacing of the RPE defect occurs in situ.
Human clinical studies indicate that periods of macular detachment up to 2 weeks are compatible with recovery of visual acuity of 20/50 or better in a substantial number of patients.166 Monkey and cat experiments indicate that many photoreceptors survive during retinal detachment periods of several weeks duration, although some photoreceptors definitely die.167,168 Approximately 80% of the cat outer nuclear layer survives during 3 days of detachment.169 The numbers of photoreceptor nuclei in detached cat retina do not begin to decline significantly (i.e., >20% decline in density) until detachment periods longer than 13 days.170 Data from cat retinal detachment studies indicate that 14-day detachments followed by 30-day reattachment is associated with rod and cone outer segment length similar to that observed after 5-day detachments.168 The cat retina, however, is rod dominated. Cones are more prone to apoptosis with detachment (versus rods)171 although in one study, approximately 75% of S- and M-cones survived a 1-day detachment (followed by retinal reattachment) in the ground squirrel.172 Thus, at this time, published experimental data do not indicate clearly what the exact survival of cones is after 2-week periods of retinal detachment, but clinical data indicate that it is good enough to warrant developing methods to promote RPE resurfacing. In addition to duration of detachment, the height of detachment influences photoreceptor survival.172,173 The macular detachments that would arise from CNV excision are quite shallow (<1–2 mm height), which also favors photoreceptor survival during a 2-week RPE resurfacing period. Thus, we believe that the RPE resurfacing approach is feasible despite its possible association with photoreceptor death during the resurfacing process.
Alternate source of rpe: stem cells
Stem cells are undifferentiated cells that have the capacity for self-renewal (i.e., can divide indefinitely to produce more karyotypically stable stem cells) and pluripotency (i.e., produce cells that are destined to differentiate into more than one type of cell).174–176 (See Chapter 35, Stem cells and cellular therapy.) During the course of differentiation, stem cells form intermediate populations of increasingly committed progenitor cells that have a decreasing proliferative capacity. Embryonic stem cells (ESCs) are derived from the inner cell mass of the blastocyst and are capable of differentiating into cells of all three germ layers.177,178 Embryonic and fetal tissues (e.g., retina) contain cells that vary in their differentiation status, ranging from progenitor cells, that can proliferate but have a restricted differentiation potential compared to embryonic stem cells, to committed but still immature rods and cones isolated from fetal retina. In adults, multipotent stem cells exist that can differentiate to replace lost or injured cells. Such cells have been isolated from various organs,179,180 including the eye.181,182 Stem cells isolated from a particular tissue can be induced to differentiate into cells of an unrelated tissue. Bone marrow cells, for example, can undergo metaplasia to form skeletal muscle;183 neural stem cells can develop into muscle184 (Table 125.1). It seems that stem cells exhibit plastic behavior depending upon their environment and signals from damaged tissues.185 The central nervous system contains neural progenitor cells. They exist in both embryonic and adult tissues and can differentiate into glial and neuronal lineages. The capacities for self-renewal and pluripotency render stem cells candidates to replace lost or injured cells. In summary, potential sources for replacement of RPE include non-eye-derived embryonic stem cells from the blastocyst, bone-marrow-derived stem cells, neural progenitor cells, and eye-derived progenitor cells.
Table 125.1 Potential sources of cells for retinal cell replacement*
| Cell type | Potential developmental capacity |
|---|---|
| Totipotent stem cell (mammalian zygote, blastomeres) | Can develop into cells derived from all lineages, i.e., ectoderm, mesoderm, and endoderm, as well as supporting tissues like placenta |
| Pluripotent stem cell (e.g., embryonic stem cell from inner cell mass) | Can develop into cells derived from all three germ layers (ectoderm, mesoderm, endoderm) |
| Multipotent stem cell (e.g., retinal progenitor cell) | Can develop into different cell types derived from one lineage |
| Reprogrammed cell (induced pluripotent cell) | Somatic cell reprogrammed by nuclear transfer, limited transfer of transcription factors, or cell fusion to enhance potency |
| Immature post mitotic rod precursor cell | Rod photoreceptors |
| Neural retina and retinal pigment epithelium | Rods, cones, Müller cells, or committed retinal neurons |
* Modified from Jaenisch R, Young R. Stem cells, the molecular circuitry of pluripotency and nuclear reprogramming. Cell 2008;132:567–82 and Zarbin MA. Retinal pigment epithelium-retina transplantation for retinal degenerative disease. Am J Ophthalmol 2008;146:151–3.
During development, the various retinal cells are derived from a common population of multipotent retinal progenitor cells (RPCs).186 Embryonic retina can be a potential source of such cells. Such cells exist in adults in lower species like fish and amphibians in the ciliary margin zone (CMZ), located in the retinal periphery, surrounded by a pigmented ciliary margin.187–189 These cells can regenerate the neural retina throughout the life of the organism.190 Progenitor cells from chick CMZ can regenerate retina if there is ectopic expression of fibroblastic growth factor and sonic hedgehog.191 More importantly, retinal progenitor cells have been isolated from mouse eyes and from fetal and adult human eyes.182,192–195 Efforts to induce RPCs to differentiate into RPE continue,196,197 but there are no pure, fully characterized RPE cells from RPCs yet.
Ethical considerations preclude the use of stem cells from human embryos although methods have been developed to establish stem cell lines without destruction of the embryo.198,199 It is possible to generate embryonic stem cells by somatic nuclear transfer-DNA transfer from an adult somatic cell into an oocyte, whose nucleus has been removed.200,201 The oocyte would then reprogram the adult DNA and develop in a normal embryonic pattern and form embryonic stem cells. Although the mitochondrial DNA will be derived from the oocyte, the chromosomal DNA would be from the host. Thus, a perfectly matched graft RPE could be prepared. However, ethical considerations are still an issue since it involves human life form. Alternatively, adult fibroblasts can be manipulated by transferring into these cells transcription factors like Oct4, Sox2, Klf4, and c-Myc that can reactivate developmentally regulated genes, thus inducing the cell to become an induced pluripotent stem cell (iPSCs).202,203 Nuclear transfer may be more effective at establishing the ground state of pluripotency than factor-based reprogramming, which can leave an epigenetic memory of the tissue of origin that may influence efforts at directed differentiation for applications in disease modeling or treatment.204 Although iPSCs are pluripotent stem cells, iPSCs and ESCs do differ in some important ways. iPSCs have the theoretical advantage of not being rejected by the patient from whom they are derived (versus ESCs, unless the ESCs were harvested from the patient as an embryo), but abnormal gene expression in some cells differentiated from iPSCs (both via a retroviral and episomal approach) can induce a T-cell-dependent immune response in a syngeneic recipient.205 This response is likely due to the abnormal expression of antigens (e.g., Zg16, Hormad1) not expressed during normal development or differentiation of ESCs, leading to loss of tolerance.205 Expression of these antigens is a reflection of epigenetic differences (e.g., DNA methylation) between iPSCs and ESCs.206,207 Continuous passaging of iPSCs may help attenuate these differences,208 but iPSCs clearly seem to be at greater risk for tumor formation (e.g., due to p53 suppression) than ESCs.
RPE can be derived easily from embryonic stem cells (perhaps because RPE develop early during embryogenesis)40,196,209–211 as well as from iPSCs.151,152,211–214 Transplantation of these human embryonic stem-cell-derived RPE (hES-RPE) into RCS rats also results in improved photoreceptor survival and visual function (although the xenograft does not survive long).36,40 With stem cells, there is a risk of dedifferentiation in the subretinal space and potentially developing into a teratoma. Indeed, in one study, when embryonic stem-cell-derived neural cell precursors were injected into mouse subretinal space, 50% of the eyes developed teratomas within 8 weeks.215 However, in another study, a more stringent selection of 18 different human embryonic stem cell lines did not result in teratoma formation.40 There is a risk of insertional mutagenesis due to the use of viral vectors, and teratoma formation due to the use of oncogenic factors such as c-Myc, in iPSC cell production. Newer techniques of iPSC production without the use of c-Myc216–218 or by using non-integrating episomal vectors219,220 could allay such concerns. Recently, RPE have been generated from human iPSCs (iPS-RPE) and have been shown to express RPE properties such as ion transport, membrane potential, polarized vascular endothelial growth factor secretion, and gene expression profile similar to those of native RPE.221
Bone-marrow-derived mesenchymal stem cells (BM-MSCs) are another potential source of stem cells that have been studied in rats and mice.222 Human BM-MSCs can also be induced to differentiate into retinal cells by co-culturing with RPE cells.223 It is not clear that fully functional RPE cells can be generated with this approach. RPE derived from mouse BM-MSCs have been transplanted into subretinal space of RCS rats and rescue photoreceptors.222 Interestingly, intravenously injected BM-MSCs home into areas of chemically induced RPE loss in mice.224 Autologous cells from umbilical cord tissue can also be a potential source.225
1. Proliferate extensively to generate sufficient quantities of material to serve as a “universal donor”
2. Differentiate into the desired cell type(s). hESC-derived RPE can spontaneously dedifferentiate to non-RPE-like cells and spontaneously redifferentiate into RPE-like cells, indicating phenotypic instability.226 The cultures may not retain a stable phenotype after 5–8 passages. ESCs and iPSCs vary in their tendency to differentiate into cells of a given lineage.204,226 As mentioned above, a number of different criteria should be applied to define a “differentiated” RPE175 and photoreceptor cell.227 The retinal and subretinal microenvironment can influence the differentiation and functionality of transplanted cells, including expression of developmental markers and markers of proliferation.228,229 Abnormalities in Bruch’s membrane may prevent transplanted hESC-derived RPE from surviving and differentiating long term in AMD eyes.22 Since Bruch’s membrane is derived from mesoderm, there is no expectation that hESC- or iPSC-derived RPE will manufacture Bruch’s membrane
3. Survive in the recipient after transplantation. Human iPSC-derived RPE survive ~4 months in RCS rats (xenografts)212
4. Integrate into the surrounding tissue after transplantation
5. Function appropriately for the duration of the recipient’s life.
To be useful for cell-based rescue therapy, stem cells must elaborate needed trophic factors and not proliferate in an uncontrolled manner. In rhodopsin knockout mice, bone-marrow-derived mesenchymal stem cells rescue photoreceptors.230 Subretinal bone-marrow-derived mesenchymal stem cells rescue photoreceptors in RCS rats.231 hESC-derived RPE elaborate neurotrophic substances that have been shown to support photoreceptor survival in preclinical models of retinal degenerative disease.22
RPE cell transplants are an attractive starting point for cell-based combination replacement and rescue therapy in the eye because: (1) hESCs and iPSCs can be induced to differentiate into RPE relatively easily, and, at least in the case of hES-RPE, one can generate large quantities of cells with stable genotype and appropriate phenotype; (2) RPE cells integrate easily with host photoreceptors; (3) RPE cells elaborate trophic substances that support photoreceptors22,33,232; (4) there is abundant evidence for transplant efficacy in pre-clinical models; (5) diseases in which RPE cells appear to be targeted primarily include Best disease233,234 and some forms of retinitis pigmentosa,235,236 and secondarily include Stargardt macular dystrophy237,238 and age-related macular degeneration (AMD).66,104 However, in the case of AMD eyes, RPE survival and proper differentiation on submacular Bruch’s membrane may be problematic.22
Background and rationale for photoreceptor transplantation in retinal dystrophies
Retinitis pigmentosa (RP) is a group of heterogeneous hereditary disorders with an estimated prevalence between 1 : 3000 and 1 : 5000, affecting 50 000 to 100 000 individuals in the USA and approximately 1.5 million people worldwide.239 In the vast majority of cases, the condition results from mutations in photoreceptor cell genes encoding a myriad of structural and phototransduction proteins, and, to a lesser extent, from mutations in genes of the RPE.239 Blindness ultimately results from rod and cone photoreceptor cell death.240 Inner retinal layers, however, remain relatively preserved, especially early in the course of the disease.241,242 As degeneration progresses, however, extensive synaptic rewiring of the inner retina occurs.240,243,244
A variety of approaches to restore vision are under study, including gene therapy,245–249 pharmacotherapy,250,251 visual prostheses,252 endogenous regeneration,253 and transplantation.254–260 The goal of retinal transplantation is to restore and/or maintain visual function by transplanting healthy donor photoreceptors that replace lost photoreceptors by integrating with the host inner retina to reconstruct a functional, neural retinal network. As noted above, this approach to sight restoration is termed “replacement.” Retinal transplantation may also slow down the progression of the disease by means of graft-released trophic substances that rescue residual host photoreceptors,261–264 an approach termed “rescue.” A trophic effect would be most useful, however, only while the critical number of foveal cones required to support useful vision is present.265 In late stages of the disease, when most photoreceptors have degenerated, replacement of lost photoreceptors probably would be more effective.
Results of photoreceptor transplants in experimental animals
Animal models of retinal degeneration
Early studies of retinal transplantation utilized normal animals as hosts. Subsequently, investigators turned to animal models of retinal degeneration. Animal models became available either by discovery of naturally occurring genetic mutants such as the retinal degeneration (rd) mouse,266 the retinal degeneration slow (rds) mouse,267 the RCS rat,6 and the rod–cone dysplasia (Rdy) Abyssinian cat,268 or by experimentally induced degenerations such as photoreceptor degeneration induced by phototoxicity269 and by intravitreal injection of a concentrated solution of human hemoglobin.270 In addition, genetically engineered models of homologous human retinitis pigmentosa mutations such as the transgenic rhodopsin (Pro23His) mutant mouse271 and the transgenic rhodopsin (Pro347His) mutant pig272 were developed. The advantages of these models include the rapid and nearly total loss of target cells such as RPE or rod photoreceptor cells as well as their similarity (in some cases, equivalence) to human disease, allowing for the efficient study of retinal transplantation under more clinically relevant conditions. Most work to date has employed rodent models such as the light-damaged rat retina,13,273,274 the rd mouse,256,275–279 the P23H rhodopsin transgenic mouse,280 and the RCS rat.281 The small eye and large lens of the rodent favor transscleral rather than transvitreal graft delivery. Transvitreal delivery is a less traumatic and more pertinent technique to human retinal transplantation than the transscleral approach. Furthermore, the rod to cone ratio in rodent retina282 is different than that of the relatively cone-rich human retina. To overcome these problems, animal models using larger eyes, such as the chemically ablated rabbit retina270 and the Abyssinian Rdy cat,283,284 have been employed as well as normal rabbits,270,285,286 pigs,287 and primates.118 The pig may prove to be a valuable animal model because the anatomy and size of the porcine eye are closer to that of the human,288,289 and the pig retina is well endowed with cones, has an area centralis, and is holangiotic.290 In addition, the transgenic rhodopsin (Pro347His) mutant pig272 carries a mutation similar to that found in a form of human RP and is an animal model closest to human RP, especially with regard to refining surgical procedures.291 However, the pig retina does not have a fovea.
Graft implantation sites and preparations
Retinal transplants have been placed in a surgically induced retinal lesion, in the vitreous cavity (i.e., epiretinally),12,292 and into the host subretinal space.13,293 Grafts from donors of the same species (i.e., allogeneic grafts) and from a different species (i.e., xenogeneic grafts)118,294,295 have been transplanted with or without host immune suppression.
Photoreceptor microaggregates, which are retinal fragments in which tissue integrity is disrupted by gentle trituration with cells remaining attached to each other in small clusters (<0.2 mm2), have been used in retinal transplantation.12 Isolated cell suspensions obtained by mechanical and/or enzymatic dissociation of the retina also have been used.275,277,293,296 Compared to microaggregates, isolated photoreceptors usually are more traumatized, frequently lose their outer segments, display less organization, and often are not properly oriented.255,296 Both preparations often form spherical structures or rosettes, with photoreceptors oriented towards a central lumen similar to those seen in retinoblastomas and retinal dysplasias.12,292,294 Outer segments of misoriented photoreceptors as well as those found in rosettes rapidly degenerate due to their failure to contact RPE cells.297 In addition, both retinal microaggregates and cell suspensions frequently are contaminated with non-photoreceptor retinal cells. Rosette formation prevents the reconstruction of the normal retinal anatomy and interferes with establishing contacts between graft photoreceptor terminals and host second-order neurons.33
To avoid these problems, Silverman and Hughes13 introduced the photoreceptor sheet preparation, created by vibratome-sectioning of adult mammalian retina. Sheet preparations preserve the organization and polarity of grafted photoreceptors and maintain the densely packed arrangement of photoreceptors found in situ and thought to be critical for good visual acuity.298 Furthermore, the inner retina, which could interfere with synaptic interactions between grafted photoreceptors and host second-order neurons, is removed. Sheets can be placed between the host retina and RPE in the proper orientation, and they show a much lower frequency of rosette formation.13,262,270,276 This technique has been modified further to obtain photoreceptor sheets using both vibratome-sectioning and excimer laser ablation of the inner retina.299 A similar approach has been used to transplant sheets of adult full-thickness retina, intact embryonic or fetal retinal sheets,274,281,286 and vibratome-sectioned fetal retina. Aramant and coworkers introduced the approach of co-transplanting fetal neural retina with the adjacent RPE.281 This approach might be advantageous in cases of advanced disease where both retinal cell types have degenerated.258
