Antigen-Dependent Release of IFN-γ by Cytotoxic T Cells Up-Regulates Fas on Target Cells and Facilitates Exocytosis-Independent Specific Target Cell Lysis

CD8⁺ effector T cells exert their biological activity via two quite distinct means upon specific encounter with target cells. One is the synthesis and release of cytokines such as IFN-γ (1) and TNF-α (2), which may act proximal and distal to the effector cell. The other is direct cytolysis and apoptosis of the bound target cell. These two distinct effector mechanisms are not necessarily executed by the same CD8⁺ T cell. Evidence for such is the vastly different numbers of Ag-specific CD8⁺ T cells identified as IFN-γ producers vs estimates of cytolytic T (Tc)² cells using limiting dilution techniques (3), and our recent finding that in response to flavivirus infection, Tc cells with lytic activity are not always producers of IFN-γ (4).

Apoptosis and lysis of target cells by cytolytic lymphocytes (NK and Tc cells) can be executed via at least two distinct pathways: one, the exocytosis pathway mediated by perforin (perf) and granzymes (gzm); the other, by the Fas pathway involving the Fas ligand (FasL) on the effector cell engaging the Fas receptor (Fas or CD95) on the target cell (5, 6, 7, 8, 9, 10, 11, 12). It is generally believed that the exocytosis pathway is primarily involved in the elimination and/or control of intracellular pathogens such as viruses. The Fas pathway of cytotoxicity, in contrast, was thought to be in essence immunoregulatory (11, 13, 14). However, some recent studies (15), including our own on flavivirus-induced cytotoxicity (16), suggest that in certain virus infections both pathways are operative.

We have shown previously (17) that delayed target cell lysis and apoptosis by alloreactive cytolytic T cells from perf-deficient (perf^−/−) mice proceed via the Fas pathway. This conclusion was reached for the following reasons: 1) Fas expression, on originally Fas-negative (Fas^neg) targets, after incubation with Tc cells from perf^−/− mice, was increased in a time-dependent manner; 2) target cells had to be biosynthetically active, i.e., RNA and protein synthesis was required as well as protein transport for lysis to occur; 3) Abs to Fas or soluble Fas-Fc inhibited cytolysis and nucleolysis; 4) poxvirus-encoded serpins, in particular SPI-2, a strong inhibitor of the Fas pathway (18, 19, 20), completely inhibited lysis of target cells by perf^−/− Tc cells; 5) brefeldin A treatment of Tc cells severely reduced lysis, which interferes with polypeptide transport, but only marginally with concanamycin A, an inhibitor of exocytosis (21); 6) target cells derived from the Fas-defective mutant mouse (lpr) were refractory to lysis; and finally 7) Tc effector cells from perf^−/− × gld mice, defective in both cytolytic effector pathways, did not exhibit this lytic phenotype on Fas^neg target cells.

In this study, we present evidence as to the mechanism by which Tc cells induce Fas expression on Fas^neg target cells, investigate its consequence on bystander killing, and speculate on the significance of this process in recovery from viral infections.

C57BL/6 (K^bD^b) (B6), AKR/N (K^kD^k) (AKR), B10.HTG (K^dD^b) (HTG), the perf^−/− mutant (8), the triple knockout mouse, deficient in perf plus gzm A and B (perf×gzmA×B^−/−) (17), and the IFN-γ-deficient mouse (IFN-γ^−/−) (22) were bred under pathogen-free conditions at the Animal Breeding Facility of the John Curtin School of Medical Research or the animal facilities of the Max-Planck-Institute for Immunbiology.

The mouse cell lines L929 (H-2^k), L1210 (H-2^d), and L1210.Fas (transfected with mouse Fas; kindly provided by P. Golstein, Marseilles, France) were grown in Eagle’s MEM (EMEM) supplemented with 10% FCS.

Cells were stained for Fas expression using the FITC-conjugated mAb specific for mouse Fas (Jo-2; PharMingen, Hamburg, Germany). Cells were examined with a FACScan flow cytometer (BD Biosciences, San Jose, CA).

L929 were plated in 24-well Costar (Cambridge, MA) plates and treated with 300 U rIFN-γ (Genzyme, Cambridge, MA; MG-IFN; 200 U/μl) for 4, 6, and 22 h. Cells were trypsinized and aliquots were stained for cell surface Fas expression and Fas transcript quantitation by PCR.

mRNA was isolated from cell lines using TriReagent, following the instruction of the manufacturer (T9424; Sigma-Aldrich, St. Louis, MO). After treatment with 2 μl (2 μg) DNase (Roche Molecular Biochemicals, Mannheim, Germany), mRNA was incubated with 1 μl oligo(dT)_12–18 primer (500 ng; Pharmacia, Freiburg, Germany) and Omniscript RT (4 U; Qiagen, Hilden, Germany). RT-PCR was done as described by Qiagen. Aliquots of 50–100 ng cDNA were used as template for PCR amplification of mouse Fas using the primer pairs, as described (17, 23). PCR products were amplified with 35 cycles, separated by 1.5% agarose gel electrophoresis, and visualized by ethidium bromide staining. Quantitation of PCR products was done using the NIH Image software (version 1.62; freeware from National Institutes of Health download page) by analyzing the density of the scanned bands.

For the generation of alloreactive Tc cells, 8 × 10⁷ responder splenocytes were cocultured with 4 × 10⁷ irradiated (2000 rad) allogeneic stimulator cells for 5–6 days in 40 ml EMEM, 10% FCS, plus 10⁻⁵ M 2-ME.

The methods used for ⁵¹Cr labeling of target cell lines have been described (24). Duration of the assays varied from 4 to 22 h, as indicated in Results. Percentage of specific lysis was calculated by the formula: percentage of specific lysis = ((experimental release − medium release)/(maximum release − medium release)) × 100. Data given are the means of triplicate determinations. SEM values were always <5%.

To assay DNA fragmentation, target cells (2 × 10⁵/ml) were prelabeled with 5 mCi/ml [³H]thymidine (thymidine, methyl[³H]; aqueous solution; 1 mCi/ml; DuPont NEN, Bad Homburg, Germany) in complete EMEM overnight, washed, and used as targets in cytotoxicity assays. Effector cells were mixed with 1–2 × 10⁴ labeled target cells in triplicates at the indicated E:T ratio in 200 μl EMEM supplemented with 2 mg/ml BSA. In some experiments, mAb to Fas (Jo-2) was added to cell cultures before incubation. After indicated time periods, cells were lysed with 25 μl TTE, except maximum release (2% T-X-100/80 mM Tris-HCl, pH 8.0, 5 M EDTA, pH 8.0), for 10 min at 37°C. After centrifugation (1200 rpm; 10 min), 25 μl supernatant was harvested into a solid scintillator plate (LumPlate; Packard, Dreieich, Germany), dried, and counted with TopCount (Packard). For maximum release, 25 μl EMEM was added to the wells, and 25 μl resuspended cell suspension was used. Percentage of specific lysis was calculated by the formula: percentage of specific lysis = ((experimental release − medium release)/(maximum release − medium release)] × 100. Data given are the means of triplicate determinations. SEM values were always <5%.

Anti-IFN-γ mAb (AN18) (25) or control IgG (hamster IgG, rat IgG) was added at the indicated concentrations to cultures containing Tc cells and target cells for the duration of assay.

Experiments were performed in six-well Transwell (Costar catalogue 3412) tissue culture plates. A total of 1 × 10⁶ Fas^neg L929 indicator cells was cultured in the bottom chamber separated by a 0.4-μm-pore-size polycarb membrane. Upper chamber contained 1 × 10⁶ L1210 target cells and effector alloreactive Tc cells, as indicated. After 20-h assay time, L929 cells were trypsinized and Fas expression was analyzed by FACS.

Two possibilities as to the mechanism by which alloreactive Tc cells induce Fas on Fas^neg target cells were considered: first, direct TCR/MHC class I signaling; second, cytokine-mediated up-regulation. Initially, we explored whether signaling by the Tc cell via MHC class I on the target cell was responsible for inducing Fas expression on Fas^neg targets. We failed to mimic Fas up-regulation on Fas^neg L929 (H-2^k) targets by a panel of anti-K^k or anti-D^k Abs. In addition, Fas^neg L929 target cells transfected with MHC class I K^d cDNA, lacking the cytoplasmic domain, still induced Fas after recognition by K^d-reactive Tc cells (data not shown). Thus, it was unlikely that Fas expression was induced by signaling via MHC class I directly. Therefore, we considered the second option, i.e., cytokine-induced up-regulation of Fas on target cells.

Double-chamber experiments were performed to investigate whether soluble mediators were responsible for the up-regulation of Fas on Fas^neg L929 targets. Splenocytes from perf×gzmA×B^−/− mice (H-2^b) deficient in exocytosis-mediated cytotoxicity were cocultured in vitro with B10.HTG (K^dD^b) stimulator cells and tested for their ability to lyse Fas^neg L1210 targets after 6- and 20-h assay time (Fig. 1,A). As shown previously (17), lysis of Fas^neg L1210 targets was negligible at 6 h, but became highly significant at 20 h of incubation. The same effectors and targets were used in the double-chamber experiment (Fig. 1 B), with Fas^neg L929 indicator cells in the bottom chamber. Induction of Fas expression on L929 cells was analyzed by FACS using FITC anti-Fas mAb (Jo-2). Background fluorescence of L929 cells is indicated by the left two bars (0:1, E:T), and refers to L929 cultivated with either no cells (filled) or only L1210 target cells (hatched) in the upper chamber. A substantial increase in Fas expression on L929 cells was observed when the upper chamber contained both effector and target cells (hatched bars), but not effector cells alone (filled bars), with similar levels obtained at E:T ratios of 1:1 and 3:1, and a slight decrease at 9:1 (target cell numbers were held constant). This suggests that Ag was not limiting at conditions used, and at high effector numbers the putative mediator(s) may have also been utilized by the Tc cell population.

IFN-γ and TNF-α are two dominant cytokines known to be released from Tc cells following activation (1, 2). We have argued previously that it is unlikely that TNF-α is involved in Fas up-regulation, as conditions optimal for determining TNF activity (treatment of effector cells with actinomycin D) actually inhibited Fas expression on target cells (17, 26). We thus investigated whether L929 cells, used in our study, can be induced to express Fas upon treatment with rIFN-γ in vitro, as has been shown previously (27). L929 cells were incubated with rIFN-γ for 4, 6, and 22 h, and were analyzed for cell surface Fas expression by FACS, and the presence of Fas-specific transcripts by RT-PCR (Fig. 2, A and B). Cell surface Fas expression was evident at the first (4-h) time point measured, declined slightly by 6 h, but stayed elevated till at least 22 h (Fig. 2,A). Low levels of Fas message were present in mock-treated cells, increased at 4 h after rIFN-γ treatment to double the intensity, and decreased to 1.5 times the intensity as compared with untreated control by 22 h (Fig. 2 B). The low levels of Fas message observed in mock-treated L929 target cells suggest that Fas cell surface expression is also regulated at the translational level and may facilitate a faster response after an appropriate signal.

H-2^k-reactive Tc cells from perf×gzmA×B^−/− mice were tested for their ability to induce ⁵¹Cr and [³H]DNA release on Fas^neg L929 targets in the presence of anti-IFN-γ and control Abs (Fig. 3). T cell-mediated lysis and apoptosis of targets were partially inhibited, in a dose-dependent manner, by anti-IFN-γ mAb, compared with isotype control Abs over a 6-h assay period.

To test whether IFN-γ is the sole mediator responsible for the up-regulation of Fas on Fas^neg targets, we tested whether alloreactive Tc cells from IFN-γ-deficient mice facilitate Fas expression in a double-chamber experiment. Splenocytes from B6, perf^−/−, and IFN-γ^−/− mice were stimulated in in vitro culture with HTG stimulator splenocytes. K^d-reactive effectors were tested for lysis of L1210 and L1210.Fas target cells after 6- and 20-h assay times (Fig. 4,A). Effectors from perf^−/− mice lysed L1210 targets only marginally at 6 h, but significantly at 20-h assay times, as expected. Lysis of L1210.Fas targets was above that of L1210 targets with all effectors. IFN-γ^−/−-derived effectors lysed both target cells slightly more efficiently than B6-derived effectors, an observation made previously (22). The same effector cells used in the cytotoxic assay were used in a double-chamber experiment. Indicator Fas^neg L929 cells were placed in the bottom chamber and assayed for Fas cell surface expression using anti-Fas mAb (Jo2) after a 20-h incubation (Fig. 4 B). The presence of B6 and IFN-γ^−/− effectors, in the absence of Ag, resulted only in a slight increase of fluorescence above the control (dotted line) and was similar for both cell populations. When effectors were cocultured with L1210 target cells, indicator Fas^neg L929 cells had significantly increased fluorescence from wells with B6 and perf^−/− effectors, but not with IFN-γ^−/− effectors. This indicates that IFN-γ is the sole mediator released by Tc cells, which induces the expression of Fas on Fas^neg bystander cells.

The availability of Tc cells lacking in IFN-γ gene and thus unable to induce Fas on Fas^neg L929 cells enabled us to investigate whether up-regulation of Fas on bystander cells leads to their lysis by Tc cells. H-2K^d alloreactive Tc cells from B6, perf^−/−, and IFN-γ^−/− mice were tested for lysis of target cell mixtures of equal proportions of L1210 (H-2^d, the specific target) and L929 (H-2^k, bystanders) with only one of the cell type labeled with ⁵¹Cr. Fig. 5,A shows ⁵¹Cr release from L1210 targets at three time points at an E:T of 0.7:1. Effector cells from B6 and IFN-γ^−/− mice lysed L1210 targets to a similar extent at all time points tested, whereas perf^−/− effectors did not lyse L1210 targets at this low E:T at the early time points and only marginally after 20-h assay time. Lysis of L929 cells was low and insignificant up to the 8-h point. At 20-h assay time, however, significant cross-reactivity by K^d-reactive Tc cells on H-2^k targets can be observed, predominantly with effectors from B6 and IFN-γ^−/− mice (Fig. 5 B). However, most importantly, effectors from B6 mice did not lyse L929 target cells, which by this time express Fas, to a greater extent than those cultured with effectors from IFN-γ^−/− mice. The latter effector population was actually more cross-reactive than B6 effectors. This clearly indicates that Tc cell-mediated up-regulation of Fas on bystander cells does not render them more susceptible to lysis by irrelevant (third party) effectors, even in the presence of functioning FasL/Fas and exocytosis pathways.

The data presented in this study provide clear evidence that IFN-γ is the principal mediator responsible for the Tc effector cell-mediated up-regulation of Fas on target cells. In particular, the findings that treatment of L929^neg targets with rIFN-γ up-regulates Fas and the lack of induction of Fas expression on L929^neg cells, when Tc effector cells from IFN-γ ^−/− mice were used, underpin this conclusion. The discovery of this phenomenon was made possible by the fortuitous use of Tc effector cells deficient in exocytosis-mediated cytotoxicity, in cytotoxic assays over extended assay periods (17), and reveals another possible facet of Tc cell-mediated apoptosis.

The evidence that a secretable mediator of Tc cells, rather than direct cell contact-mediated signaling, is responsible for induced Fas expression, thus able to exert its biological effect distal to the effector cell, may have additional biological implications. We speculated previously (17) that the induction of Fas expression on target cells may constitute a back-up cytolytic mechanism for granule exocytosis-mediated cytotoxicity in the event of a defective perf/gzm machinery (this is mimicked in the perf^−/− mouse) or exhaustion in effector molecules of the exocytosis-dependent killing pathway during an extended Tc cell response. Furthermore, the option of Tc cells to switch from the perf/gzm pathway to the FasL/Fas pathway may facilitate survival from infections with pathogens that have evolved means to evade the granule exocytosis-mediated mechanisms of apoptosis and cell death. One such evasion strategy has recently been described for adenoviruses, in which the adenovirus late assembly protein, L4-100K, inhibits gzmB-mediated cell death (28). This would indicate that one of the cytolytic mechanisms, exocytosis-mediated cell death or Fas-mediated cell death, is redundant. Such complementary cytolytic mechanisms operating in the recovery from primary viral infections or immunopathology would obscure clear-cut phenotypes in granule exocytosis-mediated cytotoxicity-deficient mice. This is actually observed in most studies on viral infections using perf- or gzm-deficient mice (13, 15, 16, 29, 30, 31). The one prominent exception is Ectromelia (Ect), mouse pox (32, 33, 34). In this case, the observed absolute requirement for perf- and gzm-mediated mechanisms for recovery from Ect infection is most likely the consequence of the inhibition of the Fas-mediated pathway by Ect-encoded serpin SPI-2 (32, 35) and has implications for the use of poxviruses in bioterrorism (32).

IFN-γ release by Tc cells will induce Fas on bystander cells proximal and distal to the effector T cell. In the event that these cells express the cognate Ag, i.e., are infected with the relevant parasite, this should facilitate a more efficient control of the infection. The possibility that this may lead to killing of innocent (noninfected) bystander cells has been investigated by using Tc effector cells from IFN-γ^−/− mice. If innocent bystander lysis is occurring due to Fas expression, the following prediction should be fulfilled. Tc cells from B6 mice upon Ag encounter induce Fas on bystanders as a result of IFN-γ release in a time-dependent manner. Thus, at early time points, only Fas-independent (cross-reactive) lysis takes place. With increasing assay times and Fas expression, the rate of bystander killing is predicted to increase and should result in an increase in the magnitude and kinetics of killing when compared with Tc cells from IFN-γ^−/− mice in which no Fas expression takes place. Experiments shown in Fig. 5 fail to provide evidence that IFN-γ-mediated up-regulation of Fas on target cells makes them increasingly susceptible to lysis by Tc cells without Ag specificity. In accordance with this observation are our findings that Fas-positive targets are not lysed by activated Tc cells expressing FasL in the absence of TCR ligation, i.e., expression of the relevant MHC class I/peptide complex or allogeneic MHC class I (17). However, Fas-mediated cell death in the absence of cognate Ag has been reported to be operative in peripheral T cell deletion. The mechanism by which this is controlled is not clear (36), but may be dependent on cell type (37). In addition, bystander killing mediated via the Fas pathway has been reported by other groups under conditions of simultaneous recognition of cognate Ag on one APC and Fas on the bystander cell (38, 39). However, in these experiments, lysis was only seen in cells expressing high Fas levels from the beginning of the cytotoxic assay.

IFN-γ-mediated up-regulation of Fas may also play a role in processes of feedback regulation of an immune response by acting on APCs, including dendritic cells (DC), rendering them more sensitive to FasL/Fas-mediated apoptosis by Tc cells at the early stages of the Tc cell response when cytolytic granule formation has not yet matured. However, reports that DC cells are refractory to Fas-mediated apoptosis (37) and evidence that the perf rather than the Fas pathway is responsible for elimination of Ag-presenting DC (40) suggest otherwise.

Finally, Fas up-regulation by IFN-γ may only concern regulation of Tc cells themselves by self-regulating clonal expansion. The evidence of greatly reduced apoptosis and lymphoproliferative disease in the Fas mutant mouse lpr/lpr (36), an elevated Tc cell response in IFN-γ^−/− mice (Ref. 22 and our own unpublished results), and the essential requirement for IFN-γ in activation-induced cell death of T effector cells in vitro (unpublished results) are in agreement with this hypothesis.