CTL-Mediated Killing of Intracellular Mycobacterium tuberculosis Is Independent of Target Cell Nuclear Apoptosis¹

Cytotoxic T lymphocytes have been shown to contribute to host defense against intracellular pathogens, including the global killer Mycobacterium tuberculosis (M. tb.). Two major pathways have been defined for CTL-induced lysis upon Ag recognition (1, 2, 3). One involves cell surface cross-linking of Fas and Fas ligand (FasL)⁵ and the recruitment of the Fas-associated death domain (FADD) which triggers downstream molecular events leading to apoptotic cell death. During granule exocytosis, CTL granule content is released and multimerizing perforin forms pores in the target cell membrane, allowing other granule proteins to gain access into the target cell. Both pathways lead to rapid target cell lysis, but also target cell apoptosis (4, 5, 6). CTL that kill by the granule exocytosis pathway can contribute to direct killing of intracellular microbes by the release of granulysin, a constituent of cytotoxic granules with a broad spectrum of antimicrobial activity (7). It is not clear whether this recently described antibacterial effector pathway requires apoptosis of the target cell.

Upon induction of apoptosis, cysteine proteases known as caspases are proteolytically processed in an autoactivation cascade and serve a key function during the activation phase of apoptosis (8). During Fas-dependent lysis, caspase 8 becomes activated after Fas cross-linking to FADD (9, 10). Several caspases can be activated by the serine protease granzyme B released during granule exocytosis (11, 12, 13). The requirement of caspase activation for host cell apoptotic nuclear damage has been established; however, target cell lysis differentially involves the activation of caspases: CTL that lyse via the granule exocytosis pathway are generally resistant to caspase inhibition. In contrast, Fas-dependent lysis completely depends on caspase activation (14). Apoptotic cell death is characterized by several cellular changes, including loss of membrane asymmetry and mitochondrial potential, membrane blebbing, and rapid and profound nuclear damage resulting in chromatin condensation and nuclear fragmentation (15). This is in contrast to necrosis, which involves irreversible cell membrane damage and subsequent failure to maintain osmotic regulation.

In this study, we investigated whether CTL-induced nuclear apoptosis has a role in the control of M. tb. infection. The results of our experiments indicate the differential involvement of caspase activation in CTL-induced target cell lysis by CD1-restricted CTL. Our data further indicate that the ability of CD8⁺ CTL to kill intracellular M. tb. does not require the CTL to induce nuclear target cell apoptosis.

CTL lines, CD1⁺ APC, and cell lines were cultured in RPMI 1640 (Life Technologies, Grand Island, NY) supplemented with 0.1 mM sodium pyruvate, 100 U/ml penicillin, 2 mM l-glutamine, and 50 μg/ml streptomycin (Life Technologies). The protease inhibitors N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethylketone (ZVAD-FMK) and N-benzyloxycarbonyl-Phe-Ala-fluoromethylketone (ZFA-FMK) were purchased from Calbiochem (San Diego, CA), dissolved as 50 mM stock solution in DMSO, and stored at −70°C. The fluorescent probe Di-IC16 was purchased from Molecular Probes (Eugene, OR) and dissolved as 1 mM stock solution in DMSO. The following mAb were used for blocking experiments: anti-CD1a (OKT6), anti-CD1b (BCD1b), anti-CD1c (10C3), and anti-CD95 (ZB4). Na₂⁵¹CrO₄ was purchased from ICN Pharmaceuticals (Costa Mesa, CA).

An aqueous sonicate of M. tb. strain H37Rv was prepared by sonication of γ-irradiated bacteria in PBS followed by centrifugation at 1200 × g to remove insoluble material and adjusted to 1 mg of protein/ml in PBS.

PBMC were isolated from fresh blood by Ficoll centrifugation. CD1⁺ APC were generated by culturing adherent cells in RPMI 1640/10% FCS in the presence of GM-CSF (200 U/ml) and IL-4 (100 U/ml) and harvesting adherent cells after 72 h as previously described (16).

In addition to CD1⁺ APC, the CD1b-expressing T cell tumor line Jurkat (American Type Culture Collection, Manassas, VA) and the monocytic leukemia line THP-1 (American Type Culture Collection) transfected with an expression vector encoding for CD1b (17) were used as target cells to measure CTL-induced lysis or apoptosis. The CD1b transfectant is designated as THB from here on.

CD1-restricted T cells were derived from the peripheral blood of healthy donors as previously described (18). In brief, to derive CD8⁺ T cell lines, PBMC were depleted of CD4^+, TCRγδ, and CD56⁺, and to establish double-negative (DN) T cell lines, CD8⁺ cells were also depleted by immunomagnetic beads and cultured with M. tb. Ag (10 μg/ml) in the presence of autologous CD1⁺ APC. T cell cultures were maintained with IL-2 and biweekly restimulations with M. tb. Ag in the presence of heterologous CD1⁺ APC.

HLA-A2.1-restricted CD8⁺ CTL which recognize as influenza virus matrix protein were generated from the blood of healthy donors by stimulation of total PBMC with an influenza peptide (kind gift from Alessandro Sette, Epimmune, San Diego, CA). CTL lines were maintained by weekly stimulation with the influenza peptide in the presence of irradiated autologous PBMC. Before experiments, CD8⁺ T cells were enriched by immunomagnetic depletion of CD4⁺ and NK cells.

M. tb. strain Erdman was grown in suspension in 7H9 Middlebrook broth (Difco, Detroit, MI) supplemented with 1% glycerol, 0.05% Tween 80, and 10% Middlebrook oleic acid/albumin/dextrose/catalase enrichment (Becton Dickinson, Mountain View, CA), and aliquots from logarithmically growing cultures were frozen for in vitro experiments in PBS containing 10% glycerol. Upon thawing, bacteria were enumerated for viability by plating on 7H11 Middlebrook agar plates. Excessive clumping of the bacteria was prevented by sonication to disrupt aggregates of bacteria.

CD1⁺ APC were infected with M. tb. as detailed elsewhere (16). In brief, adherent CD1⁺ APC were detached with 1 mM EDTA (Sigma, St. Louis, MO) and plated into a 6-well plate at a density of 3 × 10⁶ cells/well. Monolayers of CD1⁺ APC were pulse infected with M. tb. at a multiplicity of infection of 5:1 for 4 h. Subsequently, extracellular bacteria were removed by extensive washing. Infected APC were detached with 1 mM EDTA, washed, and replated into 96-well flat-bottom microtiter plates at a density of 1 × 10⁴ cells/well. A parallel culture of infected CD1⁺APC was stained with Ziehl-Neelsen to confirm 1) the efficiency of infection and 2) the absence of clumps.

DN or CD8⁺ CTL were coincubated with M. tb.-infected CD1⁺ APC at an E:T ratio of 10:1 for 18 or 48 h. After the indicated incubation period, cells were lysed with saponin (0.3% final concentration) and 5-fold serial dilutions of the cell lysates were plated on 7H11 Middlebrook agar plates and evaluated for CFUs after 21 days of culture.

In some experiments, HLA-A2.1-restricted CD8⁺ CTL specific for an influenza peptide were derived (CTL lines CD8.RTL and CD8.GTL) and coincubated with influenza peptide-pulsed CD1⁺APC derived from heterologous HLA-A2.1-positive donors and infected with M. tb. as described.

The Fas-dependent cytotoxicity pathway was blocked by preincubating target cells in the presence of a blocking anti-CD95 mAb (1 μg/ml) for 30 min. The granule exocytosis pathway was blocked by preincubating effector cells in the presence of 25 mM Sr²⁺ for 12 h, which has been shown to degranulate T cell and NK cell granules (18, 19), but in contrast to EGTA, not interfere with FasL expression on CTL. Target cells were incubated with M. tb. Ag or influenza peptide for 12 h, washed, and labeled with Na₂⁵¹CrO₄ as follows: 1 × 10⁶ cells target cells were incubated with 100 μCi of Na₂⁵¹CrO₄ in 50 μl of RPMI 1640/10% FCS for 1 h at 37°C, washed once, allowed to release ⁵¹Cr from damaged cells for 30 min, and washed three times and used as target cells in a standard 4-h ⁵¹Cr release assay. Percent specific lysis was calculated as follows: 100 × (experimental release − spontaneous release)/(maximum release −spontaneous release).

For inhibition experiments, percent inhibition of cytotoxicity was calculated as follows: [(specific lysis in the absence of inhibitor − specific lysis in the presence of inhibitor)]/specific cytotoxicity in the absence of the inhibitor).

Nuclear fragmentation of target cells was assessed as a hallmark of apoptosis. Target cells were Ag pulsed for 12 h and labeled with Di-IC16 at 1 μM final concentration for 15 min at 37°C. This lipophilic carbocyanine dye is localized exclusively to the plasma membrane when added to a cell suspension and does not transfer to other cells in mixed cultures. Target cells were washed extensively to remove free dye, seeded into 8-well chamber slides, and preincubated with the caspase inhibitor ZVAD-FMK or the control inhibitor ZFA-FMK at 25 μM final concentration. Subsequently, cells were coincubated with effector CTL at an E:T ratio of 5:1 and incubated for 2 h at 37°C. Cells were fixed with 4% paraformaldehyde and stained with Hoechst dye 33258 for 15 min. Finally, the slide was allowed to air dry and nuclear morphology of target cells was viewed under UV light exposure. Target cells were distinguished from effector cells under standard rhodamine excitation. For quantitative analysis, percent nuclear fragmentation was calculated based on examination of at least 100 target cells in multiple vision fields.

Our previous investigation indicates that two phenotypic subsets of human CTL, DN and CD8⁺, lyse targets by distinct mechanisms. Furthermore, only CD8⁺ CTL could mediate an antimicrobial activity against M. tb.-infected target cells (18). To learn more about the mechanisms involved in this novel antimicrobial effector pathway, we derived additional CD1-restricted, M. tb.-reactive CTL. Our goal was to investigate the involvement of caspases in lysis, nuclear apoptosis, and antimicrobial activity. T cell lines were derived from healthy donors as described previously (18) and two lines, one DN and one CD8⁺, were selected for more detailed analysis. Both lines proliferated in response to M. tb.-pulsed CD1⁺ APC, and this activity could be blocked with mAb to CD1b, but not CD1a or CD1c (Fig. 1,A). The DN and CD8⁺ T cell lines also produced similar levels of IFN-γ in an Ag-specific manner (Fig. 1,B). This was representative of a separate panel of DN and CD8⁺ T cell lines in which DN CTL and CD8⁺ CTL produced IFN-γ in response to M. tb.-infected CD1⁺ APC (DN CTL, 1.44 ± 0.13 ng/ml; CD8 CTL, 1.82 ± 0.16 ng/ml). Both the DN and CD8⁺ CTL were cytolytic against M. tb.-pulsed CD1⁺ targets, lysing CD1b-transfected THP-1 cells but not untransfected cells (Fig. 1 C). These data show that DN and CD8⁺ CTL are equally potent in Ag-specific proliferation, cytokine production, and lytic activity.

Previous investigations indicated that the CD8⁺ CTL lyse targets by the granule exocytosis pathway, whereas DN CTL use the Fas-FasL pathway. Both mechanisms of cytolysis require caspase activation for the induction of target cell apoptosis, but caspase activation is required for lysis induced by the Fas-FasL pathway, but not the granule exocytosis pathway (14). Our goal was to examine whether the activation of different pathways of cytotoxicity by CD1-restricted T cells would correlate with the involvement of caspases. Cytotoxicity experiments were performed in the presence or absence of the caspase inhibitor ZVAD-FMK. The peptide-fluoromethylketone inhibitor ZVAD-FMK irreversibly and specifically inhibits different caspases including caspase 1, 3, 4, and 7 but not serine proteases such as granzyme B (20, 21, 22). We found that caspase inhibition substantially prevented the cytolytic activity of DN CTL (Fig. 2). The cytolytic activity of CD8⁺ CTL, however, was independent of caspase activation (Fig. 2).

To extend these results, we analyzed a total of seven CTL lines, three DN CTL and four CD8⁺ CTL. The generation of some of these CTL has been reported previously (18). We found that ZVAD-FMK blocked the cytolytic activity of all three DN CTL by 70–90% (Fig. 3). Inhibition of cytotoxicity in the presence of the control inhibitor ZFA-FMK did not exceed 5%. Cytolysis by these DN CTL lines was almost completely inhibited in the presence of an anti-Fas mAb, but was not affected by pretreatment with Sr²⁺, which has been shown to release cytotoxic granules (19) (Fig. 3). In contrast, the ability of four of four CD8⁺ CTL to lyse targets was minimally affected by treatment with ZVAD-FMK (Fig. 3). Blocking of Fas did not inhibit the cytolytic activity of CD8⁺ CTL lines but treatment with degranulating Sr²⁺ almost completely inhibited their lytic effector function (Fig. 3). These data confirm the existence of two distinct subsets of CTL. In extension to earlier studies, we now show that these subsets also differ in their ability to mediate lysis of target cells by the activation of caspases.

In this study, we investigated whether CTL-induced nuclear apoptosis is also differentially regulated among DN vs CD8⁺ CTL. CTL were coincubated with Ag-pulsed target cells in the presence or absence of ZVAD-FMK or ZFA-FMK for 2 h, and target cells were monitored for nuclear fragmentation, a hallmark of the apoptotic phenotype, by applying the Hoechst staining method and analyzing the nuclear morphology of target cells. We found that both DN and CD8⁺ CTL-induced nuclear fragmentation in >40% of the target cells. The frequency of target cells with nuclear damage, a key feature of apoptosis, was diminished by >90% in the presence of ZVAD-FMK but not the control inhibitor ZFA-FMK (Figs. 4-6). These data suggest that CTL-mediated target cell nuclear apoptosis in short-term assays, whether by the Fas-FasL or granule exocytosis pathway requires caspase activation.

Previous work established an antimicrobial effector function of the CD8⁺ CTL subset that was dependent on the release of the cytotoxic granules (7, 18). Here, we investigated whether CD8⁺ CTL-mediated killing of intracellular M. tb. requires nuclear apoptosis of the infected target cell. We coincubated CD1⁺ APC infected with the virulent M. tb. strain Erdman with CTL lines in the presence or absence of the ZVAD-FMK and measured the number of viable M. tb. according to the number of CFUs. We found that the CD8⁺ CTL lines CD8.TX and CD8.945 had antimicrobial activity, reducing the number of CFUs by 56 and 46%, respectively (Fig. 7). Treatment with ZVAD-FMK did not alter the antimicrobial activity (Fig. 7). In contrast, DN CTL lines DN.780 and DN.ORR, which lack the expression of granulysin (7), yet can induce nuclear apoptosis of the target cells, had no effect on the viability of intracellular M. tb. These data indicate that nuclear apoptosis of the target cells is neither sufficient nor required in order for CD8⁺ CTL to kill intracellular M. tb.

The ability of the immune system to control infection by intracellular pathogens, in particular with M. tb., largely depends on the ability of T cells to interact with infected host cells. CD8⁺ T cells are known to be required for effective immunity (23, 24) and may contribute to host resistance by at least four mechanisms: 1) the release of IFN-γ, 2) lysis of the target cell, 3) induction of apoptosis of the target cells, and 4) by mediating a direct antimicrobial activity. In the present study, we compared the ability of DN and CD8⁺ CTL to contribute to host defense according to these mechanisms. Our data indicate that caspase activation, and hence the induction of nuclear apoptosis, is not required for the antimicrobial activity mediated by CD8⁺ CTL.

One mechanism by which CD8⁺ CTL can contribute to protection is the release of cytokines among which IFN-γ is of paramount importance. In murine models, defects in IFN-γ production or signaling are associated with impaired control of mycobacterial infection (25, 26). IFN-γ serves mainly to activate infected macrophages to induce reactive nitrogen intermediates (27). In human infection, IFN-γ does not activate macrophages to kill intracellular M. tb.; however, an intact IFN-γ signaling pathway seems to be necessary since deficiencies in the IFN-γ receptor genes are associated with increased susceptibility to mycobacterial infection (28, 29). A recent study showed that macrophages infected with M. tb. become resistant to the IFN-γ signaling pathway by inhibiting the interaction of STAT1 with transcriptional coactivators CBP and p300 (30). Here we show that CD1-restricted, M. tb.-reactive CTL are potent producers of IFN-γ. In our in vitro system, both DN and CD8⁺ CTL produced equivalent levels of IFN-γ; however, only CD8⁺ CTL could mediate an antimicrobial activity. Furthermore, previous analysis showed that pretreatment of CD8⁺ CTL with Sr²⁺ released the cytotoxic granules and prevented the antimicrobial activity but did not affect the ability of the CTL to release IFN-γ or TNF-α (18, 31). In addition, perforin-deficient mice were unimpaired in their ability to release IFN-γ upon stimulation with Ag, indicating that the release of IFN-γ is independent of the cytotoxic effector function (32). Our data do not preclude that CD8⁺ CTL released IFN-γ is required for immunity to M. tb. infection, but suggest that CD8⁺ CTL can contribute to host resistance by an IFN-γ independent mechanism.

A second mechanism of CTL effector function is to lyse target cells. CTL-induced target cell lysis can be mediated by two pathways: Fas-FasL interaction and granule exocytosis (1, 2, 3). Upon lysis of infected host cells, bacteria are released and can be taken up at low multiplicity by freshly activated macrophages that then can effectively kill the pathogen (33). The fact that both DN and CD8⁺ CTL induce lysis of the target cell but that only CD8⁺ CTL mediate an antimicrobial activity would suggest that lysis of the target is not a critical factor for host defense (7, 18). This argument is bolstered by studies of mice with defects in perforin or granzymes that indicate that these granule constituents are not required for effective immunity during the initial stages of infection (34, 35). Although an in vitro study suggests a role for the Fas-FasL pathway in antimicrobial activity (36), our data suggest that this pathway does not result in an antimicrobial effect. However, our data confirm that these two phenotypic subsets, DN and CD8⁺, lyse targets by different mechanisms, in that caspase activation was not required for CD8⁺ CTL-mediated lysis. It should be noted that the ability of CD8⁺ CTL to mediate an antimicrobial activity was dependent on granule release since pretreatment with Sr²⁺ blocked the antimicrobial effect. In addition, perforin-deficient mice become susceptible in the later stages of M. tb. infection (34, 35). Together, these data indicate that CTL-mediated lysis is not sufficient for CTL-mediated antimicrobial activity, but may be required.

A third mechanism by which CD8⁺ CTL could contribute to an antimicrobial mechanism is by induction of target cell apoptosis. The ability of M. tb.-specific CTL to induce lysis has been extensively examined (37, 38, 39, 40), but little is known about their ability to induce apoptosis. Here, we provide evidence that M. tb.-specific CTL that lyse targets by either the Fas-FasL pathway or the granule exocytosis pathway induce apoptosis of targets as measured by nuclear fragmentation. We confirm that both the Fas-FasL (DN CTL) and granule exocytosis (CD8⁺ CTL) pathways induce apoptosis of target cells by a caspase-dependent mechanism, since ZVAD-FMK but not ZFA-FMK inhibited apoptosis. The ability of CD8⁺ CTL to mediate an antimicrobial activity was not blocked by caspase inhibition although apoptosis was significantly reduced. These data indicate that induction of nuclear apoptosis is not required for CTL-mediated antimicrobial effector function. Our data do not preclude that the induction of nuclear apoptosis under different conditions can result in an antimicrobial effect. Several stimuli appear to induce apoptosis and also result in killing of intracellular mycobacteria, including ATP (41, 42, 43) or hydrogen superoxide (44). There have been different reports concerning the role of Fas-induced apoptosis on the viability of intracellular mycobacteria (36, 42). High-dose M. tb. infection causes apoptosis as an early event and is associated with pathogen survival rather than killing (45).

A fourth mechanism by which CTL contribute to host defense is their ability to directly kill microbial pathogens. Direct killing has been proposed as CTL effector mechanism based on findings that CTL were able to directly mediate antimicrobial activity against parasites, fungi, and bacteria (46). More recently, the human cytolytic granule protein granulysin present in NK and CD8⁺ CTL granules has been characterized (47), and investigations from our laboratory demonstrated that granulysin has a direct antimicrobial activity against a broad spectrum of pathogens (7). Structure function analysis of granulysin revealed that the antimicrobial activity of this polypeptide was dependent on intact α-helices which are enriched for positively charged amino acid residues which interact with negatively charged bacterial cell membranes (W. A. Ernst, unpublished). Granulysin interacts with M. tb. by disrupting the integrity of the bacterial cell wall and cell membrane (Ernst, unpublished observations). CD8⁺ CTL-mediated killing of intracellular M. tb. was dependent on granule exocytosis (7). However, granulysin itself was not able to kill intracellular M. tb., but was able to do so when coadministered with perforin, the cytolytic molecule present in CTL granules.

We believe that CTL can contribute to host defense against M. tb. principally by two pathways. First, CTL release IFN-γ that can activate macrophages to kill intracellular pathogens. Second, CTL release cytotoxic granules, which by the synergistic action of perforin and granulysin, and independent of target cell nuclear apoptosis can mediate an antimicrobial effector pathway. This dual role of CTL in contributing to host defense against M. tb. infection suggests that monitoring of CTL activity during vaccination and therapy should be a powerful tool in developing strategies to combat the global epidemic of tuberculosis.

We thank Marcus Horwitz for allowing us to use the P3 unit in his laboratory.