We examined the role of perforin and FasL in corneal allograft rejection mediated by CD8+ and CD8− T cells. BALB/c corneas were transplanted orthotopically into vascularized, ‘high-risk’ graft beds in C57BL/6 mice, perforin knockout mice and FasL-defective gld/gld mice. CD8+ and CD8− T cells were collected following graft rejection and adoptively transferred to SCID mice, which were then challenged with BALB/c corneal allografts. In every case, CD8− T cells could mediate graft rejection when adoptively transferred to SCID mice that received BALB/c corneal allografts. Although CD8+ T cells also mediated graft rejection, the tempo was slower. Moreover, CD8+ T cells collected FasL-defective donors that had rejected corneal allografts, mediated corneal allograft rejection in only 50% of the SCID mice that received the adoptively transferred cells. In some cases, CD8+ T-cell-mediated rejection occurred in the absence of delayed-type hypersensitivity and cytotoxic T-lymphocyte activity, but was associated with CD8+ T-cell-mediated apoptosis of BALB/c corneal cells in vitro. The results demonstrate the redundancy in immune mechanisms of corneal allograft rejection. Either CD8+ or CD8− T cells can produce corneal allograft rejection, however functional FasL is necessary for optimal rejection, even in a high-risk setting.
Keratoplasty is the oldest, most common, and arguably, the most successful form of solid tissue transplantation (1). In uncomplicated cases, up to 90% of the corneal transplants will succeed, even in the absence of HLA matching and systemic immunosuppressive drugs (2). However, rejection rises steeply when corneas are transplanted into vascularized or inflamed corneal graft beds or into hosts who have previously rejected a corneal allograft (3,4). The presence of blood vessels in the corneal graft bed is perhaps the most widely recognized risk factor for corneal graft rejection. Stimuli that induce the formation of new blood vessels in the cornea also stimulate lymphangiogenesis (5). Vascularization of the corneal graft bed is believed to increase the risk for rejection by promoting the generation of alloimmune responses by providing lymphatic and vascular conduits for the migration of corneal alloantigens and antigen presenting cells (APC) from the graft site to regional lymphoid tissues. The presence of blood vessels in the corneal graft is also thought to facilitate the delivery of circulating immune effector elements to the graft. Animal studies have shown that long-term survival of corneal allografts transplanted to avascular graft beds is associated with the absence of cytotoxic T lymphocyte (CTL) and delayed-type hypersensitivity (DTH) responses to corneal alloantigens (6–10). In fact, there is compelling evidence that the success of corneal allografts transplanted into avascular graft beds is directly correlated with the graft's ability to induce antigen-specific down-regulation of DTH responses, a phenomenon termed anterior chamber-associated immune deviation (1,11)
Although more than 50 years have passed since the elegant animal studies of Maumenee established the immunologic basis for corneal allograft rejection (12), we still do not have a clear understanding of the T-cell-dependent mechanisms that are involved in this process. T cells are absolutely necessary for corneal graft rejection, but the roles of CD8+ and CD4 + T-cell subsets are unclear (10,13). Corneal allograft rejection occurs unabatedly in CD8 knockout (KO) mice, perforin knockout (PKO) mice and wild-type mice depleted of CD8+ T cells via systemic treatment with monoclonal antibody (7,9,14). By contrast, corneal allograft rejection is significantly reduced in CD4 KO mice and in both rats and mice treated with anti-CD4 depleting monoclonal antibodies (10,13–15). However, even though corneal allograft rejection is dramatically reduced in the absence of CD4+ T cells, corneal allografts fail in approximately 50% of CD4 KO mice and 33–65% of the wild-type mice and rats depleted of CD4 cells with antibody (10,14,15). The notion that CD4+ Th1 cell-dependent immunity is an absolute requirement for corneal allograft rejection is further questioned by recent findings indicating that interferon-γ (IFN-γ) KO mice, which cannot mount classical Th1 immune responses, reject 100% of their corneal allografts (16).
In the present study, we examined the independent roles of CD8+ and CD8− T lymphocytes in corneal allograft rejection in high-risk hosts. By using an adoptive transfer model, we were able to dissect the role of the high-risk graft bed in the induction of immune effector elements and its role in promoting the expression of allodestructive immunity at the host/graft interface.
Materials and Methods
C57BL/6 (H-2b), FasL deficient (B6Smn.C3H-Fasl<gld>; gld/gld H-2b) and PKO (C57BL/6-Pfp<tm1Sdz>, H-2b) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). BALB/c (H-2d) mice were purchased from Taconic Farms (Germantown, NY). C.B.-17 C57BL/6 SCID mice were purchased from Charles River (Willmington, MA) and Taconic Farms. Animals used in grafting experiments were female, 8–12 weeks in age. All animals used in these experiments were housed and cared for in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research.
Full-thickness penetrating orthotopic corneal grafts were performed as previously described (17). Donor grafts and recipient graft beds were scored with 2.5 and 2.0 mm trephines, respectively, and the corneas were excised with Vannas scissors. Donor grafts were sewn into place using interrupted 11-0 nylon sutures (Ethicon, Sommerville, NJ), which were not buried. All sutures were removed on day 7 post-transplantation. No immunosuppressive drugs were used, either topically or systemically.
High-risk, vascularized corneal allografts
A vascularized, high-risk graft bed was produced as previously described (3). Briefly, three interrupted sutures (11-0 nylon, 50-μm diameter needle; Sharpoint, Vanguard, Houston, TX) were placed through the central cornea of the right eye of recipient mice 2 weeks prior to orthotopic corneal transplantation.
Clinical evaluation of grafted corneas
Corneal grafts were examined 2–3 times a week with a slit-lamp biomicroscope (Carl Zeiss, Oberkochen, Germany). Graft opacity was scored using a scale of 1–3 as previously described (18). Corneal grafts were considered rejected upon two successive scores of 3.
BALB/c corneal epithelial and endothelial cell cultures were established and propagated in MEM (Bio Whittaker, Walkersville, MD) as described previously (19). Transformed corneal cells proliferate indefinitely, maintaining their original morphologic characteristics and expressing the same histocompatibility antigens as their nontransformed counterparts (20). Tissue-cultured corneal cells were used for in vitro studies rather than the usual lymphoid cells, as corneal cells are the relevant target cells in vivo during corneal allograft rejection.
CD8+T-cell enrichment and depletion
CD8+ splenic T cells were harvested by positive selection using rat anti-mouse CD8-conjugated magnetic microbeads (MACS system, Miltenyi Biotec Inc., Auburn, CA) as described elsewhere (21). To eliminate any CD4+ cells that may have been retained in the column, cells were treated with anti-mouse CD4 (BD Pharmingen) and baby rabbit complement (Cedarlane, Hornby, ON) (21). The enriched CD8− T-cell suspension that was not retained by the magnetic bead treatment was used as a CD4+ T-cell-enriched spleen cell suspension and was treated with anti-CD8 antibody plus complement to deplete any contaminating CD8+ T cells. Cells were washed three times in HBSS and assessed for CD8+ and CD4+ cells by flow cytometry. The CD8-enriched cell suspensions contained 88–92% CD8+, 93%αβ TCR+ and 2.4%γδ TCR+ and 0.42–0.43% CD4+ cells. The CD4-enriched cell suspensions >50% CD4+, 63%αβ TCR+, 5%γδ TCR+ and <2% CD8+ cells.
Adoptive transfer of CD8+and CD8− cells
Spleen cell suspensions were enriched or depleted of CD8+ T cells as described above and were transferred intravenously to each C57BL/6 SCID mouse. In order to make physiologically relevant comparisons between CD8+ and CD8− T-cell populations, one-donor equivalent of either the CD8+ or CD8− T-spleen-cell population from a single mouse was transferred to each C57BL/6 SCID mouse, which then received a BALB/c corneal allograft. We elected not to transfer the same absolute number of CD8+ and CD8− spleen cells, because we consistently found that there were 4–5 times as many CD8− cells as CD8+ cells in the spleen of the average C57BL/6 mouse that had rejected a BALB/c corneal allograft. For CD8− cells, one donor-equivalent was 20 ± 5 × 106 cells, which was adoptively transferred to each SCID mouse. For CD8+ cells, one donor-equivalent was 4.4 ± 1.5 × 106 cells, which was transferred to each C57BL/6 SCID mouse. All C57BL/6 SCID mice were challenged with BALB/c corneal allografts 1 day after the adoptive cell transfers.
In vitro boost of effector splenocytes with alloantigen
Boosting cultures are frequently used to maximize the in vitro detection of cell-mediated immune responses that are initially induced in vivo (22). Accordingly, spleens were collected from C57BL/6 or PKO mice that had rejected BALB/c corneal allografts 7–14 days earlier and were boosted in vitro with γ-irradiated (3000 rad) BALB/c spleen cells as previously described (13). Positive controls consisted of C57BL/6 mice that were subcutaneously immunized with 1 × 107 BALB/c spleen cells 14 days earlier. Negative controls consisted of naïve C57BL/6 mice that had been exposed to BALB/c alloantigens in vivo. The in vitro boosted effector cells were washed once with HBSS and then resuspended in CRPMI and used as effector cells in the apoptosis assays (below).
Three-color Annexin V staining to evaluate T-cell-induced apoptosis in vitro
CD8+ and CD8− T cells were collected from C57BL/6 or PKO mice 1 week after the rejection of BALB/c corneal allografts. Following in vitro boosting (described above), the CD8− and CD8+ T cells were incubated with 5 μM Syto-59 red fluorescent nucleic acid stain (Molecular Probes, Eugene, OR) in CRPMI for 45 min at 37°C. Syto-59-stained C57BL/6 CD8− and CD8+ effector cells were cultured with BALB/c corneal endothelial cells overnight. Apoptosis of the corneal cells was then measured at various time points with the TACS Annexin V-FITC apoptosis detection system (R&D Systems, Minneapolis, MN). Briefly, the target cells were harvested with 0.05% trypsin/EDTA and washed once with ice-cold PBS. The cells were then resuspended in 100 μL Annexin V Incubation Reagent (10 μL 10X Binding Buffer, 10 μL propidium iodide (PI), 1 μL Annexin V-FITC and 79 μL of de-ionized water) and incubated in the dark at room temperature for 15 min. A total of 300 μL of ice-cold 1X binding buffer were then added to each sample and the samples were analyzed by fluorescence-activated cell sorting (FACS) within 1 h for maximal signal. During FACS analysis, Syto-59-stained CD8+ and CD8− cells were gated out to allow only BALB/c corneal cell apoptosis to be detected. Annexin V-FITC staining was used as a marker for apoptotic cells, while PI staining was used as an indicator of necrotic cells that had lost membrane integrity. Cells that stained positively with both PI and Annexin V-FITC were considered necrotic.
Delayed-type hypersensitivity assay
DTH was measured using a local adoptive transfer (LAT) assay (21). Spleens were collected 1 week after mice rejected BALB/c corneal allografts. Single cell suspensions were prepared and treated with anti-CD8 treated microbeads as described previously (21). CD8+ or CD8− spleen cells were mixed with normal BALB/c spleen cells, which served as the alloantigens. Ears of naïve C57BL/6 mice were measured with a Mitutoyo engineer's micrometer and 25 μL (2.5 × 106 cells) of the mixed cell suspension was injected into the right ear pinna of each mouse. The left ear pinna of each mouse was injected with BALB/c spleen cells alone, and served as the background control. Positive controls consisted of a mixture of spleen cells from C57BL/6 mice that were immunized subcutaneously with BALB/c spleen cells 2 weeks previously plus naïve BALB/c spleen cells, which served as the alloantigens. Negative controls consisted of spleen cell suspensions containing a mixture of naïve C57BL/6 cells plus naïve BALB/c spleen cells. Ears were measured again 24 h later.
Results were expressed as specific ear swelling = (24-h measurement − 0-h measurement) for experimental ear − (24-h measurement −0-h measurement) for negative control ear.
Statistical significance of DTH and apoptosis results was determined using Student's t-test. Median survival times (MST) for corneal allografts were determined and compared using the Mann-Whitney U-test.
Two categories of orthotopic corneal allografts were examined in the following experiments. Corneas transplanted to normal, avascular graft beds are categorized as ‘normal risk.’ By contrast, corneas transplanted to prevascularized or ‘high-risk’ corneal graft beds are typically rejected swifter and at a higher incidence than corneas transplanted to avascular graft beds. Moreover, corneas transplanted to normal-risk graft beds do not elicit donor-specific CTL, even in hosts that reject their corneal allografts (9). By contrast, corneal allografts transplanted to high-risk graft beds induce robust CTL responses (23). In order to evaluate the contribution of CD8+ and CD8− T cells in corneal graft rejection, spleen cells were collected from mice that had rejected BALB/c corneal allografts. The splenic CD8+ and CD8− T-cell populations were isolated and adoptively transferred to C57BL/6 SCID mice. C57BL/6 SCID mice were challenged with BALB/c corneal allografts after they received either CD8+ or CD8− spleen cells from mice that had rejected BALB/c corneal allografts. C57BL/6 SCID mice do not have functional T-cell populations and thus, cannot reject corneal allografts unless they are reconstituted with T cells.
Corneal allograft rejection mediated by either CD8+or CD8− T cells
Adoptive transfer experiments were performed to determine the role of CD8+ and CD8− T-cell subsets in corneal allograft rejection and to ascertain the effect of a vascularized, high-risk graft bed on the capacity of immune effector elements to gain entry to the graft bed and mediate rejection of corneal allografts. The results demonstrate that rejection occurred in 100% of the corneal allografts transplanted into either high-risk or normal-risk graft beds in the C57BL/6 SCID mice that received adoptively transferred spleen cells (Figure 1A). However, the tempo of rejection was significantly swifter in the recipients of the CD8− T-cell suspension (p = 0.007). By contrast, none of the BALB/c corneal grafts were rejected in SCID mice that did not receive spleen cells from euthymic donors (data not shown; N = 10).
The cell suspensions used for adoptive transfers were also assessed for immune effector functions that might contribute to corneal allograft rejection. A LAT assay was used to evaluate DTH in both the CD8+ and CD8− T-cell suspensions. Although CD8− T-cells suspension from C57BL/6 mice that rejected BALB/c corneal allografts produced significant DTH responses to BALB/c alloantigens, the CD8+ T-cell suspension failed to produce significant DTH activity (Figure 2A). By contrast, CTL activity was not detected in either the CD8− or the CD8+ cell suspension (data not shown). The CD8+ T-cell suspensions isolated from mice that had rejected corneal allografts that had been placed into normal risk (i.e. avascularized) graft beds did not produce demonstrable DTH or CTL activity. However, both the CD8+ and CD8− T cells induced significant in vitro apoptosis of BALB/c corneal cells (Figure 3).
High-risk graft beds and the generation of CD8− and CD8+effector cells
Experiments were performed to analyze the effect of a high-risk, vascularized graft bed on the generation of immune effector cells. BALB/c corneal allografts were transplanted into prevascularized graft beds in C57BL/6 hosts. CD8+ and CD8− spleen cell suspensions were prepared 1 week after corneal allograft rejection and were adoptively transferred to SCID mice, which received BALB/c corneal allografts that were placed into avascular graft beds. All of the corneal allografts underwent swift rejection in both groups of mice that received adoptively transferred cells (Figure 1B). As in the previous experiment, graft rejection was significantly faster in the recipients of CD8− T cells compared to mice receiving CD8+ T cells (p = 0.01). Unlike the CD8+ T-cell suspensions collected from normal-risk hosts, CD8+ T cells isolated from mice with prevascularized, high-risk graft beds mediated positive DTH responses to BALB/c alloantigens (Figure 2B).
Perforin-independent corneal graft rejection in high-risk and normal-risk hosts
Experiments were performed to determine if corneal allografts transplanted into high risk, vascularized graft beds in PKO hosts would induce the generation of CD8+ T cells that were capable of mediating corneal graft rejection, even though the perforin-deficient mouse is incapable of mounting CTL responses. BALB/c corneal allografts were transplanted orthotopically into vascularized graft beds of C57BL/6 PKO mice. CD8− and CD8+ spleen cell suspensions were collected 1 week after corneal allograft rejection and were adoptively transferred to SCID mice, which then received BALB/c corneal allografts placed into avascular (normal risk) graft beds. This created a situation in which immune cells were collected from mice that received corneal grafts placed into high-risk graft beds and were then transferred into SCID mice that were then challenged with corneal allografts placed into normal-risk graft beds. The corneal allografts underwent rejection in both the recipients of CD8− cells and recipients of CD8+ T cells from PKO donors that had rejected BALB/c corneal allografts (Figure 4A). A LAT assay was used to evaluate DTH in both the CD8− and CD8+ T-cell suspensions from similar PKO donors that had previously rejected BALB/c corneal allografts. As before, CD8− T cells from rejector mice produced positive DTH responses to the BALB/c alloantigens, while CD8+ T cells failed to produce detectable responses (Figure 5). However, both CD8− and CD8+ T-cell suspensions produced significant in vitro apoptosis of BALB/c corneal endothelial cells (Figure 6).
Additional experiments were performed to determine if a vascularized graft bed affected the capacity of perforin-deficient T cells to mediate graft rejection. In these experiments, corneal grafts were transplanted orthotopically into high-risk, prevascularized graft beds in C57BL/6 PKO mice. One week following corneal allograft rejection, CD8− and CD8+ spleen cell suspensions were collected and adoptively transferred to SCID mice, which contained prevascularized corneal graft beds into which. BALB/c corneal allografts were then transplanted. As in the previous experiment, either CD8− or CD8+ cells from PKO donors mediated the rejection of 100% of the corneal allografts (Figure 4B). Although the tempo of rejection was brisker in recipients of CD8− cells, the MST was not significantly different from that found in recipients of CD8+ cells (p > 0.05).
Effect of FasL deficiency on corneal allograft rejection in high- and low-risk graft beds
We have recently shown that FasL-defective C57BL/6 gld/gld mice reject only 50% of BALB/c corneal allografts transplanted into avascular graft beds (13). By contrast, normal C57BL/6 mice reject 95–100% of BALB/c corneal allografts placed into avascular graft beds. It was important to determine if the reduced incidence of rejection in the FasL-defective mice could be overcome by creating a high risk, vascularized graft bed. Accordingly, BALB/c corneal allografts were transplanted orthotopically into prevascularized graft beds in C57BL/6 gld/gld mice. The results indicated that the high-risk graft bed produced a steep increase in the incidence of corneal allograft rejection in the FasL-defective gld/gld mice; 100% of the BALB/c corneal allografts (18/18) underwent rejection with an MST of 17 days. Adoptive transfer experiments were performed to determine if the increased rejection that occurred in vascularized graft beds in C57BL/6 gld/gld mice was mediated by CD8− or CD8+ T cells. Accordingly, CD8− and CD8+ spleen cells were isolated from high risk C57BL/6 gld/gld mice 1 week following the rejection of BALB/c corneal allografts and were adoptively transferred to SCID mice that had either prevascularized or nonvascularized corneal graft beds. One day later, orthotopic BALB/c corneal allografts were transplanted orthotopically to the adoptive cell transfer recipients. The results show that all of the BALB/c corneal allografts underwent rejection in both the high-risk and normal-risk hosts that received CD8− T cells (Figure 7). By contrast, only 50–58% of the corneal allografts underwent rejection in hosts that received CD8+ T cells. It was noteworthy that the presence of a vascularized graft bed did not enhance rejection mediated by CD8+ T cells from C57BL/6 gld/gld mice, even though these cells were capable of producing DTH responses against BALB/c alloantigens in a LAT assay (Figure 8).
The presence of blood vessels in the graft bed at the time of corneal transplantation is arguably the leading risk factor for corneal allograft rejection. Several interesting conclusions can be drawn from the present results relating to the effect of a vascularized corneal graft bed on the generation and expression of cellular immune effectors of corneal graft rejection. In all of the hosts studied, both CD8− and CD8+ T cells were capable of mediating corneal allograft rejection. In every case, CD8− T cells—presumably CD4+ cells—were able to adoptively transfer graft rejection. By contrast, CD8+ T-cells-mediated rejection, but in many cases the tempo of rejection was slower, and in the case of FasL-defective mice, CD8+ T cells were only able to mediate rejection in 50% of the mice, even the CD8+ T cells that were collected from high-risk donors and transferred to high-risk recipients. These results, along with previous reports, which demonstrated that corneal allograft rejection occurs in approximately 50% of CD4 KO mice (10), contradict the notion that CD8+ T cells play no role in corneal graft rejection. Even in the case of FasL-defective hosts, CD8+ T-cells-mediated corneal allograft rejection in at least 50% of the hosts.
Although CD8+ T cells are best known for their CTL activity, this immune effector mechanism is not required for corneal allograft rejection, as corneal allograft rejection proceeds unabatedly in both PKO mice and CD8 KO mice (7,9). In the present study, we found that either CD8− or CD8+ T cells from perforin-deficient mice could mediate the rejection of 100% of BALB/c corneal allografts. Although, CD8+ T cells from PKO did not display CTL or DTH activity, they did induce in vitro apoptosis of BALB/c corneal cells. Apoptosis of donor corneal cells is a plausible mechanism to account for corneal graft rejection by CD8+ PKO T cells and is consistent with previous findings, which noted the presence of apoptotic keratocytes and corneal endothelial cells in rejected corneal allografts in humans and rats, respectively (24,25).
The role of FasL in corneal allograft rejection is complicated. FasL is constitutively expressed on corneal epithelium and endothelium and is necessary for the long-term survival of murine corneal allografts. Two independent studies have shown that corneal allografts from FasL-defective gld/gld mice experienced a 2-fold increase in the incidence of immune rejection compared to corneal allografts prepared from donors with normal FasL expression (26,27). However, corneal epithelium and endothelium also express Fas receptor and can undergo FasL-induced apoptosis (13,28). Previous findings indicate that FasL-defective gld/gld C57BL/6 mice display an impaired capacity to reject BALB/c corneal allografts (13). Our results suggest that the CD8+ T-cell population in gld/gld hosts is significantly less capable of mediating corneal allograft rejection, even when CD8+ T cells are collected from donors that have rejected high-risk corneal allografts and are transferred to high-risk recipients. Moreover, CD8+ T cells from high-risk gld/gld hosts produce donor-specific DTH responses, yet only half of the recipients of CD8+ T-cells-rejected BALB/c corneal allografts. Although FasL-defective gld/gld mice can mount allospecific CTL responses (29), no CTL activity was detected in the CD8+ T-cell suspensions from high-risk gld/gld corneal allograft rejector mice.
The CD8− cell suspensions used in the present studies might have contained natural killer (NK) cells and CD4+ NK T cells, which have been implicated in the rejection of some categories of allografts. However, we have previously shown that corneal allografts are not rejected by athymic, nude mice, which have a normal NK cell repertoire (13). The possibility that NK T cells contribute to corneal allograft rejection is unlikely, as Sonoda and co-workers have shown that NK T cells are necessary for corneal allograft survival and mice deficient in CD1d-reactive NK T cells have an increased incidence of corneal allograft rejection (30).
In summary, these results demonstrate that corneal allograft rejection: (a) can be mediated by either CD8+ or CD8− T cells; (b) can occur in the absence of demonstrable CTL or DTH; (c) does not require functional FasL or perforin; (d) is impaired in SCID mice that receive CD8+ T cells from FasL-defective donors, even if the CD8+ T-cell donor is a high-risk graft rejector and the corneal graft is placed into a prevascularized, high-risk graft bed in the SCID mouse recipient and (e) in some cases, corneal allograft rejection correlates with CD8+ T-cell-mediated in vitro apoptosis of donor-specific T cells. The present findings along with previous reports demonstrating CD4+ T-cell-independent mechanisms of corneal allograft rejection demonstrate the redundancy in the immune effector mechanisms that can be invoked in the rejection of corneal allografts (10,13–15).
This work was supported by NIH grants EY07641 and EY016664 and an unrestricted grant from Research to Prevent Blindness, Inc., New York, NY.