Dynamics of Donor Cell Persistence and Recipient Cell Replacement in Orthotopic Corneal Allografts in Mice

Abstract

purpose. To determine the extent to which donor cells persist and recipient cells repopulate each of the three cell layers of orthotopic corneal grafts in mice.

methods. BALB/c, C57BL/6, and enhanced green fluorescence protein (EGFP) transgenic mice (B6 background) were used as donors and recipients for orthotopic syngeneic and allogeneic corneal grafts. Graft-bearing eyes were harvested at 5, 10, 15, 28, and 56 days, stained with propidium iodide, and observed (layer by layer) by confocal microscopy. Bone marrow–derived cells in the grafts were assessed immunohistochemically.

results. Donor epithelium was totally replaced by recipient epithelial cells within 15 days in both syngeneic and allogeneic grafts, whereas donor stromal keratocytes and endothelial cells were retained virtually intact in syngeneic grafts and in accepted allografts. In rejected allografts, neither donor-derived keratocytes nor endothelial cells were detected, and, instead, recipient-derived stromal fibroblasts, neovessels, and infiltrating leukocytes were heavily represented. The posterior surface of rejected grafts was devoid of corneal endothelium and was covered incompletely with bone marrow–derived cells of recipient origin.

conclusions. Whereas in mice graft-derived epithelium is largely irrelevant to corneal allograft outcome, persistence of donor-derived endothelium and keratocytes correlates perfectly with graft acceptance. Recipient endothelium is incapable of covering the posterior surface of accepted or rejected corneal grafts, whereas bone marrow–derived cells of recipient origin come to occupy this site in rejected grafts.

Although everyone acknowledges that there is a high rate of acceptance of orthotopic corneal allografts in humans ¹ and in animal models, ^{2 3} a significant number of such grafts are rejected, especially in eyes that are so-called high risk. ⁴ Therefore, it is important to understand the pathogenesis of rejection of orthotopic corneal allografts so that therapeutic strategies to prevent rejection can be developed. 

Ever since the seminal work of Khodadoust and Silverstein, ⁵ experimental attention has been drawn to the relative importance of the role that each cell layer of the cornea plays in inducing sensitization to donor alloantigens and in serving as a target of alloimmune rejection. Much evidence points to a potent role for corneal epithelial cells as a source of immunizing alloantigens in corneal grafts placed orthotopically ^{6 7} and heterotopically (cutaneous surface, ⁸ subcutaneous pouches, ⁹ and beneath the kidney capsule ^{10 11} ). Although the corneal stroma has also been found to be alloimmunogenic, corneal endothelial cells appear to play only a minor role as graft-derived immunogens. On the contrary, corneal endothelium has been found to confer immune privilege on corneal tissues, protecting them from immune rejection by virtue of constitutive expression of CD95 ligand. ^{10 11 12}  

When corneal allografts are accepted for prolonged periods in rodent model systems, the extent to which each cellular layer of the graft persists is an important consideration. For example, the corneal endothelial cells of rodents are much more capable of proliferation than are their human counterparts. ^{13 14} Thus, in long-accepted corneal allografts in rodents, it is unclear whether donor endothelium persists or is replaced by recipient endothelium. Similarly, it is not known to what extent graft-derived epithelium and stromal keratocytes persist in accepted corneal allografts in rodents. Because it is unclear whether cells of each corneal layer can and do persist in accepted grafts, it cannot be decided whether the nonrejection of long-accepted corneal allografts in rodents is due to the absence of an effective alloimmune response or to the absence of suitable alloimmunogenic targets. 

The availability of transgenic mice that express enhanced green fluorescence protein (EGFP) in all their cells ¹⁵ offers an opportunity to resolve these important issues. In the following experiments, GFP⁺ corneas were transplanted into eyes of GFP⁻ recipients, and vice versa. By examining the grafted eyes with confocal and fluorescence microscopy, we determined that donor epithelium was rapidly replaced by recipient epithelium, a process that was completed, even before an immune rejection response had emerged. In addition, we found that donor-derived keratocytes and endothelium persisted to a high level in accepted corneal allografts, whereas these cells were eliminated from grafts that were rejected. In the latter instance, bone marrow–derived cells of recipient origin incompletely covered the posterior graft surface. 

Materials and Methods

Mice and Anesthesia

Male BALB/c (H-2^d) and C57BL/6 (B6, H-2^b) mice were purchased from Taconic Farm (Germantown, NY). EGFP transgenic mice (B6 background) ¹⁵ were kindly provided by Masaru Okabe (Osaka University, Japan) and were bred in our animal colony. All mice were used at 8 to 12 weeks of age and were treated according to the ARVO Statement for 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. 

Tissues of EGFP mice are green under excitation light, with the exception of hair and erythrocytes. In these mice, EGFP is expressed in the cytosol. Because the excitation optimum for EGFP is close to 488 nm, cells from EGFP transgenic mice are suitable for analysis by fluorescence microscopy. ¹⁵  

Orthotopic Corneal Transplantation and Graft Evaluation

Penetrating keratoplasty was performed as described previously. ^{2 7} Briefly, donor corneas 2 mm in diameter were placed in the same sized recipient bed with eight interrupted sutures (11-0 nylon). Sutures were removed at 8 days after grafting. Orthotopic grafts were observed by slit lamp microscopy at weekly intervals, and assessment of orthotopic corneal graft survival was performed according to a previously described scoring system ^{2 7} : 0, clear graft; 1+, minimal superficial nonstromal opacity; 2+, minimal deep stromal opacity with pupil margin and iris vessels visible; 3+, moderate deep stromal opacity with only pupil margin visible; 4+, intense deep stromal opacity with the anterior chamber visible; and 5+, maximum stromal opacity with total obscuration of the anterior chamber. Grafts with opacity scores of 2+ or greater after 3 weeks were considered to have been rejected. 

Assessment of Fate of Corneal Cells by Confocal Microscopy

Survival of donor cells was evaluated in corneal grafts from normal eyes of EGFP mice to the eyes of syngeneic wild-type C57BL/6 and allogeneic BALB/c mice. Replacement of donor cells by recipient cells was assessed in grafts derived from normal eyes of syngeneic wild-type C57BL/6 and allogeneic BALB/c mice and placed in eyes of EGFP mice. Graft-bearing whole corneas excluding limbus were harvested from eyes of recipients at 5, 10, 15, 28, and 56 days. At each observation period, groups of at least 5 grafts were evaluated. The corneal tissue was fixed with 4% paraformaldehyde for at least 1 hour at room temperature. After a wash with PBS, the sample was mounted on a slide with mounting medium containing propidium iodide (PI), according to the manufacturer’s instruction (Vectastain; Vector Laboratories, Inc., Burlingame, CA), to indicate cell nuclei. Each layer of the cornea samples was observed by confocal microscopy. Endothelium and stroma were examined in the grafts at both center and periphery (100–300μ m from graft margin) and in the recipient bed of each sample. Cell density of corneal endothelial cells was evaluated by counting cells in ×40 magnification confocal microscopic views of three individual areas at both center and periphery of the grafts and in the recipient bed of each sample. 

Evaluation of Epithelium Replacement

Corneas from normal BALB/c and C57BL/6 mice were orthotopically grafted into the eyes of normal EGFP mice. Epithelial replacement by EGFP-recipient cells was observed by confocal microscopy, and the extent of epithelial replacement was assessed semiquantitatively, according to the following scoring system: 0, no green cells in graft surface; 1, epithelial sheet of green cells in one or two quadrants of graft surface, not exceeding 1 mm from the graft margin; 2, epithelial sheet of green cells in three or four quadrants of graft surface, not exceeding 1 mm from the graft margin; 3, epithelial sheet of green cells in one or two quadrants of graft surface, exceeding 1 mm from the graft margin; 4, epithelial sheet of green cells in three or four quadrants of graft surface, exceeding 1 mm from the graft margin; 5, epithelial sheet of green cells covering the entire surface of the graft. 

Assessment of Bone Marrow–Derived Cells in Corneal Grafts

To study the presence of recipient bone marrow–derived cells on the posterior surface of corneal grafts, C57BL/6 donor corneas were placed in BALB/c recipient beds. Immunohistochemical studies for I-A^d (BALB/c recipient-derived major histocompatibility complex [MHC] class II antigens) and F4/80 (marker on macrophages and dendritic cells) were performed on both accepted and rejected corneal allografts, using FITC-labeled rat anti-mouse I-A^d (PharMingen, San Diego, CA), and FITC-labeled rat anti-mouse F4/80 (Caltag, Burlingame, CA). Graft-bearing whole corneas were removed at 8 weeks after orthotopic corneal grafting from C57BL/6 donors to BALB/c recipients, fixed in acetone for 10 minutes, and incubated in the monoclonal antibody, diluted to 4 mg/ml, for 2 hours at room temperature. After a wash with PBS, the sample was mounted on a slide with mounting medium according to the manufacturer’s instructions (Vectastain; Vector Laboratories), and the posterior surface of the corneal graft was observed by confocal microscopy. 

Statistical Analyses

Cell density of corneal endothelial cells and scores of epithelium replacement were evaluated statistically by using a two-tailed, unpaired Student’s t-test. P < 0.05 was deemed significant. 

Results

In the following experiments, we prepared full-thickness corneas from EGFP donor mice for use as orthotopic grafts in eyes of wild-type mice. We also placed corneas from wild-type mice in eyes of EGFP recipients. By harvesting grafts and graft beds at periodic intervals, we assessed the extent to which donor epithelial, stromal, and endothelial cells persisted within grafted corneas and the extent to which recipient cells (cornea parenchymal as well as bone marrow-derived) infiltrated and/or replaced donor cells within the grafts. 

Donor and Recipient Cell Contributions to the Epithelial Surface of Orthotopic Corneal Grafts

Normal corneas from BALB/c (allogeneic) and C57BL/6 (syngeneic) donors were grafted orthotopically into normal eyes of EGFP mice (B6 background). Graft-bearing eyes were removed at 5, 10, 15, 28, and 56 days after grafting. The grafts, as well as the recipient corneal rims, were subjected to observation by confocal microscopy. When examined within the first 15 days after grafting, none of the grafts (syngeneic or allogeneic) displayed signs of rejection or inflammation. Beyond this time point, all syngeneic grafts remained perfectly clear, but a proportion (approximately 50%) of the allografts displayed evidence of rejection at 28 and 56 days. 

As determined microscopically, recipient-derived epithelial cells (GFP⁺) were observed to migrate as individual cells into the graft epithelium (GFP⁻) as early as day 5 (Figs. 1A 1B) . These GFP⁺ cells were distributed individually in a disperse fashion across the entire graft surface. At 10 days after grafting, recipient-derived GFP⁺ cells formed an intact sheet that extended from the graft periphery toward the center, although this sheet did not yet reach the graft center (data not shown). By 15 days after grafting, all GFP⁻ grafts (both syngeneic and allogeneic) were totally resurfaced by recipient GFP⁺ epithelium (Figs. 1C 1D) . Semiquantitative estimates of the extent of graft surface that was replaced over time by recipient GFP⁺ cells are displayed in Figure 2 . In companion experiments, corneal grafts from EGFP mice were placed orthotopically in eyes of syngeneic GFP⁻ mice. In grafts harvested from eyes at 15 days after grafting, no GFP⁺ cells were detected within the epithelium (data not shown). Together, these findings indicate that recipient epithelium rapidly replaced donor epithelium after orthotopic corneal grafting, a process that was essentially complete within 15 days for both syngeneic and allogeneic corneal grafts. 

Persistence of Donor Cells among Stroma and Endothelium of Orthotopic Corneal Grafts

The stroma and endothelium of syngeneic and allogeneic grafts were also examined for retention of graft-derived cells and for penetration of recipient-derived cells. For these studies, the excised tissues were fixed immediately, and a mounting medium containing PI was used as a means of identifying the distinctive patterns of nuclei in cells of stroma and endothelium. For purposes of comparison, confocal images of stroma and endothelium of the normal EGFP cornea are presented in Figures 3A and 3B . A single layer of densely packed, polygonal GFP⁺ cells (Fig. 3A) represents the normal corneal endothelium. Small regions of apparent cell drop-out represent areas damaged by forceps in preparing the tissue for examination. Loosely distributed GFP⁺ cells with slender cell bodies and process (Fig. 3B) in the corneal stroma were keratocytes. GFP⁺ images of neuronal axons were also evident in the stroma. The density of GFP⁺ endothelial cells in these corneas (approximately 1200 cells/mm²) is displayed in Figure 4 . Similar images were found when syngeneic and allogeneic grafts were examined microscopically at 5, 10, and 15 days after grafting. These findings indicate that little if any exchange between donor and recipient cells in the graft stroma and endothelium was taking place during this interval, even though the entire graft epithelium was replaced during the same interval. 

By 28 and 56 days after grafting, a subset of EGFP corneal allografts gave clinical evidence of rejection, whereas the other EGFP allografts remained perfectly clear (i.e., they were accepted). When the latter were examined by confocal microscopy (Figs. 3C 3D 3E 3F) , GFP⁺ endothelial cells were observed to pave the entire posterior surface of the graft, and GFP⁺ keratocytes at normal density persisted within the stroma. By counting and calculating the average cell densities, we found that donor-derived endothelial cells persisted at virtually unchanged density in the center of accepted allografts at 28 days, whereas donor-derived endothelial cell density was reduced in the periphery of these grafts (Fig. 4A) . By 56 days, the density of endothelial cells was reduced approximately 50% in both the center and the periphery of accepted allografts (Fig. 4B) . As the density of endothelial cells decreased, there was a compensatory increase in volume of residual cells, enabling the posterior corneal surface to retain its complete lining with functioning endothelium. Similar patterns of endothelial cell persistence and density reduction were observed for syngeneic grafts examined at 28 and 56 days (data not shown). This outcome implies that the changes in donor endothelium observed in accepted corneal allografts may have been unrelated to a deleterious alloimmune response. 

When the posterior surface of rejected EGFP corneal allografts was examined at 28 days (Figs. 3G 3H 3I 3J) , the density of GFP⁺ endothelial cells was drastically reduced (Fig. 4A) . In fact, virtually no GFP⁺ endothelial cells were detectable when rejected grafts were examined at 56 days (Fig. 4B) . Moreover, PI⁺ nuclei (representing GFP⁻ cells of presumed recipient origin) were readily observed on the posterior surface of 56-day rejected grafts, but the pattern of these nuclei was inconsistent with the presence of an intact, normal endothelial cell layer. When the stromas of rejected EGFP corneal allografts were examined at 28 days, GFP⁺ keratocytes existed only in the central cornea. These cells were not detected at 56 days. Nonetheless, the stroma of these rejected grafts contained numerous PI⁺ nuclear images of heterogeneous size and shape, implying that cells of recipient origin (GFP⁻) had penetrated into this layer of the graft. These findings indicate that immune rejection of orthotopic corneal allografts achieved the complete elimination of donor-derived corneal endothelium and stromal keratocytes. By contrast, in the absence of immune rejection, accepted orthotopic corneal allografts retained a large proportion of their original content of donor keratocytes and endothelium, an observation that also applied to syngeneic grafts. 

Presence of Recipient-Derived Cells in Stroma and Endothelium of Corneal Grafts

To determine the extent to which recipient-derived cells are able to replace donor cells and/or to infiltrate orthotopic corneal grafts, corneas from BALB/c and C57BL/6 donors were grafted into eyes of EGFP mice. Graft-bearing corneas (excluding limbus) were removed, fixed, and mounted in PI-containing mounting medium and then observed by confocal microscopy. Among accepted allografts at 56 days, cells suggestive of recipient-derived endothelium were never found, although a few GFP⁺ recipient cells of uncertain lineage were present on the posterior surface of these grafts (Fig. 3K) . The density of these recipient-derived cells was extremely low (Fig. 5) . Microscopic examination of the recipient bed surrounding accepted grafts revealed that the density of recipient endothelial cells was reduced to less than 50% of the density found in normal cornea (Fig. 6) , and that the endothelial cell nuclei were spaced widely (Fig. 3P) . 

With respect to the stroma of accepted allografts, only rare GFP⁺ cells were found (Fig. 3L) . The stroma was replete with PI⁺ nuclei associated with GFP⁻ cells, indicating a typical array of donor-derived keratocytes. Moreover, at the graft bed margin, recipient stromal elements were readily visible in the recipient bed, but virtually none was seen to have migrated into the stroma of the graft (Fig. 3O) . 

The posterior surface of rejected allografts contained no GFP⁺ cells (recipient in origin) with a morphology suggestive of corneal endothelium (Fig. 3M) . Instead, the posterior layer of these grafts was partly resurfaced by numerous recipient cells of uncertain lineage (Fig. 5) . The stroma of rejected allografts contained a high density of recipient-derived activated fibroblasts, infiltrating leukocytes, and profiles of GFP⁺ vessels (Fig. 3N) . In rejected grafts, recipient cells appeared to have migrated across the graft bed margin and infiltrated the graft (Fig. 3Q) . The density of recipient endothelial cells on the recipient side of the graft bed margin was reduced to a degree similar to that found in the recipient bed of accepted allografts (Fig. 6) . Recipient keratocytes at normal density persisted in the recipient side of the graft bed margin (Fig. 3R) . 

Together, these results indicate that recipient-derived inflammatory, fibroblast, and vascular endothelial cells infiltrated the stroma and endothelium of rejected corneal allografts. By contrast, cells of recipient origin rarely penetrated the stroma and endothelium of accepted corneal allografts. These findings also reveal that the integrity of recipient endothelium was compromised at the margin where the graft bed meets the graft and that this occurred even in the absence of alloimmune rejection of the graft (as in syngeneic grafts). 

Characterization of Bone Marrow–Derived Cells of Recipient Origin on the Posterior Surface of Grafted Corneas

Our next experiments were designed to characterize the nature and, if possible, the histogenetic origin of the recipient cells that were found on the posterior surface of corneal allografts. Normal corneas of C57BL/6 donors were grafted orthotopically into the eyes of normal BALB/c mice. Once the fate of these grafts was established as accepted or rejected at 8 weeks, the grafts were removed, fixed, stained with monoclonal antibodies directed at I-A^d (MHC class II of BALB/c recipient) or F4/80 (a marker of macrophages and dendritic cells), and then mounted in PI-containing mounting medium. The posterior surface of accepted and rejected grafts was then observed by confocal microscopy. As displayed in Figure 7 , the posterior surface of accepted allografts contained a few (<50/mm²) scattered I-A^d-positive cells (recipient in origin) and a similar distribution of F4/80⁺ cells (Figs. 7A 7B) . In both instances, the labeled cells displayed small, rounded cell bodies from which several rather short dendritic processes projected into the spaces between adjacent endothelium. 

When rejected allografts were examined microscopically, the posterior surface contained large numbers of I-A^d- positive cells (recipient in origin) characterized by slender cell bodies with dendritic processes that were much longer and more elaborate than those associated with cells found on the posterior surface of accepted allografts. An even higher density of F4/80⁺ cells was observed on the posterior surface of rejected allografts. Although some of these cells were dendritic in form (resembling the cells expressing I-A^d), other F4/80⁺ cells had larger cell bodies, were round or oval, and often displayed short, stubby processes—an image typical of activated macrophages. Taken together, these results indicate that bone marrow–derived cells of recipient origin were capable of taking up residence on the posterior surface of allografts. In accepted grafts, these cells appeared to be dendritic cells that express class II MHC molecules and were sparsely intercalated among putatively healthy, donor-derived endothelium. In rejected grafts, the posterior surface, denuded of typical endothelium, was incompletely covered by large numbers of both class II MHC⁺ dendritic cells and activated macrophages. 

Discussion

The results of the studies reported here provide several new pieces of information that are relevant to the fate of orthotopic corneal allografts in mice. First, graft epithelium was rapidly replaced (within 15 days) by recipient epithelium in all grafts: syngeneic, accepted allogeneic, and rejected allogeneic. Second, the stroma and endothelium of accepted allografts contained significant numbers of donor keratocytes and endothelial cells, with little evidence of replacement of these cell types with cells of recipient origin. Third, the stroma and endothelium of allografts that were rejected were completely depleted of their complement of donor-derived keratocytes and endothelium. Moreover, the posterior surface of rejected allografts was not only devoid of recognizable endothelial cells, but was covered by recipient leukocytes (predominately dendritic cells and macrophages). The meaning of each of these observations for our understanding of mechanisms of acceptance and rejection of orthotopic corneal allografts will be considered separately. 

Although we have never observed evidence of epithelial rejection in orthotopic mouse cornea grafts (J. Hori, unpublished observations, 1999), we fully expected that our results would confirm the view that the epithelium of orthotopic corneal allografts offers a significant barrier to graft acceptance. These expectations derived from several previous observations: (1) Epithelial cells of the cornea strongly express MHC-encoded as well as minor histocompatibility antigens. ^{16 17 18} (2) Full-thickness allografts of corneal tissue placed heterotopically (beneath the kidney capsule, even within the anterior chamber of the eye) induce vigorous donor-specific delayed hypersensitivity in their recipients, and this immunogenicity has been traced largely to the epithelial layer of these grafts. ^{11 19} (3) Composite corneal grafts composed of recipient epithelium and donor stroma plus endothelium fail to sensitize their recipients and are highly resistant to rejection. ⁷ (4) Although uncommon, epithelial rejection is an important cause of failed keratoplasty in humans ²⁰ and has been described in rabbits. ^{6 21}  

Yet, the current observations ran counter to these expectations. Donor epithelium was completely replaced on our orthotopic cornea allografts within 15 days. Because these grafts were placed in normal (low-risk) eyes and because the earliest that donor-specific delayed hypersensitivity can be detected after grafting is at 3 weeks, we infer that the epithelium of allogeneic corneas in mice makes little or no antigenic contribution to the ability of these grafts to sensitize their hosts. Moreover, because the earliest rejection reactions detected in corneal allografts in these mice occurred after 3 weeks, and because donor epithelium was absent at that time, we further infer that, at least in mice, corneal epithelium does not serve as the target of alloimmune rejection of orthotopic corneal allografts. Because graft-derived epithelium contributes neither immunogenicity nor allogeneic targets to corneal allografts placed in normal eyes of mice, the mouse model is not a good one for the study of epithelial rejection in humans. 

It is worth commenting on the finding that the rate of replacement of donor epithelium by recipient epithelium was virtually identical for syngeneic and allogeneic grafts. This result rules out a role for alloimmunity in encouraging rapid replacement of donor epithelium. We presume that the process of resurfacing the epithelium of orthotopic corneal grafts is an exaggeration of the normal process by which the normal corneal epithelial surface is constantly renewed. ^{22 23 24 25 26 27}  

Donor endothelial cells in significant numbers were found to persist on the posterior surfaces of syngeneic corneal grafts and of accepted orthotopic corneal allografts. Moreover, recipient endothelial cells rarely migrated onto the posterior surface of these grafts. Thus, the clarity of long-accepted orthotopic corneal allografts in mice correlates perfectly with the persistence of donor endothelium. However, we observed that the density of the endothelial layer of accepted grafts gradually declined during the 8-week observation period. Because this process was first evident at the graft periphery and within the recipient bed, we assume that the trauma of surgery contributes to reduced density of endothelial cells. However, the slow reduction in the density of endothelial cells in the center of accepted grafts cannot be so easily explained. Because it was observed in both syngeneic and allogeneic grafts, we doubt that immunity contributes to this decline. 

Donor keratocytes also persisted in accepted corneal grafts. Moreover, very few recipient-derived cells were detected in the stroma of accepted grafts. The relative absence of recipient cells and the strong persistence of donor cells in accepted orthotopic corneal allografts argue that the clarity of these healthy grafts depends on the persistence and functional vitality of donor keratocytes and endothelium. 

This concept is fortified by the results of microscopic analysis of rejected corneal allografts. Neither donor-derived keratocytes nor endothelial cells were detected in opaque allografts observed at 8 weeks after grafting. The stroma of rejected grafts was heavily invaded with recipient-derived cells, ranging from fibroblasts, to vascular endothelial cells, to bone marrow–derived cells, but never were recipient endothelial cells observed to migrate onto the posterior surface of rejected grafts. This is a surprising failure and warrants discussion. One possibility is that migration of corneal endothelium requires a suitable substrate. As immune rejection progressively effaces donor endothelium from the posterior graft surface, the denuded Descemet’s membrane may be molecularly altered so that recipient endothelial cells are unable to secure a migratory foothold. A second possibility is that the local microenvironment may contain inhibitory factors that suppress endothelial cell migration. The presence of an incomplete layer of recipient bone marrow–derived dendritic cells and macrophages on the posterior surface of rejected grafts may be relevant. 

The presence of recipient leukocytes on the posterior surface of rejected grafts implies a key role for these cells in the rejection process. On the one hand, activated macrophages of the type observed in our samples have been implicated as the proximate mediators of the graft rejection reaction that is triggered by donor-specific effector T cells. ²⁸ On the other hand, recipient class II MHC⁺ dendritic cells have been suspected of capturing, processing, and presenting donor-derived alloantigens. ^{29 30} We speculate that the cells observed on the posterior surface of rejected allografts may serve as the local targets of donor-specific effector CD4⁺ T cells. We wonder whether factors released by activated macrophages and T cells act to inhibit endothelial cell migration. ^{31 32 33 34}  

Finally, it is worth commenting on the observation that small numbers of scattered class II⁺ recipient dendritic cells were present in the donor endothelium on the posterior surface of accepted allografts. Several explanations can be considered for this unexpected finding. First, recipient dendritic cells that were recruited to the site during a transient (subclinical) rejection episode may have persisted, even though the episode itself resolved, and the graft survived. Second, dendritic cells presenting graft-derived antigens may present tolerogenic signals that promote the T-cell tolerance that is characteristic of mice bearing accepted orthotopic corneal allografts. ^{35 36 37} Experiments to examine these possibilities are currently under way. 

 Supported by National Institutes of Health Grant EY10765. JWS is a recipient of a Research to Prevent Blindness Senior Scientific Investigator Award.

 Submitted for publication September 22, 2000; revised January 31, 2001; accepted February 23, 2001.

 Commercial relationships policy: N.

 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.

 

Epithelium replacement by GFP⁺ recipient cells in orthotopic corneal allografts. BALB/c corneal allograft–bearing corneal buttons were harvested from eyes of recipient EGFP mice and observed by confocal microscopy on days 5 (A, B) and 15 (C, D). (A, D; arrowheads) Recipient–graft junction. R, recipient bed; D, donor cornea.

Figure 2.

 

Extent of epithelium replacement after orthotopic corneal transplantation. Syngeneic wild-type C57BL/6 (top) and allogeneic wild-type BALB/c (bottom) corneal grafts were grafted orthotopically in EGFP recipients. Graft-bearing corneas were harvested at 5, 10, 15, 28, and 56 days and were observed by confocal microscopy and scored for content of GFP⁺ cells in the epithelium by a semiquantitative method. The y-axis indicates the extent to which recipient epithelium replaced donor epithelium. Each data point represents the score of an individual corneal graft.

Figure 3.

 

Confocal images of endothelium and stroma of orthotopic corneal grafts. Normal corneal endothelium (A) and stroma (B) of a normal EGFP mouse. (C-J) Green cells indicate donor GFP⁺ corneal cells that were grafted orthotopically in non-EGFP mice. (K-R) Green cells indicate recipient GFP⁺ cells that infiltrated non-EGFP donor corneas. Red staining with PI indicate cell nuclei in all images except (H), which displays an image of the FITC-channel alone. (C-F) Accepted allografts from EGFP mice to BALB/c mice. Endothelium (C) and stroma (D) of the center of allografts at 4 weeks. Endothelium (E) and stroma (F) of the center of allografts at 8 weeks. (G-J) Rejected allografts from EGFP mice to BALB/c mice. Endothelium (G) and stroma (H) of the center of allografts at 4 weeks. Endothelium (I) and stroma (J) of the center of allografts at 8 weeks. Endothelium (K) and stroma (L) of the center of accepted grafts from wild-type C57BL/6 mice to EGFP mice at 8 weeks. Endothelium (M) and stroma (N) of the center of rejected allografts from BALB/c mice to EGFP mice at 4 weeks (M) or 8 weeks (N). Endothelium and stromal layers of graft (O) and endothelium at recipient bed (P) of an accepted graft from wild-type C57BL/6 in EGFP recipient at 8 weeks. Endothelium and stromal layers (Q) and stroma at recipient bed (R) of a rejected allograft from a BALB/c donor in an EGFP recipient at 8 weeks. (H, O, and Q; Arrowheads) Recipient–graft junction. R, recipient bed; D, donor cornea.

Figure 4.

 

Cell density of GFP⁺ positive endothelial cells in orthotopic corneal grafts from EGFP mice to non-EGFP recipients. Accepted EGFP corneal grafts in syngeneic wild-type C57BL/6 eyes, accepted EGFP corneal allografts in wild-type BALB/c eyes, and rejected EGFP corneal allografts in wild-type BALB/c eyes were harvested at 4 (A) or 8 weeks (B), fixed and mounted in PI-containing medium, and observed by confocal microscopy. EGFP⁺ donor endothelial cells were counted at three individual areas in both the center and the periphery of each of five test grafts per experimental group in ×40 confocal views, and average density of GFP⁺ endothelial cells was calculated. P < *0.05, **0.01, ***0.001: significantly less than normal corneas of EGFP mice.

Figure 5.

 

Resurfaced endothelial layer of grafts by GFP⁺ recipient cells of uncertain lineage. Accepted syngeneic C57BL/6 corneal grafts in EGFP eyes and rejected BALB/c corneal allografts in EGFP eyes were harvested at 4 or 8 weeks, fixed and mounted in PI-containing medium, and observed by confocal microscopy. GFP⁺ recipient-derived cells that were present in the posterior surface of the grafts were counted at three individual areas of each of five test grafts per experimental group in ×40 confocal views, and the average density of GFP⁺ cells was calculated.

Figure 6.

 

Cell density of GFP⁺ recipient endothelial cells in the recipient bed. EGFP-recipient beds of accepted syngeneic C57BL/6 corneal grafts and rejected BALB/c corneal allografts were observed by confocal microscopy. GFP⁺ recipient-derived endothelial cells in the recipient beds were counted in three individual areas of each of five recipient beds per experimental group in ×40 confocal views, and average density of GFP⁺ cells was calculated.*** P < 0.001: significantly less than normal EGFP corneas.

Figure 7.

 

Recipient bone marrow–derived cells on the posterior surface of accepted and rejected allografts. Full-thickness C57BL/6 corneas that were accepted or rejected during 8 weeks as orthotopic transplants in eyes of BALB/c recipients were harvested and stained with FITC anti-mouse I-A^d antibody or FITC anti-mouse F4/80 antibody and PI, and the endothelial layer was observed by confocal microscopy. Presence of I-A^d-positive dendritic cells in the posterior surface of accepted (A) and rejected (C) allografts. F4/80⁺ cells were also present in accepted (B) and rejected (D) allografts.

The authors thank Jacqueline M. Doherty and Jian Gu for support in these experiments and Michael Young and Marie A. Ortega for providing EGFP transgenic mice.