Lesions of T-Cell-Mediated Kidney Allograft Rejection in Mice Do Not Require Perforin or Granzymes A and B

Authors


*Corresponding author: Dr. Philip F. Halloran, phil.halloran@ualberta.ca

Abstract

Organ allograft rejection is strongly associated with the presence of alloreactive cytotoxic T cells but the role of cytotoxicity in the pathologic lesions is unclear. Previous studies showed that the principal lesions of kidney rejection – interstitial infiltration, tubulitis, and endothelial arteritis – are T-cell-dependent and antibody-independent. We studied the role of cytotoxic granule components perforin and granzymes A and B in the evolution of the T-cell-mediated lesions of mouse kidney transplant rejection. By real-time RT-PCR, allografts rejecting in wild-type hosts at days 5, 7, 21, and 42 showed massively elevated and persistent expression of perforin and granzymes A and B, but evolution of tubulitis and arteritis did not correlate with increasing granzyme or perforin expression. Allografts transplanted into hosts with disrupted genes for perforin or granzymes A and B showed no change in tubulitis, arteritis, or MHC induction. Thus the development of the histologic lesions diagnostic of T-cell-mediated kidney transplant rejection are associated with but not mediated by perforin or granzyme A or B. Together with previous graft survival studies, these results indicate that the granule-associated cytotoxic mechanisms of T cells are not the effectors of T-cell-mediated allograft rejection.

Introduction

Kidney transplant rejection is characterized by the rapid development of an inflammatory infiltrate, first described by Carrel as ‘nephritis’ or inflammation (1), followed by damage to the parenchyma and arteries and loss of function. The pathogenesis of tissue injury is poorly understood but remains relevant, because rejection remains a strong predictor of long-term graft survival (2). The histopathologic lesions defined in the Banff consensus process (3) and the CCTT process (4) are the standard by which human kidney rejection is diagnosed and treated, and correlate with changes in function. The main histologic features in the Banff classification are interstitial infiltrate, invasion of tubules (tubulitis) and invasion of the arterial intima by mononuclear cells (endothelial arteritis). Organ injury correlates with these invasive lesions rather than interstitial inflammation (5–11). The Banff process also recognizes the lesions of antibody-mediated rejection, which is characterized by damage to the microcirculation with deposition of complement component C4d, and in some cases arterial lesions (12).

Mechanistic studies in experimental transplantation are increasingly focused on the development of pathologic lesions (13). We developed a model system in which the evolution of the lesions of kidney rejection can be studied across full MHC plus non-MHC disparities, similar to the usual human transplant (14). The lesions are absent in syngeneic hosts or in allogeneic hosts lacking T cells. By day 5 these kidneys develop edema, high MHC expression, and interstitial infiltrate dominated by CD8 lymphocytes. These features peak by days 7–10, after which the interstitial infiltrate and MHC expression stabilize or regress slightly. However, edema, tubulitis and arteritis progress, and hemorrhage, and patchy necrosis develop by day 21. Studies in T-cell-deficient and B-cell-deficient hosts showed that all early lesions (days 5 and 7) and all tubulitis and much of the arteritis were independent of B cells and immunoglobulin (14). However, at day 21 hosts deficient in B cells/immunoglobulin failed to develop a number of features of microcirculation damage: hemorrhage, peritubular capillary congestion, epithelial necrosis, and some edema (14). Thus we could distinguish three patterns of lesions in mice: early T-cell-mediated interstitial infiltration, edema, and MHC induction; slower but progressive T-cell-mediated tubulitis; and late alloantibody-mediated microcirculation injury (14). Late arteritis was partially dependent on B cells and immunoglobulin, a conclusion similar to the observations in human kidney rejection where antibody-mediated rejection can produce arterial lesions (12,15). Thus the mouse and human studies are compatible with tubulitis, being a prototypic T-cell-mediated lesion.

The critical question is the role of the cytotoxic molecules, particularly perforin and granzymes (16,17), in the development of the T-cell-mediated lesions during kidney rejection. In humans and experimental transplants, rejection is strongly associated with the presence of alloreactive cytotoxic T lymphocytes (CTLs). During rejection, granzyme B and perforin expression are demonstrable in blood (18), urine (19), and biopsies (20–22), and in cells grown from biopsies (19). Perforin is expressed in tubulitis lesions in biopsies (23). However, Clark et al. using allogeneic tumors concluded that 'cell-mediated cytotoxicity results from, but may not be critical for, primary allograft rejection' (24). In perforin-deficient mouse hosts, rejection across full MHC barriers is not impaired despite severe reduction in CTL activity (25,26). Fas/Fas ligand (FasL) interactions also mediate T-cell cytotoxicity, and FasL is up-regulated in rejecting allografts (25,26). However, FasL is not required for murine cardiac allograft rejection (27) and graft survival (28), although it may play a role in the slow rejection of minor histocompatibility-antigen-mismatched skin grafts (29).

In the present study, we compared the time course of CTL molecules expression to the evolution of lesions in our mouse kidney rejection model (14), and studied the evolution of the lesions in kidney allografts in hosts lacking perforin or granzymes A and B. The results indicate that kidney lesions of rejection are associated with perforin and granzyme A and B expression but are not mediated by these activities. Thus the effectors that mediate lesion development in T-cell-mediated kidney rejection must be distinct from the cytotoxic granule mechanisms, but could include other cytotoxic pathways or delayed-type hypersensitivity.

Materials and Methods

Mice

Male CBA/J (CBA), C57Bl/6 (B6), mice were obtained from Jackson Laboratory (Bar Harbor, ME). Perforin-deficient (PerfKO) and granzyme AB-deficient (GzmABKO) mice on a B6 background were obtained from Dr Chris Bleackley (Department of Biochemistry, University of Alberta). The perforin-deficient mice originated from Dr W. R. Clark at the University of California and were bred at the University of Alberta. The granzyme B knockout mice came from Jackson Labs and the granzyme A knockout mice came from Dr T. Ley, Washington University. The A and B knockouts were crossed at the University of Alberta for 3–4 generations until a double knockout resulted. For detection of mutants, DNA was analyzed by PCR (30). Lymphocytes isolated from GzmABKO mice possessed reduced cytotoxic activity, as determined by chromium release assay, and displayed negligible granzyme A or B activity. PerfKO lymphocytes were deficient in killing. Animals were maintained in the Health Sciences Laboratory Animal Services at the University of Alberta. Mouse maintenance and experimentations were in conformity with approved animal care protocols. The following mouse strain combinations were studied across full MHC and non-MHC disparities: CBA (H-2K, I-Ak) into C57Bl/6 (B6; H-2KbDb, I-Ab), CBA into PerfKO/B6 and CBA into GzmABKO/B6.

Renal transplantation

Donor mice 9–11 weeks of age were anesthetized and the abdomen was opened through a midline incision. The right kidney was excised and preserved in cold lactate Ringer's solution. The host mice were similarly anesthetized and the right native kidney was excised. The donor kidney was anastomozed heterotopically to the aorta, inferior vena cava and bladder on the right side, usually without removing the host's left kidney (non life-supporting kidney transplantation). The mice were allowed to recover and were killed at days 5, 7, 21 and 42 post-transplant following anesthesia and cervical dislocation. No transplant hosts received immunosuppressive therapy. Mice with technical complication or pyelonephritis at the time of harvesting were removed from the study. GzmABKO hosts and 20 wild-type hosts received prophylactic antibiotics (Cefazolin, 0.3–0.5 mg per animal) to prevent wound and urinary tract infections. Eighteen wild-type hosts and eight PerfKO hosts did not receive antibiotics. No change in the histological lesions owing to antibiotics treatment was observed.

Histopathology

Tissue sections (2 μm) were stained with periodic acid-Schiff (PAS). The lesions, such as interstitial infiltrate, tubulitis, peritubular capillary congestion (PTC) and graft necrosis, were scored as percent of the total parenchyma involved. Glomerulitis was scored from 0 to 3 based on the percentage of parenchymal involvement, with 0 representing no changes, 1 representing less than 25%, 2 representing 25–75%, and 3 representing greater than 75% of the total parenchyma involved. Arteritis and venulitis lesions were counted in each specimen and the mean number of affected vessels per kidney section was calculated. We also counted the number of arteries with mononuclear cells adherent to the endothelium mononuclear cells.

Antibodies

Hybridoma cell lines were obtained from ATCC (Rockville, MD). Cell lines producing mAb 11–5.2.1.9 (anti I-Ak), 11–4.1 (anti H-2K) and 20–8-4S (H-2KbDb) were maintained in tissue culture in our laboratory. Antibodies for use in the radiolabeled antibody binding assay (see later) were purified from the respective supernatants by protein A affinity chromatography. Antibody concentration was determined by a modified Lowry method, adjusted to 1  mg/mL, and preparations were kept frozen at −70 °C. Anti I-Ab was purchased from Serotec (Cedarlane Laboratories, Hornby, Ontario, Canada). Radio-iodination was performed by the Iodogen method (Pierce Chemical Co., Rockford, IL).

Radioactive antibody-binding assay

This technique has been previously reported, and its quantitative characteristics have been described (31). Briefly, kidney transplant specimens were homogenized, washed in PBS and centrifuged at 3000 r.p.m. for 20 min. The pellets were suspended in PBS at 20 mg/mL. Next, 5 mg of extract was centrifuged and then resuspended in 100 μL of radiolabelled mAb solution (1000 c.p.m./μL) and incubated on ice for 60 min. After that time incubates were diluted with 1 mL of PBS and centrifuged at 3000 r.p.m. for 20 min. The pellets were counted in a gamma counter. The results are expressed as specific c.p.m. bound by the tissue homogenate after subtracting the background (the c.p.m. absorbed by the negative control tissue). The rate of rise in c.p.m. underestimates the degree of change in antigen expression: each twofold change in c.p.m. corresponds to an approximate threefold change in antigen expression (32).

Immunoperoxidase staining

Frozen kidney sections, 4 microns thick, were fixed in acetone and washed twice in PBS. The slides were then blocked in normal goat serum, washed three times in PBS and incubated with anti CD45, CD3, CD4, CD8 or MHC class I and II monoclonal antibodies as well as an IgG isotype control. After washing in PBS, the slides were incubated with goat antirat or goat antimouse peroxidase (ICN, Costa Mesa, CA) and washed three times in PBS. The color reaction was developed by 3′3 diaminobenzidine and the slides were counterstained with hematoxylin.

Real-time RT-PCR

Granzyme A (GzmA), granzyme B (GzmB), perforin, Fas ligand (FasL), IFN-γ, TNF-α, MIP-1α, MIG, I-TAC, MCp-1, Ip-10 and hypoxanthine phosphoribosyltransferase (HpRT) expression were assessed by TaqMan real-time pCR (RT-PCR). Total RNA was extracted from three individual kidney grafts obtained at various times post-transplant. Two micrograms of RNA was transcribed using M-MLV reverse transcriptase and random primers. Quantification of gene expression was performed using the ABI prism 7700 Sequence Detection System (PE Applied Biosystems). TaqMan probe/primer combinations were designed using Primer Express software version 1.5 (PE Applied Biosystems) and are listed in Table 1. Threshold cycle numbers (Ct) were determined and transformed using the ΔCt or ΔΔCt methods as described by the manufacturer, using HPRT as the calibrator. cDNA was amplified in a multiplex system using hypoxanthine phosphoribosyltransferase HPRT as the control gene. The results were shown as fold increase over normal mice.

Table 1.  Sequences of real-time PCR primers and probes
 GenesSequence
Gzm AForward5′-ATCTGTGCTGGCGCTTTGA-3′
 Reverse5′-ACTTAGATCTCTTTCCCACGTTACAGT-3′
 Probe5′-TGAAAAGAACTGGGTGTTGACTGCTGCC-3′
Gzm BForward5′-CGATCAAGGATCAGCAGCCT-3′
 Reverse5′-CTTGCTGGGTCTTCTCCTGTTCT-3′
 Probe5′-TGCTGCTCACTGTGAAGGAAGTATAATAAATGTCACT-3′
PerforinForward5′-GAAGACCTATCAGGACCAGTACAACTT-3′
 Reverse5′-CAAGGTGGAGTGGAGGTTTTTG-3′
 Probe5′-ACCAGGCGAAAACTGTACATGCGACACT-3′
IFN-γForward5′-AGCAACAGCAAGGCGAAAAA – 3′
 Reverse5′-AGCTCATTGAATGCTTGGCG – 3′
 Probe5′-ATTGCCAAGTTTGAGGTCAACAACCCACA-3′
Fas-LForward5′-GAAGGAACTGGCAGAACTCCG-3′
 Reverse5′-CCCTGTTAAATGGGCCACACT-3′
 Probe5′-AAAGCAAATAGCCAACCCCAGCACACC-3′
TNF-αForward5′-TCGAGTGACAAGCCCGTAGC-3′
 Reverse5′-CTCAGCCACTCCAGCTGCTC-3′
 Probe5′-CGTCGTAGCAAACCACCAAGCGGA-3′
IP-10Forward5′-CTATCCTGCCCACGTGTTGAG-3′
 Reverse5′-CTTAGATTCCGGATTCAGACATCTC-3′
 Probe5′-CATTGCCACGATGAAAAAGAATGATGAGC-3′
MIGForward5′-CCTTTTGGGCATCATCTTCCT-3′
 Reverse5′-AGGAGCATCGTGCATTCCTT-3′
 Probe5′-AGCAGTGTGGAGTTCGAGGAACCCTAGTG-3′
MIP-1α Obtained from ABI
MCP-1 Obtained from ABI
I-TACForward5′-CCCTGGCTGCGATCATCT-3′
 Reverse5′-CAAAGACAGCGCCCCTGTT-3′
 Probe5′-CCACAGCTGCTCAAGGCTTCCTTATGTTC-3′
HPRTForward5′-TGACACTGGTAAAACAATGCAAACT-3′
 Reverse5′-AACAAAGTCTGGCCTGTATCCAA-3′
 Probe5′-TTCACCAGCAAGCTTGCAACCTTAACC-3′

Statistical analysis

Data was evaluated using the SPSS 11.5 statistical software package (SPSS Inc., Chicago, IL, USA). Means were compared using the Mann–Whitney U test.

Results

Studies of the evolution of histologic lesions require avoidance of unexpected death of the hosts from renal failure, as post-mortem deterioration of the tissues compromises the quality of the histology and RNA. In our model the contralateral host kidney is left in place to preserve renal function during rejection. The only exclusions were grafts judged to be early surgical complications or bacterial sepsis: of 79  mouse transplants, eight were excluded because of anesthetic deaths or death of unknown causes, and six were excluded because of pyelonephritis (many polymorphonuclear leukocytes, clumps of bacteria), leaving 65 kidney transplants for analysis in the present studies.

Expression of T-cell-effector genes in rejecting kidneys

As in previous studies, the lesions of rejection in CBA/J kidneys transplanted into wild-type (WT) B6 hosts evolved rapidly, emerging at day 5 and persisting through day 42 (14) (also see later). In the WT transplants we compared the expression of genes characteristic of CTL (i.e. granzymes A and B, perforin) with a number of other representative cytokine and chemokine genes. RNA isolated from rejecting kidney allografts in WT hosts at days 5, 7 and 21 and 42 post-transplant was analyzed by real-time RT-PCR (Figure  1). Expression pattern of granzymes A and B, perforin, Fas ligand, TNF-α, and IFN-γ are shown in Figure 1 (A), compared to a number of other chemokines and cytokines in Figure 1(B). Expression of all these mediators was already prominent at day 5. The highest relative increases were in granzyme B and IFN-γ. Granzymes A and B mRNA underwent a significant spontaneous decline by day 21, by approximately 50%. Expression of TNF-α and IFN-γ did not decline, while perforin and Fas-ligand increased through day 42.

Figure 1.

Real-time RT-PCR of selected T-cell-specific and kidney-specific transcripts in rejecting graft. (A) Expression of granzyme A (GzmA), GzmB, IFN-γ, perforin, TNF-α, and Fas/Fas ligand (FasL). (B) Expression of MIP-1α, MIG, I-TAC, MCP-1 and IP-10. Transcripts were evaluated in CBA kidneys grafted into B6 hosts at days 5, 7, 21 and 42 post-transplant. mRNA levels are expressed as fold increase over normal donor CBA kidneys. Each analysis was carried out on RNA isolated from three individual kidneys, and assays were run in duplicates.

We also studied the time course of the expression of a number of chemokine/cytokine genes: MIP-1α, MCP-1, and the IFN-γ inducible chemokines I-TAC, IP-10 and MIG (Figure 1B). All these tended to peak at day 5 and then either stabilized (MIP-1α, MCP-1 and MIG) or declined (I-TAC and IP-10) as rejection progressed.

Histology of rejecting transplants in hosts deficient in perforin or in both granzyme A and granzyme B

We evaluated histologic lesions in a total of 65 kidney allografts (Table 2), after excluding six kidneys for pyelonephritis (four WT/total, one PerfKO/total, and one GzmABKO/total). Compared with control kidneys (142 ± 32 mg, n = 21), rejecting kidney allografts increased markedly in weight by day 7 (245% on average compared with control kidneys), and continued to gain weight through day 21 (300%) (Table 2). The weight gain in kidneys in PerfKO and GmzABKO hosts was similar to that of kidneys in B6 hosts.

Table 2.  Histopathology of wild-type (i.e. CBA) grafts in control (B6) and genetically modified hosts (i.e. PerfKO, GzmABKO)


Host
B6
Day 7
(n = 22)
PerfKO
Day 7
(n = 7)
GzmABKO
Day 7
(n = 5)
B6
Day 21
(n = 16)
PerfKO
Day 21
(n = 9)
GzmABKO
Day 21
(n = 6)
  1. Control kidneys mean weight was 142 ± 31 mg of wet tissue (n = 21). Scoring system of lesions is explained in Materials and Methods. Values are means ± SE. Number of mice used in experiments is shown in parentheses.

  2. aSignificant difference between WT and PerfKO, GzmABKO; (Mann–Whitney, p < 0.05).

Lesions
Interstitial infiltrate percentage 49 ± 2 56 ± 2a 56 ± 2 44 ± 3 56 ± 2a 52 ± 3
Venulitis2.0 ± 0.32.3 ± 0.64.2 ± 0.4a1.1 ± 0.21.2 ± 0.42.3 ± 0.3a
Tubulitis percentage 25 ± 2 24 ± 4 14 ± 2 64 ± 5 73 ± 4 76 ± 2
Arteritis0.1 ± 0.10.8 ± 0.3a  02.1 ± 0.41.2 ± 0.23.8 ± 1.1
Glomerulitis1.2 ± 0.22.1 ± 0.3a2.0 ± 02.4 ± 0.22.3 ± 0.33.0 ± 0
Necrosis percentage  0 ± 01.3 ± 1.3  0 10 ± 4 16 ± 9 13 ± 10
PTC percentage  3 ± 1  8 ± 3  2 ± 2 18 ± 6 23 ± 8 20 ± 7
Kidney weight (mg)309 ± 11353 ± 15334 ± 18366 ± 37458 ± 33a442 ± 43

We compared the histopathology of CBA kidneys transplanted into wild-type B6 mice with those transplanted into PerfKO and GzmABKO hosts (Table 2, Figure 2). Interstitial infiltrate was extensive by day 7 and persisted through day 21, and was mildly increased at day 21 in kidneys in PerfKO and GzmABKO hosts. Venulitis was present in all rejecting kidneys, but tended to peak at day 7 and then decline by day 21. Tubulitis was increased at day 21 (Figure 2D–F) compared with day 7 (Figure 2A–C), and was unaffected by the absence of perforin or granzymes A and B. Transplants rejecting in PerfKO hosts (Figure 2G) or GzmABKO hosts (Figure 2H) showed a similar pattern of arteritis to those in WT hosts, with increased scores at day 21 but no difference between WT and KOs. Glomerulitis was present in all rejecting grafts and was generally similar in KO vs. WT hosts. No lesions were attenuated in the rejecting transplants in the PerfKO or GzmABKO hosts, but some lesions were increased in the GzmABKO hosts (venulitis at days  7 and 21) and in the PerfKO hosts (glomerulitis and arteritis day 7). This effect was not consistent, in that other lesions were similar in rejecting transplants in WT and KO hosts. Necrosis and peritubular capillary congestion were increased at day 21 vs. day 7 in all three types of host, but were similar in the knockouts and WT hosts.

Figure 2.

Histopathology of rejecting CBA kidney mouse allografts. (A) Transplant in a wild-type (WT) host at day 7, showing interstitial infiltrate (PAS, ×400). (B) Transplant in a perforin-deficient (PerfKO) host at day 7, showing interstitial infiltrate (PAS, ×400). (C) Transplant in a granzyme AB-deficient (GzmABKO) host at day 7, showing interstitial infiltrate (PAS, ×400). (D) Transplant in a WT host at day 21, showing severe tubulitis lesion (→) and edema. (PAS, ×600). (E) Transplant in PerfKO hosts at day 21, showing severe tubulitis (→) and edema (PAS, ×600). (F) Transplant in GzmABKO hosts at day 21, showing severe tubulitis (→) and edema (PAS, ×600). (G) Transplant in PerfKO hosts at day 21, showing arteritis with lymphocytes within the intima (→) (PAS, ×600). Transplant in GzmABKO at day 21, showing arteritis with lymphocytes within the intima (→) (PAS, ×600).

The contralateral kidneys were evaluated in all hosts and were within normal limits.

Intragraft MHC expression and cytology of infiltrating cells

We analyzed the induction of donor and host MHC expression at days 7 and 21 semiquantitatively by immunostaining of tissue sections and by radio-labeled antibody binding (31). Donor class I and II expression was strongly induced in both WT and PerfKO hosts at days 7 and 21, localized to the epithelium, particularly its basolateral aspect, as previously reported (33). Host MHC class I and II expression were also increased, localized to the interstitial infiltrate. There was no difference in donor or host MHC class I or II induction between the experimental groups (PerfKO and GzmABKO hosts) and the WT hosts at days 7 and 21.

The cellular composition of the infiltrate was also examined by immunostaining for CD4, CD8, CD3, and CD45, by counting 10 fields using high-power magnification. The number of infiltrating cells was similar between the experimental groups (PerfKO and BrABKO) and the WT controls. Cell counts were higher at day 7, and decreased by 30–40% by day 21.

Discussion

The present studies explored whether the granule-associated cytotoxic activity of effector CTL is necessary for the T-cell-mediated lesions of kidney allograft rejection. We had previously established that all lesions are T-cell-dependent, and that the infiltrate, venulitis, tubulitis and much of the arteritis in rejecting kidney grafts were independent of antibody and B cells (14). The tubulitis is of particular interest as the principal lesion of human T-cell-mediated kidney rejection. Grafts rejecting in wild-type hosts displayed typical patterns and time courses of tissue deterioration, as described earlier (14), associated with increased mRNA for perforin, granzyme A, granzyme B, and with massively elevated MHC expression. The expression of inflammatory cytokines and chemokines, such as IFN-γ, MIG, I-TAC, MCP-1, MIP1-α, IP-10, peaked early, coinciding with the interstitial infiltrate. The granzyme A and B mRNA levels peaked at day 7 post-transplant and then declined, whereas perforin transcripts increased steadily through day 42. We transplanted kidney allografts into GzmABKO or PerfKO hosts, both of which are severely deficient in CTL lytic activity [reviewed in (34). In these hosts, rejection lesions developed normally with no apparent reduction in speed or severity. The infiltrate tended to be greater in PerfKO and GzmABKO but the cytology was not changed. Thus the T-cell-mediated lesions of kidney allograft rejection are associated with but not mediated by T-cell cytotoxic activity, suggesting that the principal effects of T cells on the allograft parenchyma are not cytotoxic.

Target T-cell lysis by CTL effector T cells is highly dependent on granzymes and perforin (34). Granzymes released by CTL disrupt target cells by activating suicide pathways including caspases, leading to PCD (35). Mitochondria play an important role in the initiation of apoptosis and control of substrate cleavage by granzyme B (36). Granzyme uptake is facilitated by mannose 6-phosphate/insulin-like growth factor II receptor (37). Perforin, originally conceived as a membrane 'pore-forming' activity analogous to complement, is critical in releasing granzymes from their intracellular vesicles and thus activating them (35). In the human, CTL express yet another lytic activity, granulysin, which is absent in mice. Granzymes and perforin also mediate lysis of targets by NK cells but to date no evidence indicates a role for that mechanism in kidney rejection lesions.

Although the CTL mechanism has been invoked to explain T-cell-mediated rejection for more than 30 years (38), many studies have cast doubt on this explanation. Perforin deficiency severely reduces target T-cell lysis but does not affect rejection of allogeneic skin grafts (39). Rejection of mouse heart transplants is not affected by perforin, Fas ligand, or alloantibody (25–27), although grafts with minimal incompatibility reject slower in perforin-deficient hosts (25). CTLs mediate corneal allograft rejection (40) but do not require perforin, CD8, or immunoglobulin (41). Neither perforin nor Fas is required for the rejection of islets (42). Fas ligand is also a potential effector but to date the absence of Fas ligand does not abrogate cardiac allograft rejection, even rejection via CD4 T cells (27). Granzyme B has been less well studied: one system of rejection of allogeneic fibroblasts required receptors for granzyme B on the fibroblasts but it is not clear how this system related to conventional rejection mechanisms (37).

The fact that the cellularity of the infiltrate in hosts with disruption of the perforin or GzmAB mechanisms was not greatly different from that of wild-type hosts does not support a major role of the perforin-granzyme mechanisms in T-cell homeostasis, perhaps reflecting the multiplicity of homeostatic pathways. There was a tendency toward increased cellularity in some of the PerfKO and GzmABKO hosts – increased venulitis, glomerulitis, and tubulitis at some times in the PerfKO and GzmABKO hosts – but these effects were small and inconsistent. Nevertheless these trends support the concept of a homeostatic role for perforin or GzmAB. Lymphocyte homeostasis is a major issue in organ transplants. Renal transplants in some mouse strain combinations spontaneously survive despite early rejection (43), as do some human kidney transplants (44). This finding may be related to a fundamental characteristic of clinical transplants: that the host–graft relationship adapts with a reduced probability of rejection over time (45). Homeostasis in effector T cells involves both antigen-driven and antigen-independent programs (46,47), and involves perforin and IFN-γ (48). Granzyme-perforin mechanisms are involved in experimental tolerance induction systems, probably through apoptosis of activated T cells (49), supporting a role for granule mechanisms in lymphocyte homeostasis.

The evolution of typical rejection lesions in GzmABKOs and PerfKOs despite the severe defect in cytotoxic function requires us to invoke other mechanisms for lesion development in T-cell-mediated graft rejection. Effector T cells have activities independent of perforin and granzymes that can deliver a signal to parenchymal and endothelial cells, e.g. the FasL : Fas cytotoxic pathway and the delayed-type hypersensitivity mechanism of activating the effector functions of macrophages and other inflammatory cells. CD103 expression may be involved in this: T-cell integrin CD103 (αEβ7) is characteristic of intraepithelial CD8 T cells, and is inducible by TGF-β and IL-15, both of which are expressed in rejecting grafts (50,51). CD103 may enable CD8 T cells, through the interaction with E-cadherin, to engage kidney epithelium (52). CD103 is needed for rejection of islet allografts (53), apparently because T-cell entry into islet epithelium requires CD103. Previous studies in heart transplants, however, have not shown a role for the Fas pathway in graft rejection and survival (26,28), as it is likely to be more homeostatic than an effector function (27). Thus the molecular mechanism of T-cell-mediated organ graft rejection across MHC disparities remains unknown. The fact that the lesions that define rejection in human kidney transplantation are accompanied by striking expression of perforin and granzymes but not mediated by these activities should help us to zero in on the actual molecular effectors.

Ancillary