Perforin and Fas Act Together in the Induction of Apoptosis, and Both Are Critical in the Clearance of Lymphocytic Choriomeningitis Virus Infection

Abstract

In this report we questioned the current view that the two principal cytotoxic pathways, the exocytosis and the Fas ligand (FasL)/Fas-mediated pathway, have largely nonoverlapping biological roles. For this purpose we have analyzed the response of mice that lack Fas as well as granzyme A (gzmA) and gzmB (FasxgzmAxB^−/−) to infection with lymphocytic choriomeningitis virus (LCMV). We show that FasxgzmAxB^−/− mice, in contrast to B6, Fas^−/−, and gzmAxB^−/− mice, do not recover from a primary infection with LCMV, in spite of the expression of comparable numbers of LCMV-immune and gamma interferon-producing cytotoxic T lymphocytes (CTL) in all mouse strains tested. Ex vivo-derived FasxgzmAxB^−/− CTL lacked nucleolytic activity and expressed reduced cytolytic activity compared to B6 and Fas^−/− CTL. Furthermore, virus-immune CTL with functional FasL and perforin (gzmAxB^−/−) are more potent in causing target cell apoptosis in vitro than those expressing FasL alone (perfxgzmAxB^−/−). This synergistic effect of perforin on Fas-mediated nucleolysis of target cells is indicated by the fact that, compared to perfxgzmAxB^−/− CTL, gzmAxB^−/− CTL induced (i) an accelerated decrease in mitochondrial transmembrane potential, (ii) increased generation of reactive oxygen species, and (iii) accelerated phosphatidylserine exposure on plasma membranes. We conclude that perforin does not mediate recovery from LCMV by itself but plays a vital role in both gzmA/B and FasL/Fas-mediated CTL activities, including apoptosis and control of viral infections.

Studies with perforin (perf)-deficient mice have shown that perf is a key element in NK cell- and cytotoxic T-lymphocyte (CTL)-dependent recovery from viral infections (20, 28, 63). However, these studies did not address the question of whether NK cells and CTL can mediate their exocytosis-mediated function in vivo solely via perf or whether they depend on additional effector molecules for maximal effector function, in particular components of the exocytosis pathway, such as granzymes (gzm) (58), and/or those of other cytolytic processes, such as Fas ligand (FasL) or tumor necrosis factor (TNF) (15, 21, 58). A signature of all three effector pathways is induction of DNA fragmentation (nucleolysis) in vitro, a marker of apoptosis and target cell death (31, 64).

In vitro studies using either purified perf (18, 38, 65) or mice lacking both gzmA and gzmB (45, 46) have shown that NK cells and CTL can lyse target cells—as monitored by the ⁵¹Cr release assay—solely by means of perf. However, it is still unclear whether this in vitro cytotoxicity is of biological relevance for processes underlying NK cell- and CTL-mediated host defenses in vivo. In vitro induction of programmed cell death (termed apotosis) by the granule exocytosis pathway is strictly dependent on the concerted action of perf and functional active gzm (17, 42, 43, 45, 46). According to recent studies, gzm, in particular gzmA and gzmB, which are released by NK cells and CTL upon encountering antigens, are bound to and internalized by target cells, are subsequently delivered to the cytosol via perf, and initiate cell death by caspase-dependent (gzmB) and/or caspase-independent (gzmA, gzmB) pathways (7, 12-14, 27, 33, 40, 55, 58).

Recent studies with mouse strains deficient in one or more components of the granule exocytosis pathway indicate that the concerted action of perf and gzm is essential for recovery from primary infection with a number of pathogens, such as ectromelia virus (EV) (29), mouse cytomegalovirus (39), herpes simplex virus (36), and Trypanosoma cruzi (30), as well as for the control of certain tumors (34). On the other hand, a number of related studies suggest that gzms and/or the FasL/Fas pathway is dispensable for perf-mediated control of tumor growth (9, 49, 50, 60, 61) and virus infection (6, 20, 63). This conjecture is fostered by the finding that both gzmAxB^−/− and Fas^−/− mice are able to recover from lymphocytic choriomeningitis virus (LCMV) infection with kinetics similar to that of wild-type (wt) C57BL/6 (B6) mice (6). However, the possibility that in gzmAxB^−/− mouse strains the perf-facilitated defense process is operative only in the presence of Fas was not formally excluded.

In order to address this question, we have bred a triple-knockout (triple-ko) mouse strain which is deficient in Fas and both gzms but not perf. The analysis of its lymphoid compartments, the development of NK and CTL effector cells, and its cytotoxic potential, as well as its ability to control primary LCMV infection, is described.

MATERIALS AND METHODS

Mouse strains and genetic analysis.

Inbred B6 and BALB/c mice as well as Fas^−/− (1), gzmAxB^−/− (46), perfxgzmAxB^−/− (47), and FasxgzmAxB^−/− (30, 51) mice (all on the B6 background) were maintained under pathogen-free conditions in the animal facilities of the Max-Planck-Institut für Immunbiologie, Freiburg, Germany. In some experiments B6xgzmA (F₁) mice were used as controls, since we have not observed any differences between B6xgzmA (F₁) and B6 mice. Mice of both sexes between the ages of 7 and 9 weeks were used.

To verify the respective phenotypes, genomic DNA was prepared and analyzed by PCR as described previously (46) and was complemented for Fas-specific PCR (mFas primers, 5′-GCT TGA GTA AAT ACA TCC CG-3′ and 5′-GAT TCG CAG CAC ATC GCC TT-3′; neo-primer, 5′-TGT CCA TCA ATG GAA GGT CT-3′).

Virus.

LCMV (strain WE), kindly provided by O. Utermöhlen (Cologne, Germany), was expanded in L929 cell culture (fibroblasts). The virus-specific epitope gp33 (KAVYNFATC; Neosystem, Strasbourg, France) for H-2D^b was used for target labeling in the cytotoxicity assays.

Disease model.

Mice were infected intraperitoneally (i.p.) with 10⁵ PFU of LCMV WE according to established protocols (6, 22, 24) and were analyzed for virus titers in the liver and spleen, histopathological alterations, the cytolytic and nucleolytic potential of ex vivo-derived CTL, and perforin mRNA and protein expression in splenocytes.

Virus titers.

Aliquots of liver and spleen tissues were homogenized and used for determination of virus titers as described elsewhere (8) with minor modifications: The LCMV nucleoprotein-specific antibody VL-4 (8) was detected with an alkaline phosphatase-conjugated goat anti-rat immunoglobulin (secondary) antibody (Biozol, Eching, Germany). 5-Bromo-4-chloro-3-indolylphosphate (BCIP) (Kirkegaard & Perry Laboratories, Gaithersburg, Md.) was used as the alkaline phosphatase substrate.

Effector cells.

For in vivo generation of LCMV-immune CTL, mice were injected i.p. with 10⁵ PFU of LCMV WE. Effector cells were derived from spleen and lymph nodes (LN) on day 8 postinfection (p.i.).

Enrichment of CD8⁺ cells and in vitro restimulation.

Splenocytes from LCMV-infected mice (day 8 p.i.) were labeled with anti-CD8 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's instructions or with a fluorescein isothiocyanate (FITC)-conjugated anti-CD8 monoclonal antibody (MAb) (clone 53-6.7). CD8⁺ cells were positively selected with an autoMACS (Miltenyi Biotec) or MoFlo (Cytomation, Freiburg, Germany) cell sorter and resuspended in minimal essential medium with 2 mg of bovine serum albumin/ml prior to use in cytotoxic assays. The purity of selected CD8⁺ cells was assessed by fluorescence-activated cell sorter (FACS) staining and found to be between 95 and 98%. Alternatively, splenocytes from LCMV-infected mice (day 8 p.i.) were restimulated in vitro with gp33 (10⁻⁶ M) and ConASN (+αMM, 10%) for 5 h prior to analysis for intracellular gamma interferon (IFN-γ) levels in gp33-specific CD8⁺ CTL.

Target cells.

The mouse cell line MC57G (fibroblasts; H-2^b) as well as MBL.2 and MBL.Fas (lymphoma; H-2^b; kindly provided by M. van den Broek [61]) were cultivated in minimal essential medium with 10% heat-inactivated fetal calf serum (FCS) as previously described (11).

Flow cytometry.

All MAbs were derived from BD Pharmingen, Heidelberg, Germany. Splenocytes or LN cells were incubated with an anti-FcR antibody (clone 2.4G2) prior to the first staining step. The cell surface markers CD3, CD4, CD8, CD19, B220, and Thy1.2 were detected by single-color (CD19), three-color (CD3, CD4, CD8) or four-color (detection of Thy1⁺ B220⁺ CD4⁻ CD8⁻ cells) staining. Splenocytes were incubated with biotin-conjugated Thy1.2 (clone 30-H12), biotin-labeled anti-CD3ɛ (clone 500A2), FITC-labeled anti-CD4 (clone H129.19), phycoerythrin (PE)-conjugated anti-B220 (clone RA3-6B2), PE-labeled anti-CD19 (clone 1D3), or allophycocyanin (APC)-conjugated CD8 (clone 53-6.7). A cell-bound, biotin-conjugated antibody (anti-Thy1.2 or anti-CD3) was stained with streptavidin-conjugated peridinin chlorophyll protein (Becton Dickinson; Heidelberg, Germany). All fluorescence-conjugated MAbs were diluted in phosphate-buffered saline (PBS), 1% FCS, and 0.1% azide to final concentrations of 2-5 μg/ml. Stained cells were fixed in PBS containing 1% paraformaldehyde, examined in a FACSCalibur (Becton Dickinson), and analyzed with CellQuest software (Becton Dickinson). LCMV-immune CTL were detected by double staining with FITC-conjugated anti-CD8 (clone 53-6.7) and gp33-labeled tetramers (H-2D^b; peptide KAVYNFATM, R-PE conjugate; ProImmune, Oxford, United Kingdom) (referred to below as tetramer-PE) (4).

For analysis of intracellular IFN-γ levels, in vitro-restimulated splenocytes from infected mice (day 8 p.i.) were first double stained with APC-conjugated anti-CD8 (clone 53-6.7) and tetramer-PE and then fixed in PBS containing 1% paraformaldehyde for 15 min at 4°C. Subsequently, cells were incubated in permeabilizing buffer (100 μl of PBS, 5% FCS, 0.1% sodium azide, and 0.1% saponin; ROTH, Karslruhe, Germany) for 10 min at room temperature and then stained for 45 min with FITC-conjugated IFN-γ (CALTAG Laboratories, Hamburg, Germany) diluted 1:40 in permeabilizing buffer. Cells were washed twice in permeabilizing buffer and examined by FACS analysis as described above.

Apoptosis of target cells as well as changes in their mitochondrial transmembrane potential (Δψm) and the levels of reactive oxygen species (ROS) were measured by three-color flow cytometry. Accordingly, MBL cells were incubated, in the presence or absence of gp33, with CD8⁺ cells from B6 and mutant mice (4 h; effector-to-target cell [E:T] ratio, 10:1) that had been sorted by magnetic cell sorting (MACS). Subsequently, mixed cell populations were first stained with an anti-CD8 MAb and then, after a wash, either with FITC-conjugated annexin V and propidium iodide (PI) according to the manufacturer's protocol (Pharmingen) or, alternatively, with 2 μM hydroethidine (Molecular Probes, Leiden, The Netherlands) (for detection of ROS) and 2 μM 3,3-dihexyloxacarbocyanine (DiOC₆; Molecular Probes) (for determination of the level of Δψm) for 30 or 15 min, respectively. Data were examined by using a FACSCalibur and analyzed with CellQuest software (both from Becton Dickinson).

Cytotoxicity assays.

All cytotoxicity assays were performed in cell culture medium supplemented with 2 mg of bovine serum albumin/ml instead of FCS (47). MC57G (H-2^b; fibroblasts), MBL.2, and MBL.Fas (H-2^b; lymphoma) cells, target cells for virus-specific killing, were pulsed with 10⁻⁶ M gp33 (1 h) prior to the assay. For the ⁵¹Cr release assay, serial dilutions of spleen or LN effector cells from LCMV-infected mice were incubated with 2 × 10^{4 51}Cr-labeled target cells/well in a final volume of 0.2 ml for the indicated times. Plates were centrifuged, and 25 μl of the supernatant was removed for counting (TopCount; Canberra-Packard, Dreieich, Germany). Percent specific lysis was calculated as (sample − medium control)/(Triton lysate of targets − medium control) × 100. The nucleolytic potential of effector cells was determined as described previously (47). In some experiments, effector cells were preincubated for 30 min with 20 μg of the anti-FasL-antibody MFL-3 (Pharmingen)/ml or with an isotype control (hamster immunoglobulin G [IgG]; Dianova, Hamburg, Germany) prior to their addition to target cells.

Probing for mRNA transcription.

Total RNA was extracted from 2 × 10⁶ LCMV-infected spleen cells by using the Tri Reagent system (Sigma, Taufkirchen, Germany) according to the manufacturer's instructions. mRNA was transcribed by incubating total RNA with an oligo(dT)_12-18 primer (500 ng; Pharmacia, Freiburg, Germany) and Omniscript reverse transcriptase (4 U; QIAGEN, Hilden, Germany) as advised by the manufacturer. The resulting cDNA was used as a template for hypoxanthine-guanine phosphoribosyltransferase (HPRT) and perf amplification in the LightCycler system (Roche Diagnostics, Mannheim, Germany) by using FastStart DNA Master SYBR Green I (Roche). Primers used for amplification and probing of perf-specific cDNA are described in reference 6; primers for HPRT probing were as follows: sense, 5′-GCT GGT GAA AAG GAC CTC C-3′; antisense, 5′-CAC AGG ACT AGA ACA CCT GC-3′. As a perf standard, cDNA from the T-cell line 1.3E6SN (48) was amplified and cloned into a plasmid vector (pGEMT easy; Amersham Biosciences, Freiburg, Germany) by using standard protocols. The plasmid was expressed in, and extracted from, Escherichia coli JM109.

Western blot analysis.

Intracellular perf levels in lysates from MACS-sorted CD8⁺ cells were determined by Western blotting under reducing conditions, as described elsewhere (11). The perf-specific rat anti-mouse MAb CB5.4 was obtained from Alexis Biochemicals (Grünberg, Germany), and the actin-specific goat anti-mouse MAb C-11 was obtained from Santa Cruz Biotechnology (Santa Cruz, Calif.). Staining with the secondary antibodies was performed in a two-step process, by incubating first with an alkaline phosphatase-conjugated swine anti-goat antibody (Caltag, Burlingame, Calif.) and subsequently with an alkaline phosphatase-conjugated goat anti-rat antibody (Southern Biotechnology Associates, Birmingham, Ala.). BCIP (Kirkegaard & Perry Laboratories) (see above) was used as a phosphatase substrate. In vitro-stimulated cells from B6 and perf^−/− mice (anti-BALB/c, fifth stimulation) were used as control lysates, as described elsewhere (46).

Histopathological analysis.

Liver tissue was prepared as described elsewhere (6). Paraffin sections were stained with hematoxylin-eosin (H/E) or by terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) (In Situ Cell Detection kit; Roche Molecular Biochemicals) according to the manufacturer's instructions and were embedded in Entellan (Merck, Darmstadt, Germany).

RESULTS

Life span, lymphoid compartments, and organ pathology of FasxgzmAxB^−/− mice.

A triple-ko mouse strain, which lacks Fas, gzmA, and gzmB (FasxgzmAxB^−/−) (Fig. 1), was bred in order to analyze the function of perf in the absence of both gzms and the FasL/Fas system in NK cell- and CTL-mediated processes in vivo and in vitro. The average life span of FasxgzmAxB^−/− mice was 30 ± 10 weeks, compared to those of Fas^−/− mice (34 ± 12 weeks) and wt B6 mice (>24 months). The sizes of LN and spleens from FasxgzmAxB^−/− mice, including ratios of T-cell (CD4⁺ or CD8⁺) and B-cell subsets, were comparable to those of B6 and Fas^−/− mice up to the ages of 6 to 8 weeks (data not shown). At ages of >10 weeks, FasxgzmAxB^−/− mice, like Fas^−/− mice, suffered from splenomegaly and enlarged LN. Mononuclear infiltrations appeared in lung, heart, and ear tissues of moribund animals (data not shown). The pathological alterations observed in LN and spleens of aging FasxgzmAxB^−/− mice were associated with the accumulation of an unusual population of CD3⁺ B220⁺ cells that lacked CD4 and CD8 (CD4⁻ CD8⁻ B220⁺ Thy1⁺ cells), as shown previously for mice with deficiencies in the FasL/Fas system (Fas^−/−; lpr and gld mutations [1, 2, 32] [data not shown]).

FIG. 1.

Perf alone is not sufficient to eliminate LCMV. B6, B6xgzmA (F₁), Fas^−/−, FasxgzmAxB^−/−, and gzmAxB^−/− mice were infected i.p. with 10⁵ PFU of LCMV WE and were killed on the days indicated. Virus titers in livers and spleens from individual mice were determined as described in Materials and Methods. The combined results from four independent experiments are shown. DL, detection limit; d8, day 8; d15, day 15.

FasxgzmAxB^−/− mice are unable to recover from primary LCMV infection.

Studies with perf^−/− (6, 20, 63) and gzmAxB^−/− (6) mice suggested that perf is essential and sufficient for recovery from primary LCMV infection. In order to verify this assumption, we compared the course of virus infection in 6- to 10-week-old B6, B6xgzmA (F₁), Fas^−/−, gzmAxB^−/−, and FasxgzmAxB^−/− mice upon i.p. infection with 10⁵ PFU of LCMV WE. Virus titers in livers and spleens were estimated on days 8 and 15 p.i. Figure 1 summarizes the results of four independent experiments. On day 8 p.i, at the peak of virus infection in wt B6 mice (22), virus titers were comparable in both organs of infected B6, B6xgzmA (F₁), and Fas^−/− mice but were higher, though not significantly, in livers and spleens from FasxgzmAxB^−/− mice (Fig. 1). At day 15 p.i., virus was undetectable in the livers of all B6 mice (seven of seven), all but one of the B6xgzmA (F₁) mice (four of five), and seven of the nine Fas^−/− mice. In addition, no virus was found in the spleens of any B6 or B6xgzmA (F₁) mice, but virus was detectable in the spleens of four of nine Fas^−/− mice. In contrast, high virus titers in the range of ∼10⁵ (spleen) and ∼10⁶ (liver) were present, with one exception, in livers (seven of eight) or spleens (eight of eight) of FasxgzmAxB^−/− mice. In this respect the triple-ko mouse strain resembles perf^−/− mice (6). As shown previously (6), on day 15 p.i., virus was undetectable in the livers of all (eight of eight) infected gzmAxB^−/− mice.

LCMV infection in B6 mice is associated with virus-immune CTL-mediated liver injury, due to apoptosis of hepatocytes and release of liver enzymes (alanine aminotransferase, aspartate aminotransferase) into blood. This is dependent on the simultaneous action of the Fas/FasL system and perf plus gzmA and gzmB (6). Figure 2 shows liver tissue sections from virus-infected (10⁵ PFU i.p.) B6, Fas^−/−, and FasxgzmAxB^−/− mice taken on day 8 or day 15 p.i. in situ and viewed for apoptosis by using the TUNEL assay or by staining with H/E (Fig. 2; Table 1). At day 8 p.i., apoptotic hepatocytes were seen predominantly in tissue sections of infected B6 mice, and to a lesser extent in Fas^−/− or FasxgzmAxB^−/− mice. At day 15 p.i., no apoptotic cells were seen in any of the mouse strains tested. Inflammatory infiltrates in B6 liver tissue were pronounced on day 8 p.i. and had increased moderately by day 15 p.i. In comparison, inflammatory infiltrates were moderate on days 8 and 15 p.i. in liver tissue from Fas^−/− mice, whereas in FasxgzmAxB ^−/− mice, inflammatory infiltrates in liver tissue were similar to those seen in B6 mice on day 8 p.i. but were pronounced on day 15 p.i.

FIG. 2.

Histopathological analysis of liver tissues taken from LCMV-infected (10⁵ PFU, i.p.) B6, Fas^−/−, and FasxgzmAxB^−/− mice on days 8 and 15 p.i. Liver sections were stained with H/E (upper panels) or by using the TUNEL assay (lower panels) as described in Material and Methods. Arrows indicate apoptotic cells. Magnification, ×360.

TABLE 1.

Histopathological findings in liver tissues of LCMV-infected mice

Mouse strain	Liver infiltrationa at:		No. of apoptotic cells/ 10 HPFb on:
	Day 8 p.i.	Day 15 p.i.	Day 8 p.i.	Day 15 p.i.
B6xgzmA (wt)	++	+	10	1
B6xgzmA (wt)	++	++	9	0
B6xgzmA (wt)	++	++	17	0
Fas−/−	+	+	0	0
Fas−/−	+	+/−	6	0
FasxgzmAxB−/−	++	++	1	0
FasxgzmAxB−/−	++	++	3	0

^aInflammatory infiltrates were assessed on an arbitrary scale as follows: +/−, slight; +, moderate; ++, marked.

^bHPF, high-power fields (magnification, ×400). Similar amounts of apoptotic cells were calculated by using H/E staining and TUNEL staining.

Cytotoxicity of ex vivo-derived LCMV-induced CTL from B6, Fas^−/−, and FasxgzmAxB^−/− mice on Fas-negative target cells.

To further elucidate the putative cellular and/or molecular basis for the impaired potential of FasxgzmAxB^−/− mice to recover from LCMV infection, the numbers of virus-immune CTL present in spleens or LN of B6, Fas^−/−, and FasxgzmAxB^−/− mice at day 8 p.i. were determined, and their perf contents and in vitro nucleolytic and cytolytic activities were analyzed. Cells were stained for surface expression of CD8 and of gp33/D^b-binding T-cell receptors by using gp33-specific PE-conjugated tetramer complexes. Figure 3B shows the numbers of cells staining for both CD8 and tetramers, which were similar for Fas^−/− (2.4%) and FasxgzmAxB^−/− (2.3%) splenocytes but lower for B6 splenocytes (1.5%). In LN populations of infected mice, the percentages of CD8-tetramer double-positive cells were 1.1% for Fas^−/− mice, 0.8% for FasxgzmAxB^−/− mice, and 2.1% for B6 mice (Fig. 3C). FACS-enriched CD8⁺ splenic T cells of the three mouse strains retained the relative percentages of tetramer-positive cells when compared to each other: Fas^−/−, 4.1%, FasxgzmAxB^−/−, 3.5%; B6, 2.7% (Fig. 3D).

FIG. 3.

Cytolytic and nucleolytic activities of LCMV-immune CTL are impaired in FasxgzmAxB^−/− mice compared to those in B6, Fas^−/−, and perfxgzmAxB^−/− mice. (A) Splenocytes from B6 (squares), Fas^−/− (triangles), FasxgzmAxB^−/− (circles), and perfxgzmAxB^−/− (diamonds) mice previously infected i.p. (8 days earlier) with 10⁵ PFU of LCMV WE were incubated with gp33-pulsed (filled symbols) or untreated (open symbols) MC57G target cells for 5 h at the indicated E:T ratios, and cytolysis (⁵¹Cr release) (left panels) and nucleolysis ([³H]thymidine release) (right panels) were determined. Results of two independent experiments are shown. (B) Numbers of gp33-specific CTL in splenocytes from the mouse strains listed in the legend to panel A were determined by FACS following double staining with anti-CD8 and gp33-labeled tetramers (H2-D^b) as described in Materials and Methods. Representative data for 1 mouse/group are shown (the average proportions of CD8-tetramer double-positive CTL in spleen populations of 5 individual mice/group were as follows: for B6 mice, 1.5% ± 0.5%; for Fas^−/− mice, 2.3% ± 0.9%; for FasxgzmAxB^−/− mice, 2.3% ± 0.4%). (C) LCMV-immune LN cells (2 mice/group) from B6, Fas^−/−, FasxgzmAxB^−/−, and perfxgzmAxB^−/− mice were tested for their cytolytic (⁵¹Cr release) (left panel) and nucleolytic ([³H]thymidine release) (right panel) activities on gp33-pulsed (filled symbols) or untreated (open symbols) MC57G target cells (5 h) at the indicated E:T ratios. Percentages of LN cells double staining for CD8 and the gp33 tetramer are given on the right. (D) LCMV-immune CD8⁺ cells, enriched by FACS from splenocytes (2 mice/group) of B6, Fas^−/−, and FasxgzmAxB^−/− mice, were tested for cytolytic activity (⁵¹Cr release) on gp33-pulsed (filled symbols) or untreated (open symbols) MC57G target cells. Percentages of CD8⁺ gp33 tetramer⁺ cells within the three responder populations are given on the right.

Furthermore, as seen in Fig. 4A, no significant differences in copy numbers of perf-specific mRNA transcripts, ranging between 10⁶ and 10⁷/10⁶ cells, were observed in enriched CD8⁺ cell populations from the three mouse strains. In addition, similar copy numbers of HPRT-specific mRNA were found in the three cell populations (Fig. 4B).

FIG. 4.

CD8⁺ splenocytes of LCMV-immune B6, Fas^−/−, FasxgzmAxB^−/−, and gzmAxB^−/− mice contain similar levels of perf- and HPRT-specific transcripts and protein. Total mRNA was isolated from CD8⁺ cells, previously enriched (by MACS) from LCMV-immune spleen cells (10⁵ PFU, i.p; day 8 p.i.), of B6, Fas^−/−, FasxgzmAxB^−/−, and gzmAxB^−/− mice as described in Materials and Methods. Perf- and HPRT-specific transcripts were quantified from the respective cDNA probes by real-time PCR, as described in Materials and Methods. Numbers refer to copies of perf (A)- and HPRT (B)-specific transcripts per 10⁶ cells. (C) Log₂ dilutions (starting with 2 × 10⁶ cell equivalents) of lysates from enriched (MACS of splenocytes) CD8⁺ T cells from LCMV-infected B6, Fas^−/−, and FasxgzmAxB^−/− CD8⁺ mice (10⁵ PFU p.i.; day 8 p.i.) were analyzed by Western blotting for perf and actin, as described in Materials and Methods. Lysates from in vitro-propagated CD8⁺ alloreactive T cells (B6 anti-BALB/c and perf^−/− anti-BALB/c, fifth stimulation) served as positive and negative controls. Lanes 1 to 4, B6 anti-BALB/c; 5, perf^−/− anti-BALB/c; 6 to 9, LCMV-immune CD8⁺ B6 cells; 10 to 13, LCMV-immune Fas^−/− cells; 14 to 17, LCMV-immune FasxgzmAxB^−/− cells.

Levels of perf protein expression in enriched CD8⁺ T cells from infected B6, Fas^−/−, and FasxgzmAxB^−/− mice (day 8 p.i.) were analyzed by Western blot analysis of serial dilutions of cell lysates. As depicted in Fig. 4C, specific bands of similar intensities were seen with all three effector cell populations, suggesting the expression of comparable amounts of perf protein. Thus, no major differences in the numbers of LCMV-immune CTL and in their expression of perf were seen among the mouse strains studied.

To further verify that the inability of FasxgzmAxB^−/− mice to control LCMV infection in the observed period of 15 days p.i. was not due to an incomplete activation state of CTL, including expression of IFN-γ, splenocytes from infected B6, gzmAxB^−/−, and FasxgzmAxB^−/− mice were restimulated in vitro with gp33 (5 h) and analyzed for IFN-γ expression in gp33-specific and CD8⁺ CTL. As shown in Fig. 5, all three CTL populations expressed similar amounts of tetramer-PE staining of CD8⁺ T cells, with comparable intracellular IFN-γ levels.

FIG. 5.

Ex vivo-derived LCMV-immune CD8⁺ spleen cells of B6, gzmAxB^−/−, and FasxgzmAxB^−/− mice express similar amounts of IFN-γ. Splenocytes from B6, gzmAxB^−/−, and FasxgzmAxB^−/− mice previously (8 days earlier) infected i.p. with 10⁵ PFU of LCMV WE were restimulated in vitro with gp33 for 5 h and then analyzed by FACS for numbers of gp33 and CD8⁺ CTL and their intracellular IFN-γ content as described in Materials and Methods. Spleen cells from uninfected B6 mice served as controls.

In contrast, when the capacity of LCMV-induced CTL (10⁵ PFU; day 8 p.i.) from B6, Fas^−/−, and FasxgzmAxB^−/− mice to induce ⁵¹Cr and [³H]thymidine release in gp33-pulsed Fas-negative MC57G fibroblast target cells was determined in vitro, major differences were seen. Both the cytolytic and the nucleolytic potential of spleen (Fig. 3A, two independent experiments) and LN (Fig. 3C) populations from FasxgzmAxB^−/− mice were about 8 times lower than those from Fas^−/− mice and 3 to 5 times lower than those of B6 mice, in spite of comparable numbers of CD8-tetramer double-positive CTL. A reduction in cytolytic and nucleolytic potential similar to that of FasxgzmAxB^−/− mice was seen with spleen cells from gzmAxB^−/− mice (data not shown). The cytotoxic activity of Fas^−/− CTL differed between independent experiments, in that their target cell lysis was at times higher than or similar to, but never lower than, that of B6 CTL (Fig. 3A, B, and C). Although not clear so far, it is possible that the increased cytolytic and nucleolytic potential of Fas^−/− versus B6 CTL, observed in some cell-mediated lympholysis but not others, is due to Fas-mediated activation-induced cell death (AICD)/antigen-induced cell death (AgICD) in the latter but not in the former effector cell population (51).

Similar results were obtained when spleen cell populations were enriched for CD8⁺ T cells by FACS and tested for cytotoxic activity on MC57G (H-2^b) target cells (Fig. 3D). A possible contribution of Fas-mediated processes to cytolysis and/or nucleolysis by B6, Fas^−/−, or FasxgzmAxB^−/− CTL, all of which express FasL (data not shown), was excluded by showing that virus-immune perfxgzmAxB^−/− LN cells were unable to lyse gp33-pulsed MC57G target cells (Fig. 3A and C).

perf enhances FasL/Fas-mediated nucleolytic activity on syngeneic Fas^high target cells in vitro.

The finding that gzmAxB^−/− but not FasxgzmAxB^−/− mice can control LCMV infection (6) (Fig. 1) suggested that perf contributes to FasL/Fas-mediated CTL activities, including induction of apoptosis. Accordingly, splenocytes from LCMV-infected (10⁵ PFU) gzmAxB^−/− and perfxgzmAxB^−/− mice were tested in vitro (day 8 p.i.) for their cytolytic and nucleolytic potential on Fas-transfected MBL cells (H-2^b) previously pulsed with gp33 (61) for various times (2, 4, or 8 h) (Fig. 6). As expected from previous studies (46), ⁵¹Cr release was more pronounced with gzmAxB^−/− (expressing normal levels of perf [6]) than with perfxgzmAxB^−/− effector cells at all time points tested (data not shown). More importantly, gzmAxB^−/− effector cells also expressed higher nucleolytic activity than perfxgzmAxB^−/− effector cells after 2, 4, and 8 h of incubation, as revealed by DNA-fragmentation (Fig. 6A). Pretreatment of gzmAxB^−/− and perfxgzmAxB^−/− effector cells with an anti-FasL MAb drastically reduced or abolished the nucleolytic activities of these cells compared to those of controls (no addition or hamster Ig) (Fig. 6B). This suggests that the CTL-mediated apoptosis observed is elicited mainly via FasL-Fas interaction and that this process is amplified by perf.

FIG. 6.

Fas-mediated nucleolysis of MBL.Fas cells by LCMV-immune CTL is enhanced in the presence of perf. (A) Splenocytes from LCMV-infected (10⁵ PFU, day 8 p.i.) gzmAxB^−/− (diamonds) or perfxgzmAxB^−/− mice (circles) were incubated with gp33-pulsed (filled symbols) or untreated (open symbols) MBL.Fas cells at the indicated E:T ratios and time points, and nucleolysis ([³H]thymidine-release) was determined. (B) gzmAxB^−/− (open bars) and perfxgzmAxB^−/− (filled bars) splenocytes were incubated for 4 h with MBL.Fas target cells (E:T ratio, 100:1) in the presence of either an anti-FasL MAb, the isotype control (hamster IgG [ha Ig]), or no antibody, and nucleolysis ([³H]thymidine release) was determined. Results of two independent experiments are shown. Data are expressed as relative percentages of [³H]thymidine release.

To investigate the mechanism by which perf enhances Fas-mediated cell death, we measured the changes in the differential Δψm and ROS generation in Fas-transfected MBL cells upon incubation with LCMV-immune MACS-sorted CTL from B6, gzmAxB^−/−, or perfxgzmAxB^−/− mice (10⁵ PFU, day 8 p.i.; E:T ratio, 10:1; t = 4 h). In Fig. 7A and B, results of one representative experiment (out of three experiments that gave identical results) are shown. The numbers of gp33-pulsed MBL.Fas cells staining Δψm^low ROS^high were similar after incubation with either B6 or gzmAxB^−/− CTL but significantly reduced after incubation with perfxgzmAxB^−/− CTL (Fig. 7A). This increase in target cell numbers expressing Δψm^low ROS^high correlated with another early apototic feature, i.e., phosphatidylserine (PS) exposure on plasma membranes, in cultures containing B6 and gzmAxB^−/− CTL compared to perfxgzmAxB^−/− CTL (Fig. 7B). Proportions of tetramer-CD8⁺ T cells in LCMV-immune B6, gzmAxB^−/−, and perfxgzmAxB^−/− CTL populations were 8.2, 12.7, and 15.5%, respectively.

FIG. 7.

Suppression of Δψm, ROS generation, and annexin V-PI staining in MBL.2 and MBL.Fas cells treated with LCMV-immune B6, gzmAxB^−/−, and perfxgzmAxB^−/− CTL. CD8⁺ T cells enriched (by MACS) from spleens of LCMV-infected (10⁵ PFU, day 8 p.i.) B6, gzmAxB^−/−, or perfxgzmAxB^−/− mice were incubated (E:T, 10:1) with gp33-pulsed (left panels) or untreated (right panels) MBL.Fas (A, B) or MBL.2 (C, D) cells for 4 h. Cell mixtures were first stained with anti-CD8-APC and then double stained with the membrane potential-sensitive dye DiOC₆ plus hydroethidine (A, C) to assess ROS generation. Alternatively, cells were double stained by annexin V-PI to assess apoptosis as described in Materials and Methods. Only CD8-negative cells (MBL.2, MBL.Fas) were used for analysis. Proportions of CD8-tetramer (gp33-D^b-binding T-cell receptors) double-positive CTL within the enriched cell populations were 8.2, 12.7, and 15.5% for B6, gzmAxB^−/−, and perfxgzmAxB^−/− mice, respectively. x̄, mean fluorescence intensity on x axis.

To support the contention that the increase in the number of Δψm^low ROS^high target cells after incubation with gzmAxB^−/− compared to perfxgzmAxB^−/− CTL was due to perf, Fas-negative MBL.2 cells were incubated with B6, gzmAxB^−/−, or perfxgzmAxB^−/− CTL (Fig. 7C). Mitochondrial perturbation (Fig. 7C) and PS exposure on plasma membranes (Fig. 7D), induced via granular exocytosis (B6 CTL-MBL.2 target cells), were much lower in Fas-negative (Fig. 7C and D) than in Fas-positive (Fig. 7A and B) cells. Incubation of MBL.2 cells with gzmAxB^−/− CTL, which act via perf, led to a small but significant increase in the Δψm^low ROS^high population (Fig. 7C). In addition, only a little PS exposure on plasma membranes was seen under these conditions (Fig. 7D). The fact that only small changes, if any, in the loss of Δψm, ROS production, and PS exposure on plasma membranes, were seen in MBL.2 cells following a 4-h incubation with perfxgzmAxB^−/− CTL indicates that the Fas pathway did not contribute to the death of Fas-negative MBL.2 cells.

DISCUSSION

The current view, that the two principal cytotoxic pathways, the exocytosis and the FasL/Fas-mediated pathway, have largely nonoverlapping biological roles, i.e., host protection from intracellular pathogens versus immune regulation, respectively, can no longer be upheld. This contention is strongly supported by the present study as well as by a number of previous reports on virus disease models (6, 10, 23, 35), in which mice with deficiencies in one or more components of the two cytolytic pathways were used. The important finding reported here is that mice expressing perf alone but neither Fas nor the two gzms, gzmA and gzmB, do not recover from primary LCMV infection, whereas mice which express perf plus gzmA and gzmB (Fas^−/−) or perf plus Fas (gzmAxB^−/−) do recover (6). The possibility that this is due to an incomplete activation state of FasxgzmAxB^−/− CD8⁺ T cells was excluded by showing that they express IFN-γ levels similar to those of B6 mice. Thus, perf seems to be an essential requirement not only for efficient gzm-mediated clearance of virus but also for effective Fas-mediated virus clearance. The finding that only coexpression of either perf plus gzmA and -B or perf plus Fas endows NK cells and CTL with optimal and/or accelerated nucleolytic potential in vitro indicates that apoptosis of infected cells is key to recovery from LCMV infection.

The finding that up to the ages of 6 to 8 weeks, FasxgzmAxB^−/− mice were comparable with Fas^−/− and B6 mice regarding their lymphoid compartments and numbers of mature B and T cells, including ratios of CD4 to CD8 T cells, ensured that a meaningful comparison of NK cell and CTL response to LCMV infection could be undertaken. The similarity of the immune statuses of the three mouse strains at early ages is further supported by identical peaks of virus-immune CTL responses (day 8 p.i.), as revealed by similar numbers of CD8-tetramer-positive T cells (Fig. 4) with comparable levels of intracellular perf and IFN-γ in wt and mutant mice (Fig. 5).

The inability of FasxgzmAxB^−/− mice to recover from LCMV infection mirrors the phenotype observed with perf^−/− mice, which are unable to clear virus even though they express functional gzm and Fas (6, 20). This demonstrates that the critical role of perf in antiviral defense strictly depends on the simultaneous activation of either gzm and/or of Fas. The data with FasxgzmAxB^−/− mice also indicate that the absence of Fas and gzm cannot be compensated for by other mediators such as gzmK (linked to gzmA but expressed in gzmA^−/− mice [44]), or by IFN-γ (which is expressed at normal levels in the infected triple-ko mouse [Fig. 5])—and/or TNF, both of which are known to be critical in the clearance of LCMV (15). However, a role for gzms other than gzmB, but linked to it, in the control of LCMV infection cannot be ruled out, since the respective genes are disrupted in addition to gzmB in gzmB^−/− mice (37).

perf-mediated clearance of LCMV is most probably the result of rapid apoptosis and/or removal of virus-infected targets by phagocytic cells. Accordingly, recovery of Fas^−/− mice from LCMV infection is presumably the result of target cell nucleolysis executed by perf and gzm of virus-immune CTL. Such a process has been shown to occur readily in vitro (17, 19, 42, 46). The Fas pathway, on the other hand, is in itself insufficient to compensate for a lack of perf and gzm function, as perfxgzmAxB^−/− mice do not recover from LCMV, despite their ability to readily induce apoptosis on receptive targets in vitro via the FasL/Fas pathway (6, 47). The finding that gzmAxB^−/− but not FasxgzmAxB^−/− mice are able to clear LCMV implies that perf is also essential for a timely and effective Fas-mediated effector execution. This is also supported by the in vitro observation that target cell death, as revealed by DNA fragmentation (Fig. 3 and 7), is induced more rapidly and at a higher rate by virus-immune gzmAxB^−/− CTL than by perfxgzmAxB^−/− CTL, in particular in the early stages of cytolysis. The enhancement of Fas-mediated apoptosis by perf seems to be due, at least partially, to the ability of perf alone to induce Δψm reduction and ROS production (shown in Fig. 7A and C and reference 25). Further support for this contention derives from the fact that a reduction in Δψm is associated with increased Ca²⁺ signaling, secretion of cytochrome c, and the release of proapoptotic factors, such as apoptosis-inducing factor (AIF), and/or inhibitors of apoptosis (IAP) from mitochondria (3, 27, 52-56). Thus, the data suggest that the simultaneous action of Fas and perf in target cells leads to an amplification of proapoptotic signals that result in enhanced cell death. The role of ROS in the latter process(es) is elusive. Although increased oxidative stress eventually leads to apoptosis and finally necrotic cell death, depending on the cell type and the amount of ROS, its putative role in apoptosis is still controversial (5, 16, 41, 52, 59, 62). In the present study, the positive correlation between loss of mitochondrial transmembrane potential and loss of plasma membrane integrity, as revealed by PS exposure on plasma membranes of target cells (Fig. 7), suggests supporting roles for both processes in cell death.

The previous proposal that perf alone can control noncytopathic LCMV in the absence of both gzm and the Fas pathway, probably by interfering with virus budding (6), is questioned by the present study. Although it is not clear so far how virus replication and/or the transmission of LCMV is blocked by the concerted action of either perf plus gzm or perf plus Fas, and whether the respective effector molecules must be coexpressed by the same NK cell or CTL, apoptotic processes seem to be critical. This does not exclude an additional direct effect of perf on virus replication or spreading. Thus, to some extent, LCMV infections resemble EV infections, in that in the absence of a functional Fas pathway, perf and both gzms are absolutely required for effective control of infection. However, in contrast to EV, which is protected against Fas-mediated attack by inhibition of caspases via virus-encoded serpins (26, 28, 57), LCMV is susceptible to an additional effector pathway, involving perf and Fas.

This is the first report which identifies a synergistic effect of perf and Fas in the induction of target cell death and recovery from a virus infection. This process, which may also be relevant for CTL-mediated control of tumor growth, highlights the great complexities and interrelatedness of functions of diverse effector molecules in the context of a life-threatening disease, such as natural infection and cancer, which would not have become evident in more artificial antigenic systems.