Epithelium-Deficient Corneal Allografts Display Immune Privilege beneath the Kidney Capsule

Abstract

purpose. To determine whether corneal tissue as an allograft displays immune privilege in a nonprivileged site and, if so, whether CD95 ligand expression contributes to the privileged status.

methods. Syngeneic and allogeneic corneal tissues deprived of epithelium were transplanted beneath the kidney capsule of normal mice. Syngeneic BALB/c, allogeneic C57BL/6, and allogeneic B6Smn.C3H-gld (CD95 ligand–deficient) mice were used as donors for BALB/c recipients, and syngeneic C3H/HeJ-gld (CD95 ligand–deficient) mice were used for normal C3H/HeJ recipients. Allogeneic conjunctival segments served as positive grafting controls. Graft fate was assessed by visual inspection at 4, 7, 14, and 21 days and was confirmed by histologic study. Viability of graft endothelium was assessed by immunocytochemical analysis.

results. Syngeneic corneas and C57BL/6 corneas survived almost indefinitely beneath the kidney capsule. Both the stroma and the endothelial layers remained healthy and intact. Allogeneic conjunctiva evoked a strong inflammatory response attended by neovascularization. Similarly, B6-gld corneas deficient in CD95 ligand expression showed acute destruction beneath the kidney capsule. Circumstantial evidence implicates alloimmune rejection as the mechanism.

conclusions. Epithelium-deprived corneas from normal mice possess inherent immune privilege that protects them from alloimmune rejection even at nonprivileged sites. Constitutive expression of CD95 ligand contributes to the privileged status. It is inferred that the extraordinary success of orthotopic corneal allografts owes as much to the qualities of the graft as an immune-privileged tissue as to the qualities of the eye as an immune-privileged site.

Penetrating keratoplasties are by far the most successful solid organ transplantations performed in humans today. ^{1 2 3} In most uncomplicated cases, topical rather than systemic immunosuppression is all that is required to secure the graft’s survival. This positive clinical experience is matched by the results obtained in experimental models in which orthotopic corneal transplants have been performed in nonimmunosuppressed mice and rats. Under these conditions, the corneal grafts enjoy considerable success. ^{4 5} The usual explanation for the extraordinary success of orthotopic corneal allografts, either in humans or in experimental animals, relates to the phenomenon of immune privilege. ^{4 6} In orthotopic grafting, the corneal graft is sutured into an avascular (both blood and lymph) rim of recipient cornea. Moreover, the graft necessarily forms the anterior surface of the anterior chamber. The anterior chamber is a well-described immune-privileged site where grafts of a variety of foreign tissues are accepted for prolonged and often indefinite intervals of time. ^{7 8} Acceptance of corneal allografts at this site is no exception. 

Although the site of engraftment is immune privileged, the cornea itself has also been considered to be an immune-privileged tissue. Early experiments by Medawar ⁹ and Barker and Billingham ¹⁰ have indicated that the cornea has the capacity to avoid or escape destruction by the alloimmune rejection process. Although these data suggest that the cornea is immune privileged, corneal allografts transplanted heterotopically to the cutaneous surface or into the subcutaneous space readily induce systemic alloimmunity, and these grafts are swiftly destroyed. ¹¹ The one exception to this outcome is in the case of corneal grafts derived from donors who are histoincompatible with their recipients only at the class II region of the major histocompatibility complex (MHC). Class II–only disparate corneal grafts display prolonged survival when placed on body wall skin of naive mice. ¹² However, even in this highly unusual genetic situation, Rubsamen et al. ¹³ demonstrated that if the donor cornea contained Langerhans’ cells at the time of grafting, class II–only disparate grafts were swiftly rejected. Thus, at this time, the literature contains claims for and against the immune-privileged status of the cornea as a tissue. 

Within the recent past, it has been learned that the normal cornea expresses CD95L constitutively and that this expression promotes the survival of orthotopic corneal allografts. ¹⁴ Allografts prepared from mice deficient in CD95L expression are uniformly and promptly rejected when placed orthotopically in the eyes of normal mice. ^{15 16} These observations imply that the normal cornea possesses inherent immune privilege. However, the matter cannot be resolved if the grafts are placed orthotopically in the eye, because this site itself is immune privileged. 

The space beneath the kidney capsule has long been used by transplantation immunologists to study the fate of various solid tissue grafts. The survival patterns of heterotopic allografts of skin, kidney, heart, islets of Langerhans, and other solid tissues show that the subcapsular space of the kidney is a conventional, rather than a privileged, site. ^{17 18 19} Allografts placed beneath the kidney capsule induce typical systemic alloimmunity, and the grafts are rejected with an acute tempo that is commensurate with a conventional alloimmune response. In fact, Bellgrau et al. ²⁰ used the kidney capsule as a conventional site to test the putative immune-privileged status of testis allografts. These investigators reported that normal allogeneic testis tissue placed beneath the kidney capsule survived indefinitely, whereas CD95L deficient allogeneic testis tissue was rejected. 

In the experiments to be described, allogeneic corneal tissue was implanted under the kidney capsule and its fate examined clinically and histologically. By taking advantage of corneas from CD95L-deficient mice, we were able to establish that corneal tissue can be successfully grafted to this heterotopic site and that normal cornea displays at this heterotopic, putatively conventional site, the properties of an immune-privileged tissue. 

Materials and Methods

Mice and Anesthesia

Male BALB/c (H-2^d), C57BL/6 (B6, H-2^b), and C3H/HeJ (H-2^k) mice were purchased from Taconic Farm (Germantown, NY). Male B6Smn.C3H-Faslgld (B6-gld) and C3H/HeJ-Faslgld (C3H-gld) mice were purchased from Jackson Laboratories (Bar Harbor, ME). All mice were used at 8 to 10 weeks of age and were treated according to the ARVO Statement on the Use of Animals in Ophthalmic and Vision Research. Each mouse was anesthetized by intramuscular injection of a mixture of 3.75 mg ketamine and 0.75 mg xylazine before all surgical procedures. 

Heterotopic Corneal Transplantation under the Kidney Capsule

BALB/c and C3H mice were used as recipients, and BALB/c, normal B6, B6-gld, and C3H-gld were used as donors. Wedge-shaped segments of corneal tissue with or without epithelium were transplanted beneath the left kidney capsule. Conjunctival segments of similar size served as positive controls. Donor cornea was excised by a 2.0-mm trephine and vannas scissors and divided in half (∼2 × 1 mm). Corneal epithelium was removed by scraping in some experiments. Documentation of complete removal of epithelium was judged histologically. A skin incision was made in the left flank of recipient mice, and the muscle wall was incised and the kidney exteriorized. A subcapsular pocket was created between the kidney and the kidney cortex, and the graft was placed in the pocket. The kidney was replaced in the abdominal cavity, and skin was closed with 7-mm clips. In general, four to six animals per group served as recipients at each time point examined. These recipients were killed for clinical inspection and all the graft-bearing kidneys were subjected to histologic examination. 

Orthotopic Corneal Transplantation

Penetrating keratoplasty was performed as described previ-ously, ⁴ using normal C3H mice as recipients and C3H-gld mice as donors. Briefly, both donor and recipient central corneas (2-mm diameter) were excised by trephine and vannas scissors. Donor corneas were placed in the recipient bed with eight interrupted sutures (11-0 nylon). Sutures were removed at 8 days after grafting. 

Clinical and Histologic Evaluation of Heterotopic Corneal Grafts

Heterotopic corneal graft survival was assessed by visual inspection by a single observer (JH) under the operating microscope at 4, 7, 14, and 21 days after implantation. All the graft-bearing kidneys were removed for histologic assessment, fixed with 10% formalin, embedded in paraffin, sectioned, and stained with hematoxylin and eosin. Approximately 40 to 60 sections were prepared from each graft-bearing kidney. 

Evaluation and Scoring of Orthotopic Corneal Transplantation

Orthotopic grafts were observed with slit lamp microscopy at weekly intervals, and judgment of orthotopic corneal graft survival was performed in masked fashion according to a previously established scoring system ⁴ : 0, clear graft; 1+, minimal superficial nonstromal opacity; 2+, minimal deep stromal opacity; 3+, moderate deep stromal opacity; 4+, intense deep stromal opacity; and 5+, maximum stromal opacity. Grafts with opacity scores of 2+ or greater at 8 weeks were considered to have been rejected. 

Evaluation of Corneal Endothelial Cell Integrity

To discern the integrity of the endothelial monolayer of corneal grafts under the kidney capsule, immunocytochemistry was performed using a tight junction–associated protein marker, zonula occludens (ZO)-1. This anti-ZO-1 antibody has been demonstrated to identify the integrity of the corneal endothelial monolayer. ^{21 22 23} ZO-1 protein is associated with the apical junctional complex that forms in areas of cell–cell attachment. When normal cornea is sectioned transversely, a linear array of cells representing the corneal endothelium is readily observed. ZO-1 staining in cells with an intact monolayer appears in a regular, periodic pattern, representing points of cell–cell contact. Cells within the stroma and epithelium also stain positively with this antibody but do not show this specific pattern that is characteristic only of the endothelium. In injured or dead endothelial cells, cell–cell attachment is lost, the proteins forming the apical junctional complex dissociate, and ZO-1 staining is no longer discernible. The graft-bearing kidney was removed at 14 days, frozen in optimal cutting temperature (OCT) compound in acetone-dry ice, and stored at −80°C. The frozen specimens were sectioned at 5 μm by cryostat, fixed in acetone, and air dried. After incubation with 2% BSA for 10 minutes, the sections were incubated with rabbit polyclonal anti-ZO-1 antibody diluted to 4 μg/ml (Zymed Laboratories, San Francisco, CA) for 2 hours. Section were washed with phosphate-buffered saline (PBS) and incubated with fluorescein isothiocyanate (FITC)–conjugated donkey anti-rabbit IgG as a secondary antibody (6 μg/ml; Jackson ImmunoResearch, West Grove, PA) for 1 hour. Sections were again washed with PBS, mounted with medium containing propidium iodide (PI), according to manufacturer’s instructions (Vecta-stain; Vector Laboratories, Burlingame, CA), and observed by confocal microscopy. Tissue incubated in secondary antibody alone served as negative control. 

Immunocytochemistry and Immunohistochemistry

Immunocytochemical and immunohistochemical studies for MAC-1, CD3, CD4, CD8, and CD95L were performed on allografts under the kidney capsule, using phycoerythrin (PE)-labeled rat anti-mouse Mac-1 (Caltag, Burlingame, CA), FITC-labeled rat anti-mouse CD3, FITC-labeled rat anti-mouse CD4, FITC-labeled rat anti-mouse CD8, and PE-labeled mouse anti-mouse CD95L (Kay-10) monoclonal antibodies (Pharmingen, San Diego, CA). Graft-bearing kidneys were removed at 14 days, and most of the kidney tissue was cut away, leaving the corneal graft in place. For confocal microscopy, after fixation in acetone for 10 minutes, the small piece of graft-bearing kidney was incubated in PE-anti-Mac-1 or FITC-anti-CD3 antibodies, diluted to 4 μg/ml for 1 hour at room temperature. The sample was washed with PBS, mounted on a slide, and observed by confocal microscopy. For immunohistochemical studies for CD4, CD8, and CD95L, the graft-bearing kidneys were frozen in OCT compound in acetone-dry ice and stored at −80°C. The frozen specimens were sectioned at 5 μm by cryostat, fixed in acetone, and air dried. After they were washed in PBS, the sections were incubated in FITC-anti-CD4, FITC-anti-CD8, or PE-anti-CD95L antibodies diluted to 4 μg/ml for 2 hours at room temperature. Sections were again washed with PBS and observed by fluorescence microscopy. 

Results

Fate of Syngeneic Corneal Grafts Placed beneath the Kidney Capsule

It was first necessary to determine whether corneal tissue can survive in the heterotopic site beneath the kidney capsule. To explore this point, corneas devoid of limbus were removed from eyes of normal BALB/c mice, cut in half, and inserted (endothelial side down; i.e., adjacent to the cortex of the kidney) beneath the left kidney capsule of normal BALB/c mice. The grafts were inspected by dissecting microscope and removed for histologic examination on days 4, 7, 14, and 21 after implantation. In our initial experiments, we learned that the epithelial layer of full-thickness syngeneic corneal grafts rapidly proliferated and formed a keratinizing cyst that gave the graft an opaque appearance when examined with a dissecting microscope at 14 days (Fig. 1A ). The number of grafts examined is listed in Table 1 . Moreover, when examined histologically, the keratinized cyst, which derives from graft epithelium, was surrounded by, and presumably had incited, an inflammatory reaction within the kidney capsule itself (see Fig. 1B ), which was responsible for the clinical appearance of opacity. Despite this reaction, however, microscopic examination revealed that most of the stroma was free of infiltrating cells, and the endothelium appeared to be intact. No similar capsular response was observed when syngeneic corneas deprived of epithelium were paced beneath the kidney capsule. To avoid the capsular inflammatory response, all subsequent corneal grafts were performed after the epithelium of the donor cornea had been stripped away. Sixteen such grafts were performed. When examined at 4, 7, 14, and 21 days, the grafts were found to be clear by clinical inspection through a dissecting microscope (Fig. 1C) . It was easily possible to see features of the parenchyma of the kidney cortex through these grafts. Histologic examination revealed the stroma to be free of infiltrating cells, and the endothelium appeared to be intact (Fig. 1D) . 

To verify the integrity of corneal endothelium in grafts 14 days after implantation, frozen sections of grafts removed at this time were stained with an antibody to the zonula occludens (anti-ZO) normally expressed by intact endothelial cells (see Fig. 1E ). ZO-1 is a protein associated with tight junctions and can act as an indicator of corneal endothelial monolayer integrity. In Figure 1F , PI staining in fixed tissues indicates the position of cell nuclei. As revealed in this figure, the endothelial layer of the graft stained strongly with ZO-1 antibody on days 14 and 28 after grafting, indicating that corneal endothelium survived for an extended time and was intact on corneal grafts placed at this heterotopic site. Although the epithelium of convoluted tubules were also stained with ZO-1 in normal kidney (data not shown), the linear staining pattern found between corneal stroma and kidney tissue was easily identified as a typical pattern of a corneal endothelial monolayer. Moreover, the absence of PI staining of the endothelial cell layer in unfixed tissues examined by confocal microscopy confirmed that most ZO-1⁺ endothelial cells were viable (data not shown). 

Fate of Allogeneic Corneal Grafts Placed beneath the Kidney Capsule

Having confirmed that syngeneic corneal grafts, deprived of epithelial layers, survive with intact stromal and endothelial cell layers for extended intervals when placed beneath the kidney capsule of syngeneic recipients, we next examined the fate of allogeneic grafts placed at this heterotopic site. Corneas were excised from eyes of normal C57BL/6 mice, denuded of epithelial layer, and placed (endothelial side down) beneath the kidney capsule of normal BALB/c mice. When these grafts were inspected directly with the aid of a dissecting microscope at selected intervals thereafter, the grafts displayed a clear appearance. The parenchyma of the kidney cortex could be easily viewed through the corneal grafts 4 days after implantation. Moreover, this appearance was maintained throughout the entire observation period (up to 8 weeks). No evidence of new blood vessel formation was detected around the periphery of the grafts at any time after implantation, nor could neovessels be detected within the stroma of the grafts. In all these parameters, the clinical appearance of the grafts was indistinguishable from that of syngeneic grafts placed at this heterotopic site (see Fig. 2A ). During histologic examination of hematoxylin and eosin–stained sections of corneal allografts, the stroma and endothelial layers were observed to be intact at 4 days, and these layers remained so at all subsequent times examined. Moreover, there was no histologic evidence of blood vessel invasion of the stroma. Inflammatory cells, either within the stroma, or at the graft periphery, were only very infrequently observed (Fig. 2B) , and the incidence and intensity of these cellular infiltrates were similar to those of syngeneic grafts. Fourteen-day allografts were snap frozen, and sections were stained with anti-ZO antibody. As in syngeneic grafts, an endothelial layer that stained positively for zonula occludens was easily observed (Fig. 2C) . In addition, anti-CD95L antibody identified a layer of positively staining cells representing corneal endothelium (Fig. 2D) . Together, these results indicate that allogeneic corneas survive for prolonged intervals, perhaps indefinitely, when placed beneath the kidney capsule of normal mice. Our studies were terminated at 8 weeks. That allogeneic corneas survived in the subcapsular sinus of the kidney and that the fate of these corneas was strikingly similar to that of syngeneic corneal grafts implies that the allografts were actively avoiding alloimmune rejection—that is, they were displaying inherent immune privilege. 

Fate of Allogeneic Conjunctiva Placed beneath the Kidney Capsule

One interpretation of these results is that the kidney capsule site is itself immune privileged and thus incapable of promoting the rejection of allogeneic tissues. Although this interpretation is at odds with considerable evidence to the contrary in the literature, we tested this possibility directly by placing allogeneic conjunctival grafts beneath the kidney capsule. Conjunctiva is not expected to be immune privileged. It contains blood vessels and lymphatics, and the epithelium includes Langerhans’ cells, whereas the stroma includes macrophages and dendritic cells. Moreover, explants of conjunctiva do not secrete immunosuppressive factors when cultured in vitro. ²⁴ Accordingly, conjunctiva was removed from the ocular surface of eyes of normal C57BL/6 mice. These conjunctival segments were then placed beneath the kidney capsule of naive BALB/c mice. The implanted grafts were examined by dissecting microscopy and evaluated histologically. At 4 days after implantation, conjunctival allografts displayed a hazy, opalescent appearance, and at 7 days the grafts were profoundly opaque. Viewed through a dissecting microscope, conjunctival grafts were surrounded by a halo of swollen tissue, representing intense edema within the superficial kidney cortex. Neovessels were observed to arise from the capsule and the kidney cortex, to course across the cortical surface, and to penetrate into the substance of the graft (Fig. 3A ). Histologic examination of these grafts at 7 days revealed that the conjunctival epithelium had been destroyed and that the graft was neovascularized and heavily infiltrated with leukocytes, ranging from neutrophils to macrophages (Fig. 3B) . These findings confirm that conventional solid tissue allografts (in this case, conjunctiva) placed in the kidney subcapsular space elicit vigorous alloimmunity and undergo swift and intense immunologic rejection. Thus, the prolonged survival of allogeneic corneal grafts beneath the kidney capsule appears to reflect the immune-privileged status of the cornea as a tissue, rather than the immune privilege of the graft site itself. 

Fate of CD95 Ligand-Deficient Corneal Allografts beneath the Kidney Capsule

Expression of CD95 ligand by normal corneal cells has been demonstrated to contribute to the capacity of corneal allografts to survive when placed orthotopically in the eyes of normal mice. To determine whether CD95L also contributes to the immune-privileged displayed by allogeneic corneas placed under the kidney capsule, corneas were excised from the eyes of B6-gld mice (tissues of these mice do not display a functional CD95L molecule). The epithelial layer was removed, and the corneal tissues were implanted (endothelial side down) beneath the kidney capsule of normal BALB/c mice. When examined 4 days later, five of five grafts were clear, similar in appearance to normal allogeneic and syngeneic grafts. However, at 7 days after grafting, 3 of 6 gld grafts were opaque, and by 14 and 21 days, 14 of 14, and 6 of 6 gld grafts, respectively, were opaque (Fig. 4A ). Neovessels were also observed emerging from the kidney and penetrating into these grafts. Histologic examination showed a mononuclear cell infiltrate as early as 4 days after implantation in gld allografts (Fig. 4B) . At 7 days, and thereafter, the leukocytic infiltration of the stroma was intense, and numerous profiles of neovessels were observed in the stroma. The stroma itself had a “ground-glass” appearance, and Descemet’s membrane was shriveled and irregular (Fig. 6E) . These results indicate that allogeneic corneal grafts can be rejected in the subcapsular space of the kidney and that the condition for rejection is that the grafts fail to express CD95L. These results are summarized in Table 1 . 

Although the hematoxylin and eosin–stained photomicrographs clearly documented infiltration and disruption of the heterotopic gld corneal grafts, it was necessary to determine whether parenchymal cells of these grafts—corneal endothelium and keratocytes—showed the deleterious consequences of the presumed immune attack. To investigate, gld grafts at 14 days in residence were harvested and stained for ZO-1 to determine the integrity of the endothelial monolayer. Immunolocalization of ZO-1 in gld cornea is displayed in Figure 4C . There appeared to be a loss of integrity of the layer of corneal endothelium, with linear deposits separated by space without deposits. Similarly, profiles of ZO-1–positive keratocytes in the graft stroma in Figure 4C were less frequent and distinct than those displayed in Figure 2C . These findings indicate that gld grafts undergoing an inflammatory, presumed immune, insult have a deficit in corneal endothelial cells and stromal keratocytes, implying that these cells have been destroyed. 

Fate of CD95 Ligand-Deficient Syngeneic Corneas Placed beneath the Kidney Capsule

Because CD95L-deficient allogeneic corneas were rejected beneath the kidney capsule, but CD95L-bearing allogeneic corneas were not, the possibility existed that a deficit of CD95L might be prejudicial to the survival of any corneal graft, even syngeneic grafts. Moreover, there are no published reports of the fate of syngeneic CD95L-deficient grafts placed orthotopically in mice. To examine this possibility, corneas were removed from C3H-gld mice. Mice of the C3H background were chosen because the anterior chambers of these mice are deeper than those of C57BL/6 mice, and for this reason it is easier to perform successfully orthotopic transplants in C3H mice. Excised corneas were deprived of epithelium and either placed under the kidney capsule or grafted orthotopically into the normal eyes of normal C3H mice. Gld corneas, whether placed beneath the kidney capsule (n = 8), or grafted orthotopically (n = 8), showed no deleterious effects (see Table 2 and Fig. 5A ). Both sets of grafts healed in place and survived without clinical (data not shown) or histologic evidence (see Fig. 5B ) of inflammation or rejection for the observation period (8 weeks). Thus, a deficit of CD95L expression did not negatively affect the ability of corneal grafts to survive at either heterotopic or orthotopic sites. This supports the conclusion that the destruction of CD95L-deficient corneal allografts placed beneath the kidney capsule is immunologically mediated. 

Phenotypic Characterization of Leukocytes Infiltrating Rejecting CD95L-Deficient Corneal Allografts

We wanted to determine whether infiltrating cells within rejected gld grafts belonged to the classes of immune effectors typically involved in solid organ allograft rejection: macrophages and T lymphocytes. To that end, corneal allografts from gld donors were harvested at 14 days after implantation beneath the kidney capsule of BALB/c mice. The tissues were snap frozen, and sections were stained with monoclonal antibodies directed at marker molecules on macrophages (Mac-1) and T cells (CD3, CD4, CD8). In preliminary experiments, we examined the tissues of the normal kidney capsule, before graft implantation, for expression of these marker molecules. Mac-1⁺ cells (macrophages) were observed in the capsule itself, but no T cells were detected (data not shown). Examination of grafted gld tissues revealed that Mac-1⁺ cells were present in the stroma of the grafted cornea (Fig. 6A ), Mac-1⁺ cells were also observed in the infiltrate surrounding the graft and in the vessels of the kidney cortex (data not shown). Similarly, CD3⁺, CD4⁺, and CD8⁺ cells were detected in the stroma of these grafts (Figs. 6B 6C and 6D) , as well as in the surrounding infiltrate, the kidney cortex, and cortical vessels (data not shown). Neither macrophages nor T cells were observed in CD95L-bearing allogeneic corneal grafts placed beneath the kidney capsule (data not shown). These results indicate that CD95L-deficient allografts of cornea accumulate within and around the macrophages and T cells during the time the grafts are destroyed, supporting the contention that the rejection is immune mediated. 

Discussion

From an immunologic standpoint, the normal cornea is a remarkable, unusual tissue. At the cellular level, the epithelium resembles epidermis in that it is stratified, but it fails to keratinize, and it contains neither Langerhans’ cells nor (in mice) dendritic epidermal T cells. The stroma resembles the dermis of skin, but it has no blood and lymph microvessels, and it displays no adventitial cells of bone marrow origin, such as macrophages, dendritic cells, and mast cells. At the molecular level, although corneal epithelium expresses class I MHC molecules quite well, these molecules are poorly expressed on keratocytes and (especially) corneal endothelium. ^{25 26 27} Moreover, upregulation of class II molecules on keratocytes and corneal endothelial cells in response to interferon-γ is sluggish at best. ²⁸ In addition, explants of corneal tissue (but neither conjunctiva nor skin) secrete immunosuppressive factors that suppress T cell activation in vitro. ^{29 30} Although transforming growth factor (TGF)-β is present in supernatants of cultured corneal tissue, other as yet undescribed immunosuppressive factors are also present. Finally, corneal cells, especially endothelium and epithelium, constitutively express CD95L, a molecule that functions as a receptor for CD95. ¹⁴ CD95 is expressed on many cells but particularly on activated T lymphocytes. ³¹ Engagement of CD95 on T cells by CD95L triggers apoptosis among the CD95-bearing cells, and this mechanism of deletion has been implicated in the ability of orthotopic corneal allografts to resist immune rejection. ^{15 16} Some, or all, of these facets of the normal cornea contribute to its reputation as an immune-privileged tissue. 

The capacity of allogeneic corneal tissue to survive, and indeed thrive, when implanted beneath the kidney capsule of normal mice is striking testimony to the privileged nature of the cornea. Clinical and histologic comparisons of heterotopically placed allogeneic corneal tissue with syngeneic corneal tissue revealed virtually no differences. Based on these findings, we conclude that cornea is an immune-privileged tissue. In addition, our results confirm the long-held belief that the inherent immune-privileged nature of the cornea helps to account for the extraordinary success of orthotopic corneal allografts. 

There has been one previous study of heterotopic allogeneic corneal transplants placed beneath the kidney capsule in mice. Guymer and Mandel ³² have reported that corneal grafts survive longer beneath the kidney capsule than do grafts of skin or islets of Langerhans. However, their study differed from ours in several respects. First, their grafts were full-thickness corneas, including epithelium. Second, their grafts also contained limbal tissue, which is known to contain many Langerhans’ cells and elements of both blood and lymph vessels. We used grafts that were devoid of limbus, and we removed the epithelium that obscured our ability to judge the quality of the graft by clinical examination. Moreover, the scoring system Guymer and Mandel used for evaluating the fate of heterotopic corneal grafts was not described sufficiently to allow any direct comparison with our results. Although the results of our present study resemble the findings reported by Guymer and Mandel, the differences between the two studies make a direct comparison difficult. 

Although many factors are thought to contribute to the privileged nature of the cornea as a tissue, expression of CD95L is critical. In our experiments, the simple exclusion of CD95L expression on corneal allografts was sufficient to render these grafts vulnerable to intense immunologic rejection. Whereas neither syngeneic nor normal allogeneic corneas displayed any evidence of infiltration with macrophages and T cells, CD95L-deficient corneal grafts contained macrophages and T cells that were easily detectable during the destruction of the grafts. These findings support the view that CD95L expression on cornea provides the tissue with a “shield” that prevents attacking CD95⁺ leukocytes from damaging the cells within the graft. It is too early in our analysis to determine whether other factors thought to be important in corneal immune privilege are as important as CD95L expression. Experiments to test other factors are currently under way. 

It is of considerable interest that MHC plus minor histocompatibility antigen–disparate corneas from CD95L⁺ donors placed beneath the renal capsule are not rejected, whereas genetically identical grafts placed on the skin or subcutaneously are. ⁸ The transplantation immunology literature indicates that both the subcapsular space of the kidney and the skin are“ conventional” sites for alloimmune rejection. Specifically, the subcapsular space resembles skin in its possession of lymphatics that drain through a superficial capsular system and a deeper hilar system to the para-aortic nodes. ³³ The capsular microcapillary network is supplied by interlobular arteries of the kidney. ³⁴ Our finding that the unperturbed kidney capsule contains Mac-1⁺ cells, as does the dermis of skin, supports evidence of others that these potential antigen-presenting cells are constitutively present in the kidney capsule. Despite these similarities, however, our findings reveal that the kidney capsule and the skin are not equivalent “conventional” sites. We are not the first to observe this difference. In a personal communication, Richard Duke states that both CD95L⁺ and CD95L⁻ testis allografts placed subcutaneously are rejected, whereas only the latter are rejected beneath the kidney capsule. Beyond the detection of macrophages and microvessels in the kidney capsule, our experimental results give us little insight into why there should be a difference between the skin and the kidney capsule as graft sites—at least for immune-privileged tissues. Among transplantation immunologists, there is general agreement that, compared with other solid tissues, orthotopic skin allografts arouse the most vigorous alloimmune responses, and there is experimental evidence that suggests that this is an inherent property of skin as a graft. Our results suggest that the skin—as a graft site—may also be an exceptionally strong promoter of alloimmunity. Studies to explore this point are needed to resolve this issue and to understand the biologic bases. 

We were unable to use corneal grafts with intact epithelial layers for implantation beneath the kidney capsule. The ability of the epithelium to proliferate extensively in these corneal implants and to form epithelial cysts with keratinized pearls rendered it difficult to assess the viability and inflammatory qualities of the grafts. For these reasons we must qualify our statement that “the cornea possesses inherent immune privilege” by restricting the statement to corneal tissue devoid of epithelium. Pertinent to this issue, in the recent past we have reported that implants of epithelium-containing allogeneic corneas implanted into the anterior chamber elicit a transient, systemic, donor-specific delayed hypersensitivity. ³⁵ By contrast, corneal implants devoid of epithelium (similar to the grafts we placed beneath the kidney capsule) failed to induce donor-specific delayed hypersensitivity when placed in the anterior chamber. Thus, corneal epithelium by itself may not actually display immune privilege. However, in orthotopic corneal allografts, such as we and others have performed in mice, the graft epithelium is rapidly eliminated and replaced by recipient epithelium, so the relevance of privilege for corneal epithelium may be moot with regard to the ultimate success or failure of the graft. ³⁶  

It is rather surprising that corneal endothelium (on both syngeneic and allogeneic grafts) was able to survive for prolonged intervals under the kidney capsule. It could have been predicted that this cell type, which is regarded both clinically and experimentally as particularly vulnerable to injury by trauma and inflammation, would have rapidly deteriorated at this heterotopic site. On the contrary, corneal endothelial cells survived the trauma of transplantation, and the cells were able to maintain their morphologic integrity during the graft’s tenure, sufficient to retain tight junctions between and among themselves. We have no direct information concerning the “pump” activity of these cells, but the findings that long-standing grafts beneath the kidney capsule retained their clarity and that the stroma of these grafts displayed no histologic evidence of edema suggest that at least some of the physiologic properties of the graft endothelium were still operative. 

Confirmation of the immune-privileged nature of corneal tissue encourages us to continue to search for, and assign relative importance to, the multiple factors thought to be involved. Clearly, CD95L expression is critical, but other properties of the cornea may also contribute. Moreover, we are eager to determine whether a heterotopically implanted cornea can create an immunosuppressive microenvironment around itself and perhaps even promote immune deviation to the alloantigens it expresses. Experiments to test these and other possibilities are currently under way. 

 Supported in part by National Institutes of Health Grant EY10765.

 Submitted for publication February 24, 1999; revised July 30, 1999; accepted September 3, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

Clinical and histologic finding of syngeneic (BALB/c) corneal segments with epithelial layer (A, B) or without epithelial layer (C, D) at 14 days. Arrows: graft margins in (A) and (C). Arrowheads: the epithelial mass shed from the surface of the proliferating epithelium; inset: high-power image of stroma in (B). Arrow: nuclei of stromal cells; arrowhead: nucleus of endothelium with intact Descemet’s membrane in (D). (B, D) Hematoxylin and eosin. (E) ZO-1 staining pattern in a normal BALB/c cornea. Arrows: linear deposit staining pattern of corneal endothelium. Confocal image. Inset: there was no positive staining for ZO-1 when the tissue was incubated with secondary antibody alone. (F) Immunolocalization of ZO-1 in epithelium-deprived corneal graft beneath the syngeneic kidney capsule at 14 days, stained with FITC-anti-ZO-1 antibody (green) and PI (red). Confocal image. Arrows: linear deposit of ZO-1. Magnification, (B, D) ×17; inset, ×100. K, kidney; CS, corneal stroma; DM, Descemet’s membrane; EP, corneal epithelium.

Table 1.

 

Fate of Corneal Grafts under the Kidney Capsule of BALB/c Mice

Corneal Graft*	Days, †	Clear/Grafts, ‡
BALB/c (epithelium+, n = 21)	4	0/5
	7	0/6
	14	0/6
	21	0/4
BALB/c (epithelium−, n = 16)	4	2/2
	7	3/3
	14	7/7
	21	4/4
C57BL/6 (epithelium−, n = 27)	4	4/4
	7	6/6
	14	10/11
	21	5/6
B6-gld (epithelium−, n = 31)	4	5/5
	7	3/6
	14	0/14
	21	0/6

*  Corneas excised by trephination from donor eyes. Epithelium was denuded by scraping with a cotton swab. Stroma and endothelium were cut in half, and each half was placed beneath the kidney capsule of BALB/c mice.

†  Days after implantation when graft was inspected via dissecting microscope.

‡  Grafts without opacity, through which renal parenchyma could be visualized were considered “clear.” Data presented as number of clear grafts and total number of grafts examined.

Figure 2.

 

Clinical and histologic finding of allogeneic (C57BL/6) corneal segments without epithelial layer (A, B) at 14 days. Arrows: graft margins; inset: high-power image of stroma. Hematoxylin and eosin. (C) Immunolocalization of ZO-1 at 14 days, stained with FITC-anti-ZO1 antibody. Confocal image. Arrows: linear deposit; inset: no positive staining for ZO-1 when the tissue was incubated in secondary antibody alone. (D) Expression of CD95L on C57BL/6 corneal endothelium at 14 days, stained with PE-anti-CD95L antibody. Confocal image. Arrows: CD95L⁺ endothelium. Inset: no staining for CD95L when the tissue was incubated with isotype control alone. Magnification (B) ×19; inset (B)× 100. K, kidney; CS, corneal stroma.

Figure 3.

 

(A) Clinical findings of allogeneic (C57BL/6) conjunctiva grafted in the kidney at 7 days. Arrows: margins of the graft; large arrowheads: rim of swollen tissue (halo) that surrounded the graft; small arrowheads: neovessels that formed in the periphery of the graft bed. (B) Histologic findings of allogeneic (C57BL/6) conjunctiva at 7 days. Arrowheads: destroyed conjunctiva grafts. Hematoxylin and eosin; magnification, ×21.

Figure 4.

 

(A) Clinical findings of CD95 ligand-deficient corneal allograft (B6-gld) without epithelial layer implanted beneath the kidney capsule at 14 days. Arrows: margins of the graft; arrowheads: neovessels that formed in the periphery of the graft bed. (B) Histologic findings at 4 days. Inset: high-power image of stromal infiltrating cells. (C) Immunolocalization of ZO-1 at 14 days. Arrows: spaces where linear deposit was missing. Magnification, (B) ×43; inset, ×100. K, kidney; CS, corneal stroma.

Table 2.

 

Fate of Syngeneic Corneal Grafts under the Kidney Capsule

Corneal Graft	Days	Clear/Grafts
FasL-positive; BALB/c → BALB/c (n = 12)	4	2/2
	7	3/3
	14	7/7
FasL-negative; C3H/HeJ-gld → C3H/HeJ (n = 8)	14	4/4
	21	4/4

 See footnotes to Table 1 for explanation of columns’ contents.

Figure 5.

 

(A) Opacity score of CD95 ligand-deficient syngeneic corneal grafts placed orthotopically. (B) Histologic appearance of CD95 ligand-deficient syngeneic corneal grafts placed beneath the kidney capsule at 14 days. Inset: high-power image of stroma. Hematoxylin and eosin; magnification, ×19; inset: ×100.

Figure 6.

 

Stroma of B6-gld corneal graft beneath BALB/c kidney capsule at 14 days. Confocal images of en face sections stained with (A) αMac-1 or (B) αCD3. Conventional fluorescent microscope images of cross sections of graft stroma stained with (C) αCD4 or (D) αCD8. (E) Large arrowhead: neovessel in stroma of the graft; small arrowheads: disrupted and curled Descemet’s membrane. Hematoxylin and eosin. Magnification, (A through D) ×200; (E) ×80.

The authors thank Deshea L. Harris for help with confocal microscopy and graphic support, Peter Mallen for expert assistance with preparation of the figures, and Jacqueline Doherty for contributions made to the research efforts.