Though CD8+ T lymphocytes are important cellular mediators of islet allograft rejection, their molecular mechanism of rejection remains unidentified. Surprisingly, while it is generally assumed that CD8+ T cells require classic cytotoxic mechanisms to kill grafts in vivo, neither perforin nor FasL (CD95L) are required for acute islet allograft rejection. Thus, it is unclear whether such contact-dependent cytotoxic pathways play an essential role in islet rejection. Moreover, both perforin and CD95L have been implicated in playing roles in peripheral tolerance, further obscuring the role of these effector pathways in rejection. Therefore, we determined whether perforin and/or FasL (CD95L) were required by donor MHC-restricted (‘direct’) CD8+ T cells to reject islet allografts in vivo. Islet allograft rejection by primed, alloreactive CD8+ T cells was examined independently of other lymphocyte subpopulations via adoptive transfer studies. Individual disruption of T-cell-derived perforin or allograft Fas expression had limited impact on graft rejection. However, simultaneous disruption of both pathways prevented allograft rejection in most recipients despite the chronic persistence of transferred T cells at the graft site. Thus, while there are clearly multiple cellular pathways of allograft rejection, perforin and FasL comprise alternate and necessary routes of acute CD8+ T-cell-mediated islet allograft rejection.
T-cell-dependent islet allograft rejection is a complex process that remains challenging to characterize due to the multiple cellular and molecular pathways involved in the response. To date, most evidence indicates that both CD8+ and CD4+ T cells participate in islet allograft rejection (1–4), but the relative contribution of each cell type to the rejection response is often ambiguous. Furthermore, the molecular effector mechanisms required for the actual execution of the islet allograft by host T cells remain surprisingly elusive. For example, despite the generally accepted notion that CD8+ T cells use cytolytic mechanisms in vivo, there is surprisingly little evidence that this is, in fact, the case in allograft rejection in vivo. In fact, published studies challenge this concept. For example, while studies show that increased perforin or granzyme gene expression clearly correlates with graft rejection (5,6), including islet allograft rejection (7), gene disruption studies show that perforin is not required for rejection (6,8,9). Indeed, neither of the classical CTL effector mechanisms, perforin or FasL/Fas, appear to be necessary for islet allograft rejection (8,9), or indeed for some forms of CD8+ T-cell-mediated pathology in vivo (10,11). Such results have led to the speculation that cytotoxic pathways are dispensable for acute allograft rejection (8,12). In addition, both perforin and Fas have been shown to play major roles in peripheral lymphocyte homeostasis and tolerance (13–17), further obscuring the role of such cytolytic pathways as effectors of graft rejection. One could even argue that such pathways play a greater role in immune regulation than in allograft rejection. Based on these varied issues, we reexamined the contribution of the perforin- and FasL-dependent cytolytic pathways in CD8+ T-cell-mediated islet allograft rejection. We reasoned that the difficulty in defining rate-limiting effector mechanisms of rejection may be explained by the presence of multiple compensatory effector pathways in immuno competent animals, making rejection difficult to study without limiting the model system to study a single arm of the response (CD8+ T cells).
Previous studies implicate CD8+ cells as being central to the final destruction of islet allografts in immune-competent animals. CD8+ T cells can be required for efficient acute islet rejection (4,18) and, once activated, are sufficient to mediate islet allograft rejection (9). Also, MHC class I expression is important for islet rejection (19–21), further implicating MHC class I-directed reactivity in terminal graft injury. In vitro, activated CD8+ T cells are classically known to mediate target cell death primarily by using two independent but complementary cytolytic effector mechanisms—perforin and FasL (22). Such results also have been extended to the in vitro killing of islet cells by reactive CD8+ T cells (23). In perforin-dependent cell killing, activated CD8+ T cells release cytotoxic granules containing perforin and granzymes upon T-cell receptor (TCR)–MHC engagement with target cells. Results show that perforin acts by permeabilizing the target cell membrane and allowing granzyme B to enter the target cell cytoplasm, where the latter activates caspases and other apoptotic pathways (reviewed in (24)). In Fas (CD95)-directed killing, activated CD8+ T cells upregulate cell surface FasL (CD95L) expression, which binds to its cognate ligand Fas (CD95), a pro-apoptotic signaling molecule expressed on the surface of susceptible target cells, ultimately triggering caspase activation and target cell death (reviewed in (25)). Because ‘direct’ (donor MHC-restricted) alloreactive CD8+ T cells require TCR–MHC contact with graft tissue for rejection to occur, we hypothesized that the perforin and FasL/Fas pathways might be essential mechanisms of CD8+ T-cell-mediated islet allograft destruction. In light of these data, we chose to investigate the effector mechanisms used by primed, alloreactive CD8+ T cells in vivo in isolation from CD4+ T-cell participation. Results show that, indeed, contact-dependent alloreactive CD8+ T cells predominantly require the alternative use of either perforin- or Fas-dependent effector pathways to mediate efficient acute islet allograft rejection in vivo.
Materials and Methods
Female BALB/c (H-2d), BALB/c β2m (H-2d), BALB/c DM2 (H-2d), C3H/HeJ (H-2k; ‘C3H’) and C3.MRL-Tnfrsf6lpr (H-2k; C3H-Fas deficient), and male B6.129S7-Rag1tm1Mom (H-2b; ‘B6 rag’) and C57B6 (H-2b) and C57B6-Pfptm1 (H-2b; ‘B6-Perforin deficient’) were obtained from The Jackson Laboratory (Bar Harbor, ME) and housed in the Barbara Davis Center rodent facility. B6 2C transgenic mice were obtained from Dr. Richard Miller (Ontario, Canada) and bred for >25 generations on the B6 background at the Barbara Davis Center rodent facility (Denver, CO). All animals and procedures were performed in accordance with the National Institutes of Health guidelines.
CTL (51Cr release) assay
In vitro lysis was determine by a standard 51Cr release assay as described (26). Briefly, target cells were generated by culturing splenocytes with Concavalin A for 48 h. Targets were then washed and labeled with 100 μCi 51Cr, then cocultured with primed T lymphocytes from mixed lymphocyte cultures for 4 h. Spontaneous Lysis was determined by culturing labeled targets in culture medium and Maximum Lysis was determined by culturing labeled targets in 1% SDS. After 4 h, 20 μL of the culture supernatants was transferred to a LumaPlate, and the plate was read on a Wallac Top Count gamma counter. Each sample was cultured in quadruplicate, and average counts were calculated. Percent specific lysis was calculated by the formula:% Specific Lysis = (Sample Lysis – Spontaneous Lysis)/(Maximum Lysis – Spontaneous Lysis).
Induction of diabetes
Diabetes was chemically induced in recipient mice via a single i.v. injection of streptozotocin (180 mg/kg; Calbiochem, La Jolla, CA). Non-fasting whole blood glucose was measured with a OneTouch Ultra glucose meter (LifeScan, Inc., Milpitas, CA). Chemically-induced diabetes was defined by a minimum of two consecutive blood glucose readings ≥20 mM.
Islet isolation and transplantation
Donor islets were isolated from female mice by collagenase (Sigma, St. Louis, MO) digestion and Histopaque (Sigma, St. Louis, MO) purification, as reported previously (26). Four hundred and fifty islets were handpicked and transplanted under the left kidney capsule of male diabetic recipients (26). Blood glucose measurements were taken three times per week. Graft rejection was defined as two consecutive blood glucose measurements >10 mM, which corresponds to approximately three SDs above the mean normal blood glucose levels in untreated animals. In mice with grafts functioning >60 days, removal of the graft-bearing kidney verified graft-dependent maintenance of normoglycemia. Where indicated, recipients were depleted of peripheral CD8+ T cells with the depleting monoclonal antibody 2.43 (27) at a dose of 20 mg/kg on days −1,0,2,4 relative to transplantation. Controls were treated with the same dose of non-immune rat IgG.
CD8+ T Lymphocyte Isolation
Lymph nodes and spleens were homogenized to a single-cell suspension, and red blood cells were removed by centrifugation over a Lympholyte M (Cedarlane Laboratories, Hornby, Ontario, Canada) gradient. CD8+ lymphocytes were enriched by passing the cells over a Cellect Immunoaffinity Column (Biotex, Edmonton, Canada) to remove IgG+ and CD4+ cells. Purity of eluted CD8+ T cells was verified by flow cytometry (regularly <0.2% CD4+ and <0.2% CD19+ cells, data not shown).
Mixed lymphocyte culture
Column purified CD8+ T lymphocytes were cocultured upright in 25-mL flasks with irradiated allogeneic stimulator cells at a 1:3 ratio in Eagle's Minimal Essential Medium Plus 10% FCS. Cells were incubated at 37°C, 10% CO2 for either two (in vitro51Cr release assay) or 5 (adoptive transfer into allograft recipients) days.
Adoptive transfer of primed CD8+ cells
Islets were transplanted into immunodeficient B6 rag1−/- recipient mice, and grafts were allowed to normalize for at least 14 days before adoptive transfer of primed CD8+ cells (9). Ten million in vitro-primed CD8+ cells from mixed lymphocyte cultures were injected i.v. into transplanted animals, and blood glucose was measured to determine graft survival.
Histological examination of islet grafts
Islet-graft bearing kidneys, either removed after graft rejection or by survival nephrectomy at least 60 days after transplant, were fixed in 10% (v/v) formalin in PBS for paraffin embedding. Tissue sections were stained with Harris' hematoxylin-eosin (Fisher Scientific, Pittsburgh, PA) or stained for insulin by immunoperoxidase staining with a polyclonal guinea pig anti-swine insulin antibody (DakoCytomation Inc., Carpinteria, CA). The degree of mononuclear cell infiltration and islet tissue damage was determined for each section.
Statistical evaluation of transplant survival data was performed with the ranked sum Mann-Whitney U test.
Donor MHC class I expression is required for efficient islet allograft by activated CD8± T cells
We first examined the importance of target MHC Class I expression for in vitro cytolysis in a model of defined, ‘direct’ (donor MHC-restricted) CD8+ cells, the 2C CD8+ TCR transgenic model (28). We assessed the ability of primed TCR transgenic 2C T cells to lyse BALB/c wild type, DM2 and MHC class I-deficient β2M−/- (29) splenocyte-derived ConA-induced blast cell targets in vitro. The DM2 strain is a spontaneous deletion mutant of Ld, serving as a specificity control for the 2C transgenic T cell (30). It is important to note that β2m-deficient cells can potentially express low-level MHC class I heavy chains (independent of β2m) on the cell surface (31) that can potentially be recognized by primed CD8+ T cells, so it was not clear whether or not β2m-/- cells or allografts could be targeted by CD8+ T cells. Results show that although activated 2C cells potently lyse wild-type BALB/c target cell, they fail to exert detectable lysis of either DM2 or β2M−/- targets in vitro (Figure 1A). This result illustrates the concept that target expression of MHC Class I is, indeed, necessary for cytolysis by an activated ‘direct’ CD8+ T cell in vitro. These data further indicate that the BALB/c β2M−/- animal provides a useful model of MHC Class I deficiency for subsequent in vivo examination.
Next we determined whether donor MHC Class I recognition is necessary for graft destruction by donor-reactive CD8+ T cells in vivo. Specifically, we determined whether a direct engagement of donor MHC Class I of the islet allograft target was necessary for islet allograft destruction by these cells. To test this, we transplanted either BALB/c wild type or β2M−/- (class I deficient) islets into diabetic B6 Rag-/- recipients. Fourteen days after transplant, we adoptively transferred ten million in vitro-primed B6 CD8+ T lymphocytes into the immunodeficient allograft recipients to reconstitute anti-donor immunity. In addition to initial purification of CD8+ T cells and flow cytometric analysis following activation, Rag-/- hosts were also treated with anti-CD4+ (GK1.5) to further ensure that potentially contaminating CD4+ T cells would not participate in the response. While wild-type grafts were rapidly rejected, three of four β2M deficient islet allografts survived indefinitely in the presence of primed allospecific CD8+ T cells (Figure 1B, p < 0.05). Flow cytometric analysis of peripheral blood verified the presence of CD8+ cells in these animals, 60 days after adoptive transfer (data not shown). These results defined our system as a model of primarily direct CD8+ T-cell killing in vivo. These direct CD8+ T cells, by definition, require a TCR–MHC engagement with the allograft in order to injure the donor cells.
Primed CD8±s require either perforin or FasL/Fas to destroy islet allografts
Although previous studies show that neither perforin nor FasL were required to mediate islet allograft rejection, we tested the hypothesis that perforin and FasL/Fas represent alternate but essential mechanisms of CD8+ T-cell-mediated graft rejection. Thus, we eliminated perforin from reactive CD8+ T cells, Fas expression from the islet allograft target, or disrupted both pathways simultaneously. Because perforin deficient (pfp-/-)/FasL deficient double-mutant mice are too sickly to be used as CD8+ T-cell donors (32), we chose to disrupt the FasL/Fas pathway in this system by transplanting Fas deficient islet allografts. Specifically, we transplanted C3H wild type or Fas deficient (lpr) islets into diabetic B6 rag recipients. Fourteen days after transplant, we transferred 10 million in vitro primed B6 wild type or perforin deficient CD8+ T cells into grafted animals, and blood glucose was measured to determine graft survival. Wild-type CD8+ T cells rejected all wild-type C3H islet allografts within 20 days post T-cell transfer. Periodically, rejecting islet allografts were dissociated into single cell suspensions and analyzed by flow cytometry to confirm the presence of CD8+ T cells. Such analysis confirmed the presence of CD8+ T cells without detectable populations of either CD4+ T cells or B220+ B cells. Thus, primed CD8+ T cells are sufficient to trigger the rejection of islet allografts, consistent with our previous results using this model system (9).
The individual disruption of either perforin from the responding CD8+ T cells or Fas expression by the allograft target led to modestly delayed allograft rejection (Figure 2), but this was not statistically different from wild-type controls (p = NS). These results confirmed that neither perforin nor Fas, alone, is necessary for CD8+ T-cell-mediated islet allograft rejection. However, we found that simultaneous disruption of T-cell-derived perforin and allograft Fas expression resulted in marked reduction of CD8+ T-cell-mediated rejection relative to wild-type controls (Figure 2; p < 0.001). Indeed, 11 out of 13 islet allografts in this group maintained normoglycemia for greater than the 60 day observation period. These data indicate that the perforin and FasL/Fas pathways are essential for efficient acute islet allograft rejection by direct (donor MHC-restricted) CD8+ T cells in vivo. Furthermore, competence of either pathway in the absence of the other is sufficient for allograft rejection.
Pfp-/- T cells migrate to and persist at the graft site despite a failure to reject Fas-deficient islets
After 60 days post T-cell transfer, graft-bearing kidneys were removed to verify graft-dependent maintenance of normoglycemia, and for histological analysis. Though these long-term surviving allografts appeared grossly intact macroscopically, we were surprised to discover a florid infiltrate of pfp-/- CD8+ cells amidst intact Fas deficient islets with relatively undisturbed islet architecture (Figure 3A) and normal insulin staining (Figure 3B). Strikingly, these lymphocytes persisted in the graft even 60 days after lymphocyte transfer, despite their inability to destroy the transplanted islets. Together, these data indicate that primed perforin deficient CD8+ T lymphocytes are capable of infiltrating Fas deficient islet allografts, but they are impotent to mediate graft destruction.
Perforin-deficient mice show only limited dependence on CD8+ T cells for islet allograft rejection
The finding that alloreactive CD8+ T cells rely largely on perforin and/or FasL to mediate islet rejection would appear to conflict with previous studies indicating neither perforin nor Fas are required for islet allograft rejection (8). This would be especially perplexing if CD8+ T cells are indeed a major mediator of islet allograft rejection. However, as mentioned above, there are potentially multiple cellular and molecular pathways of allograft rejection, some of which are likely to be CD8+ T cell independent. To determine whether perforin-deficient mice potentially utilize alternate, CD8+ independent routes of rejection, wild type and perforin-deficient mice were depleted of peripheral CD8+ T cells by monoclonal antibody therapy in vivo (2.43). This depletion resulted in >98% depletion of CD8+ T cells in both wild-type and pfp-/- recipient mice (data not shown), consistent with previous studies utilizing this antibody (17,21,33). Although anti-CD8+ treatment led to a significant prolongation of islet allograft survival in perforin-deficient mice (p = .02), this result was quite modest (Figure 4) with a mean extended allograft survival of only approximately 10 days relative to control treated animals (mean survival of 20.3 ± 3.2 days in anti-CD8+ treated versus 10.3 days ± 1.0 days in control treated perforin-deficient mice). Interestingly, anti-CD8+ treatment was apparently more effective in prolonging islet allografts in wild-type mice, but this prolongation was not significantly different to results using pfp-/- recipients. Thus, while there is a degree of CD8+ T-cell dependence in islet allograft rejection in perforin-deficient hosts, this does not appear to be a major, rate-limiting pathway of rejection in these animals.
Despite extensive characterization of acute islet allograft rejection, surprisingly little is known defining rate-limiting pathways whereby T lymphocytes actually inflict islet graft injury in vivo. Though it is generally accepted that effector CD8+ T cells utilize the cytotoxic pathways perforin and FasL/Fas to mediate target cell injury in vitro (22), there is little indication that this is the case for graft rejection in vivo. In fact, there has been considerable evidence to the contrary. For example, despite the strong correlation of perforin gene expression during acute rejection in vivo (5–7), perforin is not actually required for acute allograft rejection (6,15), including the rejection of islets (8,9,17) Specifically, Ahmed et al. found that Fas deficient islet allografts were rejected with normal kinetics by perforin deficient mice (8). These data supported the conclusion that these pathways may correlate with rejection, but are not required for the terminal execution of the graft. Such results have led to the speculation that cytotoxic pathways are dispensable for acute allograft rejection (8,12). In addition, we had previously found the CD8+ T cells require IFNγ production, rather than perforin or FasL expression, to reject islet allograft rejection, further casting doubt on the requirement for cytotoxic activity in CD8+ mediated rejection (9).
In the current study, we show that graft destruction by activated CD8+ T cells predominantly requires donor MHC Class I expression, indicating an interaction with the graft via a ‘direct’ cognate (TCR-mediated) recognition of the donor. Additionally, data show that individual disruption of either T-cell-derived perforin or allograft Fas expression does not significantly delay islet allograft rejection by these direct CD8+ T cells. However, simultaneously interrupting both of these classical cytolytic pathways largely attenuates CD8+ T-cell-mediated graft rejection. Such results suggest that islet allograft rejection mediated by ‘direct’ CD8+ T cells is largely achieved by alternative, but obligate utilization of perforin and FasL dependent effector mechanisms. Thus, results indicate that this defined CD8+ T-cell pathway of rejection is indeed dependent on overlapping, but limited pathways of cell-mediated cytotxicity in vivo. It is important to note that there was a trend towards graft prolongation with the individual disruption of either graft Fas expression or T-cell-derived perforin expression (though such prolongation was not statistically significant). Thus, while the majority of acute CD8+ T- cell-mediated rejection is circumscribed within the function of perforin and/or FasL, it is possible that these two pathways have an additive impact on graft injury during unmodified acute rejection.
Importantly, the mechanism of T-cell-dependent graft injury may be intimately related to the precise nature of graft target cell recognition. For example, two distinct CD8+ T-cell recognition pathways potentially contribute to target destruction: ‘Direct’ (donor MHC-restricted) and ‘Indirect’ (host MHC-restricted) (reviewed in (34)). Though most available evidence favors the contribution of CD8+ cells that directly engage donor MHC class I molecules on islet allografts (19–21), there can also be a role for indirect (cross-primed) host MHC class I/ V-restricted CD8+ T cells in alloreactivity (35,36). Of course, alloreactive CD4+ T cells may likewise trigger an ‘indirect’ response to donor-derived antigens. This distinction between the direct and indirect pathways of graft recognition may be critical regarding the corresponding mechanism of graft injury. If primed T lymphocytes require a cognate TCR-mediated engagement with allograft target cells in order to successfully destroy them, we would envision that contact-dependent killing mechanisms delivered with great precision may be involved, with perforin and FasL being the most likely candidate effector pathways. If, however, primed T lymphocytes are capable of orchestrating graft destruction without directly engaging donor MHC, we imagine that a ‘bystander’ type of inflammatory injury by a number of potential inflammatory cytokines and other mediators are likely to be involved in the death of the target (37,38). Thus, the simultaneous involvement of both direct and indirect mechanisms of graft injury is likely to obfuscate the role of classical contact-dependent cytolytic pathways in rejection. Notably, current results indicate that perforin-deficient mice show only a modest reliance on CD8+ T cells for rejection, further suggesting that other redundant pathways of alloreactivity are sufficient to mediate rejection independently of contact-dependent CD8+ T cells. Though not statistically significant, there appears to be reduced CD8+ T-cell dependency for acute islet rejection in pfp-/- mice relative to wild-type controls, further supporting the concept of heightened immune dysregulation found in pfp-/- animals (14,15,17).
Interestingly, results suggest a minor component of CD8+ T-cell-mediated rejection that is both perforin and Fas independent (Figure 2). It is possible that pro-inflammatory cytokines, such as TNF and IFN, produced by alloreactive T cells can be sufficient to mediate islet injury (39,40). We had previously found that IFNγ production by CD8+ T cells was also important for islet allograft rejection (9). However, in our experimental model, such cytokine-mediated inflammatory injury appears to comprise only a minor role in CD8+ T-cell-mediated acute rejection relative to the perforin and FasL/Fas-dependent cytotoxic pathways. Unfortunately, the identity of such minor pathways would be quite difficult to assess since only a small proportion of allografts appear to reject by such mechanisms in this model. Nonetheless, data indicate that perforin and Fas are the predominant mechanisms of CD8+ T-cell-mediated islet allograft destruction, and that competence of either pathway is sufficient for relatively normal graft rejection.
Based on current and previous findings, we would propose a ‘two-hit’ model of efficient CD8+ T-cell-mediated islet allograft rejection in vivo. Since IFNγ plays an important contribution to CD8+ T-cell-mediated rejection (9), it is possible that such cytokines or other mediators upregulate the expression of islet target receptors rendering them susceptible to subsequent cytolytic effector molecules, such as Fas and perforin/granzymes. Islet cells are known to upregulate Fas in response to IL-1α, IL-1β, IFN-γ (41) and nitric oxide (42). Also, other findings indicate perforin/granzyme mediated killing is facilitated by target cell-surface receptors, such as the platelet activating factor receptor (PAFR) for perforin (43) and the mannose-6-phosphate receptor (M6PR) receptor for granzyme B (44). Importantly, inflammatory cytokines, such as IFN-γ, can induce PAFR and M6PR expression on some target cells (43). It is thus possible that vulnerability to cytolytic mediators is not a constitutive property of allograft parenchymal cells, but rather involves induction of appropriate cell-surface ligands for cytolytic effector molecules by the inflammatory micro-environment during rejection.
In summary, the requirement for classical cytolytic molecules FasL and perforin has been controversial as effector mechanisms of allograft rejection in vivo. Furthermore, both perforin and Fas have been shown to have important roles in immune regulation (13–17), further complicating our understanding of these molecules in allograft immunity. Results show that rejection of islet allografts by direct, donor MHC-directed CD8+ T cells is largely defined by the alternative use of perforin and FasL, supporting an important role for cell-mediated cytotoxicity in this form of rejection. However, it is probable that the significance of such cytolytic effector mechanisms in rejection is masked by concurrent perforin- and FasL-independent pathways of alloreactivity in vivo.
We would like to thank Tinalyn M. Kupfer and Sarah M. Weber for critically reading this manuscript, and Travis Still for his technical assistance with the histology.