Clonal exhaustion of Ag-specific T cells is a GZB, PRF-independent event. Mice were tolerized to DST via anti-CD154 blockade and coadoptively transferred with 1 × 10⁶ CD45.1⁺ TEa Tg T cells. Five days after transfer, spleen and LN were analyzed for expression of CD44, CD62L, and up-regulation of IFN-γ. Loss of either GZB or PRF expression in the host had no impact on the tolerizing capacity of DST/anti-CD154. Data are representative of four independent experiments.

Spi-6Tg T cells are resistant to T^reg-mediated suppression via GZB. CD4⁺CD25⁺ (T^reg) and CD4⁺CD25⁻ (T^eff) cells were isolated from mice via Miltenyi Biotec magnetic selection. All populations were >90% pure. T^eff cells were labeled with CFSE and cocultured with T^reg at ratios described. GZB-resistant Spi-6Tg T^eff were resilient to suppression (A). However, Wt T^eff were fully suppressed by Wt and PRF^−/− T^reg (A and B). PRF^−/− T^eff also manifested a resistance to Wt T^reg- mediated suppression (B). Data are the average of three independent experiments. Tritium incorporation demonstrates similar trends (data not shown). Groups were normalized to T^eff alone proliferation (*, p < 0.05; **, p < 0.001; and ***, p < 0.005).

Induction of FoxP3 determines regulator/effector status. To determine the effector population mediating rejection, Sf and Wt chimera CD4⁺ and CD8⁺ cells were isolated from mice 20-30 days after graft rejection. Cells were further separated based on the CD45.1/2 expression by flow sorting. Purified T cells were then recalled by CD40-activated B cell blasts from C57BL/6 or CB6F1 mice and examined via ELISPOT for IFN-γ production. The CD4⁺ Sf-deficient derived cells in the mixed chimeras were the main component of the allospecific recall response (p < 0.0001). Moreover, the CD8 compartment also showed a skew toward the Sf compartment in allospecific memory (p < 0.005).

T^reg in tolerant skin express GZB. Sixty days postsurgery, graft tissue was excised and sectioned for histological analysis. For comparison, 7-day-old grafts from nontolerized mice were also examined. Tolerant as well as rejecting grafts showed high levels of cellular infiltration compared with syngenic skin (A). Rejecting grafts were predominantly infiltrated with CD4⁻GZB⁺ cells, whereas tolerant grafts were infiltrated with CD4⁺ cells (A–D). A large proportion of CD4⁺ cells in tolerant grafts were also expressing FoxP3 and a proportion of those FoxP3⁺ cells coexpressed GZB (arrows, A). Enumeration of the histology showed tolerant grafts contained 3-fold more CD4⁺ cells than syngeneic grafts (B). This CD4⁺ cell infiltration was mediated principally by CD4⁺FoxP3⁺ cells, as a 13 times increase in T^reg infiltration was seen, compared with syngeneic grafts (p < 0.005; B). Additionally, of these FoxP3⁺ cells, a 2-fold increase in GZB⁺ cells was observed in situ (p < 0.05; D). *, p < 0.05 and ***, p < 0.005).

GZB^−/− and Spi-6Tg mice are defective in maintaining long-term graft survival via DST/anti-CD154 costimulatory blockade. C57BL/6 mice were given i.v. injections of 3 × 10⁷ T-depleted splenocytes (DST) from CB6F1 mice in conjunction with i.p. injections of 250 μg of anti-CD154 on days −7, −5, and −3 (A). On day 0, mice were grafted with CB6F1 tail skin. Nontolerized mice, deficient in GZB, had no delay in their ability to reject skin grafts compared with Wt mice (p = 0.7; B). However, mice deficient in GZB were unable to maintain allografts to the same extent as their Wt counterparts (p = 0.016; B). PRF^−/− mice also manifested a deficiency in maintenance of allografts (p < 0.05), as has been previously reported (43) (B). Mice overexpressing Spi-6, a GZB-specific inhibitor, manifested decreased graft survival as compared with tolerized Wt mice (p < 0.01). This loss of graft survival in Spi-6Tg mouse is similar to that of the GZB^−/− mouse (B and C).

GZB must be expressed in T^reg to mediate long-term graft survival. To eliminate cytolytic function within FoxP3⁺ T^reg, we preformed mixed BM chimeras. Mice were lethally irradiated via a split dose of 1200 rad, then hematopoietically repopulated by a 3:1 transfer of Sf and Wt or knockout BM (A). After 8–10 wk of reconstitution, CD4⁺ cells (bottom right) maintained the same ratio of chimerism as was identified in the entire host (top left), whereas the CD4⁺FoxP3⁺ cells (top right) derived almost entirely from the non-Sf BM (B); some T^reg emanated from the residual host BM. Following tolerization by DST/anti-CD154 blockade, mice were grafted and maintenance of the grafts was monitored. Tolerized mice reconstituted with Sf and Wt mix were able to established extended tolerance compared with control nontolerized mice (p < 0.0001; C). However, if mice were reconstituted with Sf and GZB^−/− BM, long-term graft survival could not be maintained (p < 0.0001). This loss of tolerization was due to a defect in the T^reg compartment since reconstitution of mice with Wt and GZB^−/− BM did not show this defect in survival, instead showing improved survival compared with Sf and Wt controls (p < 0.001). This loss of graft survival is not due to a perforin-dependent defect in the T^reg themselves, as mice reconstituted with Sf and PRF^−/− BM showed no loss of suppression (p = 0.59; D). To restrict the loss of FasL function to FoxP3⁺ T^reg, we performed similar mixed BM chimeras. Mice reconstituted with Sf and GLD^−/− BM maintained grafts in duration comparable to that of controls (p = 0.27; E). Rag ^−/− mice reconstituted with 99% pure FACS-purified T^eff or T^reg also demonstrated that a loss of T^reg GZB expression was detrimental to graft survival. TEa⁺ CD4-reconstituted Rag^−/− had a significantly accelerated graft decay compared with their Wt T^reg controls (p < 0.005; F). Whereas 2C⁺ CD8-reconstituted Rag mice showed earlier rejection kinetics, but no statistical difference in graft decay compared with Wt T^reg controls (p = 0.7; G).