Inhibition of caspase activation and a requirement for NF-κB function in the Toxoplasma gondii-mediated blockade of host apoptosis

Toxoplasma gondii is arguably among the most successful protozoan parasites, establishing both acute and chronic infections in all warm-blooded animals (Tenter et al., 2000). In healthy humans, T. gondii generally causes asymptomatic infections. By contrast, in immunocompromised individuals (including those with HIV-AIDS, organ transplant recipients and those undergoing chemotherapy), acute toxoplasmosis causes a life threatening infection (Remington and McLeod, 1992; Tenter et al., 2000). In addition, congenital toxoplasmosis (transmitted vertically from an actively infected mother) can cause severe fetal damage, including spontaneous abortion (Petersen et al., 2001).

As an obligate intracellular parasite, Toxoplasma is completely dependent on the establishment of a replication-permissive niche within the infected cell (Sinai and Joiner, 1997). Within this niche, parasites must acquire nutrients and neutralize host defenses (Sinai and Joiner, 1997), among them apoptosis or programmed cell death (Heussler et al., 2001). Increasingly, viral, bacterial and protozoan pathogens have been shown to modulate the host apoptotic response in the establishment of their respective niches (Gao and Abu Kwaik, 2000; Moss et al., 1999; Roulston et al., 1999). T.-gondii-infected cells are resistant to apoptosis (Goebel et al., 1998; Goebel et al., 1999; Goebel et al., 2001; Nash et al., 1998), suggesting that this blockade is important in the biology of the parasite. A blockade of apoptosis is probably important in both the acute and the chronic phases of the infection (Heussler et al., 2001).

Apoptosis is a highly ordered physiological response to cell damage caused by external and internal insults (Hacker, 2000; Kaufmann and Hengartner, 2001). In addition to its role in limiting the growth of infectious agents, apoptosis plays crucial roles in the immune response (Baumann et al., 2002) and development (Sanders and Wride, 1995; Vaux et al., 1994; Zhivotovsky, 2002). Accordingly, imbalances in apoptotic responses contribute to several pathological states including cancers (Negoescu, 2000; Pan et al., 1997). The molecular pathways involved in the initiation, progression and execution of apoptosis have been elucidated in great detail (Kaufmann and Hengartner, 2001). There are two primary apoptotic mechanisms (Sun et al., 1999): (1) the mitochondrion-dependent pathway activated by the release of cytochrome c from the mitochondrial intermembrane space (Green and Reed, 1998); (2) the activation of surface `death receptors', including the tumor necrosis factor (TNF) family of receptors, and the signaling cascade resulting therefrom (Nagata, 1997). Regardless of the apoptogenic trigger, stimuli converge on the activation and activity of a group of cysteine proteases, called caspases, that cleave at specific aspartic-acid-containing substrate recognition sites (Budihardjo et al., 1999; Earnshaw et al., 1999). Caspases exist within the cell as inactive zymogens that are activated along a specific amplification cascade by proteolytic cleavage in response to an apoptogenic signal (Budihardjo et al., 1999; Earnshaw et al., 1999). Accordingly, the release of cytochrome c from mitochondria activates the initiator caspase, caspase 9, which in turn activates the executioner caspase, caspase 3 (Slee et al., 1999). Likewise, activation of death receptors triggers the cleavage of caspase 8, which in turn converges on the executioner caspase, caspase 3 (Nagata, 1997). Given the crucial role of caspases in apoptosis, inhibition of their activation and/or activity is a likely target for microbial intervention.

Several viral antiapoptotic factors have been identified that interfere with caspases (Roulston et al., 1999). These factors, designated IAP (inhibitor of apoptosis), have cellular homologs (cIAP) whose expression is regulated as a component of the cell survival response (Deveraux and Reed, 1999; Deveraux et al., 1997; Hawkins et al., 1998; Roy et al., 1997; Salvesen and Duckett, 2002). Key components of the survival response, including IAP expression, include genes regulated by the transcription factor NF-κB (Ghosh, 1999). Consequently, pathogen-mediated exploitation of this pathway could play an important role in the establishment of the antiapoptotic state (Tato and Hunter, 2002). In this study, we address the molecular mechanisms underlying the inhibition of apoptosis established in T.-gondii-infected cells. We focus on the effect of infection on caspase activation and activity, and the contribution of the host survival response under the regulation of the transcription factor NF-κB. Manipulation of these cellular pathways suggests an important role for the inhibition of apoptosis in T. gondii infection.

Wild-type mouse NIH 3T3 Balb/c fibroblasts (ATCC-CCL163) and p65/RelA^–/– 3T3 fibroblasts (Beg and Baltimore, 1996) were cultured in α-minimal essential medium (α-MEM) and supplemented with 100 units ml^–1 penicillin, 100 μg ml^–1 streptomycin, 7% heat-inactivated fetal bovine serum and 2 mM L-glutamine. A derivative of T. gondii strain RH with a deletion in the HXGPRT gene (RHΔHX) (Donald et al., 1996) (obtained from the NIH AIDS Research and Reference Program) was maintained by serial passage in VERO cells as previously described (Sinai et al., 2000).

3×10⁵ host cells per well were seeded in a six-well tissue culture dish (Falcon) for at least 6 hours. Designated wells were infected with 10⁶ freshly passaged RHΔHX parasites for 15-18 hours. Apoptosis was induced using either staurosporine (STS; Sigma, St Louis) or tumor necrosis factor α (TNF-α; Research and Diagnostic Systems, Minneapolis) in the presence of cycloheximide (10 μg ml^–1) (Sigma) at the concentrations noted in the figures or figure legends unless otherwise noted. Induction of apoptosis was conducted for 8 hours unless otherwise noted.

Following induction of apoptosis, both adherent and detached cells were collected by scraping and pelleted by centrifugation using a standard tissue culture centrifuge at 1800 g for 5 minutes. The cell pellets were resuspended in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM NaH₂PO₄, 1.8 mM KH₂PO₄ adjusted to pH 7.4) and protein concentration determined using the BCA protein assay (Pierce, Rockford, IL). The cell suspension was lysed directly in 4× SDS-PAGE sample buffer and subjected to polyacrylamide gel electrophoresis (Laemmli, 1970). 25 μg total protein per sample was resolved on 7% gels for α2-spectrin and poly-ADP-ribosylating polymerase (PARP) and a 12% gel for caspase 3 using Bio-Rad Mini-PROTEAN II apparatus. The gels were transferred onto nitrocellulose (Pall, East Hill) using a semi-dry blotting apparatus as recommended by the manufacturer. Transferred proteins were visualized using Ponceau S (Sigma) and blocked for at least 1 hour using PBS-Tween (PBS with 0.2% Tween 20) containing 5% skimmed milk (PBS-TM). Incubations with primary antibodies were generally conducted overnight at room temperature in PBS-TM on a reciprocating shaker. Blots were washed three times in PBS-TM for 15 minutes each prior to the addition of secondary antisera conjugated to horseradish peroxidase (HRP) for 2 hours. Following three washes as above, the immunoreactive bands were visualized using enhanced chemiluminescence (Pierce). Exposure times were generally between 10 seconds and 1 minute. Antibodies used were rabbit anti-spectrin (Janicke et al., 1998) (1:5000; J. Morrow, Yale University), rabbit anti-PARP (1:600; Cell Signaling, Beverly, MA), rabbit anti-caspase-3 (1:800; Stressgen) and goat anti-rabbit-HRP (1:5000, Jackson Laboratories, Bar Harbor, ME).

Following induction of apoptosis, cells were collected as above. Determination of caspase activity using fluorescent substrates was measured by adapting the method of Mesner et al. (Mesner et al., 1999). Cell pellets were lysed in PBS containing 0.1% Triton X-100 and the protein concentration determined using a BCA protein assay (Pierce, Rockford). An appropriate volume of 2× buffer A (50 mM HEPES, 10 mM MgCl₂, 2 mM EGTA, 2 mM PMSF, 20 μg ml^–1 pepstatin A, 20 μg ml^–1 leupeptin) was added so that the final protein concentration was 0.5 μg μl^–1 in 1× buffer A. Protease inhibitors were purchased from Sigma. Caspase activity assays were performed in opaque-walled 96-well plates (Corning, Ithaca, NY). Generally, 30 μl of the cell extracts (15 μg of protein) were combined with 170 μl of buffer B (100 mM HEPES, 20% glycerol, 0.5 mM EDTA, 5 mM β-mercaptoethanol, 50 μM fluorescent caspase substrate) and incubated for 2 hours at 37°C. In a modification of the original protocol, β-mercaptoethanol is used in place of dithiothreitol to prevent the activation of a potent dithiothreitol-inducible apyrase activity present in Toxoplasma (Asai et al., 1995).

In the experiment to detect inhibitors capable of blocking activity of activated caspases (Fig. 6), mixtures of apoptotic and non-apoptotic cell extracts in the ratios indicated in the figure were incubated in buffer A on ice for 1 hour prior to the addition of buffer B in order to permit putative inhibitory interactions to establish.

Caspase activity based on the release of the fluorescent dye 4-methyl-coumaryl-7-amide (MCA) was determined using a Perkin Elmer LS 50B fluorometer using an excitation wavelength of 380 nm, an emission wavelength of 440 nm and a filter cutoff at 430 nm. The caspase substrates corresponding to the preferred cleavage sites were DEVD-MCA, LEHD-MCA, IETD-MCA for caspases 3, 8 and 9, respectively (Talanian et al., 1997) (Peptides International, Louisville, KY).

4×10⁵ wild-type 3T3 fibroblasts per well were seeded in a six-well tissue culture dish (Falcon) for at least 6 hours. Appropriate wells were either left uninfected or infected with 2×10⁶ parasites overnight. Apoptosis was induced using STS (300 nM for 5 hours) and both adherent and detached cells were collected and pelleted by centrifugation (1800 g for 5 minutes). Pellets were resuspended in 100 μl of annexin-V/FITC in the binding buffer at three times the concentration recommended by the manufacturer (Roche, Palo Alto, CA). The additional annexin-V/FITC was required because the kit is optimized for lymphocytes, which have a much smaller surface area than fibroblasts. Propidium iodide (PI) was added to distinguish necrotic cells (PI positive) from apoptotic and live cells (PI negative). Cells were diluted into 1 ml PBS, pelleted and resuspended in 1 ml PBS and analysed immediately by fluorescence-activated cell sorting (FACS). FACS analysis was performed on a FACScaliber cell sorter (Becton Dickinson) using the 488 nm line to excite FITC (FL1) and PI (FL2) simultaneously, and analysed using CellQuest software (Becton Dickinson). Cells that were FITC/PI double positive (necrotic) were excluded from the analysis.

5×10⁴ host cells were seeded on sterile 12 mm coverslips in 24-well tissue culture dishes for at least 8 hours. Cells were infected at an m.o.i. (multiplicity of infection) of 3-5 and incubated overnight. Infected monolayers were washed three times with PBS and fixed in 3% paraformaldehyde (PFA) for 15 minutes. Monolayers were permeabilized with PBS containing 0.2% Triton X-100 for 5 minutes. Immunofluorescence staining was performed as previously described (Sinai et al., 2000). Antibodies used were: rabbit anti-activated-caspase-3 (1:400; Cell Signaling, Beverly, MA), mouse monoclonal T3H11 against GRA3 (1:1000) (Bermudes et al., 1994) and mouse monoclonal against SAG1 (1:1500; Argene, North Massapequa, NY). TUNEL assays to detect nuclear fragmentation in situ were performed using the In situ Cell Death kit (Roche) with a single modification. The reaction mixture was diluted 1:3 in 100 mM sodium cacodylate buffer to reduce the high background and nonspecific labeling. Immunofluorescence labeling of coverslips was performed after the TUNEL reaction was complete. These coverslips were stained with Hoechst dye (Molecular Probes) to visualize the nuclei. Slides were viewed on a Ziess Axiphot stand using a 100×, 1.4 numerical aperture, phase-contrast, oil-immersion objective. Images were digitally captured using a SPOT camera system and software in a grayscale mode. The merged images in Fig. 2 were generated by pseudocoloring the constituent grayscale images and merging them using Adobe Photoshop software. In all cases, adjustments for brightness and contrast, and for color balance in the merged images, were performed on the entire image.

Confluent monolayers in 10 cm tissue culture dishes (Falcon) of wild-type 3T3 cells and the p65^–/– mutants were left uninfected or infected at an m.o.i. of 5 overnight. Apoptosis was induced on appropriate samples with either 10 ng ml^–1 TNF-α in the presence of cycloheximide or 10 nM STS for 8 hours. Samples were then processed for TEM as described previously (Sinai et al., 1997). Sections were visualized using a Phillips Tecnai 12 transmission electron microscope and images digitally acquired.

Cells undergoing apoptosis can be distinguished based on several morphological and biochemical characteristics (Saraste and Pulkki, 2000). The early studies on apoptosis defined clear morphological properties associated with cell suicide [historical perspective in Rich et al. (Rich et al., 1999)]. These include changes in nuclear condensation, alterations at the plasma membrane, disrupted organellar organization and vacuolation, all of which are markers of the systematic dismantling of cellular organization leading to death (Hacker, 2000). As predicted, these features are readily observed by electron microscopy in NIH3T3 cells following treatment with the kinase inhibitor staurosporine (STS), activating apoptosis along the mitochondrial pathway, and the cytokine tumor necrosis factor α (TNFα), activating the death receptor pathway in the presence of cycloheximide (Fig. 1B,C). Most notable are changes in the density of the nucleus (Fig. 1B, arrowheads, Fig. 1C), the disorganization of organelles and vacuolation (Fig. 1B,C). By contrast, identically treated Toxoplasma-infected cells appear relatively unaffected (Fig. 1E,F) compared with the untreated control cells (Fig. 1A,C). The cellular debris evident in the fields is presumably caused by apoptosis occurring in uninfected cells in the specimen (Fig. 1E,F)

Changes in nuclear organization are caused by the activation of specific nucleases during apoptosis, resulting in DNA fragmentation (Hacker, 2000). The presence of free DNA ends thus generated is detectable using the poly dT-terminal transferase (TUNEL) assay (Sanders and Wride, 1996). TUNEL assays performed following treatment with TNFα, (Fig. 2Ae-h) and STS (Fig. 2Ai-l) as apoptogenic triggers indicate that a positive reaction is largely restricted to uninfected cells, whereas the nuclei of T.-gondii-infected cells do not incorporate the FITC-conjugated nucleotides (Fig. 2Ae,h,i,l). Infected cells can be noted by the presence of SAG1-labeled parasites (Fig. 2Ab,f,j) adjacent to Hoechst-dye-stained host nuclei (Fig. 2Ad,h,l). Notably, the use of Hoechst dye to label DNA confirms the alteration of DNA organization in the TUNEL positive cells and the relatively normal morphology of infected fibroblasts (Fig. 2c,g,k). Accordingly, TUNEL positive cells tend to have highly condensed amorphous nuclei (Fig. 2Ag,k). Other apoptotic triggers (including ultraviolet irradiation, actinomycin D and camptothecin) gave identical results (data not shown).

Finally, apoptosis is associated with the loss of plasma membrane asymmetry manifested by the translocation of phosphatidylserine from the inner to the outer leaflet of the plasma membrane (Martin et al., 1995). This event can be readily detected by the calcium-dependent binding of fluorescently labeled Annexin V to the plasma membrane followed by FACS analysis (van Engeland et al., 1996; Vermes et al., 1995). As expected, treatment of uninfected NIH3T3 cells with STS is associated with a significant increase in Annexin-V/FITC binding, represented by the appearance of an additional peak (Fig. 2B, Uninfected, green trace). By contrast, significantly fewer Annexin-V/FITC positive cells were observed with identically treated Toxoplasma-infected cells (Fig. 2B, Infected, green trace). Within the uninfected population, 39% of the cells were apoptotic, as opposed to 17.5% in the infected cell population. Most of the apoptotic cells in the infected sample are probably uninfected cells in the population because, in the experiment presented, an infection level of ∼70% was attained (data not shown). Taken together, these results show that T. gondii infection inhibits the appearance and progression of multiple stigmata classically associated with apoptosis. In light of the role of caspases in these changes (Saraste and Pulkki, 2000; Thornberry, 1998), we examined whether their activity was altered in infected cells.

The morphological and cellular changes associated with apoptosis are due to the activation of the caspase cascade, most notably that of the executioner, caspase 3 (Earnshaw et al., 1999; Thornberry, 1998). The specific cleavage of several host targets of caspase 3 leads to highly predictable products detectable by immunoblot analysis. Among these, non-erythroid spectrin (α2-spectrin, also known as fodrin) (Janicke et al., 1998) and PARP (Nicholson et al., 1995) have been used extensively as markers for caspase 3 activity. We examined whether infection of wild-type mouse embryonic 3T3 fibroblasts with T. gondii protected these cellular substrates from caspase-3-mediated cleavage following the induction of apoptosis by diverse stimuli. The selected stimuli have been shown to mediate apoptosis along both the mitochondrial (STS) (Reynolds et al., 1996; Stepczynska et al., 2001) and death receptor (TNFα) (Wallach et al., 1997) pathways. Of note, induction of apoptosis by TNFα in 3T3 fibroblasts necessitated the use of the protein synthesis inhibitor cycloheximide (CX) as indicated in the figures.

Caspase 3 activity in uninfected cells triggered to undergo apoptosis using both mitochondrion (STS) and death receptor (TNFα) pathway causes the generation of the diagnostic 120 kDa fragment of α2-spectrin (Fig. 3A,B, top) (Janicke et al., 1998). By contrast, T.-gondii-infected cells treated under identical conditions fail to generate this diagnostic cleavage product, suggesting that caspase 3 activity is inhibited (Fig. 3A,B, top). Under higher concentrations of STS (>150 nM for 8 hours), some caspase 3 activity was observed and this might be due to the toxicity of STS towards the parasite under these conditions (data not shown). In addition to α2-spectrin, we examined the cleavage of PARP, another well characterized nuclear target of caspase 3. As observed with α2-spectrin, the band diagnostic for caspase 3 activity [cleavage of the 116 kDa polypeptide to generate the 89 kDa form (Nicholson et al., 1995)] was observed in uninfected cells under apoptogenic conditions (Fig. 3A,B, bottom) but was inhibited in identically treated Toxoplasma-infected cells (Fig. 3A,B, bottom).

In order to confirm the data from the immunoblot analyses, we measured caspase 3 activity directly in cell extracts using the fluorescent substrate analog DEVD-MCA. Cellular extracts were incubated with the substrate and the release of MCA measured fluorometrically. In agreement with the immunoblot studies, an inhibition of caspase 3 activity was observed across a wide range of concentrations of the mitochondrial pathway specific agent STS (Fig. 3C) and death-receptor-specific trigger TNFα (Fig. 3D). The block in caspase activity could arise from the inhibition of the activated caspase 3, an inhibition of caspase 3 activation or both.

The activity of caspase 3 depends on the activation of the zymogen procaspase form to the activated form by proteolytic cleavage (Nicholson, 1999; Nicholson et al., 1995). We examined the activation status of caspase 3 by immunoblot and immunofluorescence analyses. Upon activation, the 32 kDa zymogen form is cleaved into 11-12 kDa and 17-20 kDa activated forms (Nicholson, 1999). Evidence for activated caspase 3 (generation of the 17-20 kDa fragment) is readily apparent in uninfected cells undergoing apoptosis in response to both STS and TNFα (Fig. 4A,B). By contrast, Toxoplasma infection blocked caspase 3 activation, as indicated by the lack of the activated form in cell lysates from infected cells (Fig. 4A,B).

We examined caspase 3 activation microscopically at the single cell level by immunofluorescence analysis using an antibody specific for the activated form of caspase 3 (Fig. 4C). As inferred from the immunoblot analysis, T.-gondii-infected cells [detected by immunoreactivity against the parasite antigen GRA3 (Fig. 4Cb,e)] were significantly less likely to label with the activated caspase-3-specific antiserum than their uninfected counterparts (Fig. 4C). Accordingly, infected cells (Fig. 4Cc,f, white boxes) were protected from apoptosis (Fig. 4Cc,f, black boxes). In addition to STS and TNFα/CX, the apoptogenic triggers actinomycin D, beauvericin, camptothecin and ultraviolet irradiation yielded similar results, indicating a generalized mechanism for the inhibition of apoptosis relies on the blockade of both caspase 3 activation and activity (data not shown).

Together, these data suggest that the inhibition of caspase 3 activity is a consequence of a blockade of activation by proteolysis. Among the possible mechanisms to explain this phenomenon are the inhibition of the initiator caspases, caspase 8 and caspase 9, both of which are capable of activating caspase 3 following an apoptogenic stimulus (Nicholson, 1999; Slee et al., 1999; Steinnicke and Salvesen, 1997; Stennicke et al., 1998).

We used fluorescent substrates for caspase 8 (IETD-AMC) and caspase 9 (LEHD-AMC) to measure the activity of these enzymes in extracts from infected and uninfected cells left either untreated or induced to undergo apoptosis along both the mitochondrial and the death-receptor pathways. As observed with caspase 3, inhibition of both the death-receptor-activation-dependent caspase 8 (Fig. 5A,B) and the mitochondrion-dependent caspase 9 were observed (Fig. 5C,D) in infected cells. Together, these results indicate that the T.-gondii-mediated blockade of caspase activity is multifactorial, affecting the activation of the caspase cascade at the level of the initiator caspases or within the signaling pathways responsible for the activation of the cascade. As discussed below, it might indicate interference with the signaling mechanisms activated within cells undergoing apoptosis.

The inhibitor of apoptosis (IAP) family of proteins have been shown to inhibit the activation and/or activity of mammalian caspases directly (Deveraux et al., 1999; Ekert et al., 1999; Villa et al., 1997). A potential basis for the inhibition of caspase activity is the presence of a parasite-derived or parasite-induced but host-derived factor in extracts of infected cells. We reasoned that such a factor should be able to inhibit caspase activity in extracts of uninfected cells undergoing apoptosis. The fluorogenic substrate assay was extended to examine if mixing extracts of infected nonapoptotic (INA) cells to extracts from uninfected cells undergoing apoptosis (UAs) would inhibit already active caspases. As a control, extracts of untreated, uninfected non-apoptotic (UNA) cell extracts were added to UA extracts.

Following co-incubation to permit the interaction between a putative inhibitor in INA extracts with the active caspases in UA extracts (Fig. 6, white bars), the net caspase activity was measured using appropriate fluorogenic substrates for caspase 3 (Fig. 6A), caspase 8 (Fig. 6B) and caspase 9 (Fig. 6C). In the control experiments, increasing concentrations of UNA extracts were added to the activated UA extract (Fig. 6A-C gray bars). We predicted that, in the presence of a specific inhibitor of active caspases, INA extracts would cause a non-linear decrease in net caspase activity (Fig. 6D, dashed line). By contrast, if no specific inhibitor were present, a linear decrease in activity reflecting the dilution of the UA extract would be observed (Fig. 6D, solid line).

Contrary to our expectations no infection-specific inhibition of activity for any of the caspases was observed (Fig. 6A-C, white bars) because the decrease was identical to that observed when the UNA control extracts were added (Fig. 6A-C gray bars). The linear decrease exactly mirrored the reduction of activity seen with increasing dilution of the apoptotic (UA) extract. The mixing of cellular extracts strongly suggests that the inhibitor of caspase activity acts at the level of the activation of these enzymes and cannot therefore reverse the activity of already activated caspases.

T. gondii inhibits TNFα-mediated apoptosis in the presence of CX when administered following an overnight infection, suggesting that de novo protein synthesis by the parasite during induction is not required (Fig. 3). The primary target of cycloheximide in promoting TNFα-mediated apoptosis is the inactivation of the pro-survival response regulated by the transcription factor NF-κB (Wang et al., 1996; Wang and Baldwin, 1998). We examined whether the activation of this response by the parasite contributes to the anti-apoptotic state in infected fibroblasts using NF-κB knockout (p65^–/–) fibroblasts (Beg and Baltimore, 1996).

The host transcription factor NF-κB is an important component of the anti-apoptotic and pro-survival responses (Ghosh, 1999). Genes it regulates encode proteins including IAPs (Stehlik et al., 1998; Wang et al., 1998) and members of the Bcl2 family (Brasier et al., 2001; Khoshnan et al., 2000; Qiu et al., 2001), which play crucial roles in apoptosis. We reasoned that, if a component of the T.-gondii-mediated blockade of apoptosis involved activation of the host survival response, infection would provide no protection to cells deficient in NF-κB activity. Mouse embryonic 3T3 fibroblasts derived from embryos of mice engineered to possess a targeted disruption of both copies of the p65 subunit of NF-κB (p65^–/–) are profoundly susceptible to apoptotic stimuli, particularly following death receptor activation (Beg and Baltimore, 1996).

In marked contrast to wild-type 3T3 fibroblasts, T. gondii infection afforded no protection to p65^–/– fibroblasts against apoptotic triggers along either the mitochondrion or the death-receptor mediated pathways using multiple assays (Fig. 7). Following the induction of apoptosis, caspase 3 activity was detected by the cleavage of the endogenous substrates PARP (Fig. 7A, top two panels) and α2-spectrin (data not shown) as well as the fluorescent peptide substrate DEVD-MCA (Fig. 7B,C). The high level of activity based on PARP cleavage observed in both infected and uninfected cells following TNFα treatment (Fig. 7A, middle) is reflected in the hyperactive cleavage of the fluorescent substrate DEVD-MCA [Fig. 7C (notice the scale)]. The basis for the generally lower values for the infected samples potentially reflects a more rapid loss of infected cells, resulting in their lysis prior to harvesting. Alternatively, it might suggest the presence of an alternative (albeit less efficient mechanism) for inhibition. To test this possibility, we examined the effect of TNFα at significantly lower concentrations (0.05-1.00 ng ml^–1) under which cytopathic milestones associated with apoptosis are not achieved as rapidly. Under these conditions, between five- and sixfold increases in caspase 3 activity were observed in the fluorescent substrate cleavage assay in both uninfected and infected p65^–/– cells in which apoptosis had been induced (data not shown). This confirms that the T.-gondii-mediated inhibition of apoptosis depends on NF-κB activity and does not merely reflect a different threshold for protection in this genetic background.

The detection of caspase 3 activity correlated with evidence for caspase 3 activation by immunoblot analysis (Fig. 7A, bottom). This in turn was found to be consistent with the presence of the active initiator caspases, caspase 8 and caspase 9 in extracts of both infected and uninfected p65^–/– fibroblasts using the fluorogenic substrate cleavage assay (data not shown).

The failure of Toxoplasma infection to block apoptosis was confirmed morphologically using electron microscopy. Treatment of uninfected p65^–/– cells resulted in the classical morphological signs of apoptosis (Fig. 8B,C). The effects were particularly pronounced and frequent in the case of TNFα/CX-treated cells (Fig. 8C), consistent with the hyperactivity of caspase 3 following this treatment (Fig. 7). In marked contrast to observations with wild-type fibroblasts (Fig. 1), T.-gondii-infected p65^–/– cells displayed all the same stigmata of apoptosis as uninfected cells when triggered using either STS or TNFα/CX. These include profound nuclear condensation, vacuolation and disruption of organellar architecture (Fig. 8E,F). Of note, these features manifest even in the absence of CX (data not shown), as reported previously for TNFα mediated death in p65^–/– cells (Beg and Baltimore, 1996). Taking these resuts together, we conclude that a crucial component of the T.-gondii-enforced blockade of host apoptosis depends on the activation of the host antiapoptotic machinery.

The inhibition of apoptosis appears to be a common theme in the biology of obligate intracellular pathogens (Gao and Abu Kwaik, 2000; Heussler et al., 2001) as well as several viruses (Roulston et al., 1999). Their complete dependence on the metabolic potential of the host cell for propagation is likely to have contributed to evolution of elaborate strategies to block cell suicide as a way to ensure the continued supply of nutrients. Several obligate intracellular bacteria [e.g. Chlamydia (Airenne et al., 2002; Fan et al., 1998; Geng et al., 2000; Perfettini et al., 2002), Rickettsiae (Clifton et al., 1998)] and protozoa [T. gondii (this study) (Goebel et al., 1998; Goebel et al., 1999; Goebel et al., 2001; Nash et al., 1998), Theileria spp. (Dobbelaere et al., 1999; Heussler et al., 1999; Heussler et al., 2001)] actively suppress host apoptosis. In addition, modulation of apoptosis is observed for several facultative intracellular organisms (Gao and Abu Kwaik, 2000) and those with distinct extracellular and intracellular phases (including Leishmania spp., Trypanosoma cruzi) as components of their life cycle (Heussler et al., 2001).

The blockade of apoptosis by Toxoplasma manifests at the level of apoptotic machinery as well as signaling cascades that power it. Accordingly, we find that inhibition of both the initiation (Fig. 5) and execution (Figs 3, 4) phases of the caspase cascade are observed. Although the initiator caspases (8 or 9) are generally required to activate caspase 3, the fact that T.-gondii-infected cells are resistant to granzyme-mediated apoptosis (Nash et al., 1998) suggests that direct inhibition of caspase 3 activity is also instituted. Granzyme from cytotoxic T cells bypasses the conventional activation schemes by directly activating both initiator and executioner caspases (Thornberry et al., 1997; Van de Craen et al., 1997). Interestingly, the primary mechanism of inhibition of all the caspases does not appear to manifest at the level of activity but rather at their activation as extracts from infected cells fail to inhibit pre-activated caspases (Fig. 6).

The activation of the initiator caspases by the upstream signals has been extensively studied. Activation of the caspase 8 in addition to its activity (Fig. 5) upon engagement of death receptors (TNFα) in T.-gondii-infected cells might be inhibited (data not shown) (Goebel et al., 2001). In addition we (L. Hardi, J. Carmen and A. P. Sinai, unpublished) and others (Goebel et al., 2001) have found the triggering event in mitochondrial apoptosis (activating caspase 9) – the release of cytochrome c – is inhibited in T.-gondii-infected cells. This is particularly intriguing given the intimate association between the T. gondii vacuolar membrane and host cell mitochondria suggesting a potential direct link to the inhibition of apoptosis (Sinai and Joiner, 2001; Sinai et al., 1997). The molecular basis of the interference with both the mitochondrial and death receptor cascades remains to be elucidated. Such factors, could be parasite derived, triggered in the host cell by the parasite subverting existing pro-survival and anti-apoptotic regulatory circuits, or both. Our data [see accompanying article in this issue (Molestina et al., 2003)] suggest that the parasite-mediated regulation of host genes involved in apoptosis and cell survival responses are key components in the blockade of apoptosis.

Our finding that T. gondii is unable to block apoptosis in cells deficient in NF-κB function (p65^–/–) by both biochemical (Fig. 7) and morphological (Fig. 8) criteria argues that the host pro-survival response is essential for the parasite-mediated blockade of apoptosis. A role has been documented in viruses, bacteria and protozoa for NF-κB in anti-apoptotic states established by infectious agents (Tato and Hunter, 2002).

The crucial role of NF-κB in our experiments is diametrically opposed to the finding of Goebels et al., who noted no role for NF-κB function in the blockade of apoptosis (Goebel et al., 2001). This conclusion was based on the observation that, in the human promyelocytic cell line HL-60, nuclear extracts from infected cells treated with actinomycin D exhibit a loss of NF-κB binding activity by electrophoretic mobility shift assay in a manner no different from uninfected cells. We have extended the results in the present manuscript to demonstrate that T. gondii infection of wild-type murine fibroblasts not only triggers NF-κB binding activity but causes the specific upregulation of several anti-apoptotic and pro-survival genes (Molestina et al., 2003). Among the anti-apoptotic genes induced by T. gondii infection are members of the IAP family (Deveraux and Reed, 1999) as reported in a comprehensive microarray study (∼22,000 cDNAs) of parasite-induced host gene expression (Blader et al., 2001; Molestina et al., 2003).

The IAP family was initially identified as baculoviral gene products capable of directly interfering with caspase activity and activation (Deveraux and Reed, 1999). Interestingly, many of the cellular IAPs are known to be under NF-κB regulation (Stehlik et al., 1998; Wang et al., 1998). This provides a potential mechanism for the inhibition in the activation and activity of caspases in Toxoplasma-infected cells based on the parasite-mediated overexpression of host cell IAPs. Subversion of the host pro-survival/anti-apoptotic machinery would be energetically favorable to the parasite because one or a few factors could initiate a host cascade simultaneously blocking multiple steps in the signaling and execution of apoptosis.

One of the more potent activators of NF-κB is TNFα. Upon binding its cell surface receptor, TNFα simultaneously transmits two opposing signals (Wallach et al., 1996; Wallach et al., 1997). One activates caspase 8 while the other potently activates NF-κB-dependent gene expression and thereby the pro-survival response (Wallach et al., 1996; Wallach et al., 1997). Although the activation of the caspase cascade does not require de novo protein synthesis (Earnshaw et al., 1999), the pro-survival arm of TNFα response does. Based on our data, one would predict that, in the presence of CX, one should mimic the p65^–/– phenotype and fail to see protection from apoptotic stimuli. Instead, we find high levels of infection-mediated protection blocking the activity and activation of the caspase cascade. We conclude from these observations that the overnight infection allows the accumulation of sufficient steady state levels of antiapoptotic factors such that they do not need to be replenished by de novo protein synthesis during the course of apoptotic induction. Because many of these factors are NF-κB regulated, the infected p65^–/– fibroblasts fail to synthesize the protective factors succumbing to TNFα treatment in both the presence (Figs 7, 8) and the absence (data not shown) of CX. Because CX also inhibits parasite protein synthesis, we can conclude that continual replenishment of parasite factors is not crucial once the antiapoptotic state is established. This reinforces the suggestion that interference at the levels of activation of the caspase cascade is crucial in the blockade of apoptosis.

Does the apparent induction of the host pro-survival pathway imply that Toxoplasma-derived products do not interfere directly with caspases? A direct inhibitor of caspase activity does not appear to be likely because mixing cellular extracts from infected cells with extracts containing activated caspases failed to elicit a specific inhibitory effect (Fig. 6). This experiment, however, in no way rules out interference with the machinery involved in the activation of the initiator caspases. Identification of the molecular basis of the inhibition of caspase activation by T. gondii is a current focus of our laboratory.

The observed blockade of apoptosis at multiple levels suggests interference with cell death is important in T. gondii pathogenesis. In the present study, we focused on the rapidly growing tachyzoite form of the parasite associated with acute infection (Dubey et al., 1998). In vitro, T. gondii tachyzoites grow rapidly within the infected cell, lysing it within 36-72 hours, depending on the cell type (data not shown). A reason to block apoptosis is somewhat paradoxical given that this timescale approximates most natural apoptotic events (Hacker, 2000). We argue, however, that there is a vital role for the inhibition of apoptosis in the tachyzoite stage for several reasons. As an obligate intracellular parasite, T. gondii is auxotrophic for several nutrients provided by the biosynthetic machinery of the cell (Sinai and Joiner, 1997). A cell in the process of apoptosis, devotes its energetic resources to systematically dismantling the cell as opposed to engaging in biosynthetic activities, thereby becoming a poor provider. An apoptotic cell would therefore be a poor provider to an intracellular pathogen. A second, perhaps equally important, reason to block apoptosis is as a mechanism to modulate the immune response given the crucial role of apoptosis in immune function (Baumann et al., 2002). Accordingly, Wei et al. (Wei et al., 2002) have demonstrated that T.-gondii-infected dendritic cells are resistant to apoptosis but trigger a contact-dependent apoptosis in cytotoxic T cells. Modulation of the balance between apoptosis and necrosis at a site of tissue infection could modulate the inflammatory response, thereby promoting either growth as a tachyzoite or triggering stage conversion to the long-term bradyzoite form associated with chronic infection (Alexander et al., 1997; Yap and Sher, 1999). This conversion is crucial for the long-term survival of T. gondii within the host as it mounts an effective immune response against it (Alexander et al., 1997; Yap and Sher, 1999). Bradyzoites possess considerably lowered metabolic activity and persist for long periods of time (years to decades), with their intracellular localization affording them protection from immune attack (Dubey, 1998; Dubey et al., 1998). Given their long-term intracellular residence, a blockade of apoptosis is likely to be a crucial function. It is therefore likely that, by establishing an anti-apoptotic state during tachyzoite infection, stage conversion to the bradyzoite can proceed within an environment that is primed to block apoptosis and ensure long-term residence. These questions are under investigation.

In summary, these studies indicate Toxoplasma gondii inhibits the apoptotic cascade by manipulating the activity of the caspase cascade. This blockade is critically dependent on the host pro-survival machinery regulated by the transcription factor NF-κB. We hereby establish a foundation upon which to define molecular interactions responsible for the manipulation of this critical physiological pathway. Current and future studies are focused on defining the specific parasite and host components involved in these interactions between the parasite and host. Furthermore, elucidation of these pathways will provide insights not only in the area of microbial pathogenesis but also, more broadly, in our understanding of the balance between survival and death in mammalian cells.

We thank D. Baltimore and J. Morrow for their gifts of the p65^–/– cells and the affinity-purified anti-α2-spectrin antibody, respectively. We are indebted to the staff of the University of Kentucky College of Medicine Electron Microscopy and Imaging Suite, and the Flow Cytometry Core Facility for their excellent technical assistance and advice. This work was supported by new faculty startup funds from the University of Kentucky Research Challenge Trust Fund (RCTF), American Cancer Society Institutional Research Grant (850001-13-IRG) and a grant from NIH/NIAID RO-1 AI49367 awarded to A.P.S.