Transplantation Survival Is Maintained by Granzyme B⁺ Regulatory Cells and Adaptive Regulatory T Cells

The goal of transplantation immunotherapy is to induce and maintain a state of donor-specific immunologic tolerance. Billingham et al. (1) was the first to demonstrate that donor-specific transfusion (DST)² can lead to acceptance of skin grafts, while maintaining reactivity to third-party allografts (1). This tolerance was determined to be due to mixed chimerism in infant mice leading to thymic negative selection (2). We now know that peripheral tolerance mechanisms of clonal anergy, activation-induced cell death (AICD), and most importantly active suppression contributes extensively to graft persistence (reviewed in Ref. 3). However, an understanding of the mechanism of this active suppression is still lacking.

Acquired immune tolerance to an allogeneic transplant has been repeatedly demonstrated by the blocking of costimulatory molecules. Although intervention via costimulatory blockade alone shows significant prolongation of graft survival (4, 5), when used in conjunction with DST, long-lived graft tolerance can be established (6, 7, 8). In graft-versus-host disease and bone marrow (BM) transplantation systems, costimulation blockade has been shown to enhance the establishment of chimerism (9) and synergize with certain pharmacological reagents offering a possible bridging intervention between current therapies and future therapies for transplantation (10). This adoptive transfer of noninflammatory Ag in the presence of costimulation blockade leads to the induction of anergized and clonally exhausted host CD4⁺ and CD8⁺ effector T cells (T^eff) (11, 12). Efficient establishment and maintenance of this immunosuppressive state requires the CD4⁺CD25⁺ regulatory T cell (T^reg) (13).

T^reg have been demonstrated to be a potent inducer of peripheral immunosuppression (reviewed in Ref. 14). To date, the most definitive marker of this regulatory population is the transcription factor FoxP3 (15). This suppressor population is selected for high-affinity reactivity to self-peptide during thymic selection and to help maintain tolerance to self in the periphery (16, 17). The mechanism of suppression by thymically derived natural T^reg (nT^reg) has been shown to be in part, cell contact-dependant, as separation via Transwell assay ablates suppressive capacity of this population (18, 19). A second population of T^reg cells can be induced in the periphery upon encounter with Ag under noninflammatory conditions. The mechanisms of suppression by these adaptive T^reg (aT^reg) vary based on their origin of differentiation. Tr1 aT^reg exclusively suppress in an IL-10-dependent bystander manner (20). TH3/Tr2 cells, characterized from the mesentery, are induced by TGF-β and in turn produce TGF-β to mediate suppression (21). However, aT^reg which up-regulate FoxP3, also appear to have the capacity for contact-mediated suppression, similar to their nT^reg brethren (22). Thus, we can implicate FoxP3 expression as directly instilling the ability to mediate contact suppression. Potential mechanisms of contact-mediated suppression have been proposed, although most are delineated via in vitro assays or reconstitution experiments in lymphopenic hosts (23, 24, 25, 26, 27, 28). T^reg-specific loss of these effector mechanisms in an intact in vivo system has not been demonstrated. In this report, we establish that the loss of granzyme B (GZB), in the FoxP3⁺ compartment, leads to the loss of long-lived survival in a model of skin graft rejection.

GZB is a serine protease found primarily in CD8⁺ T cells and NK cells (29), although recent reports have demonstrated expression of granzyme in CD4⁺ T cells (30, 31). The expression of granzyme has been documented in both human and murine nT^reg (27, 28, 32) as well as human aT^reg (33). The endogenous physiologic inhibitor of GZB, serine protease inhibitor 6 (Spi-6), helps to protect these cells from autoapoptosis due to the leakiness of granulytic vesicles (34, 35). Overexpression of Spi-6 leads to immune escape from GZB-regulated cell death (36). In the present study, we show that overexpression of Spi-6 by the host can undermine T^reg-dependent tolerance, further substantiating a role of GZB in dominant peripheral tolerance. Additionally, we extend this mechanism in vivo in the context of transplantation and demonstrate that GZB expression in the FoxP3⁺ population is required for the maintenance of long-lived graft survival.

Congenic strains CD45.1 or CD45.2 C57BL/6 as well as (C57BL/6 × BALB/c)F₁ (CB6F1), 5–8 wk old, were purchased from Taconic Farms. Mice deficient in Rag1 and perforin were purchased from The Jackson Laboratory. Scurfin (Sf) mice, harboring a naturally occurring point mutation in exon 8 of FoxP3, were obtained from The Jackson Laboratory and used immediately upon arrival. The TEa CD4⁺ TCR-transgenic (Tg) mice were provided by Dr. A. Rudensky (37). The TEa Tg expresses a TCR that recognizes the peptide ASFAQGALANIAVDKA presented in the context of I-A^b. This peptide is derived from the I-Eα^d chain corresponding to amino acids position 52–68. The Spi-6Tg mice were provided by Dr. P. Ashton-Rickardt (38). C57BL/6 GZB^neo/neo knockout mice were provided by Dr. T. Ley (39). C57BL/6 2C⁺ Tg mice were provided by Dr. D. Loh (40). The FoxP3-GFP mice were provided by Dr. A. Rudensky (15). Mice were bred and maintained in pathogen-free microisolator cages in our facility at Dartmouth Medical School.

Single-cell suspensions from wild-type (Wt), PRF^–/– GZB^–/–, or Spi-6Tg mice were prepared from 5- to 8-wk-old mice and applied to CD4 enrichment. CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells were further purified by magnetic separation with MACS (Miltenyi Biotec) according to the manufacturer’s instructions. Enriched cell populations and purified cells were phenotypically analyzed by FACS. The purities of CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells were >90–95%, respectively. In a polyclonal T^reg suppressor assay, CD4⁺CD25⁻ T^eff cells (5 × 10⁴) were cocultured with irradiated T-depleted splenocytes (1.5 × 10⁵), 5 μg/ml anti-CD3 and the indicated numbers of CD4⁺CD25⁺ cells for 3 days. Proliferation was assessed by both 5 μM CFSE dye dilution, assessing the full 3 days of proliferation, in conjunction with [³H]thymidine incorporation (1 μCi/well), which was added for the last 8 h of culture. Duplicates were analyzed via CFSE dilution and triplicate wells were assessed for tritium incorporation.

Skin grafting was done following the procedure as previously described (11). Briefly, 1-cm² full-thickness tail skins were excised from the CB6F1 allogeneic donor or C57BL/6-syngenic donor mice and stored on PBS-soaked gauze. The following day skins were applied to the dorsal surface of age- matched C57BL/6 hosts. Seven days before grafting, mice were tolerized to alloantigen via the i.v. adoptive transfer of 3 × 10⁷ T-depleted splenocytes in conjunction with an injection of 250 μg of anti-CD154 (clone MR1) followed by further injections on days −5 and −3 pre-graft.

To monitor the alloreactive expansion of cells in vivo, the adoptive transfer of TEa Tg cells was implemented. Lymph nodes (LN) and spleens from CD45.1 TEa Tg mice were harvested and processed through a CD4⁺ Miltenyi Biotec selection as described above. After purification, an aliquot of cells was stained for anti-Vα2 and anti-CD4 (BD Pharmingen) to determine purity. Cells were labeled with 10 μM CFSE. Labeled cells were mixed with the DST and given i.v. at 1 × 10⁶ TEa cells/mouse. Several groups received three shots of anti-CD154 as described above. Five days after adoptive transfer, LN and spleen from CD45.2 host mice were harvested and stained for anti-CD45.1-PE (clone A20), anti-CD4-PerCP (clone RM4-5, anti-CD44-allophycocyanin (clone IM7), anti-CD62L-allophycocyanin (clone Mel14; BD Pharmingen). For intracellular staining, cells were restimulated for 2 h with PMA/ionomycin (Sigma-Aldrich) with the last 1.5 h in the context of monensin (eBioscience). Cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences) and stained for intracellular anti-IFN-γ-PE (clone XMG1.2; eBioscience).

Donor mice were killed via CO₂ and BM was extracted from the femur and tibia bones. BM was filtered through 100-μm cells strainers and lysed of RBC via ACT treatment. BM was then treated with anti-Thy1.2 for T depletion. Cells were counted and mixed in the described ratios. Host C57BL/6 CD45.1 mice were lethally irradiated with a split dose of 1200 rad and i.v. injected with a total of 1–2 × 10⁶ BM cells. Engraftment was allowed to take place over the next 8–10 wk. Mice were analyzed pre- and postexperiment for the ratio of congenic marker. Where assessable, residual host BM contributed <5% of total to the chimerism.

For reconstitution of Rag^−/− mice, FoxP3-GFP was bred onto the TEa Tg and GZB^−/− backgrounds. T^reg and T^eff were flow sorted to >99% purity on a FACSAria and 1–1.25 × 10⁵ cells of T^eff and T^reg 1:1 were adoptively transferred into Rag^−/− mice. The following day, CB6F1 skin grafts were applied to reconstituted mice.

MultiScreen_HTS immunoprecipitation filter plates (Millipore) were coated overnight with 1 mg/ml capture Ab clone AN18 (Mabtech). Plates were blocked for 2 h in complete RPMI 1640 at 37°C and washed three times with PBS/Tween 20 (0.05%). Mice for stimulator cells were injected 24 h before with 100 μg of anti-CD40 (FGK.115). B cells were isolated from spleens of C57BL/6 and CB6F1 mice via CD19⁺ magnetic beads (Miltenyi Biotech). Total lymph node cells were isolated and splenic T^eff cells were isolated from Sf and (CD45.2) Wt (CD45.1/2) chimeric mice via flow cytometric sorting (FACSAria), and plated at 2 × 10⁵ cells/well. Cells were allowed to settle for 30 min, then overlayed with 2 × 10⁵ purified B cells. Plates were then incubated at 37°C/5% CO₂ for 24 h. After washing with PBS/Tween 20, detection Ab (1 mg/ml clone R4-6A2-biotin; Mabtech) were added for 2 h at 37°C. Plates were washed three times with PBS/Tween 20 and then incubated with streptavidin-HRP for 20 min. Plates were then further washed three times with PBS/Tween 20, three times with PBS, and incubated with 2-amino-9-ethylcarbazole chromingenic substrate per the manufacturer’s recommendations (Sigma-Aldrich). ELISPOTs were enumerated via an ImmunoSpot software (CTL Analyzers) or by direct visual counting using a dual-axis light-dissecting microscope.

Histological analysis of skin graft infiltration was performed at days 7 and 60 after grafting. For immunohistology, previously grafted skins were shaved, excised, and snap frozen, cryocut (50-μm-thick sections), and acetone fixed. Slides were blocked with normal mouse serum. Tissue sections were stained for anti-CD4-FITC, anti-FoxP3-PE (eBioscience), and anti-GZB-allophycocyanin (Caltag Laboratories). Ten randomly chosen images from each of three individual grafts per group were quantified by two independent researchers. Researchers were not blinded to the origin of the tissue since treatment group is apparent based on cellular infiltrate.

Analysis of proliferation assays and real-time expression between the various treatment groups were analyzed by a two-tailed, paired Student’s t test. Values of p < 0.05 were considered significant.

CD4⁺CD25⁺ T cells are profoundly suppressive in vitro via contact-mediated mechanism(s) (18, 19). We have previously demonstrated that T^reg from GZB^−/− mice have a significantly diminished ability to suppress T cell proliferation in vitro (27). Use of GZB pharmacological inhibitors has also implicated GZB as an important effector molecule in T^reg-mediated, contact-dependent suppression (28). To further evaluate the role of T^reg GZB-dependent immune suppression, we assessed the impact of overexpression of a GZB inhibitor in the target cells. The Spi-6Tg mouse has been reported to have a 10-to 30-fold increased expression of this GZB-specific inhibitor in all hematopoietic cells based on real-time PCR analysis (38). Moreover, a massive accumulation of Spi-6 protein can be visualized in purified T cells via Western blotting (38). When Spi-6Tg CD4⁺CD25⁻ T cells were purified and cocultured 1:1 with Wt nT^reg, a 50% increase in proliferation was evident compared with Wt T cells (Fig. 1 A). We have previously reported that GZB^−/− T^reg exhibit a 50% loss of suppressive capacity (27). When compared directly, the overt proliferation of Spi-6Tg T^eff in coculture with Wt T^reg is identical to the loss of suppression by GZB^−/− T^reg in coculture with Wt T^eff (data not shown).

One caveat to the use of Spi-6Tg T cells is the possible propensity to be hyperproliferative compared with Wt T cells. In fact, this has been described for PRF^−/− CD8 T^eff upon in vivo challenge with CMV (41, 42). We have therefore normalized proliferation to that of T^eff alone for each group. We found that there was a significant difference in in vitro regulation of these four populations (Spi-6, GZB^−/−, PRF^−/−, and Wt). Although Wt and GZB^−/− T^eff were fully suppressed upon coculture with T^reg, PRF^−/− and Spi-6Tg T^eff were resistant to suppression of proliferation (Fig. 1 and data not shown). As previously reported, PRF^−/− T^reg did not show any defect in their ability to suppress (Fig. 1 B) (27).

Previously, we have shown that coadministration of anti-CD154 and DST induces long-term allograft tolerance (13). To elucidate the impact of GZB on an in vivo model of tolerance, we grafted GZB^−/− mice with CB6F1 full-thickness tail skin on the dorsal surface. Mice were pretolerized to F₁ Ag via i.v. injection of 3 × 10⁷ T-depleted splenocytes (DST) and anti-CD154 (250 μg anti-CD154, given i.p. on days −7, −5, and −3; Fig. 2,A). Mice were then grafted on day 0 and graft survival was monitored. Despite the well-accepted role of GZB as a mediator of graft rejection, there was no observable delay of F₁ skin rejection by the GZB^−/− mouse (Fig. 2 B). However, when GZB^−/− mice were tolerized via DST/anti-CD154, there was a significant decrease in the ability of these mice to maintain their grafts over time (p = 0.016). As had been previously described, the PRF^−/− mice also exhibit this loss of graft tolerance (p < 0.05) (41, 43).

Verification that the loss of tolerance was due to an impairment of GZB function was further demonstrated through the use of Spi-6Tg mice. Spi-6Tg mice or Wt controls were tolerized to F₁ Ag with costimulatory blockade and grafted with an allogeneic skin graft. As compared with tolerized Wt controls, Spi-6Tg mice rejected their skin grafts (p < 0.01) with accelerated kinetics similar to those seen with GZB^−/− mice (Fig. 2 C). Thus, we conclude that GZB is playing a pivotal role in transplantation tolerance.

It is clear that GZB expression is required to maintain graft survival. However, long-term tolerance requires successful induction of clonal anergy, exhaustion, and active suppression by T^reg. Therefore, it is possible that GZB impacts the establishment of tolerance via anergy and clonal exhaustion. To this end, a Tg CD4⁺ T cells system was used which allowed us to trace a population of alloreactive T cells responding to a given alloantigen (I-Eα). Previous publications have shown that alloantigen administered as tolerogenic DST severely blunts the subsequent proliferation and activation status of an allospecific T cell population (13). Administration of DST and anti-CD154 further extinguishes the proliferative response of the alloantigen-specific T cell population upon challenge. However, if T^reg are deleted via administration of anti-CD25(PC61) (44, 45), the suppression instilled by costimulation blockade is abolished sufficiently to permit rapid graft rejection (13).

To evaluate the loss of GZB and PRF on the establishment of allotolerance, we adoptively transferred 1 × 10⁶ CFSE-labeled CD45.1⁺ TEa⁺ T cells i.v. and tolerized the mouse as described above (Fig. 2,A). Five days later, proliferation, phenotype, and function were examined. To establish a role of GZB and PRF in this clonal exhaustion, we conducted a similar set of experiments in GZB^−/− and PRF^−/− mice (Fig. 3). We identified no significant differences in the level of TEa proliferation, activation, or effector function between Wt and GZB^−/− or PRF^−/− host mice. Furthermore, Spi-6Tg TEa exhibited normal clonal exhaustion upon adoptive transfer into tolerized Wt mice (data not shown). Thus, GZB and PRF are not required during the initiation of tolerance to facilitate clonal exhaustion.

T^reg have been shown to infiltrate and accumulate in graft tissue and are capable of directly mediating dominant tolerance (46, 47). Moreover, a loss in the ability to home to skin via the chemokine receptor CCR4 ablates inducible transplantation tolerance (48). To determine whether graft-infiltrating T^reg expressed GZB, grafts were directly examined by immunofluorescence. As previously reported, tolerant grafts are highly infiltrated with FoxP3⁺ cells (Fig. 4, A and B) (47). Syngeneic grafts maintain the 10:1 T^eff:T^reg ratio, as is reported in the lymphoid organs and periphery of mice (Fig. 4,B). For comparison, grafts before rejection were highly infiltrated by CD4⁻ GZB⁺ cells (Fig. 4 C).

Tolerant skin grafts exhibited a 3-fold increase in total CD4⁺ cell-infiltrating grafts due primarily to a 13-fold increase in the number of FoxP3⁺ cells in the graft tissue (Fig. 4,B). A significant increase in the number of GZB⁺FoxP3⁺ cells was seen in conjunction with this increase of FoxP3⁺ cells (Fig. 4 D). These results demonstrate that tolerant skin grafts are highly infiltrated with FoxP3⁺CD4⁺ T cells. Furthermore, GZB expression in FoxP3⁺ T cells correlates with graft tolerance.

The results presented thus far suggest a role for GZB in T^reg-dependent suppression. However, these studies do not causally implicate GZB expression by T^reg in mediating graft survival. To establish a direct link of GZB expression by T^reg to tolerance, we used a mixed BM chimeric system. Sf mice lack the ability to produce T^reg cells due to a point mutation in FoxP3. Using a mixture of BM from Sf and Wt or GZB^−/− or PRF^−/− mice, 3:1, respectively (Fig. 5,A), we reconstituted lethally irradiated CD45.2 host mice and repopulated their hematopoietic compartment with a mixed chimera, except in the T^reg compartment which reconstitutes primarily from the knockout BM (Fig. 5 B). When quantified, the FoxP3⁺ compartment was composed of 90–95% GZB^−/−, PFR^−/−, or Wt BM. Whereas in the FoxP3⁻ compartment, the Sf BM predominated by 60–70% vs the GZB^−/−, PFR^−/−, or Wt BM. Minor contamination of the T^reg compartment (5–10% of FoxP3⁺) was seen from residual host-derived BM. Thus, we greatly biased the loss of GZB in FoxP3⁺ cells (15:1) and created a mixed chimera in all other hematopoietic lineages (1:3).

Mice were allowed to reconstitute for 8–10 wk and then tolerized with our DST/anti-CD154 protocol. Following transplantation, nontolerized mice rejected their grafts with normal kinetics and control Sf and Wt→B6 BM chimeras exhibited a significant delay (p < 0.0001; Fig. 5,B). However, when Sf BM was mixed with GZB^−/− BM (Sf and GZB^−/−→B6), extended graft survival was lost compared with Sf and Wt controls (p < 0.0001; Fig. 5,C). In contrast, Sf and PFR^−/−→B6 BM chimeras displayed no accelerated decay of graft survival, further substantiating the PRF-independent manner in which T^reg function (p = 0.59; Fig. 5 D), as previously reported (27). Interestingly, we also noted that Sf and Wt chimeras rejected skin grafts faster then the alternative Wt and GZB^−/− and Wt and PRF^−/− controls (p < 0.005). An explanation for this hastened rejection in these groups will be presented below.

The Fas:Fas ligand (FasL) pathway is a second pathway that mediates T cell death and is highly influential in regulating cell numbers, particularly through AICD (49). Additionally, Fas-FasL interaction has been demonstrated to play a role in DST-induced transplant biology (50). Furthermore, several groups have implicated FasL expression by T^reg as a mechanism of contact-mediated suppression leading to T^eff cell death (51). To address the impact of FasL expression on T^reg as a mechanism of suppression in transplant tolerance, Sf and GLD^−/−→B6 BM chimeric mice were created. Mice deficient in FasL in their T^reg compartment and tolerized by DST/anti-CD154, exhibited no impairment in allograft survival (p = 0.27; Fig. 5 E) Our results concur with other in vitro reports that FasL expression by T^reg is not a mechanism of suppression (28, 32), and we have now extended these findings to a T^reg-specific in vivo model.

We have previously demonstrated that cotransfer of T^reg with either CD4 or CD8 T cells was able to extend the survival of skin allografts in Rag^−/− recipients (47). To evaluate the role of GZB in T^reg-dependent suppression of CD4 or CD8-mediated rejection, RAG^−/− mice were adoptively transferred alloreactive CD4⁺ TEa Tg T cells or CD8⁺2C⁺ Tg T cells with Wt or GZB^−/− T^reg. Grafts were applied the following day and monitored for rejection. We found using CD4⁺ TEa T^eff, GZB^−/− T^reg were unable to maintain F₁ skin grafts compared with Wt T^reg (p < 0.005; Fig. 5,F). However, when 2C⁺ CD8 cells mediated rejection, there was no statistical difference between GZB^−/− and Wt T^reg extending graft survival (p = 0.7; Fig. 5 G).

Infectious tolerance is a term coined by Waldmann and coworkers (52, 53, 54, 55, 56, 57) and describes a process whereby T^eff are induced to convert to T^reg under certain microenvironmental conditions. The data presented in Fig. 5 show that in mice where the hematopoietic compartment was mixed, with regard to the ability to express FoxP3, long-term graft tolerance was impaired. We hypothesized that the inability of some of the CD4⁺ T^eff to convert to aT^reg because of the Sf genotype may contribute to the hastened rejection and the inability to sustain tolerance. In this case, we would predict that the majority of the T^eff responsible for graft rejection should be derived from the Sf donor. Alternatively, it was also possible that in the mixed BM chimeric mice, overall suppression is impaired and rejection is due to cells derived from both the Sf and Wt BM. To address this hypothesis, the donor origins of the cells responsible for rejection were determined. Following the rejection of skin grafts in the Sf and Wt chimeric mice, LN and spleen were harvested and flow sorted based on CD4/8 and CD45.1/2 expression. Fractionated cells were then recalled to alloantigen. When CD4⁺ cells were recalled to donor alloantigen, most of the allospecific memory response resided within the cells derived from the Sf donor BM (p < 0.0001; Fig. 6). We also identified a slight but significant bias of the CD8 compartment toward dominance by the Sf-deficient derived cells (p < 0.005; Fig. 6). These data suggest that long-term tolerance cannot be sustained if some defined frequency of allogeneic-specific T cells are incapable of FoxP3 expression.

The novel findings of the study presented are: 1) mice deficient in GZB are resistant to tolerization after being treated with a profound tolerance inducing regimen of anti-CD154 and DST; 2) mice which overexpress a specific inhibitor of GZB, Spi-6, are resistant to tolerization after being treated with a profound tolerance-inducing regimen of anti-CD154 and DST; 3) loss of GZB expression within the T^reg compartment greatly impairs the ability of T^reg to sustain an allograft; and 4) failure to undergo infectious tolerance within the CD4 compartment imparts hastened graft rejection.

Although a convincing role for T^reg expression of GZB has been demonstrated in vitro for both humans and mouse (27, 28, 32, 33), only recently has GZB been implicated in vivo as an effector mechanism of suppression (58). A number of lines of evidence presented herein substantiate a role of GZB in vivo in graft tolerance and, furthermore, GZB expression by T^reg as being critical in graft survival. Studies in both GZB^−/− and Spi-6Tg mice demonstrate their resistance to tolerance induction and rapid graft rejection when administered anti-CD154 and DST. The GZB^neo/neo knockout mouse used in these studies was reclassified as a cluster-deficient mouse, harboring defects in the production of granzymes C and F (59). However, the concerns of specificity are assuaged as Spi-6 is a GZB-specific inhibitor, and the Spi-6Tg mouse demonstrates the same phenotype of graft rejection as that seen in the GZB^−/−. Additionally, T cells isolated from Spi-6Tg T^eff are refractory to Wt T^reg suppression as was seen when GZB^−/− T^reg were used to suppress Wt T^eff (27). These data demonstrate GZB and not granzyme C or granzyme F are mediating T^reg suppression in vitro and in vivo.

We demonstrate that mice deficient in GZB had no defect in their ability to reject grafts, but did exhibit significant deficiencies in maintaining tolerance to an allograft. To sequester the loss of GZB to the FoxP3⁺ cellular compartment, we used a BM reconstitution approach. This mixed chimeric mouse contains GZB-deficient FoxP3⁺ while the remainder of host is predominantly GZB sufficient. We demonstrate that T^reg loss of GZB decreases graft survival, whereas T^reg loss of PRF or FasL does not impact establishment of transplant persistence. Thus, the mechanism by which T^reg induce graft survival is GZB dependent and PRF independent.

Previous literature has demonstrated a defect in AICD in the CD8 T cell compartment of PRF^−/− mice (41, 42). However, our results with CD4 TEa T cell adoptive transfer into PRF^−/− or Wt hosts identified no differences in their initial clonal exhaustion. Moreover, in vitro studies showed no significant difference between PRF^−/− and Wt T^eff proliferation. However, PRF^−/− T^eff were partially resistant to Wt T^reg-mediated suppression in vitro and this could account for the early loss of grafts during maintenance of tolerance in vivo, as has been previously described but attributed to insufficient AICD (43). Further research in our laboratory is elucidating the enigmatic importance of perforin in the transplant system.

To refine the necessity of T^reg expression of GZB in graft survival, we explored a three-component reconstitution system in Rag^−/− mice. TEa⁺ Tg CD4 cells were able to cause rejection of CB6F1 skin grafts. As shown in Fig. 5 C, adoptive transfer of polyclonal FoxGFP⁺ T^reg were able to delay this rejection only if they were capable of expressing GZB. Interestingly, the absence of T^reg GZB expression was not seen to significantly impact CD8 2C⁺ rejection. This observation is consistent with the immediate activation-induced up-regulation of Spi-6 in CD8 T cells to protect themselves from autolysis (60), thus rendering activated CD8 resistant to the effects of T^reg GZB-mediated suppression. Other groups have also recognized that the extent of T^reg ability to suppress proliferation and/or function differ between CD4 and CD8 (61, 62, 63). Collectively, these data indicate that the role of GZB-mediated suppression is a CD4-restricted tolerance mechanism.

A recent set of studies by Cao et al. (58) indicate that GZB expression by T^reg impacts tumor clearance. Their ex vivo data indicate that T^reg, GZB, and PRF are required for killing of host CD8 and NK cells. However, these studies do not address T^reg impact on the host CD4 compartment. In a graft tolerance model, our data indicate that GZB-sufficient T^reg are able to inhibit Ag-specific CD4 but not CD8-mediated graft rejection. Moreover, our BM chimeric system indicates loss of PRF expression by T^reg does not impact their ability to mediate graft survival in vivo.

Resistance to infectious tolerance by the T^eff appears to accelerate graft rejection. Our data show that mice in which some of the T^eff are unable to express FoxP3 graft rejection is accelerated, leading to the early decay of Sf and Wt transplants. In our system, those cells that have an inability to express functional FoxP3 dominate the CD4-mediated alloreactive memory pool. In accord with these data, Williams et al. (64) demonstrated that cells meant for the T^reg lineage, but unable to express FoxP3, are the principal mediators of autoimmune disease upon transfer to lymphopenic hosts. Collectively, these data indicate a pivotal role for the aT^reg population in establishing and maintaining tolerance. Interestingly, we also noticed a statistically significant bias of the CD8⁺ memory pool to be derived from the Sf BM. Studies examining the role of FoxP3 expression in CD8 cells have shown this transcription factor to be a marker of regulatory CD8 cells (65, 66, 67, 68). However, loss of FoxP3 expression predominantly impacts the CD4 cells, indicating a greater importance for CD4⁺ aT^reg as compared with CD8⁺FoxP3⁺ cells in mediating transplantation tolerance.

It has been demonstrated that aT^reg have the ability to mediate suppression via cell death mechanisms. A recent report demonstrated that TGF-β-induced aT^reg are able to up-regulate levels of FasL following activation but lack Fas expression, whereas activated T^eff express both Fas and FasL. Cytotoxicity by these aT^reg was maximal at 14 h of coculture and inhibitable with anti-FasL (69). Similar results were identified on CCR4⁺ human T^reg (70). However, in our Sf and GLD^−/− chimera, both nT^reg and aT^reg of the FoxP3⁺ lineage would be unable to mediate FasL killing, yet the chimera maintains allografts to a similar extent as Wt chimeras. Thus, we can conclude that T^reg are of extreme importance in transplantation tolerance and that neither aT^reg nor nT^reg mediate graft survival via Fas-FasL interactions in this system.

Collectively, these results demonstrate that GZB is a pivotal player of T^reg-mediated tolerance, particularly in the context of transplantation. Loss of GZB in the T^reg as well as resistance to GZB in the target relieves contact-mediated suppression in vitro. In vivo, following DST/anti-CD154 costimulatory blockade, GZB^−/− and Spi-6Tg mice have an inability to maintain tolerance to allografts. The loss of regulation is not found during the initiation of tolerance as T^eff transferred into GZB^−/− and PRF^−/− mice exhibit normal clonal exhaustion. Additionally, the lack of any defect in PRF^−/− mice during the initiation of tolerance shows that PRF-dependent AICD of CD4 T cells is not the main mechanism of tolerance, despite previous reports suggesting this (43). The accumulation of GZB⁺FoxP3⁺ cells in the graft itself implicates in situ suppression via GZB as a potential mechanism of tolerance. This was demonstrated using BM chimeric mice lacking GZB expression in T^reg, as well as in Rag^−/− reconstitution systems of graft rejection. Additionally, these data also show that the development of aT^reg is a key mechanism of maintaining tolerance and avoiding chronic rejection. Thus, future therapies for transplant tolerance must be capable of both aT^reg induction and instigation of T^reg effector functions to facility long-term graft tolerance.

The authors have no financial conflict of interest.