CD95 Ligand Expression on Corneal Epithelium and Endothelium Influences the Fates of Orthotopic and Heterotopic Corneal Allografts in Mice

Abstract

purpose. To determine the extent to which CD95 ligand (CD95L) expressed on corneal epithelium and endothelium influences survival of cornea grafts placed orthotopically and heterotopically in the anterior chamber (AC), an immune-privileged site.

methods. Corneas from eyes of C57BL/6 (B6) and B6.gld (CD95L deficient) mice were (1) rendered into small full-thickness fragments, with or without an epithelial layer, and placed behind recipient corneas in the ACs of BALB/c eyes, while BALB/c corneas were similarly implanted in eyes of B6 and B6.lpr (CD95 deficient) mice; or (2) corneas were grafted orthotopically in BALB/c eyes as intact corneas or as composite corneas in which epithelium from one donor source was layered in vitro onto epithelium-deprived stroma+endothelium from another donor source before grafting. The fate of the grafts was assessed clinically and histologically, and the capacity of the grafts to sensitize recipient mice to donor alloantigens (delayed hypersensitivity, DH) was evaluated.

results. Allogeneic, full-thickness B6.gld corneal fragment grafts placed in the AC of BALB/c mice were rejected and sensitized their recipients, whereas epithelium-deprived B6.gld cornea fragments survived indefinitely and failed to sensitize their recipients. BALB/c corneal fragment grafts placed in the AC of C57BL/6 or B6.lpr eyes were rejected. Orthotopic cornea grafts composed of B6.gld epithelium layered onto wild-type B6 stroma and endothelium were rejected at a tempo and incidence similar to full-thickness B6 grafts, whereas orthotopic composite cornea grafts containing B6 epithelium layered onto B6.gld stroma+endothelium were vigorously rejected.

conclusions. CD95L expression on epithelium of full-thickness cornea fragment grafts placed in the anterior chamber of BALB/c eyes protects these heterotopic grafts from rejection but has only a trivial role to play in determining the fate of orthotopic corneal grafts. In the latter type of corneal grafts, CD95L expression on the endothelium plays an essential role in preventing graft rejection.

The success of penetrating keratoplasty in low-risk eyes and the high rate of failure of orthotopic corneal allografts in high-risk eyes are dramatic expressions of the importance of immune privilege in the fate of corneal transplants. ^{1 2} An impressive array of individual factors—molecular, cellular, anatomic—have been implicated in creating and maintaining ocular immune privilege. Among these factors, the constitutive expression of CD95 ligand (CD95L, Fas ligand) on numerous ocular cell types (including corneal parenchymal cells) has garnered considerable attention. ^{3 4 5} CD95L binds to its coreceptor CD95 (Fas) on target cells. If the target cell is a CD95⁺ effector T lymphocyte, apoptosis is triggered and the cell dies. Progressive elimination of CD95⁺ T cells by antigen-bearing CD95L⁺ cells has been strongly implicated in peripheral immunologic tolerance in adult animals. In the case of the cornea, constitutive expression of CD95L on corneal cells has been firmly linked to the high rate of success of orthotopic corneal allografts placed in low-risk (normal) eyes of BALB/c mice. ^{6 7} Circumstantial evidence suggests that elimination of effector T cells that migrate to and threaten the integrity of cornea allografts is responsible for graft acceptance. CD95L expression on corneal endothelial cells has also been demonstrated to confer immune privilege on epithelium-deprived corneal allografts placed in the non–immune-privileged subcapsular space of the kidney of BALB/c mice. ^{8 9} These experiments are of interest because the expression of CD95L on corneal endothelium in these heterotopic grafts not only promotes survival, but also prevents the induction of donor-specific immunity. 

Beyond lymphocytes, CD95L acts on other CD95⁺ leukocytes, in particular polymorphonuclear leukocytes (PMN). Whereas CD95L interactions with CD95⁺ T cells leads to lymphocyte elimination and is thought to promote graft acceptance, CD95L interaction with CD95⁺ PMNs leads to activation and is known to promote graft rejection. Tissue grafts genetically manipulated to express CD95L have been found to be particularly vulnerable to destruction by a PMN-dependent process. ¹⁰ Thus, CD95L expression on nonlymphoreticular tissue grafts, such as the cornea, can have countermanding effects, promoting deletion of CD95⁺ effector T cells (which is protective of the graft) and promoting inflammation by CD95⁺ PMNs (which is deleterious to the graft). 

Not only do epithelial and endothelial corneal cells constitutively express CD95L on their surfaces, but aqueous humor contains sizable amounts of soluble CD95L. ^{11 12} This is relevant to ocular immune privilege because soluble CD95L has been found to neutralize CD95L in its membrane form. ^{13 14} Although this effect protects CD95⁺ T cells from CD95L-triggered apoptosis, it also prevents CD95⁺ PMNs from being activated and triggering destructive inflammation. The ocular anterior segment, therefore, contains both membrane bound and soluble forms of CD95L. This makes it important to determine which of these molecular forms has the dominant influence in the eye, especially with respect to the fate of corneal transplants. 

To gain insight into this question, we have implanted fragments of full-thickness and epithelium-deprived corneas into the anterior chamber of eyes of syngeneic and allogeneic recipient BALB/c and C57BL/6 mice. We have used this model systemic previously to examine the immunologic consequences and fate of allogeneic and xenogeneic cornea fragments placed in the anterior chamber of mouse eyes. ^{15 16 17} In the present experiments we assessed the survival of fragments of allogeneic cornea grafts from donors with defective CD95L genes (C57BL/6.gld), the impact these grafts make on the recipient BALB/c immune response, and whether CD95L expression on corneal epithelium or endothelium is more important in governing graft outcome. Our results indicate that CD95L expression on corneal epithelium prevents rejection of corneal fragment grafts in the anterior chamber of BALB/c mice and interferes with recipient sensitization to donor alloantigens. By contrast, similar allografts placed in the anterior chamber of C57BL/6 mice were swiftly rejected. Although CD95L is expressed on both corneal epithelium and endothelium, the latter expression is more important in determining the survival of orthotopic corneal allografts in eyes of BALB/c mice. 

Materials and Methods

Mice and Anesthesia

Eight- to 10-week-old female BALB/c and C57BL/6 mice were purchased from Taconic Farm (Germantown, NY). These strains are disparate at the murine major histocompatibility complex, H-2, and at multiple minor histocompatibility loci. Seven- to 10-week-old female B6Smn. C3H-Faslgld (B6.gld) and B6.lpr (lpr ^−/−) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). All animals were treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. In all experiments, BALB/c (H-2^d) mice received intracamerally cornea fragments from eyes of C57BL/6 (H-2^b) donors, and C57BL/6 and B6.lpr mice received intracamerally cornea fragments from eyes of BALB/c (H-2^d) donors. Each animal was deeply anesthetized with an intraperitoneal injection of 3 mg ketamine and 0.007 mg xylazine before all surgical procedures. 

Insertion of Cornea Fragments into the Anterior Chamber of the Eye

Central corneas from donor eyes were excised and cut by Vannas scissors into wedge-shaped fragments measuring approximately 0.3 × 1.5 mm, as described previously. ¹⁵ In some experiments the corneal epithelium was removed from the anterior surface of the graft by scraping before preparation of epithelium-deprived stroma+endothelium fragments. Individual fragments were then inserted into the anterior chamber of one eye of anesthetized recipients through an oblique incision near the limbus. The fragment was placed with its epithelial or stromal surface adjacent to the recipient endothelium, and the fragment was held in place temporarily with a small air bubble. 

Orthotopic Cornea Transplantation

The central 2-mm portions of donor corneas were excised and secured in recipient graft beds with eight interrupted 11-0 nylon sutures, as described previously. ¹⁸ All grafted eyes were examined after 72 hours for technical difficulties and grafts displaying these problems (e.g., infection, hemorrhage into the anterior chamber) were excluded from further consideration (fewer than 5% of grafts displayed technical difficulties that warranted exclusion of the animal from the study). Sutures holding the graft in place were removed on day 7 after surgery. 

Evaluation and Scoring of Orthotopic and Heterotopic Corneal Transplants

The qualities of corneal grafts were evaluated by slit lamp biomicroscopy once a week by two independent observers, one of whom was blinded to the experimental design. When scoring conflicts arose, the score of the nonblinded observer was discarded. At each time point, intracameral fragment grafts were scored for degree of opacification on a scale of 1+ to 3+. Grafts scored 1+ were clear, and one could readily observe the pupil margins behind the graft; grafts scored 2+ were hazy, making it difficult to see the pupil margin behind the graft; and grafts scored 3+ were opaque, and the pupil margin was not visible through the graft. Orthotopic cornea grafts were scored according to a grading system (0–5+) described previously. ¹⁸ Orthotopic grafts were judged to be rejected if a clinical score of 3+ or more was achieved within 2 weeks. Rejection was also diagnosed if orthotopic grafts achieved a score of 2+ that persisted throughout the remainder of the 8-week observation interval. Orthotopic grafts that experienced transient opacification that subsequently cleared were considered to be accepted. Results of representative experiments are presented in the figures. Each experiment was repeated at least twice with a comparable number of animals. The results of repeated experiments were very similar. 

Creation of Composite Corneal Grafts for Orthotopic Transplantation

Excised corneas were incubated in 20 mM EDTA at 37°C for 40 minutes, after which the corneal epithelial layer was easily removed with a forceps. The epithelial layers were then washed with saline and layered onto the denuded surface of stroma+endothelium, prepared from a different corneal source. To prepare the denuded stroma+endothelium, corneas were removed, and the epithelium was removed by scraping, as described previously. ^{19 20} Composite corneas, composed of epithelium from one source layered over stroma+endothelium from a different source, were then carefully grafted orthotopically into eyes of recipient mice. The fate of these grafts was evaluated as described for intact, full-thickness orthotopic corneal grafts. 

Assessment of Donor-Specific Delayed Hypersensitivity

At selected time intervals after grafting, 1 × 10⁶ γ-irradiated (2000 rad) C57BL/6 spleen cells in 10 μL medium were injected intradermally into the right ear pinnae of mice bearing orthotopic corneal allografts, or fragments of allogeneic corneas implanted in the anterior chamber. As the positive control, full-thickness allogeneic cornea fragments were implanted into the subcutaneous space of the dorsum of hind footpads of mice 7 days before ear challenge. Ear-swelling responses were measured with a low-pressure engineer’s micrometer (Mitutoyo; MTI Corp., Paramus, NJ) at 24 and 48 hours. Ear swelling, as a measure of delayed hypersensitivity (DH) was expressed as follows: specific ear swelling = (24-hour measurement of left ear – 0-hour measurement of right ear) – (24-hour measurement of left ear – 0-hour measurement of left ear) × 10⁻³ mm. Ear swelling responses at 24 hours after intrapinnae challenge are presented as mean swelling responses ± SEM for panels of five mice. Each experiment was repeated at least twice with similar results. 

Statistical Methods

We constructed Kaplan-Meier survival curves and used the Breslow-Gehan-Wilcoxon test to compare the rates of corneal allograft survival among different groups. P < 0.05 was deemed to indicate a significant difference in survival rates. DH results were analyzed by ANOVA. Mean results were considered to be significantly different when P < 0.05. 

Results

Fate of Full-Thickness CD95L-Deficient Allogeneic Cornea Fragments Implanted in the Anterior Chamber of BALB/c Eyes

In previous experiments, ¹⁵ we have determined that fragments of allogeneic corneas (C57BL/6) implanted onto the central posterior surface of corneas of normal mouse eyes (BALB/c) are routinely accepted for prolonged intervals of time. Virtually all such grafts display a clear appearance (the recipient papillary margin can be visualized easily through the graft). Because normal corneas express CD95L on epithelium and endothelium (but not apparently on stromal cells), ²¹ we investigated whether CD95L expression is important in the survival of corneal fragment grafts in this immune-privileged site. Full-thickness cornea fragments (0.3 × 1.5 mm), obtained from B6.gld and normal B6 donors, were inserted into the anterior chamber of BALB/c mice through a lateral corneal incision. The graft was placed against the central region of the posterior surface of the recipient cornea with graft endothelium facing posteriorly. At weekly intervals thereafter, the quality of the graft was assessed clinically by slit lamp biomicroscopy. Grafts were graded on an arbitrary scale of 1 to 3+, according to the criteria described in the Materials and Methods section. Grafts with a score of 1+ were clear, whereas grafts with a score of 3+ were opaque. The results of this experiment, conducted over an 8-week interval, are presented in Figure 1A . Whereas the scores of most of the fragments prepared from B6 corneas remained at 1+ throughout this observation interval (data not shown), an increasing fraction of B6.gld grafts acquired scores in excess of 3+. With the exception of two of nine B6.gld fragment grafts, all the remaining grafts displayed a 3+ score indicative of rejection at 3 weeks after grafting, and the opacity persisted through the observation interval, indicating irreversible rejection. The clinical appearance of a 3+ opaque B6.gld fragment is displayed in Figure 1B . Not only was the graft opaque and swollen, it was surrounded by a vigorous neovascular inflammatory response within the recipient stroma. The limbal vessels near the graft were engorged. These findings indicate that rejection of B6.gld cornea fragments is accompanied by intense inflammation and neovascularization, a response that was not observed when wild-type B6 cornea fragments were implanted in the anterior chamber of BALB/c mice. We conclude that the absence of expression of CD95L on full-thickness allogeneic cornea fragments correlates strongly with rejection, implying that CD95L protects heterotopic wild-type grafts in the anterior chamber from inflammation, neovascularization, and rejection. 

Fate of Epithelium-Deprived CD95L-Deficient Allogeneic Cornea Fragments Implanted in the Anterior Chamber

It is relatively simple to remove the epithelial layer from the surface of cornea grafts by scraping the surface gently with a surgical knife. We used this approach to prepare allogeneic cornea fragments deprived of epithelium as a means of examining the role that CD95L expression on corneal epithelium plays in determining graft survival. Epithelium-deprived cornea fragments from B6 and B6.gld donors were implanted in the anterior chamber of BALB/c eyes. The grafts were examined clinically as described earlier. As the results presented in Figure 2 reveal, none of the grafts became opaque, whether from B6 donors (data not shown) or B6.gld donors. Moreover, inflammation was not detected, and neovessels did not appear in the adjacent recipient stroma during the tenure of these epithelium-deprived, CD95L-deficient grafts. A small minority of B6.gld fragments acquired a haze that persisted throughout the 8-week observation interval, but failure to progress to complete opacity suggests that these grafts were not rejected. This finding, in conjunction with the fate of full-thickness B6.gld fragments, strongly suggests that CD95L expression on the epithelium of corneas has the capacity to promote acceptance of cornea fragment grafts when placed in the immune-privileged anterior chamber. Moreover, neovascularization and inflammation are only triggered in recipient corneal stromata by fragment grafts that contain an epithelium that is deficient in CD95L expression. 

Fate of Full-Thickness and Epithelium-Deprived Allogeneic Cornea Fragments Implanted in the Anterior Chamber of Eyes of CD95-Deficient C57BL/6 Mice

We next wished to confirm that CD95L interaction with its typical ligand, CD95, was responsible for preventing rejection of allogeneic full-thickness (epithelium-containing) cornea fragments implanted in the anterior chamber of BALB/c eyes. Because a CD95-deficiency has not been described in BALB/c mice, we resorted to placing full-thickness and epithelium-deprived cornea fragments from BALB/c mice into the AC of wild-type B6 mice and B6.lpr mice (deficient in CD95 expression). Allogeneic fragments devoid of epithelium were fully accepted through the 8-week observation interval by both C57BL/6 (5/5 accepted) and B6.lpr (5/5 accepted) recipients. However, full-thickness cornea fragments from BALB/c mice were rejected within 4 weeks in the eyes of both C57BL/6 (0/5 accepted) and B6.lpr (2/5 accepted) mice. Thus, C57BL/6 and BALB/c mice as recipients differ in their ability to accept intracameral, full-thickness corneal allografts. Only BALB/c mice are able to accept these grafts. Because of this strain difference in immune reactivity to allogeneic corneal tissue, we were unable to prove with these experiments that CD95 is the relevant ligand for CD95L when allogeneic C57Bl/6 cornea fragments are accepted in the anterior chamber of BALB/c eyes. 

Histologic Appearance of CD95L-Deficient Allogeneic Cornea Fragments Implanted in the Anterior Chamber of BALB/c Eyes

BALB/c eyes bearing full-thickness or epithelium-deprived cornea fragments from B6 and B6.gld donors were subjected to histologic analysis. As reported previously, ¹⁵ both full-thickness and epithelium-deprived allogeneic cornea fragments from B6 mice in anterior chambers of BALB/c mice eyes displayed healthy appearances with intact cornea layers, and absence of neovascularization and leukocyte infiltrates (data not shown). By contrast, similarly placed, clinically opaque, full-thickness B6.gld cornea fragments appeared as cysts, surrounded by a thin, uneven layer of epithelium in which nuclei of inflammatory leukocytes were dispersed. No recognizable corneal stroma was observed, and no corneal endothelium was seen. The cyst was filled with necrotic cells, inflammatory cells, and amorphous debris. The portion of the graft that projected into the anterior chamber was covered with a loose layer of eosinophilic material in which inflammatory cells were prominent (Fig. 3A) . The portion of the graft that abutted onto the posterior surface of the recipient cornea appeared to rest on recipient Descemet’s membrane that is devoid of adjacent endothelial cells. The recipient stroma immediately above Descemet’s membrane contains inflammatory cells and profiles of neovessels. B6.gld grafts with this histologic picture and lacking any semblance of the architecture of the original cornea, were regarded as rejected. Quite a different histologic pattern was observed in sections of epithelium-deprived B6.gld cornea fragments that had been scored clinically as 1+ (Fig. 3B) . Regularly arrayed donor stromal elements were present and rested on the posterior surface (adjacent to Descemet’s membrane) of the recipient cornea. Descemet’s membrane of the graft was intact, albeit somewhat folded, and an intact layer of endothelium covered the posterior surface of the donor fragment. Thus, the clinical clarity of epithelium-deprived B6.gld cornea fragments correlates positively with the histologic image of an architecturally intact donor fragment devoid of inflammation and neovascularization. We conclude that the inflammatory and neovascular destruction that befalls full-thickness B6.gld cornea fragments in eyes of BALB/c mice is determined by the lack of CD95L on the epithelium. When the epithelium was artificially removed from the fragments, neither inflammation nor neovascularity appeared in or around the allograft, implying that CD95L expression on donor endothelium, unlike that on epithelium, is irrelevant to the graft’s fate: These grafts were all accepted. 

Evidence of Donor-Specific DH in Recipients of Intraocular CD95L-Deficient Allogeneic Cornea Fragments

Orthotopic corneal allografts are rejected in mice almost exclusively by CD4⁺ T cells of the type that mediate DH. ^{22 23} In previous studies we have found that full-thickness B6 cornea fragments placed in the anterior chamber of BALB/c eyes sensitized the recipients (DH) to donor alloantigens, but fragments deprived of epithelium evoked no detectable immune response. ¹⁵ We next examined whether BALB/c mice bearing intracameral B6.gld cornea fragments, with or without an epithelial layer, acquired donor-specific DH. Panels of mice bearing B6 or B6.gld cornea fragments received intradermal injections of γ-irradiated B6 spleen cells (1 × 10⁶ /10 μL) into their ear pinnae at 2 or 4 weeks after grafting. Positive control mice were immunized by implantation of full-thickness B6 fragments placed in the subcutaneous space on the dorsum of the hind foot 7 days before ear challenge. Naive mice that were merely ear challenged with B6 spleen cells served as negative control subjects. The results of mice challenged at 2 weeks after grafting are presented in Figure 4 . Mice bearing full-thickness B6.gld cornea fragments mounted positive ear swelling responses, indicating the presence of DH directed at B6 alloantigens. Donor-specific DH was also detected when assayed at 4 weeks after grafting (data not shown). By contrast, mice bearing epithelium-deprived B6.gld cornea fragments mounted feeble ear-swelling responses, indistinguishable from negative control subjects. Mice bearing full-thickness B6 corneas at 2 weeks displayed negative ear-swelling responses (data not shown), but these responses turned positive at 4 weeks, as expected and as reported previously. ¹⁵ Epithelium-deprived B6 fragment grafts failed to sensitize their recipients (data not shown). We conclude that full-thickness B6.gld cornea fragment grafts promote their own rejection by engendering an early and intense donor-specific DH in recipient mice. This capacity to induce recipient allosensitization is a special property of CD95L-deficient epithelium, because epithelium-deprived fragments from B6.gld donors induced neither specific DH nor their own rejection. Thus, in the special context of full-thickness allogeneic cornea fragments placed in the anterior chamber, expression of CD95L on the epithelium serves to delay the onset of sensitization to donor alloantigens, helping to explain why these grafts are not rejected. 

Fate of Orthotopic Corneal Grafts Composed of Either CD95L-Deficient Epithelium or Stroma+Endothelium

Composite cornea grafts can be assembled readily in vitro by placing an intact layer of corneal epithelium from one source on the epithelium-deprived surface of stroma+endothelium from a different source. This approach has been used experimentally to demonstrate that corneal epithelium has a powerful role to play in suppressing angiogenesis and inflammation in the stroma of a graft placed orthotopically. ^{19 20} We used this technique to inquire into the relative importance of CD95L expression on corneal epithelium or endothelium in determining the fate of orthotopic corneal allografts. Composite grafts were prepared in vitro in which B6.gld epithelium was layered onto epithelium-deprived B6 corneas, B6 epithelium was layered onto epithelium-deprived B6.gld corneas, or control composite grafts in which B6 epithelium was layered onto epithelium-deprived B6 corneas. These composite grafts were then placed orthotopically into normal eyes of BALB/c mice. Graft survival was assessed by clinical examination. The results of these experiments are presented in Figure 5 . Composite grafts composed of B6 epithelium layered onto B6 stroma+endothelium displayed an incidence and rate of rejection in BALB/c eyes comparable to that of intact, full-thickness B6 cornea grafts. Composite grafts in which B6 epithelium was replaced with B6.gld epithelium layered on B6 stroma+endothelium had a relatively high rate of rejection during the first 4 weeks, but eventually the proportion of accepted composite grafts was similar to that of the intact, full-thickness control (Fig. 5A) . The extent of neovascularization triggered by these composite grafts was similar to that elicited by intact full-thickness B6 allografts. This finding implies that CD95L expression on corneal epithelium has a relatively minor role to play in prejudicing the survival of orthotopic corneal allografts. A very different outcome was observed with composite grafts in which B6 epithelium was layered onto CD95L-deficient stroma+endothelium (Fig. 5B) . Almost all these composite grafts suffered rejection, and the tempo of rejection for these grafts was rapid, being largely completed by 14 days. This finding indicates that CD95L expression on endothelium of composite corneal allografts, unlike CD95L expression on epithelium in such grafts, plays a major role in promoting the survival of orthotopic corneal allografts. 

Discussion

Although there is considerable evidence to support the view that CD95L expression on the cornea promotes graft survival, the precise mechanisms by which this protection is achieved are unresolved. First proposed by Stuart et al. ⁶ and supported by the results reported by Yamagami et al., ⁷ the most obvious explanation is that CD95L expressed on the corneal allografts deletes CD95⁺ T cells before these effector cells have a chance to destroy the grafts. Yet others ¹⁰ have reported that tissue and cellular grafts genetically engineered to express high levels of CD95L are more vulnerable to rejection than unengineered counterpart grafts, raising an important note of caution. In the latter examples, adaptive alloimmune T cells did not destroy the grafts. Instead, CD95⁺ PMNs appeared to be the proximate mediators of graft destruction. Because of the uncertainty surrounding the mechanism by which CD95L protects immune-privileged grafts from destruction, we have examined the impact of deficient CD95L expression on the fates of cornea grafts placed orthotopically and heterotopically in the immune-privileged anterior chamber of the eye of BALB/c mice. Our results reveal that CD95L expression on the corneal endothelium protected orthotopic corneal grafts from rejection. In this setting, CD95L expression on corneal epithelium had little effect on graft outcome. Alternatively, CD95L expression on corneal epithelium protected heterotopic corneal fragment grafts from rejection in the anterior chamber, and CD95L expression on corneal endothelium of these fragments was largely irrelevant to graft acceptance. A summary of the results using cornea fragments implanted into the anterior chamber of BALB/c mice is presented in Table 1 . 

We suspect that the reason CD95L on corneal epithelium dictates fragment graft survival in the anterior chamber, whereas CD95L on corneal endothelium dictates orthotopic graft survival, is that the anterior chamber is an immune-privileged site. Our suspicion derives from our previously reported findings concerning the fates of full-thickness and epithelium-deprived B6 and B6.gld corneal grafts placed beneath the kidney capsule (a non–immune-privileged site) of BALB/c mice. ⁹ Full-thickness B6 grafts were accepted in the anterior chamber, but rejected beneath the kidney capsule. This comparison implies that the nature of the graft bed can dictate the graft’s fate; immune-privileged sites promote graft survival. This implication is supported by the findings that epithelium-deprived, CD95L-deficient grafts were rejected beneath the kidney capsule, but accepted in the anterior chamber. Many immunosuppressive and anti-inflammatory factors have been described in the anterior chamber and aqueous humor, ²⁴ and are candidates for protecting corneal fragment allografts placed therein from attack by effector CD4⁺ T cells. Even in the absence of CD95L, full-thickness B6 cornea fragments and epithelium-deprived B6.gld cornea fragments could be protected in the anterior chamber by TGFβ, α-MSH, and/or VIP. This logic assumes that the capacity of the immune privileged site to protect cornea grafts from rejection is directed at the T cells presumed to mediate graft rejection. In addition, our previous studies on cornea grafts placed heterotopically at non–immune-privileged sites suggest that CD95L expression on the graft inhibits alloimmunization of the graft recipient. Thus, epithelium-deprived CD95L-deficient corneal allografts placed beneath the kidney capsule readily sensitize their recipients (they develop donor-specific DH), ⁹ whereas similar grafts placed in the anterior chamber fail to induce graft-specific sensitization. These considerations suggest that the nature and tissue-localization of CD95L expression plays more than one role in determining the type of immunity engendered by allogeneic tissues and cells. 

While CD95L is expressed on antigen-presenting cells (APCs), among other types of normal lymphoreticular cells, the evidence cited above implicates the absence of CD95L on heterotopic grafts as responsible for promoting recipient sensitization. The finding that CD95L-deficient, epithelium-deprived grafts in the anterior chamber were accepted and failed to induce donor-specific DH in BALB/c mice suggests that the immune privilege of this ocular compartment interferes with allosensitization. Perhaps it is relevant that conventional APCs that acquire ACAID-inducing properties after in vitro treatment with TGFβ2 can suppress induction of DH in vivo only if the APCs express CD95L. ²⁵ Antigen-pulsed, TGFβ-treated APCs obtained from B6.gld donors failed to induce ACAID when injected intravenously into naive recipient mice. Moreover, similarly treated APCs from normal mice failed to induce ACAID when administered to CD95-deficient B6.lpr mice. We have reported that CD95L⁺ APC-bearing allogeneic cornea fragments placed within the anterior chamber promote graft acceptance by inducing ACAID. ¹⁵ Perhaps this is why full-thickness B6.gld cornea fragments placed in the anterior chamber of BALB/c mice are rejected and induce donor-specific DH. 

The fate of full-thickness allogeneic cornea fragments in the anterior chamber of C57BL/6 eyes was much poorer than allogeneic fragments in BALB/c eyes. This difference has been noted previously: The incidence of rejection of orthotopic C57BL6 corneal grafts in eyes of BALB/c mice is repeatedly found to be approximately 50%, ¹⁸ whereas the incidence of rejection of orthotopic BALB/c cornea grafts in eyes of C57BL/6 mice is greater than 80%. ²⁶ We have speculated on the possible reasons for this disparity in these two mouse strains, including the evidence that the affinity for CD95 of the CD95L of C57BL/6 mice is significantly less than the affinity for CD95 of the BALB/c CD95L. Whether this accounts for why C57BL/6 mice reject full-thickness BALB/c cornea fragments in their anterior chamber is unclear. The disparity between BALB/c and C57BL/6 mice as cornea graft recipients reemphasizes that both graft and host factors are involved in dictating the fate of the graft. 

Experiments in which fragments of cornea are grafted into the anterior chamber are of interest immunologically, but are only indirectly relevant to clinical corneal transplantation. Our studies on the differential importance of CD95L expression on corneal epithelium and endothelium in composite corneas grafted orthotopically in BALB/c eyes may have more direct clinical meaning. A summary of our experience and that of other laboratories with full-thickness, intact, and composite cornea B6 allografts placed in eyes of BALB/c mice is presented in Table 2 . Approximately 50% of normal B6 corneas placed orthotopically survive indefinitely in BALB/c eyes. By contrast, virtually 100% of orthotopic B6.gld corneas are rejected. Composite grafts in which CD95L expression is deficient only on the epithelium display approximately the same rate of acceptance as full-thickness B6 grafts (50%). This contrasts sharply with the fate of composite grafts on which CD95L expression is deficient only on the endothelium—approximately 90% of these grafts are rejected. Together these findings reveal that CD95L expression on corneal epithelium has little influence on the fate of orthotopic grafts and that CD95L expression on the endothelium is much more important. The high rate of rejection of grafts in which the endothelium was derived from B6.gld donors strongly resembles the rate of rejection of full-thickness B6.gld grafts. Thus, in orthotopic corneal grafts, it is the capacity of the corneal endothelium to express CD95L that confers immune privilege on the graft. 

We have discussed the results reported herein with respect to the roles that CD95/CD95L interactions have been found, or are presumed, to play in ocular immune privilege, with special reference to effects on ocular immune responses. But CD95/CD95L interactions also play important roles in regulating angiogenesis, ^{27 28} independent of immune responses. In that context, we observed that full-thickness B6.gld cornea fragments placed in the anterior chamber were destroyed by a process in which robust neovascularization occurred in the recipient stroma and limbus, focusing on the area surrounding the graft. No such neovascular response was observed with full-thickness B6 cornea fragment grafts. ¹⁵ Thus, the possibility exists that CD95L on corneal epithelium contributes to acceptance of corneal allografts by inhibiting angiogenesis in the graft bed, as well as within the stroma of the graft itself. Caution about full acceptance of this possibility arises, however, from our finding that orthotopic composite grafts comprising B6.gld epithelium and B6 stroma+endothelium experienced the same incidence and rate of rejection in BALB/c eyes as did intact B6 corneal grafts. It remains to be determined whether the salutary effect of CD95L expression on corneal epithelium revealed in the fate of B6.gld fragment grafts bears on the ability of CD95/CD95L interactions to influence immune, inflammatory, or angiogenic responses. 

²  Present affiliation: Fujieda Municipal General Hospital, Shizuoka, Japan.

 Supported by Grant EY10765 from the National Eye Institute.

 Submitted for publication May 22, 2003; revised July 17 and November 9, 2003; accepted November 26, 2003.

 Disclosure: H. Osawa, None; K. Maruyama, 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.

 

Fate of full-thickness B6.gld cornea fragments implanted into anterior chamber of BALB/c eyes. Fragments of full-thickness B6.gld corneas were placed into the anterior chamber of eyes of normal BALB/c mice. (A) The clinical appearance of these grafts was assessed by biomicroscopy at weekly intervals, and graft clarity was scored on a 1+ to 3+ scale. Data represent scores of individual grafts that are (○) clear or (•) considered to be rejected. (B) Photograph of a rejected B6.gld fragment immediately behind the recipient cornea. The pupil margin is not visible through the graft, and the recipient cornea adjacent to the graft is inflamed and neovascularized.

Figure 2.

 

Fate of epithelium-deprived B6.gld cornea fragments implanted into anterior chamber of BALB/c eyes. Fragments of B6.gld corneas from which the epithelium had been removed were placed into the anterior chamber of eyes of normal BALB/c mice. The clinical appearance of these grafts at weekly intervals is presented as described in the legend to Figure 1 .

Figure 3.

 

Histologic images of B6.gld cornea fragments implanted into the anterior chamber of BALB/c eyes. (A) Photomicrograph of a rejected full-thickness B6.gld corneal fragment. The graft is represented by a large cyst, filled with inflammatory cells, other necrotic cells, and debris, and surrounded by an incomplete layer of epithelium. The adjacent recipient stroma contains inflammatory cells and profiles of new vessels. (B) Photomicrograph of an accepted epithelium-deprived B6.gld corneal fragment. The graft displays the regular lamellar arrays of corneal stroma and its posterior surface is covered with an intact layer of endothelium. Both graft and recipient stroma are devoid of inflammatory cells and profiles of new blood vessels.

Figure 4.

 

Donor-specific DH in BALB/c mice bearing intracameral B6 cornea fragments. BALB/c mice with full-thickness (gld) or epithelium-deprived (gld wo E) B6.gld cornea fragments in the anterior chamber for 2 weeks were ear challenged with an intradermal injection of 1 × 10⁶ γ-irradiated B6 spleen cells into the ear pinnae. Positive control BALB/c mice (Imm) were immunized subcutaneously 2 weeks previously with 10 × 10⁶ C57BL/6 spleen cells. Naive BALB/c mice received only ear challenge. Mean ear-swelling responses ± SEM of panels of five mice each at 24 hours are presented. *Mean responses significantly greater than naive control (P < 0.05).

Figure 5.

 

Fate of intact, full-thickness B6 and composite B6.gld/B6 cornea grafts placed orthotopically in eyes of BALB/c mice. Intact or composite grafts were placed orthotopically in normal eyes of BALB/c mice. The grafts were assessed clinically, as described in the legend to Figure 1 . (A) Comparison of graft clinical scores for full-thickness, intact B6 grafts (solid line) and composite grafts containing B6.gld epithelium layered on B6 stroma+endothelium (dashed line). P < 0.05, difference between survival patterns of intact B6 and composite B6.gld epithelium on B6 stroma+endothelium. (B) Comparison of graft clinical scores for full-thickness, intact B6 grafts (solid line) and composite grafts containing B6 epithelium layered on B6.gld stroma+endothelium (dashed line). P < 0.05, difference between patterns of intact B6 and composite B6 epithelium layered on B6.gld stroma+endothelium.

Table 1.

 

Fate and Allosensitizing Capacity of Heterotopic Corneal Allografts in Immune and Non–immune-Privileged Sites

Type of Cornea Graft	Site of Engraftment
		Beneath Kidney Capsule			Anterior Chamber
		Fate	DH	Fate	DH
Full-thickness, B6	Rejected*	++, †	Accepted, ‡	++
Full-thickness, B6.gld	Rejected	++	Rejected	++
Epi-deprived, B6, §	Accepted		Accepted
Epi-deprived, B6.gld , §	Rejected	++	Accepted

 Data summarized in this table are derived from results reported in this study and in Refs. ⁹ , ¹⁶ , and ²⁰ .

*  Cornea fragment grafts beneath kidney capsule were determined by histologic analysis to be rejected.

†  Ear-swelling responses 24 hours after intrapinnae challenge with B6 spleen cells that were significantly greater than in negative control.

‡  Cornea graft fragments in the anterior chamber were determined to be rejected on clinical evidence of opacity as judged by biomicroscopy.

§  Corneas from which the epithelium had been removed after EDTA treatment.

Table 2.

 

Fate of Intact and Composite Orthotopic Corneal Allografts

Type of Graft	Fate in BALB/c Eyes
Intact B6	50% rejected
Intact B6.gld	100% rejected
B6.gld epithelium/B6 stroma+endothelium*	50% rejected
B6 epithelium/B6.gld stroma+endothelium*	90% rejected

 Data summarized in this table are derived from results reported in this study and in Refs. ¹⁹ , ²⁰ , and ²⁸ .

*  Composite grafts containing epithelial layer from one source layered over epithelium-deprived stroma+endothelium from a different source.

The authors thank Jacqueline Doherty for excellent management of the laboratory; Marie Ortega for expert management of the vivarium; and Bruce Ksander, Zdenka Haskova, Kazumi Tanaka, and Carolina Arancibia for helpful discussions.