Vulnerability of Allogeneic Retinal Pigment Epithelium to Immune T-Cell–Mediated Damage In Vivo and In Vitro

Abstract

purpose. Because retinal pigment epithelium (RPE) constitutively expresses class I major histocompatibility complex (MHC) molecules, and CD95 ligand and secretes immunosuppressive factors, the vulnerability of these cells to attack by immune T cells is open to question. This study was conducted to determine the vulnerability of allogeneic RPE to damage by specifically sensitized T cells, both in vivo within the subretinal space, and in vitro.

method. BALB/c lymphocytes presensitized to C57BL/6 antigens were injected into the subretinal space of eyes of C57BL/6 and gld/gld mice, and the eyes were examined clinically and histologically. RPE eyecups were produced from mouse eyes by removing the anterior segment and neuronal retina, leaving an intact monolayer of RPE. Sensitized BALB/c lymphocytes were placed in the RPE eyecup and incubated for 4 hours. The RPE layer of the eyecups was assessed by confocal microscopy for viability, after staining with propidium iodide and acridine orange.

result. Eyes that received T cells sensitized to C57BL/6 antigens displayed a circumscribed patch of persistent choroidal “whitening” clinically and a disrupted RPE cell layer histologically at the injection site at 5 days after injection. By 14 days, only RPE cells at the injection site were lost. RPE in eyecup preparations was relatively resistant in vitro to cytolysis by sensitized T cells, whether the eyecups were obtained from CD95-deficient or wild-type mice.

conclusions. RPE monolayers, both in vivo and in vitro, are relatively resistant to immune-mediated attack by specifically sensitized T cells. This relative lack of vulnerability is independent of the expression of CD95 ligand by target RPE cells and implies that immune barriers to acceptance of allogeneic RPE transplants may be less than if transplanted cells are from nonocular tissues.

Retinal transplantation is viewed with hope as a solution to the loss of vision that results from irreversible damage to the neural retina and to the retinal pigment epithelium (RPE). Neural retina and RPE are vastly different types of tissues, and each confronts the experimentalist with a unique set of barriers to successful transplantation. Our laboratory has been particularly interested in determining the immunologic barriers to the successful transplantation of allogeneic retinal pigment epithelium into eyes of rodents. We and others have demonstrated that RPE as a tissue expresses class I alloantigens encoded within the murine major histocompatibility complex (MHC, H-2), ¹ and we presume, but have no direct evidence, that RPE also displays, on these class I molecules, the peptides derived from minor histocompatibility (minor H) antigens. Under normal circumstances, the RPE expresses no class II MHC molecules, although the expression of these molecules can be induced by exposure to IFN-γ. ² Thus, in principle, allografts of RPE and intact RPE should represent suitable targets for CD8⁺ T cells (TCs) of the cytotoxic type. At present, there is no direct information in the literature concerning the vulnerability of RPE to destruction by allospecific cytotoxic TCs. 

RPE grafts, as single-cell suspensions of cultured cells and as intact sheets prepared from neonatal mouse eyes, have been implanted in the anterior chamber, the vitreous cavity, and even into the subretinal space of mouse eyes. ^{3 4} Because all these intraocular compartments are immune privileged, ^{5 6} interpretation of the fate of allografts of RPE is complicated by the contribution to graft survival made by the site itself. Wenkel and Streilein ⁷ investigated this question when they transplanted sheets of allogeneic neonatal RPE tissue beneath the capsule of the kidney—a non–immune-privileged site. They reported that allogeneic grafts of RPE survived as well as syngeneic RPE grafts beneath the kidney capsule. The transplants showed no gross or histologic evidence of rejection for at least 8 weeks after implantation. This finding offers formal proof that neonatal RPE tissue has the inherent property of immune privilege. These investigators also showed that the privileged status of RPE allografts could be aborted if the grafts failed to express CD95 ligand (CD95L). Thus, similar to the corneal endothelium and to the Sertoli cells of the testes, RPE functions as an immune-privileged tissue in part through the constitutive expression of CD95L. ^{8 9}  

At present, it is not technically feasible to transplant sheets of RPE orthotopically into the subretinal space of mouse eyes. For this reason, we have embarked on a series of studies to determine the vulnerability of RPE to immune destruction by cytotoxic TCs using strategies that do not involve orthotopic transplantation of the tissue. We injected into the subretinal space suspensions of allogeneic lymphocytes containing TCs primed for class I antigens expressed on recipient eyes, and we layered fully functional primed allogeneic TCs on intact monolayers of RPE in posterior eyecups in vitro. Our results reveal a marked resistance of RPE to lysis by primed TCs, a resistance that appears to be unrelated to the expression of CD95L. 

Methods

Animals

Adult male BALB/c, C57BL/6, and C57BL/6 gld/gld (B6.gld) mice, aged 6 to 8 weeks were obtained from the animal facilities at the Schepens Eye Research Institute or from Taconic (Germantown, NY). Mice were kept in a common room of the vivarium. All experimental procedures conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Inoculations, injections, and clinical examinations were performed under anesthesia induced by intraperitoneal injection of ketamine (Ketalar; Parke Davis, Paramus, NJ) at 0.075 mg/g of body weight, and xylazine (Rompun; Phoenix Pharmaceutical, St. Joseph, MO) at 0.006 mg/g of body weight. Enucleation of the eyes and removal of spleen and lymph nodes were performed after the animals were killed by cervical dislocation. Five mice were used in each experimental group, and all experiments were repeated at least twice with similar results. 

Preparation of Allosensitized Effector TCs

BALB/c mice were immunized in the flank by injecting 1 × 10⁷ C57BL/6 splenocytes on day 0. They were killed 1 week later, and their spleens were removed and rendered into a single-cell suspension. Red blood cells were eliminated using red blood cell lysing solution. The splenocytes were cocultured at a ratio of 5:1 effector to stimulator with irradiated (2000 rad) naïve lymphocytes from C57BL/6 mice for 5 days in complete RPMI medium (RPMI plus HEPES, 2-mercaptoethanol [ME], glutamate, penicillin-streptomycin, and 10% fetal calf serum). After 5 days, the cells were washed three times in Hanks’ balanced salt solution (HBSS). The live cells were counted and used as BALB/c anti-C57BL/6 effector cells in subsequent experiments. As a control, C57BL/6 anti-BALB/c effectors were produced in a similar manner. 

Assay for Cytotoxic Activity

BALB/c anti-C57BL/6 effector cells were tested for cell-mediated cytotoxicity in a standard 4- hour ⁵¹Cr-release assay. Either P815, a mastocytoma cell line derived from DBA2 mice, or EL4 cells, a lymphoma cell line derived from the C57BL/6 strain, were used as target cells. In addition, cell mixtures containing BALB/c anti-C57BL/6 effector cells were tested for presence and functionality of natural killer (NK) cells by using the YAC-1 lymphoma cell line as targets. All target cells were labeled with 300 μCi Na₂ ⁵¹CrO₄ (New England Nuclear, Boston, MA) per 3 × 10⁶ cells for 2 hours in a water bath at 37°C. Thereafter, target cells were washed three times and placed in a microtiter plate at a concentration of 2 × 10⁴ per well. Either the BALB/c anti-C57BL/6 effector cells or C57BL/6 anti-BALB/c effector cells were added to the wells of the microtiter plate at the following effector-to-target cell ratios (E:T): 100:1, 10:1, 2:1, and 1:1. Culture plates were centrifuged at 1000 rpm for 5 minutes and incubated for 4 hours at 37°C. After incubation, 25 μL of the supernatant was removed and counted for radioactivity. The percentage of specific chromium release was calculated using the standard formula. The spontaneous release of ⁵¹Cr from the target cells was determined by counting the supernatant from three wells containing only target cells and no effector cells. The total release was determined by measuring chromium released from the supernatant of three wells containing the target cells and 1N HCl. Spontaneous release of chromium never exceeded 20% of the total release. 

Injection into Subretinal Space

Subretinal space injections were performed according to the procedure of Whiteley et al. ¹⁰ on anesthetized animals that, in addition, received topical proparacaine to anesthetize the ocular surface and tropicamide 1% to dilate the right pupil. Injection into the subretinal space was performed using very fine, bevelled, pulled-glass micropipettes that were connected to a 10-μL syringe (Hamilton, Reno, NV) by a fine polyethylene tube. The entire apparatus was filled with HBSS, but an airlock was produced before the volume to be injected, thereby preventing dilution of the injected material. The bore of the glass needle was coated (Sigmacoat; Sigma-Aldrich, St. Louis, MO) to prevent adherence of cells. The injections were made under direct visualization by transscleral approach through the peripheral retina, using a binocular surgical microscope and a coverslip held on the cornea. The glass needle was advanced carefully until it reached the subretinal space where cells in a volume of 1 μL were injected. 

Histologic Examinations

Mice were killed by cervical dislocation at various time intervals, and their eyes were enucleated and fixed immediately in 4% paraformaldehyde and embedded in methacrylate. Five-micrometer sections were cut and stained with hematoxylin and eosin. Tissue sections were then examined by light microscopy. 

Preparation of Posterior Eyecups

Eyes were enucleated from C57BL/6 and B6.gld mice and placed in Ca²⁺/Mg⁺-free HBSS on ice for 30 minutes, and then removed, and the muscles, connective tissue, and conjunctiva were excised with microscissors. A circumferential incision was performed below the level of the ciliary body, and the entirety of the anterior segment, including the cornea, iris, ciliary body, and the lens, which were discarded. The remaining tissue (posterior eyecup) was placed in 0.01 U/mL of chondroitinase ABC ¹¹ for 30 minutes at 37°C, placed on ice, and washed three times in HBSS. The neural retina was gently lifted off the RPE layer by microsurgical forceps. Posterior eyecups consisting of sclera, choroid, and a healthy, intact monolayer of RPE were placed in individual wells of microculture plates (S plate; Nunc Brand Products, Nalge Nunc International Corp., Naperville, IL) for further experiments. 

In Vitro Cytotoxicity Assay

Posterior eyecups from eyes of C57BL/6 and B6.gld mice were placed in microwell plates (S plate) and 10,000 effector BALB/c anti-C57BL/6 TCs in 10 μL of DMEM (fortified with glutamate, antibiotics, 2-ME, and 1% HEPES) were gently layered onto the eyecups. After incubation for 4 hours at 37°C, the eyecups were washed twice and stained with 1 μL/mL propidium iodide (PI) and 0.5 μL/mL acridine orange (AO) for 15 minutes. The cups were mounted on glass slides, covered with glass coverslips, and observed by confocal microscopy. For some experiments, the eyecups were incubated with 100 U/mL recombinant IFN-γ (BD Biosciences, San Diego, CA) for 4 or 12 hours before exposure to effector BALB/c splenocytes. To assess the level of class I MHC, some eyecups were stained with FITC-conjugated anti-MHC class I antibody (R&D Systems, Minneapolis, MN). 

Results

Resistance of RPE Cells to Immune Destruction by Alloreactive Effector T Cells Placed Subretinally

Primed H-2^b–specific cytotoxic TCs were generated by immunizing BALB/c mice subcutaneously with C57BL/6 spleen cells. One week later, spleen cells were harvested from these mice and restimulated in vitro with x-irradiated C57BL/6 spleen cells. In vivo primed TCs restimulated in vitro in this manner contain cytotoxic TCs specific for MHC and minor histocompatibility (minor H) antigens. After 5 days, the responding lymphocytes were harvested and assayed for their capacity to lyse EL-4 cells, a target tumor cell line derived from the C57BL/6 mouse strain. Control effector cells were prepared similarly by immunizing C57BL/6 mice with BALB/c spleen cells. As revealed in Figure 1A , BALB/c effector cells lysed target EL4 cells expressing H-2^b class I and minor H alloantigens. EL4 cells were not lysed by C57BL/6 anti-BALB/c effector cells, demonstrating that the cytotoxic attack was specifically directed at C57BL/6 alloantigens. To demonstrate that C57BL/6 effector (C57BL/6 anti- BALB/c) cells had the ability to lyse BALB/c targets allospecifically, C57BL/6 effector cells were used in a cytotoxicity assay with P815 mastocytoma cells from DBA2 mice. Even though these target cells differ from BALB/c at numerous minor H antigens, they share the same MHC antigens. Figure 1B demonstrates that C57BL/6 effector cells lysed target P815 cells expressing H-2^d class I alloantigens. However, P815 cells were not lysed by BALB/c anti-C57BL/6 effector cells, demonstrating that both effector cell types were capable of allospecific cytotoxic killing. Because activated NK cells may also be present in the preparations of effector TCs generated, the ability of NK cells to lyse target cells was tested by using YAC-1 cells as targets. Figure 1A demonstrates that very limited NK cell killing was detected, even at high effector-to-target ratios. 

Effector cells prepared in this manner were injected (10,000 cells per 1 μL inoculum) into the subretinal spaces of eyes of normal C57BL/6 mice. As the control, eyes of some mice received a 1 μL inoculum of PBS alone into the subretinal space. In one set of recipients, the eyes were examined clinically using fundoscopy at 2, 5, and 14 days after injection. In another set, injected eyes were removed at 1, 6, 24, and 48 hours and at 5 days and 14 days after injection. These eyes were prepared and sectioned for histologic examination. 

Clinical examination after 2 days revealed that the retina was edematous in an area of one disc diameter (diameter of the optic nerve head), at the site of injection. This edematous appearance was detected irrespective of the contents of the subretinal injection. At the same time, the anterior chamber was normal, the lens was clear, and there was no evidence of intraocular inflammation. These findings indicate that there is a minimal amount of trauma associated with subretinal injections of this type and that the focus of this response is restricted to the site of injection. At 5 days after injection, in eyes that received BALB/c anti-C57BL/6 lymphoid cells subretinally, a distinct pallor developed at the injection site that was readily distinguishable from surrounding tissue and was approximately two disc diameters in size. This defect in the RPE layer was still present when these eyes were examined at 14 days after injection, but had not changed in size during this time interval. Of particular importance, no area of pallor was detected in the retinas of eyes that received injections of control lymphoid cells. Thus, injection of specifically sensitized lymphoid cells containing TCs created a circumscribed defect in the RPE monolayer at the site of injection. Once established, this lesion persisted through time but failed to expand in size. 

Histologic examination of eyes receiving effector cell injections supported these observations (Figs. 2 3 4 and 5) . In eyes that received effector cell injections, lymphoid cells were detected in the subretinal space at 1 hour (Figs. 2A 2B) . Among eyes that received control cells, no lymphoid cells were seen thereafter, but among eyes that received BALB/c anti-C57BL/6 effector cells, inflammatory cells persisted at the injection site through 24 hours after injection (Fig. 2C) . 

At 48 hours after injection, all eyes contained edema-induced fluid in the subretinal space and within the retina itself around the site of injection. When examined at later intervals, neither the sham-injection eyes (data not shown) nor eyes that had received C57BL/6 anti-BALB/c TCs demonstrated any edema or inflammatory cells, and the RPE layer was intact (Fig. 3) . By contrast, at 5 days the RPE layer was disrupted at the site of injection of BALB/c anti-C57BL/6 cells. Some RPE cells appeared vacuolated and had lifted off Bruch’s membrane (Fig. 4A) . At 14 days after injection, the injection site was devoid of healthy RPE cells, and there was a sharp perimeter to the lesion, formed by healthy-appearing RPE (Figs. 4B 4C) . At no time were inflammatory cells observed within the neural retina, nor was the photoreceptor layer of the retina damaged. 

Together, the results of these experiments indicate that specifically sensitized effector lymphoid cells containing TCs can attack and destroy RPE at the precise site where they are injected but that the damage remains confined to this site, with little evidence to suggest that the effector cells can systematically and sequentially destroy adjacent RPE cells. 

Resistance of CD95 Ligand Deficient RPE Cells to Immune Destruction by Effector T Cells Placed Subretinally

Because RPE express CD95L constitutively and because CD95L expression on layers of allogeneic neonatal RPE implanted beneath the kidney capsule protects these grafts from immune destruction, ⁷ we next examined whether enhanced destruction of RPE might occur when specifically sensitized lymphoid cells are injected into the subretinal space of B6.gld mice that do not express a functional CD95L molecule. C57BL/6-primed lymphoid cells restimulated in vitro were injected into the subretinal space of B6.gld mice, and the injected eyes were examined clinically and histologically as described earlier. As revealed in Figure 5 , the pattern of clinical and histologic findings in CD95L-deficient eyes that received injections of both specifically and irrelevantly sensitized lymphoid cells was identical with that found in normal eyes. In particular, the initial RPE lesion, its evolution, and final extent were no different in eyes of CD95L-deficient mice than in those of wild-type mice. Thus, expression of CD95L cannot explain why specifically sensitized lymphoid cells injected into the subretinal space failed to create circumferentially expanding lesions in which larger and large numbers of RPE cells are destroyed. 

RPE Cells Remain Resistant to Allogeneic Effector T Cells In Vitro

The results of the experiments described to this point suggest that the resident RPE in vivo is highly resistant to the destructive potential of primed cytotoxic TCs. Because the putative effector cells were injected into the subretinal space, an immune-privileged site, the possibility exists that properties of the site itself limits the lytic capacity of the injected cells. To circumvent this problem, we created a posterior eyecup by enucleating eyes from normal C57BL/6 mice by first excising the anterior segment (cornea, conjunctiva, lens, iris, and ciliary body). The neural retina was then gently pulled away after treatment with enzymes, leaving behind an intact layer of RPE resting on Bruch’s membrane, the choroid, and the posterior sclera. These eyecups were placed, RPE layer up, in microculture plates. Effector TC suspensions prepared from BALB/c and C57BL/6 donors as described earlier were layered (10,000 cells per 10 μL) gently on top of the RPE layer and incubated for 4 hours. The lymphoid cells were then rinsed away, and the RPE layers of the eyecups were assessed for viability by staining with PI and AO, after which they were transferred to coverslips and inspected by confocal microscopy. The results, displayed in Figure 6A , reveal that allogeneic RPE in posterior eyecups were impervious to lysis by TCs and that they were equally resistant to nonspecific lysis by immunologically irrelevant but activated lymphoid cells. Similar experiments were performed with posterior eyecups prepared from B6.gld donors. Once again (Fig. 6B) , BALB/c effector cells were incapable of lysing CD95L-deficient RPE. 

One possible explanation for the relative invulnerability of RPE in the posterior eyecups to lysis by TCs is that the constitutive level of MHC class I expression is insufficient to trigger effector cells. To address this point, class I expression was stimulated in posterior eyecups by exposing the tissues in vitro to IFN-γ for 4 or 12 hours, before testing the effect of specifically sensitized TCs on RPE cell lysis. Figure 7A reveals that the RPE cells in the eyecup preparation, without treatment with IFN-γ, expressed low levels of class I MHC antigens. There was a slight upregulation of class I MHC after the RPE eyecups were treated with IFN-γ for 4 hours (data not shown), but, after treatment with IFN-γ for 12 hours, class I MHC was significantly upregulated (Fig. 7B) . We determined that treatment of the RPE eyecup with IFN-γ for 12 hours without the addition of BALB/c effector TCs did not induce toxicity in the RPE layer on its own (data not shown). Prior treatment of the RPE eyecups with IFN-γ for 12 hours led to a higher uptake of PI on exposure to TCs than did untreated RPE (Fig. 6D) . A comparatively small, specific RPE cell death was detected in eyecups treated with IFN-γ for only 4 hours (Fig. 6C) . These results suggest that low-level expression of MHC class I molecules on RPE eyecups from normal eyes is at least one factor that renders RPE cells relatively resistant to lysis by allospecific TCs. 

Discussion

There are blinding diseases of the eye in which dysfunction and loss of RPE is thought to be the primary pathogenic condition. Leber’s amaurosis is genetically determined disease in which mutation of a protein uniquely expressed in RPE renders these cells incapable of supporting photoreceptors, and eventually the retina deteriorates. ¹² In age-related macular degeneration, a multifactorial disease of unknown cause, early changes and loss of RPE are thought to be central to the pathogenesis. ¹³ For these and similar diseases, transplantation of allogeneic RPE to replace defective or depleted RPE is an attractive potential clinical solution. Allogeneic tissues as grafts are vulnerable to immune rejection, and this potential exists in principle for allogeneic RPE grafts. The importance of this potential is mitigated, on the one hand, by the immune-privileged nature of the subretinal space in which RPE grafts are to be placed. Allogeneic tissues implanted into this site have extended survival, and alloantigenic material and soluble protein antigens injected into the subretinal space induce a form of immune deviation very similar to that induced by injection of antigens into the anterior chamber. ¹⁴ On the other hand, there is unequivocal evidence that RPE tissue has properties of an immune-privileged tissue. ⁷ RPE cells express transplantation antigens that should make them vulnerable to immune recognition, and there is evidence that allografts of RPE placed intraocularly, even in the subretinal space, are subject to destruction that is presumed to be immunologic. ¹⁵ The experiments reported herein explored this possibility. 

Our results indicate that explanted, intact layers of RPE resident in posterior eyecups are quite resistant to lysis by TCs that are fully able to lyse other nonocular types of target cells. Moreover this relative resistance to TC-mediated lysis displayed in vitro by RPE in posterior eyecups was mirrored to some extent when specifically sensitized TCs were injected into the subretinal space of appropriate recipients. The positive result is that effector TCs created small, circumscribed lesions at the site of injection, and eventually these sites were found to be depleted of RPE. Moreover, the constraint applied to the injected cells was not provided by CD95L. Lesions produced in the RPE layer of B6.gld eyes were similar in all respects to those produced in wild-type C57BL/6 eyes. 

The relative resistance of RPE to lysis by TCs may be due to mechanisms inherent within RPE cells, such as their low level of class I MHC expression, or the ability of the RPE to resist the lytic mechanisms of activated TCs. RPE cells are known to undergo apoptosis as a result of several stress factors, such as hydrogen peroxide (H₂O₂), ischemia, and blue-light toxicity. ^{16 17} RPE, however, has multiple mechanisms to limit the apoptotic process. Alge et al. ¹⁸ have demonstrated that RPE cells produce a heat shock protein, αβ-crystallin, that limits apoptosis due to oxidative stress. Others have noted that pigment epithelium-derived factor (PEDF) produced by RPE cells inhibited apoptosis in cultured retinal neurons. Presumably, PEDF works similarly to inhibit RPE cell apoptosis from hydrogen peroxide injury. ¹⁹ Overexpression of Bcl-2 has been shown to decrease apoptosis in human RPE exposed to H₂O₂ or blue light. ²⁰  

TCs exert their lytic effects, either by synthesis and secretion of cytokines such as IFN-γ ²¹ or TNF-α, ²² or by direct cell-to-cell contact. There are two distinct, contact-dependent mechanisms that are used by TCs to lyse their targets: (1) perforin-granzyme–mediated cytotoxicity, with directional release of granular content from TC cytoplasm toward the target cells, and (2) a receptor-mediated cytotoxicity involving interaction of FasL or TNF-α on the TCs with Fas or TNF-α receptors on the target cells. The final common pathway is the activation of caspases leading to apoptotic cell death. Perforin-mediated apoptosis in the retina has been noted in relation to viral retinitis where the apoptotic cells are found mostly in the outer nuclear layer ^{23 24} ; however, it is not clear whether RPE cells undergo apoptosis by this pathway. 

Subthreshold expression of MHC class I molecules seems to explain partially the resistance of explanted RPE to TC-mediated lysis. When posterior eyecups were treated for 12 hours with IFN-γ, class I expression on the RPE was enhanced, and specific TCs killed a modest, but significant, proportion of RPE cells treated in this manner. Low-level constitutive expression of MHC class I molecules is a characteristic feature of other ocular cells as well. Corneal endothelial cells are especially depauperate in class I expression, and this is believed to explain why donor-specific cytotoxic TCs induced in mice bearing MHC-disparate orthotopic corneal allografts play no important role in acute graft rejection. ²⁵  

RPE also express Fas antigen and can be a target for apoptosis by FasL-bearing cells. RPE eyecups from C57BL/6 mice that were treated with allosensitized effector TCs did not show any evidence of apoptosis, but once the eyecups were treated with IFN-γ for 12 hours, RPE cell lysis occurred. IFN-γ is known to increase Fas receptor expression in tumor cells, leading to increased apoptosis in these cells. ²⁶ Mullbacher et al. ²⁷ demonstrated that TCs can indeed upregulate Fas expression as a result of IFN-γ release, but do not increase apoptosis in Fas-positive target cells in the absence of TC receptor (TCR) ligation. Upregulation of Fas receptor and increased Fas-FasL interaction may be one of the mechanisms by which RPE cells undergo apoptosis. An increase in IFN-γ levels accompanies injection of TCs into the subretinal space, in part due to production of IFN-γ by the injected cells. The expectation, therefore, would be for upregulation of class I MHC and Fas receptors throughout the subretinal space, leading to widespread lysis of the RPE. Instead, the lysis of the RPE remained quite localized in our experiments. What mitigates this positive result is our observation that these lesions failed to progress and expand through time. When TCs are added to nonocular target cells in vitro, they are able through time to eliminate every target cell. ²⁸ Yet, TCs placed in the subretinal space failed to display this capacity. It should be pointed out that the lymphoid cell suspensions injected into the subretinal space in these experiments contained primed allospecific CD4⁺ TCs, as well as CD8⁺ TCs. This is important, because histologic examination of injected eyes revealed almost no nonspecific inflammation at the injection site. We interpret these results to mean that RPE themselves secrete or the subretinal space in which the cellular inocula were placed contains factors that silence the destructive potential of effector TCs. Determining the nature and mode of action of these silencing factors is a major goal of our future experiments. 

One of the strategies that may be used by RPE cells to avoid being lysed by TCs is to induce apoptosis in the activated TCs. RPE has been shown to induce apoptosis in activated TCs in both a cell-to-cell contact-dependent ²⁹ and a contact-independent manner. ³⁰ Jorgensen et al. ²⁹ reported that cultured fetal human RPE cells (HFRPE) incubated with an anti-CD3–activated TC population led to apoptosis of TCs in a Fas-FasL–dependent manner. In our experiment, the resistance to lysis was found to be unrelated to expression of CD95L, because TCs layered onto RPE of posterior eyecups prepared from CD95L-deficient mice also did not cause lysis. To our knowledge, this is the first report of an in vitro experiment designed to test whether constitutive CD95L expression protects ocular target cells from TC-mediated lysis. 

Our results provide a certain level of optimism that the presumed immunologic barriers to RPE transplantation may be low compared with other, nonocular solid tissues. Immune privilege of the subretinal space and of the RPE seems to be important. However, this optimism is tempered by previous reports that allogeneic RPE grafts placed subretinally can eventually be rejected. Caution is further warranted by the speculation that diseases, especially those of an inflammatory and angiogenic nature, that disrupt the RPE, suprachoroidal space, and the choriocapillaris, may very well disrupt immune privilege in this region. Restoration of immune privilege may be an important strategy for promoting the immunologic success of RPE grafts in the future. 

 Supported by National Eye Institute Grants EY09595 and EY05678 and in part by a grant from the Oxford Congress, Oxford University, Oxford, United Kingdom (PZ).

 Submitted for publication March 3, 2003; revised August 15, 2003; accepted September 2, 2003.

 Disclosure: P. Zamiri, None; Q. Zhang, None; J.W. Streilein, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Ability of effector BALB/c anti- C57BL/6 or C57BL/6 anti-BALB/c cytotoxic TCs to lyse allospecific target cells. (A) BALB/c spleen cells primed against C57BL/6 targets were incubated with ⁵¹Cr-labeled EL4 cells, a C57BL/6 origin lymphoma cell line, for 4 hours at 37°C. As a control, C57BL/6 splenocytes primed against BALB/c targets were similarly incubated with EL4 cells. In addition, lymphoid cell mixtures containing BALB/c effector cells were incubated with YAC-1 lymphoma cells. (B) C57BL/6 spleen cells primed against BALB/c targets were incubated with ⁵¹Cr-labeled P815 cells, a DBA2 origin mastocytoma cell line, for 4 hours at 37°C. As a control, BALB/c splenocytes primed against C57BL/6 targets were similarly incubated with P815 cells.

Figure 2.

 

Histologic appearance of subretinal space injection site of effector lymphocytes in C57BL/6 eyes. BALB/c anti-C57BL/6 effector cells (n ∼ 10,000) were injected into the subretinal space of adult C57BL/6 mice. One hour and 24 hours later, the animals were killed and the eyes were enucleated, placed in 4% paraformaldehyde, embedded in methacrylate, sectioned, stained with hematoxylin and eosin, and viewed by light microscopy. (A) Injection site and the bleb produced around the site of injection. (B) Magnified view of the injection site shows numerous lymphocytes. A similar image was observed when C57BL/6 anti-BALB/c splenocytes were injected into the subretinal space of C57BL/6 mice. (C) Section from an eye receiving BALB/c anti-C57BL/6 splenocytes 24 hours after injection. Arrow: injection site. A retinal detachment was evident throughout the length of the retina, yet few lymphoid cells were dispersed through the subretinal space. None of the injected cells were found at 24 hours in other control experiments.

Figure 3.

 

Histologic appearance of RPE and retina around the site of injection of C57BL/6 effector lymphocytes (C57BL/6 anti-BALB/c) in C57BL/6 eyes. (A) Control C57BL/6 effector cells sensitized to BALB/c alloantigen were injected into the subretinal space of adult C57BL/6 mice. Two weeks later, the eyes were enucleated and prepared for histology. The RPE layer was intact and the retina had normal cytostructure. (B) High magnification showing the RPE cell nuclei. No inflammatory cells are visible.

Figure 4.

 

Histologic appearance of RPE and retina around the site of injection of BALB/c effector lymphocytes (BALB/c anti-C57BL/6) in C57BL/6 eyes. (A) This section was from an eye enucleated 5 days after receiving BALB/c effector cells primed against C57BL/6 antigens, depicts vacuoles within the damaged RPE segment and abnormalities in the cytostructure of the RPE cells. (B, C) Sections produced from eyes enucleated 14 days after receiving BALB/c effectors against C57BL/6 alloantigens. (B) The circumscribed extent of damage to the RPE layer (between the two arrows) flanked by normal RPE. (C) Loss of RPE cell nuclei and extensive disorganization of cells are shown within the damaged portion, at higher magnification.

Figure 5.

 

Histologic appearance of subretinal space injection site of effector lymphocytes in B6.gld eyes. Sections are from eyes of B6.gld mice enucleated 5 days (A) and 14 days (B, C) after receiving BALB/c effectors against C57BL/6 alloantigens. Well-circumscribed areas of vacuolation were visible at 5 days, leading to the loss of RPE layer around the site of injection at 14 days after injection.

Figure 6.

 

Confocal appearance of RPE layers from eyes of C57BL/6 and B6.gld mice after incubation of RPE eyecups with effector TCs with or without IFN-γ pretreatment. Posterior eyecups were produced from enucleated eyes of adult C57BL/6 and B6.gld mice and were placed in individual wells of a microculture plate. BALB/c anti-C57BL/6 effector TCs were layered into the eyecups. After incubation for 4 hours, the eyecups were stained with PI and AO. The cups were mounted on glass slides, covered with glass coverslips and observed with confocal microscopy. (A) C57BL/6 eyecup and (B) B6.gld eyecup. Both show the intact monolayer of RPE bearing only the AO staining. The few yellow spots (combination of PI and AO) are from dead splenocytes (arrows) that were left after the washing process. No damage to the integrity of the RPE layer was noted. (C) RPE eyecup treated with IFN-γ for 4 hours before exposure to presensitized splenocytes. A small number of RPE cells have died as evidenced by costaining of the RPE cell nuclei with both AO and PI (arrow). (D) RPE eyecup treated with IFN-γ for 12 hours before incubation with presensitized splenocytes. A larger number of RPE cells display yellow staining nuclei (combination staining with AO and PI), indicating increased cell death (arrow).

Figure 7.

 

Histologic sections of RPE eyecup immunostained with anti-MHC class I antibodies. Posterior eyecups were produced from enucleated eyes of adult C57BL/6 mice and stained with FITC-conjugated anti-MHC class I antibody and viewed by confocal microscope. (A) Untreated posterior eyecup demonstrates minimal class I MHC staining. (B) Posterior eyecup treated with IFN-γ for 12 hours before staining depicts significant upregulation of class I MHC.

The authors thank Susan Lightman, Tongalp H. Tezel, and Henry J. Kaplan for helpful advice; Jiang Gu for technical assistance in providing methacrylate sections; Jacqueline Doherty for expert laboratory management; and Marie Ortega for maintaining high standards of animal care in the vivarium.