Toxoplasma gondii infection confers resistance against Bim_S-induced apoptosis by preventing the activation and mitochondrial targeting of pro-apoptotic Bax

Apoptosis is one of the main defence mechanisms of higher eukaryotes against intracellular pathogens. In mammalian hosts, it can counteract intracellular infection after activation of death receptors of the tumour necrosis factor (TNF) superfamily (e.g. Fas/CD95, TNF-R1) or by the targeted release of perforin and granzymes from granules of cytotoxic T lymphocytes and natural killer (NK) cells (Liebermann, 2003; Schütze et al., 2008). Microbial infection might also provide a stress signal that triggers the intrinsic apoptotic pathway (Williams, 1994; Everett and McFadden, 1999; Labbé and Saleh, 2008). If the infected cell is induced to undergo apoptosis via extrinsic signals or commits `suicide' as an innate response, the further development, multiplication and dissemination of the invader might be disturbed. Furthermore, the subsequent uptake and degradation of the apoptotic cell, including its microbial cargo by phagocytes, might initiate T-cell responses against the pathogen (Savill and Fadok, 2000). In order to circumvent these obstacles, intracellular pathogens have evolved elaborate mechanisms to counteract host-cell apoptosis (Schaumburg et al., 2006).

Toxoplasma gondii is a widespread apicomplexan parasite of warm-blooded vertebrates, infecting up to one third of the world's population (Tenter et al., 2000). Infection of immunocompetent patients is normally asymptomatic or causes mild disease, but can be life-threatening in immunocompromised patients or after infection in utero. T. gondii actively invades a wide range of host cells and subverts major cellular functions in its host (Plattner and Soldati-Favre, 2008), including apoptotic signalling (Lüder et al., 2009). Infection with T. gondii renders host cells resistant to a variety of pro-apoptotic signals (Nash et al., 1998; Goebel et al., 1999; Goebel et al., 2001; Molestina et al., 2003; Carmen et al., 2006; Vutova et al., 2007; Hippe et al., 2008) and might also counteract the innate `suicide' program of parasite-infected cells (Lüder and Gross, 2005).

We and others have shown previously that T. gondii inhibits the release of cytochrome c from the intermembrane space of host-cell mitochondria into the cytosol in response to a pro-apoptotic trigger (Goebel et al., 2001; Carmen et al., 2006). Cytosolic cytochrome c induces activation of a caspase cascade, which leads to the execution of the apoptotic death program (Srinivasula et al., 1998). Toxoplasma-mediated blockade of cytochrome c release therefore has been consistently assumed to be crucial for the avoidance of host-cell death. By contrast, opposing models have been proposed for the underlying mechanisms that could mediate the block in mitochondrial release of cytochrome c in infected cells. These include activation of the phosphoinositide 3 kinase (PI3K) (Kim and Denkers, 2006) and the nuclear factor κB (NF-κB) pathways (Molestina et al., 2003; Payne et al., 2003), the upregulation of anti-apoptotic proteins of the Bcl-2 family (Mcl-1 and A1/Bfl-1) (Goebel et al., 2001; Molestina et al., 2003) and the degradation of proapoptotic Bcl-2 family members, Bax and Bad (Carmen et al., 2006).

A conclusive model for counteracting the mitochondrial apoptotic pathway by T. gondii has been elusive due to the complexity of upstream pathways that fuel the activation of pro- and anti-apoptotic proteins of the Bcl-2 family. This protein family comprises three groups with different structural and functional characteristics (Youle and Strasser, 2008). BH3-only proteins (e.g. Bim, Bid, Bad, PUMA) sense pro-apoptotic upstream signals and their upregulation or post-transcriptional activation precedes permeabilization of the outer mitochondrial membrane (MOMP) (Letai et al., 2002; Willis et al., 2007; Häcker and Weber, 2007). MOMP is a consequence of the oligomerization of the multidomain Bcl-2 executioner proteins Bax and Bak, possibly via pore formation, thereby allowing the release of apoptogenic proteins including cytochrome c into the cytosol (Jürgensmeier et al., 1998; Annis et al., 2005) (reviewed by Lalier et al., 2007). A third group of multidomain anti-apoptotic family members (e.g. Bcl-2, Mcl-1, Bcl-xL) restrains the activity of Bax and Bak, and overexpression of Bcl-2 blocks the mitochondrial apoptotic pathway (Yang et al., 1997; Fletcher et al., 2008).

Here, we provide a detailed analysis of events that regulate cytochrome c release and caspase activation in cells infected with T. gondii. Using mutant host cells that express the BH3-only protein Bim_S under control of a tetracycline repressor-regulated promoter, we could pinpoint the step of parasite interference with the mitochondrial apoptotic pathway to the activation of Bax and its translocation to the outer mitochondrial membrane. We also provide evidence that this Toxoplasma–host-cell interaction counteracts the cell death of parasite-infected cells after activation of the cell-intrinsic pathway of apoptosis.

We employed a tetracycline-inducible expression system of Bim_S (Weber et al., 2007) to determine the level of interference of T. gondii with the mitochondrial apoptotic pathway. Treatment with tetracycline for 3 hours led to chromatin condensation in 78.6±5.6% (mean ± s.e.m., n=3) of non-infected T-REx-HeLa/Bim_S cells (Fig. 1B,C). The condensation of chromatin is characteristic for apoptotic cells and correlated with the induction of Bim_S by tetracycline (Fig. 1A). Infection with T. gondii for 21 hours prior to conditional expression of Bim_S prevented chromatin condensation in ∼55% of the total cell population (Fig. 1B,C) (P=0.001). The level of reduction in Bim_S-induced apoptosis corresponded with infection rates of 51±2% (mean ± s.e.m., n=2; data not shown) after addition of T. gondii at a parasite to host cell ratio of 20:1. Furthermore, immunofluorescence staining of T. gondii parasites in combination with Hoechst staining indicated that parasite-positive (Fig. 1B, arrows) but not parasite-negative cells (Fig. 1B, arrowheads) were largely protected from Bim_S-induced apoptosis. Thus, infection with T. gondii efficiently inhibited host-cell apoptosis downstream of the BH3-only protein Bim_S. Chromatin condensation, as induced in HeLa229 wild-type (HeLa-WT) cells by the kinase inhibitor staurosporine, was inhibited by prior infection with T. gondii to a similar extent as observed in tetracycline-treated T-REx-HeLa/Bim_S cells (supplementary material Fig. S1).

Triggering of the death receptor pathway and the perforin-granzyme pathway of apoptosis can lead to rapid egress of T. gondii and, thereby, necrotic death of the host cell (Persson et al., 2007). In order to determine whether the activation of mitochondrial apoptosis would also induce necrotic death of Toxoplasma-infected cells, we quantified cell death using staining with Annexin V and 7-AAD (7-aminoactinomycin D). Because surface labelling of phosphatidylserine by Annexin V on adherent cells is hampered owing to the isolation procedure, we used Jurkat cells for these experiments. After treatment with staurosporine, the percentage of Jurkat cells that stained positive for Annexin V and negative for AAD (Annexin V⁺/7-AAD^–) (early apoptotic) and to a lesser extent that were Annexin V⁺/7-AAD⁺ (late apoptotic and/or necrotic) increased considerably (Fig. 2A,B). Prior infection with T. gondii dose-dependently decreased the number of Annexin V⁺/7-AAD^– cells to background levels at the highest infection ratio. Importantly, this was not accompanied by a concomitant increase in 7-AAD⁺ cells. We instead observed necrotic death of Jurkat cells as a consequence of T. gondii infection; however, the parasite-mediated increase of Annexin V⁺/7-AAD⁺ cells was lower after pro-apoptotic stimulation than in untreated controls. Consequently, the number of viable (i.e. Annexin V^-/7-AAD^– cells) increased in staurosporine-treated Jurkat cells by prior infection (Fig. 2A,B).

Using mutant T. gondii parasites expressing GFP we also showed that triggering of the mitochondrial apoptotic pathway did not affect the number of parasite-infected Jurkat cells (Fig. 2C,D). Furthermore, parasite-positive (GFP⁺) Jurkat cells showed a lower proportion of Annexin V⁺ cells after staurosporine treatment than parasite-negative cells from the same culture (Fig. 2C). Together, these results established that activation of the mitochondrial apoptotic pathway was counteracted by T. gondii, thus increasing the viability of parasite-infected host cells.

The mitochondrial apoptotic pathway crucially depends on caspase 9, which in turn activates the executioner caspases 3, 6 and 7. Tetracycline-induced expression of Bim_S in T-REx-HeLa/Bim_S cells strongly activated caspase 9 (Fig. 3A) and caspases 3 and 7 (Fig. 3B). Prior infection of the cells with increasing numbers of T. gondii dose-dependently inhibited Bim_S-induced caspase 9 (P<0.05) and caspases 3 and 7 activities (P<0.01). Upon tetracycline-induced expression of Bim_S, cleavage of inactive full-length caspase 9 and caspase 3 into active p35 or p17 fragments, respectively, was also dose-dependently reduced by T. gondii (Fig. 3C). Likewise, the activity of caspases 3 and 7 that had been induced in HeLa-WT or in Jurkat T cells by treatment with staurosporine was inhibited by prior infection with T. gondii (data not shown). Together, these results established that T. gondii interferes with the mitochondrial apoptotic pathway that was triggered by conditional expression of Bim_S or by staurosporine.

Inhibition of Bim_S-triggered apoptosis after infection with T. gondii can be the result of direct interference with cytochrome-c-dependent caspase activity (Keller et al., 2006) or alterations at the level of the Bcl-2 proteins (Goebel et al., 2001; Molestina et al., 2003; Carmen et al., 2006). In order to distinguish between these possibilities, subcellular fractions from T. gondii-infected T-REx-HeLa/Bim_S cells or non-infected controls were analysed by cytochrome c immunoblot. The results showed the redistribution of cytochrome c from mitochondria into the cytosol of non-infected cells after conditional expression of Bim_S (Fig. 4A). By contrast, cytochrome c was not detected in the cytosolic fraction of tetracycline-treated cells that had been previously infected with the parasite. Control staining with an antibody recognizing cytochrome c oxidase (COX) excluded a contamination of cytosolic fractions with mitochondria (Fig. 4A). In addition, immunolabelling of actin indicated a similar protein content of the cytosolic and organellar fractions, respectively (Fig. 4A). The redistribution of cytochrome c from mitochondria to the cytosol was also strongly decreased after parasitic infection of HeLa-WT cells or Jurkat cells that had been treated with staurosporine (supplementary material Fig. S2; and data not shown). This suggested that the step of parasite interference with the mitochondrial pathway was the same after conditional Bim_S expression as after staurosporine treatment. We also confirmed that infection of T-REx-HeLa/Bim_S cells with T. gondii did not affect the conditional expression of Bim_S (Fig. 4D). Together, these data indicate that T. gondii blocks the mitochondrial apoptotic pathway at a step between Bim_S expression and cytochrome c release.

Because the activity of effector caspases might amplify upstream regulatory events of the mitochondrial apoptotic pathway (Lakhani et al., 2006), the decrease of Bim_S-induced cytosolic cytochrome c following parasite infection might also be mediated by differential activities of downstream caspases in infected and non-infected cells. To unambiguously exclude this possibility, the subcellular redistribution of cytochrome c was evaluated in the presence of pan-caspase inhibition. The results showed that infection of T-REx-HeLa/Bim_S cells with T. gondii also inhibited the redistribution of cytochrome c into the cytosol in the presence of 20 μM of the caspase inhibitor z-VAD-fmk (Fig. 4C). z-VAD-fmk at that concentration completely abrogated the DEVD-AMC cleavage activity of effector caspases 3 and 7 in response to staurosporine (Fig. 4B). We noticed in these experiments that there was some mitochondrial cytochrome c release in non-infected cells after z-VAD-fmk treatment. Such redistribution, as well as that observed after tetracycline-induced Bim_S expression, was nevertheless abrogated by T. gondii. This confirmed that T. gondii interferes with the mitochondrial apoptotic pathway upstream of MOMP, even if downstream caspases are inhibited.

Intrinsic proapoptotic signals (including the expression of Bim_S) converge on the activation of the effectors of the Bcl-2 family, i.e. Bax and/or Bak. Therefore, we evaluated the activation of Bax and Bak in infected cells and non-infected controls using antibodies that specifically recognize an epitope within the N-terminus of the proteins (Bax-NT and Bak-NT, respectively). This epitope has been shown to be exposed during activation of Bax and Bak (Hsu and Youle, 1997; Lalier et al., 2007). Flow cytometry revealed activation of Bax in ∼50% of non-infected T-REx-HeLa/Bim_S cells after expression of Bim_S whereas this was significantly diminished in parasite-infected cells (Fig. 5A,B) (P<0.01). Likewise, Bak was also activated by Bim_S but to a lesser extent than Bax (Fig. 5A,B). Such activation was again lower in parasite-infected cells than in non-infected controls although this difference was not statistically significant due to the overall low level of Bak activation in T-REx-HeLa/Bim_S cells. The effect of T. gondii on Bak was further evaluated in Jurkat cells, which lack Bax and transmit cell-intrinsic proapoptotic signals exclusively via activation of Bak (Brimmell et al., 1998). After treatment of Jurkat cells with staurosporine, the percentage of Bak-NT-positive cells increased to ∼80%, but was significantly lower in cells infected with T. gondii (Fig. 5C) (P<0.01). Importantly, immunoblot analyses of complete cellular extracts from T-REx-HeLa/Bim_S cells indicated that the amounts of total Bax and Bak were not altered following parasitic infection (Fig. 5D). This ruled out the possibility that a decrease in the total cellular levels of pro-apoptotic multidomain proteins accounted for the lower levels of active Bax or Bak, as observed following T. gondii infection.

Parasite-mediated inhibition of Bax activation was morphologically confirmed by confocal laser scanning microscopy. Conditional expression of Bim_S led to a strong labelling of the majority of non-infected T-REx-HeLa/Bim_S cells using a Bax-NT-specific antibody (Fig. 5E). The punctate staining pattern was consistent with a mitochondrial localization of active Bax (Wolter et al., 1997; Gross et al., 1998) (see also below). By contrast, prior infection with T. gondii largely abolished the tetracycline-induced Bax-NT reactivity. Parasite-positive T-REx-HeLa/Bim_S cells were particularly protected from Bax activation (Fig. 5E, arrows) whereas parasite-negative cells of infected cell populations were strongly reactive (Fig. 5E, arrowhead). This suggested that intracellular parasites are able to block Bim_S-induced Bax activation effectively.

Inactive Bax is mainly located in the cytosol and – during activation – translocates to mitochondria where it oligomerizes and leads to MOMP (Wolter et al., 1997; Zhou and Chang, 2008). We followed the cellular redistribution of a yellow fluorescent protein (YFP)-Bax fusion construct in T-REx-HeLa/Bim_S cells. In the absence of tetracycline, YFP-Bax was diffusely distributed in the cytosol and the nuclei of non-infected cells and this did not change after T. gondii-infection (Fig. 6A). After conditional expression of Bim_S, clusters of YFP-Bax became apparent in non-infected T-REx-HeLa/Bim_S cells and correlated with condensed chromatin, as visualized by propidium iodide. The punctate distribution of YFP-Bax strongly indicated translocation to the mitochondria as shown previously (Zhou and Chang, 2008). Importantly, prior infection with T. gondii prevented Bim_S-induced redistribution of YFP-Bax in parasite-positive cells, whereas YFP-Bax redistributed in parasite-negative cells of an infected culture in the same way as in non-infected control cultures (Fig. 6A). Essentially the same results were obtained after treatment of T. gondii-infected and non-infected HeLa-WT cells with staurosporine (supplementary material Fig. S3A). In order to exclude any impact of the YFP reporter, we also determined the effect of parasite infection on endogenous Bax in HeLa-WT cells treated or not with staurosporine. Endogenous Bax was evenly located in the cytosol of untreated control cells (supplementary material Fig. S3B). Remarkably, it appeared more punctate than YFP-Bax, indicating that in healthy cells Bax might weakly associate with intracellular membrane-bound structures including mitochondria as shown previously (Antonsson et al., 2001). However, a pro-apoptotic stimulus induced the redistribution of endogenous Bax into large clusters that were unevenly distributed within non-infected cells, whereas this was prominently inhibited in Toxoplasma-infected cells (supplementary material Fig. S3B). Quantification of cells with a punctate Bax staining pattern confirmed that parasitic infection significantly inhibited the subcellular redistribution of endogenous Bax following treatment with staurosporine (supplementary material Fig. S3C) (P=0.009).

To verify that T. gondii interferes with activation of Bax, we also determined its dimerization-oligomerization, which is a prerequisite for MOMP (Annis et al., 2005; Antonsson et al., 2001). Conditional expression of Bim_S led to the formation of Bax dimers and trimers in non-infected control cells, as revealed after cross-linking with EGS (Fig. 6B). Minor amounts of Bax dimers in cells not treated to express Bim_S suggested spontaneous oligomerization of Bax to a limited extent or some leakiness of the tetracycline-regulated promoter. Importantly, infection with T. gondii almost completely abolished the formation of dimers and trimers (Fig. 6B). Control western blots of proteins not cross-linked with EGS confirmed that T. gondii does not alter the total cellular content of Bax, as already indicated (Fig. 6B; Fig. 5D).

Upregulation of anti-apoptotic Bcl-2-like proteins following infection with T. gondii could keep Bax or Bak in an inactive state. To test this hypothesis the release of cytochrome c into the cytosol was evaluated in the presence of actinomycin D. Because the transcription inhibitor would also prevent the upregulation of Bim_S by tetracycline, we used in these experiments HeLa-WT cells that were treated with staurosporine. The results showed that T. gondii inhibited the translocation of cytochrome c in the presence of actinomycin D as efficiently as in its absence (Fig. 7A). RT-PCR analyses of interleukin 2 (IL-2) transcripts from Jurkat cells that had been activated by phytohemagglutinin (PHA) and phorbol 12-myristate 13-acetate (PMA) confirmed a complete blockade of transcription by actinomycin D (Fig. 7B). As described previously (Goebel et al., 2001), treatment with actinomycin D alone also led to the accumulation of cytochrome c in the cytosol of non-infected cells, whereas this was prevented in HeLa cells previously infected with T. gondii (Fig. 7A). The results indicated that blockade of mitochondrial cytochrome c release after parasite infection generally does not depend on the transcription machinery of the host cell. This is further supported by the finding that infection of T-REx-HeLa/Bim_S cells with T. gondii at increasing multiplicities-of-infection (MOIs) did not upregulate the total cellular levels of Bcl-2, Mcl-1, Bcl-x and A1/Bfl-1 as compared to non-infected controls (Fig. 7C).

In order to further exclude an impact of anti-apoptotic Bcl-2 family proteins during T. gondii-mediated inhibition of apoptosis, we employed the BH3 mimetic ABT-737. ABT-737 is a cell-permeable small compound that binds to the hydrophobic Bcl-2 homology 3 (BH3)-binding groove of Bcl-2, Bcl-X_L and Bcl-w, thereby abrogating their anti-apoptotic activity and priming cells for death (Oltersdorf et al., 2005). Remarkably, treatment of T-REx-HeLa/Bim_S cells with 0.1 or 1 μM ABT-737 dose-dependently activated effector caspases 3 and 7 in non-infected cells whereas infection with T. gondii completely abrogated such activity (Fig. 7D). The enantiomer (i.e. the stereoisomer) exhibited little activity even at the highest concentration used (Fig. 7D, open bars). ABT-737 at 1 μM also enhanced caspase activity that had been induced in non-infected cells by tetracycline-induced Bim_S expression. By contrast, it did not sensitize T. gondii-infected cells for increased Bim_S-induced activation of caspases 3 and 7 (Fig. 7D), indicating that the parasite inhibited the mitochondrial apoptotic pathway independently of anti-apoptotic Bcl-2, Bcl-X_L or Bcl-w. Treatment of Toxoplasma-infected and non-infected HeLa-WT cells or Jurkat cells with ABT-737 alone or together with staurosporine yielded similar results (data not shown).

We have shown recently that triggering of apoptosis by Bim_S requires its mitochondrial localization (Weber et al., 2007). To further elucidate the interference of T. gondii with Bim_S-triggered apoptosis we determined the subcellular distribution of Bim_S, Bax and cytochrome c. Following treatment of T-REx-HeLa/Bim_S cells with tetracycline, Bim_S levels increased within 2 hours and were further upregulated until 4 hours of treatment (Fig. 8). Importantly, it exclusively localized to the mitochondria-containing fraction and this was not altered in Toxoplasma-infected cells as compared to uninfected controls. By contrast, Bax redistributed from the cytosol to the mitochondria in non-infected but not T. gondii-infected cells (Fig. 8), thus confirming the analyses done by immunofluorescence microscopy (see above). This clearly showed that T. gondii infection largely prevented Bax translocation to the mitochondria, whereas mitochondrial Bim_S targeting was not affected by the parasite. Colocalization of Bim_S and Bax in the heavy-organelle fraction of non-infected cells coincided with an increasing release of cytochrome c into the cytosol (Fig. 8). By contrast, MOMP and cytochrome c release was largely prevented in T. gondii-infected cells.

We demonstrate here that Toxoplasma inhibits the mitochondrial apoptotic pathway via interference with mitochondrial translocation and/or activation of the effector molecules Bax and Bak. This is achieved without altering the levels of pro- or anti-apoptotic regulators of the host, thus suggesting direct parasite interference with Bax and Bak. Remarkably, the mitochondrial localization of the BH3-only protein Bim_S did not suffice to recruit Bax to the outer mitochondrial membrane nor to induce its activation in parasitized host cells. Finally, triggering of the intrinsic pathway of apoptosis does not induce cell death of parasite-infected cells, thus increasing the viability of host cells.

We pinpointed the level of manipulation of the mitochondrial apoptotic pathway by T. gondii by using the conditional expression of Bim_S. This BH3-only protein is the shortest out of three main Bim splice variants and is normally expressed at only low levels (O'Connor et al., 1998). The latter might be due to the fact that Bim_S – in contrast to its longer counterparts – is considered to be constitutively active and not subject to any post-translational regulation (Marani et al., 2002; Weber et al., 2007). By employing the enforced expression of Bim_S, we could thus largely exclude any impact of T. gondii on the various mechanisms of post-translational regulation of BH3-only proteins, i.e. phosphorylation, proteolytic cleavage or sequestration (see Youle and Strasser, 2008). Furthermore, we also excluded a parasite impact on the various upstream pathways that sense cellular stress signals and integrate them into a core signal at the level of the BH3-only proteins. Remarkably, T. gondii was nevertheless capable of efficiently inhibiting Bim_S-induced apoptosis, indicating parasite interference downstream of Bim_S expression. Because we extended the previous findings of a parasite-mediated blockade of cytochrome c release following pro-apoptotic signalling (Goebel et al., 2001; Carmen et al., 2006) to Bim_S-induced apoptosis we could narrow down the level of parasite interference to a step between Bim_S expression and MOMP. This notion was further substantiated using pan-caspase inhibition. It excluded the possibility that differential activities of effector caspases in the presence or absence of T. gondii (Keller et al., 2006) were responsible for the parasite-mediated blockade of pro-apoptotic signalling upstream of MOMP via a feed-back regulatory pathway.

It has to be stressed that the mechanisms we describe do not seem to be restricted to Bim_S-induced apoptosis because similar results were observed after infection of wild-type HeLa and/or Jurkat cells that had been treated with staurosporine. We therefore propose that it is a general mechanism of T. gondii to inhibit both the mitochondrial apoptotic pathway induced by various cell-intrinsic stress signals (Nash et al., 1998; Goebel et al., 1999; Goebel et al., 2001; Kim and Denkers, 2006) and also the mitochondrial amplification loop following death receptor ligation (Hippe et al., 2008). Importantly, quantification of cell-death phenotypes of parasite-infected Jurkat cells established that activation of the intrinsic pathway of apoptosis did not induce necrotic death of parasite-infected host cells. Following ligation of death receptors or targeted release of the content of cytotoxic granules from NK or T lymphocytes, including perforin and granzymes, Persson et al. recently described a Ca²⁺-dependent Toxoplasma egress and subsequent lytic death of the host cell (Persson et al., 2007). It was therefore proposed that absence of apoptotic markers after infection with Toxoplasma could also be due to a necrotic cell death of parasite-infected host cells (Persson et al., 2007). However, several independent lines of evidence argue against this possibility occurring after triggering of the mitochondrial apoptotic pathway. Neither did the number of parasite-infected cells decrease after pro-apoptotic stimulation, nor did the percentage of cells that died by necrosis specifically increase after treatment of parasite-infected cells with staurosporine. Conversely, the number of viable parasite-infected cells was augmented due to an anti-apoptotic activity, which suggests that this parasite-host interaction might indeed facilitate the intracellular development of the parasite. Furthermore, cytosolic levels of various proteins investigated in this study were unaffected in Toxoplasma-infected cells treated to undergo apoptosis. Finally, the morphology of parasite-infected host cells treated with pro-apoptotic stimuli was inconsistent with necrotic cell death. Thus, induction of necrosis due to T. gondii egress from host cells can be discounted as contributing considerably to the blockade of apoptosis that was induced via the mitochondrial signalling cascade.

Bax and/or Bak are crucial in permeabilizing the outer mitochondrial membrane. Following Toxoplasma infection, events that preceded MOMP induced by Bim_S expression were prominently inhibited. This included the conformational change of Bax and Bak, translocation of a YFP-Bax reporter as well as endogenous wild-type Bax from the cytosol to mitochondria, and the oligomerization of Bax. Importantly, the total protein levels of Bax and Bak were not altered after parasitic infection. This indicated that Toxoplasma targets the functionality of Bax and Bak rather than decreasing their expression and, thereby, contradicted a previous report (Carmen et al., 2006). Because a quantification of the cellular Bax levels was not provided in that study and the functional consequences on Bax activation were not delineated (Carmen et al., 2006), the impact of Bax on the reduced cytochrome c release following Toxoplasma infection remained elusive. One might also speculate that the use of different Toxoplasma strains (type I RH in the study by Carmen and coworkers; type II NTE in the current study) explains this discrepancy because both lineages have been shown to differentially modulate host-cell signalling (Saeij et al., 2007; Kim et al., 2006). The results presented herein nevertheless provide strong evidence that Toxoplasma is capable of exclusively inhibiting the activation of Bax and Bak without reducing their total protein levels.

In healthy cells, Bax is largely cytosolic and has to translocate to mitochondria in order to initiate MOMP (Wolter et al., 1997; Zhou and Chang, 2008). In addition, changes of the protein conformation, including exposure of an N-terminal domain (Hsu and Youle, 1997; Lalier et al., 2007), accompany mitochondrial translocation although the exact order of events is unknown. In contrast to Bax, inactive Bak is constitutively bound to mitochondria and a conformational change might represent the first step during activation. Blockade of Bim_S-mediated subcellular redistribution of Bax, as well as exposure of a normally hidden N-terminal epitope of Bax and Bak in T. gondii-infected cells, collectively points towards parasite interference with an initial trigger of activation. Of note, a common signal thus appears to initiate conformational changes of Bak at the mitochondrial membrane as well as translocation plus conformational changes of Bax because both processes were efficiently inhibited by Toxoplasma infection. Whereas the nature of this trigger has not yet been conclusively identified, two opposing models are controversially discussed (Youle and Strasser, 2008; Chipuk and Green, 2008).

The `displacement model' proposes that Bax and Bak are restrained inactive by anti-apoptotic Bcl-2-like proteins and are unleashed when the inhibitors are bound by BH3-only proteins (Willis et al., 2007; Youle and Strasser, 2008). Thus, upregulation of anti-apoptotic Bcl-2-like proteins following infection provides one possible means to counteract the mitochondrial apoptotic pathway. However, several independent lines of evidence argue against this possibility. First, cytochrome c release was efficiently inhibited by T. gondii in the absence of transcriptional activity by the host cell; second, the levels of Bcl-2, Mcl-1, Bcl-X_L and A1 were not altered in T. gondii-infected cells as compared to non-infected controls; and third, T. gondii inhibited Bim_S-mediated caspase activity in the presence of the BH3 mimetic ABT-737 as efficiently as in its absence. Together, these data provide clear evidence that upregulation of anti-apoptotic Bcl-2 family members after T. gondii infection is not required to prevent host-cell death.

The second model, the `direct activation model', postulates that binding by distinct BH3-only proteins, specifically Bim, truncated Bid (tBid) and Puma activates Bax and Bak (Chipuk and Green, 2008; Häcker and Weber, 2007). However, after enforced Bim_S expression as employed in this study, Bim_S quantitatively localized into the heavy-organelle fraction containing mitochondria, whereas Bax stayed in the cytosol and only translocated to mitochondria during prolonged Bim_S expression. It is thus completely unclear how mitochondrial Bim_S could directly activate cytosolic Bax to translocate to the outer mitochondrial membrane.

Of particular interest, the mitochondrial targeting of Bim_S was completely unaffected in T. gondii-infected cells, whereas translocation and activation of Bax was at the same time efficiently inhibited (see above). Thus, mitochondrial localization of Bim_S might indeed be required to induce MOMP as reported (Weber et al., 2007), but does not suffice to recruit and activate Bax in T. gondii-infected cells as shown herein. The blockade of Bax activation and redistribution can either result from direct binding of a parasite effector molecule to Bax (and Bak), or else from parasite interference with the triggering of Bax and Bak activation. After immunoprecipitation, we were unable to detect parasite proteins associated with Bax or Bak when using a polyclonal anti-Toxoplasma serum (data not shown). Furthermore, preservation of protein complexes using EGS failed to detect an increase in the molecular weight of Bax following infection.

Although these data do not exclude the direct binding of a parasite protein to Bax and Bak, they support the view that Bim_S-induced apoptosis might rely on two distinct signals: expression of mitochondria-targeted Bim_S, and the translocation and activation of Bax (the trigger remaining elusive). The fact that T. gondii targets exclusively Bax (and Bak) activation, but not Bim_S might represent a valuable model for identifying those mechanisms that initiate Bax and Bak activity, recently referred to as the `holy grail' of apoptosis research (Youle and Strasser, 2008). In addition, future characterization of the parasite effector molecules and how they block the activation of Bax and Bak will provide important insights into the interaction of an intracellular pathogen and its host cell. This knowledge might also be applicable to the development of novel therapies against apicomplexan parasites.

HeLa229 wild-type (HeLa-WT) cells were cultivated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum (FCS), 100 U/ml penicillin, 10 μg/ml streptomycin, 1 mM sodium pyruvate and 1% non-essential amino acids (complete DMEM). T-REx-HeLa cells (Invitrogen) that stably express the BH3-only protein Bim_S under control of the tetracycline repressor from the pCDNA6/TR vector have been described previously (Weber et al., 2007) and were cultivated in complete DMEM supplemented with 5 μg/ml blasticidin and 125 μg/ml zeocin. HeLa-WT and T-REx-HeLa/Bim_S cells were transiently transfected with 2 μg of the pEYFP-C1 vector (Clontech) containing the human bax gene (Zhou and Chang, 2008) using FuGene HD transfection reagent (Roche). For some experiments, Jurkat E6.1 T lymphoblasts were cultivated in RPMI (Roswell Park Memorial Institute) medium supplemented with 10% FCS and penicillin and streptomycin.

Parasites of the mouse-avirulent type II strain NTE [which was used in all experiments unless otherwise stated (Gross et al., 1991)] were propagated in L929 fibroblasts as host cells in RPMI supplemented with 1% FCS and penicillin and streptomycin (as above). For some experiments, a GFP-expressing mutant of the mouse-virulent type I strain RH [RH-LDM (Barragan and Sibley, 2002); kindly provided by Antonio Barragan, Karolinska Institutet, Stockholm, Sweden] was used. Parasites were separated from host cells by differential centrifugation (Vutova et al., 2007). Adherent HeLa-WT and T-REx-HeLa/Bim_S cells were infected at parasite-to-host cell ratios of 10:1 or 20:1, routinely yielding infection rates of 42±1 and 51±2%, respectively, at 24 hours after infection. In order to compensate for the lower infectivity of suspension cells, Jurkat cells were infected at a parasite to host cell ratio of 30:1, which led to 54.3±2.5% parasite-positive cells. If not indicated otherwise, infected cells and non-infected controls were induced to express Bim_S using 1 μg/ml tetracycline or were treated with 1 μM staurosporine at 21 hours post-infection for an additional 3 hours. In some experiments, 20 μM of the pan-caspase inhibitor z-VAD-fmk or 20 μg/ml of actinomycin D was added starting at 1 hour prior to infection. The BH3 mimetic ABT-737 or its enantiomer (Oltersdorf et al., 2005) (kindly provided by Steven Elmore, Abbott's Pharmaceutical Research and Development Division, Abbott Park, IL) was added at 0.1 or 1 μM at 2 hours prior to the induction of apoptosis.

To assess the condensation of chromatin, T. gondii-infected or uninfected cells were fixed in 4% paraformaldehyde, 0.1 M sodium cacodylate, pH 7.4 for 30 minutes and stained with 50 ng/ml Hoechst 33258 in phosphate-buffered serum (PBS) for 1 hour. In some experiments, parasites were simultaneously visualized using a rabbit anti-Toxoplasma hyperimmune serum and Cy3-conjugated F(ab′)₂ fragment anti-rabbit IgG as described below. After being washed and mounted with Mowiol (Calbiochem), at least 500 cells per sample were examined by fluorescence microscopy.

Cell death following induction of apoptosis and infection with T. gondii was quantified by flow cytometry. To this end, infected or non-infected Jurkat cells were washed twice in PBS and then stained with R-phycoerythrin-conjugated Annexin V and 7-aminoactinomycin D (AAD) using the Apoptosis Detection Kit I as recommended (BD Biosciences, Heidelberg, Germany). Cells were analysed using a FACSCalibur flow cytometer (BD Biosciences).

In order to quantify Bax and Bak, T. gondii-infected cells or non-infected controls were washed in PBS and fixed with 0.5% paraformaldehyde in PBS for 30 minutes on ice. They were then incubated in 50 μg/ml digitonin, 1% FCS in PBS containing 10 μg/ml of conformation-specific mouse monoclonal anti-Bax-NT (clone 6A7; BD Pharmingen) or anti-Bak-NT (clone TC-100; Calbiochem). Staining with irrelevant isotype control antibody was performed in parallel. After 30 minutes on ice, cells were washed and then incubated with R-phycoerythrin-conjugated goat anti-mouse IgG before analysis.

Caspase activities were determined as described (Vutova et al., 2007). Briefly, infected and non-infected cells (1×10⁶ cells per sample) were extracted with 50 μl of 1% Nonidet P40, 150 mM NaCl, 50 mM Tris-HCl, pH 8.0, and complete protease inhibitor cocktail (EDTA-free; Roche). Aliquots (10 μl) of the supernatants were mixed in triplicates with 90 μl of either 10 μM Ac-DEVD–amino-4-methyl-coumarin (AMC; caspases 3 and 7-specific; Bachem) or 50 μM Ac-LEHD-AMC (caspase-9-specific; Bachem) in 0.1 mg/ml BSA, 0.1% CHAPS, 10 mM HEPES, 50 mM NaCl, 40 mM β-glycerophosphate, 2 mM MgCl₂ and 5 mM EGTA, pH 7.0. The kinetics of substrate cleavage was recorded over a period of 60 minutes at 37°C using a Victor V multilabel counter (Perkin Elmer).

To obtain total cellular lysates, Toxoplasma-infected cells or uninfected controls were lysed in 1% Triton X-100, 150 mM NaCl, 50 mM Tris/HCl pH 8.0, 50 mM NaF, 5 mM sodium pyrophosphate, 1 mM PMSF, 1 mM EDTA, 1 mM sodium orthovanadate, and 5 μg/ml each of leupeptin, aprotinin and pepstatin. The subcellular distribution of cytochrome c and Bcl-2 family members was determined after separation of cells into digitonin-soluble and digitonin-insoluble fractions (Goebel et al., 2001). Briefly, cells (1×10⁶ cells per sample) were resuspended in PBS, and mixed with an equal volume of 150 μg/ml digitonin in 0.5 M sucrose. After 30 seconds, heavy organelles, including mitochondria, were pelleted at 14,000 g for 1 minute and supernatants saved as the cytosol-containing digitonin-soluble fraction. The digitonin-insoluble pellets were then extracted as described above. After centrifugation, soluble proteins of total cellular extracts or subcellular fractions were separated by standard SDS-PAGE and transferred to nitrocellulose by semidry blotting. Nonspecific binding sites were blocked using 5% dry skimmed milk, 0.2% Tween-20, 0.02% NaN₃ in PBS, pH 7.4. Membranes were probed overnight at 4°C with rabbit polyclonal anti-caspase-9 (directed against the p35 fragment), rabbit polyclonal anti-Mcl-1, rabbit polyclonal anti-A1/Bfl-1 (all from Santa Cruz Biotechnology), goat polyclonal anti-caspase-3 (R&D Systems), rabbit polyclonal anti-Bim, mouse monoclonal anti-Bax (clone 3), mouse monoclonal anti-Bak (clone G317-2), mouse monoclonal anti-Bcl-2 (clone 4D7), mouse monoclonal anti-Bcl-X (clone 2H12), mouse monoclonal anti-cytochrome c (clone 7H8.2C12; all from BD Pharmingen), mouse monoclonal anti-actin (clone C4; kindly provided by James Lessard, Children's Hospital Medical Center, Cincinnati, OH), or mouse monoclonal anti-COX subunit IV (clone 10G8-D12-C12; Molecular Probes) diluted in 5% dry skimmed milk, 0.05 % Tween-20 in PBS, pH 7.4. Immune complexes were labelled with horseradish-peroxidase-conjugated anti-mouse, anti-rabbit or anti-goat IgG (Dianova) and visualized by ECL chemiluminescence.

In order to preserve oligomeric Bax complexes, cells were extracted using 2% CHAPS, 137 mM NaCl, 10 mM HEPES, 2 mM EDTA, complete protease inhibitor cocktail and 1 mM sodium orthovanadate, pH 7.4 for 30 minutes on ice. Soluble proteins were cross-linked with 1 mM ethylene glycol-bis(succinimidylsuccinate) (EGS) for 30 minutes at room temperature. The reaction was stopped by adding Tris-HCl, pH 7.4, to a final concentration of 20 mM. Protein extracts were analysed by immunoblotting as described above.

Blockade of transcription by actinomycin D was assessed by analysing IL-2 transcripts after activation of Jurkat cells for 24 hours with 5 nM PMA and 1 μg/ml PHA. Briefly, total RNA was isolated and mRNA reverse-transcribed using standard protocols. IL-2 and β-actin cDNAs were then amplified by PCR using specific primer pairs and analysed by agarose gel electrophoresis.

The intracellular distribution of WT-Bax or YFP-Bax reporter as well as the invasion and intracellular development of T. gondii were determined by double or triple immunofluorescence microscopy. Infected cells or non-infected controls were fixed with 4% paraformaldehyde in 0.1 M sodium cacodylate, pH 7.4, for 1 hour and quenched for 10 minutes in 50 mM NH₄Cl in PBS. Following permeabilization of the cells with 0.1 mg/ml saponin and 1% BSA in PBS, they were incubated for 1 hour with rabbit anti-Toxoplasma hyperimmune serum alone or along with either anti-Bax-NT (clone 6A7) or anti-Bax (clone 3; both from BD Pharmingen) diluted in PBS containing saponin and BSA. After washing, cells were incubated for 1 hour with Cy3-conjugated F(ab′)₂ fragment anti-rabbit IgG alone, or Cy2-conjugated F(ab′)₂ fragment anti-rabbit IgG in combination with Cy3-conjugated F(ab′)₂ fragment anti-mouse IgG, or else with Cy5-conjugated F(ab′)₂ fragment donkey anti-rabbit IgG (Dianova). In some experiments, total cell populations were stained with propidium iodide. Cells were examined by confocal laser scanning microscopy using Leica TCS SP2.

Results are expressed as means ± s.e.m. of at least three independent experiments unless otherwise indicated. Significant differences between mean values were identified by the Student's t-test. In order to avoid false positive results due to multiple comparisons, P-values were multiplied by the number of comparisons (Bonferroni correction). Corrected P-values of less than 0.05 were considered significant.