ABSTRACT
Intracellular bacterial pathogens frequently inhibit host cell apoptosis to ensure survival of their host, thereby allowing bacterial propagation. The obligate intracellular pathogen Coxiella burnetii displays antiapoptotic activity which depends on a functional type IV secretion system (T4SS). Accordingly, antiapoptotic T4SS effector proteins, like AnkG, have been identified. AnkG inhibits pathogen-induced apoptosis, possibly by binding to the host cell mitochondrial protein p32 (gC1qR). However, the molecular mechanism of AnkG activity remains unknown. Here, we demonstrate that ectopically expressed AnkG associates with mitochondria and traffics into the nucleus after apoptosis induction, although AnkG lacks a predicted nuclear localization signal. We identified the p32 interaction region in AnkG and constructed an AnkG mutant (AnkGR22/23S) unable to bind to p32. By using this mutant, we found that intracellular localization and trafficking of AnkG into the nucleus are dependent on binding to p32. Furthermore, we demonstrated that nuclear localization of AnkG but not binding to p32 is required for apoptosis inhibition. Thus, the antiapoptotic activity of AnkG is controlled by p32-mediated intracellular trafficking, which, in turn, seems to be regulated by host cell processes that sense stress.
INTRODUCTION
Coxiella burnetii is the obligate intracellular bacterial agent of human Q fever, a worldwide zoonotic disease (1). Infection in humans occurs by inhalation of infectious material transmitted from domestic livestock, and as few as 10 bacteria can result in disease (2). After bacterial uptake into phagocytic cells, C. burnetii establishes a phagolysosome-like vacuole (3, 4, 5). Importantly, establishing this replicative niche requires bacterial protein synthesis (6, 7), suggesting direct involvement of bacterial proteins. In agreement with this assumption, the type IV secretion system (T4SS) was shown to be essential for intracellular replication (8, 9). The presence of the replicative C. burnetii-containing vacuole (CCV) within the cell most likely causes tremendous stress for the infected cell, as the CCV almost completely fills the host cell lumen (10). Eukaryotic cells often respond to intracellular pathogen invasion and stress induction by initiating the intrinsic apoptotic pathway as part of the innate immune defense (11).
Apoptosis is a programmed cell death pathway crucial for immune system maintenance and removal of damaged or infected cells (12). Two main pathways lead to apoptosis. The extrinsic cell death pathway is launched in response to stimulation of death receptor proteins at the cell surface by extracellular stimuli, while the intrinsic cell death pathway is initiated in response to intracellular stimuli (13).
Apoptosis allows pathogen clearance without inflammation and additionally leads to activation of the adaptive immune defense (14, 15). As a countermeasure, intracellular pathogens have developed multiple mechanisms to inhibit host cell apoptosis (16). C. burnetii also interferes with host cell apoptosis (17, 18). How this occurs mechanistically is incompletely understood, but effector proteins translocated into the host cell by the T4SS are required for protection against apoptosis (8). Importantly, C. burnetii possesses several antiapoptotic effector proteins, such as CaeA and CaeB (19) and AnkG (20). How exactly AnkG interferes with the host cell apoptotic machinery has been unknown to date. However, the antiapoptotic activity of AnkG correlates with binding to p32, because only the N-terminal fragment of AnkG (amino acids [aa] 1 to 69), which interacts with p32, inhibits apoptosis, while the C-terminal fragment (aa 70 to 338) neither interacts with p32 nor interferes with host cell death. Reducing the level of p32 in mammalian cells made them more resistant to apoptosis, suggesting that p32 is a proapoptotic protein and that AnkG might function by interfering with this p32-mediated proapoptotic activity (20).
Several questions regarding AnkG's function remained open. Does AnkG influence p32 expression? Is the AnkG-p32 interaction direct or indirect? Is the binding to p32 necessary for AnkG-mediated inhibition of apoptosis?
To address these questions, we have defined the p32-binding pocket within AnkG and created an AnkG mutant that does not bind to p32. Using this and several other mutants, we demonstrated that AnkG activity is controlled by p32-mediated trafficking, which, in turn, seems to be regulated by cellular stress.
MATERIALS AND METHODS
Reagents, cell lines, and bacterial strains.Unless otherwise noted, chemicals were purchased from Sigma-Aldrich. Complete protease inhibitor cocktail mixture and Xtreme Gene 9 transfection reagent were from Roche. Protein A/G Sepharose was from Santa Cruz. Staurosporine was from Cell Signaling. Cell lines were cultured at 37°C in 5% CO2 in media containing 10% heat-inactivated fetal bovine serum (Biochrom) and 1% penicillin-streptomycin (Invitrogen). CHO-FcR cells were grown in minimal essential medium alpha medium (Invitrogen); HeLa and HEK293 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen). Bone marrow-derived dendritic cells (DCs) from C57BL/6 mice were prepared as described previously (21). Escherichia coli strains DH5α and BL21(DE3) were cultivated in Luria-Bertani (LB) broth supplemented with kanamycin or ampicillin where appropriate. Legionella pneumophila serogroup 1 ΔflaA strains were grown as described previously (20).
Plasmids and primers.Plasmids and primers used are listed in Tables 1 and 2.
Plasmids used in this study
Primers used in this study
Plasmid construction.For creation of the constructs AnkG1–91-pCMV-HA, AnkG50–338-pCMV-HA, AnkG1–157-pCMV-HA, and AnkF-pCMV-HA, the genes were amplified from C. burnetii Nine Mile phase II clone 4 genomic DNA by PCR using the primers listed in Table 2, restricted with the enzymes indicated in Table 2, and ligated with likewise-restricted pCMV-HA. For creation of the constructs AnkGR23S-pCMV-HA and AnkGR22/23S-pCMV-HA, the genes were amplified from AnkGFL-pCMV-HA with primers listed in Table 2 that were 5′ phosphorylated. The PCR constructs were gel purified and ligated. For cloning of the constructs AnkG1–69-pEGFP and AnkG70–338-pEGFP, the genes were amplified from AnkG1–69-pJV400 or AnkG70–338-pJV400 using the primers listed in Table 2, restricted with the enzymes indicated, and ligated with likewise-restricted pEGFP. For creation of the constructs AnkGR22/23S-pJV400, NES-AnkG-pJV400, and NLS-AnkGR22/23S-pJV400, the genes were amplified from AnkGR22/23S-pEGFP or NES-AnkG-pEGFP using the primers listed in Table 2, restricted as indicated, and ligated with likewise-restricted pJV400. For cloning of the construct NES-AnkG-pGEFP, the gene was amplified from AnkGFL-pEGFP using the primers listed in Table 2, restricted with the indicated enzymes, and ligated with likewise-restricted pEGFP. For cloning of the construct NLS-AnkGR22/23S-pEGFP, the gene was amplified from AnkGR22/23S-pCVM-HA using the primers listed in Table 2, restricted with the indicated enzymes, and ligated with likewise-restricted pEGFP.
Confocal microscopy.CHO cells were plated on coverslips and were transfected with the plasmids indicated below. The cells were fixed with 4% paraformaldehyde (Alfa Aeser) in phosphate-buffered saline (PBS; Biochrom), permeabilized with ice-cold methanol, and quenched with 50 mM NH4Cl (Roth) in PBS. The cells were mounted using ProLong gold with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen) to visualize the nucleus. For mitochondrial staining, the cells were incubated using Mitotracker (Molecular Probes) before fixation. Confocal fluorescence microscopy was performed using a Zeiss LSM 700 confocal microscope.
Nuclear fragmentation assays.Nuclear fragmentation assays were performed as described previously (19).
Coimmunoprecipitation.HEK293 cells were transiently transfected with the plasmids indicated below. On the following day, the cells were washed with PBS and incubated with lysis buffer (20 mM HEPES [pH 7.5], 200 mM NaCl, 1 mM EDTA, 0.1% [vol/vol] Nonidet P-40, 10% [vol/vol] glycerol, 1× protease inhibitor, 1 mM dithiothreitol [DTT]) for 30 min on ice. After centrifugation, the supernatants were incubated with anti-green fluorescent protein (anti-GFP) rabbit serum from Invitrogen for 2 h at 4°C. Complexes were precipitated by adding protein A/G Plus agarose and incubated for 45 min at 4°C. The beads were washed three times with washing buffer (20 mM HEPES [pH 7.5], 100 mM NaCl, 1 mM EDTA, 0.1% [vol/vol] Nonidet P-40), and samples were analyzed.
Protein purification.E. coli BL21(DE3) cells transformed with plasmids producing glutathione S-transferase (GST), GST-AnkG, or His-p32 were grown in LB broth containing ampicillin. Isopropyl-β-d-thiogalactopyranoside (IPTG) was added to the media, and samples were incubated for 4 h at 30°C. The cells were resuspended in PBS containing protease inhibitor. After disruption by French press, the lysate was incubated in 1% Triton X-100 for 1 h at 4°C. Lysates were clarified by centrifugation at 15,000 × g for 30 min. Proteins were purified using glutathione-Sepharose or nickel-nitrilotriacetic acid (Ni-NTA) agarose columns.
GST tag pulldown.Purified GST or GST-AnkG was loaded onto glutathione-Sepharose columns (GE Healthcare), and purified His-p32 was added to the columns. The columns were washed three times with PBS, and bound proteins were eluted with 10 mM glutathione in PBS (pH 9.0). The input, the eluate, and the bead fractions were analyzed as indicated below.
His tag pulldown.Purified His-p32 was loaded onto Ni-NTA agarose columns (GE Healthcare), and GST or GST-AnkG was added to the columns. The columns were washed with increasing concentrations of imidazole in lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM DTT), and bound proteins were eluted with 500 mM imidazole in lysis buffer. Eluate and input fractions were analyzed as indicated below.
Statistical analysis.The unpaired Student t test was used for statistical analysis.
Legionella pneumophila ΔflaA strain infection.Dendritic cells derived from C57BL/6 mice were infected with L. pneumophila ΔflaA strains containing the plasmids indicated below as described previously (20). Two and 10 h after infection, cells were lysed and plated on charcoal-yeast extract plates. The plates were incubated for 3 days at 37°C, and CFU were counted. Colony number after 2 h of infection represents the infection efficiency; that after 10 h reflects the survival of the intracellular bacteria.
RESULTS
AnkG binds p32 directly.To analyze the interaction of AnkG with p32, we first determined whether the binding is direct or indirect. Typically, GST pulldown experiments are used to verify direct interactions between two proteins. Thus, we expressed and purified GST, GST-tagged AnkG, and His-tagged p32 from Escherichia coli. Purified GST or GST-AnkG was incubated with His-p32, and the putative protein complex was pulled down with glutathione-coated Sepharose beads. The eluate and bead fractions were subjected to SDS-PAGE and stained with Coomassie blue (Fig. 1A). Additionally, we analyzed eluate and bead fractions by immunoblot analysis (Fig. 1B). As shown in Fig. 1A and B, His-p32 is pulled down by GST-AnkG but not by GST alone. To confirm the direct interaction, we also performed the reverse experiment. Thus, purified GST or GST-AnkG and His-p32 were incubated and His-coupled proteins were pulled down with nickel-NTA-coated agarose beads. Immunoblot analysis revealed that His-p32 pulled down GST-AnkG (Fig. 1C) but not GST (data not shown). Therefore, binding of AnkG to p32 is direct, because no additional proteins were needed for this interaction.
AnkG binds directly to the host cell protein p32 and does not alter its steady-state protein level. (A) Glutathione-Sepharose columns with GST-AnkG or GST alone were incubated with His-p32. Eluate (E1 to E4) and bead fractions were resolved by SDS-PAGE and stained with Coomassie blue. (B) Glutathione-Sepharose columns with GST-AnkG or GST alone were incubated with His-p32. Input, eluate, and bead fractions were subjected to immunoblot analysis using anti-GST and anti-p32 antibodies. (C) Ni-NTA agarose columns with His-p32 were incubated with GST or with GST-AnkG. Eluate and input were subjected to immunoblot analysis using anti-GST and anti-His antibodies. (D) HEK293 cells were transfected with plasmids encoding GFP or GFP-tagged AnkG. Protein extracts were separated by SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and probed with antibodies directed against GFP, p32, and actin. One representative immunoblot out of at least three independent experiments is shown. (E) HeLa cells were transiently transfected with plasmids encoding GFP or GFP-tagged AnkG. The cells were treated with Mitotracker (red), followed by fixation and permeabilization. p32 was stained with a specific primary antibody and a secondary dye 405-labeled antibody (blue).
AnkG does not alter the p32 steady-state protein level.AnkG was suggested to mediate its antiapoptotic activity by blocking p32 function (20). Therefore, we first analyzed whether the expression of AnkG results in a reduced p32 protein level. Thus, the respective p32 protein level of cells ectopically expressing GFP or GFP-AnkG was analyzed by immunoblotting using an anti-p32 antibody. As shown in Fig. 1D, the p32 protein level was not altered by GFP-AnkG expression, suggesting that AnkG does not act by changing the steady-state protein level of p32. Furthermore, AnkG expression did not cause any changes in the intracellular distribution of p32 (Fig. 1E).
AnkG associates with mitochondria and traffics into the nucleus after stress induction.The host cell protein p32 is mainly found in the mitochondria (22, 23) and a small fraction in the nucleus (24). In order to address the question of where the interaction between AnkG and p32 occurs within the cell, we analyzed the intracellular localization of ectopically expressed GFP-AnkG. As demonstrated in Fig. 2A, GFP-AnkG showed vesicular staining with close association with host cell mitochondria.
Intracellular localization of AnkG. (A) Representative immunofluorescence micrographs show HeLa cells transiently transfected with a plasmid encoding GFP-tagged AnkG (green). The cells were treated with Mitotracker (red), followed by fixation, permeabilization, and staining of the nuclei with DAPI (blue). (B) CHO-FcR cells transiently transfected with GFP-tagged AnkG were incubated with 2 μM staurosporine. At the indicated time points, cells were fixed and the intracellular localization of AnkG was analyzed in at least 100 transfected cells per sample using confocal microscopy from eight independent experiments. ***, P < 0.001. n.s., not significant (P = 0.055). (C) Representative immunofluorescence micrographs show CHO-FcR cells expressing GFP-tagged AnkG, AnkG1–69, or AnkG70–338 (green). The cells were incubated with staurosporine, followed by fixation, permeabilization, and staining of the nuclei with DAPI (blue).
The distribution of p32 is altered by perturbation of the physiological state of the cell (23–25). As AnkG interacts with p32, we asked whether AnkG also alters its intracellular localization after cellular stress induction. Thus, we treated GFP-AnkG-expressing cells for different time periods with staurosporine to cause cellular stress and analyzed subsequently the intracellular localization of AnkG by immunofluorescence. Before treatment, the majority of GFP-AnkG was localized in close association with the mitochondria, with less localized to the nucleus, although AnkG does not contain a predicted nuclear localization signal. After treatment with staurosporine, the intracellular localization of GFP-AnkG changed. After 4 h, GFP-AnkG was mainly present within the nucleus, and only a minority remained in close association with the mitochondria (Fig. 2B). These results demonstrate that AnkG traffics into the nucleus after apoptosis induction. Furthermore, it suggests that AnkG requires binding to p32 or another host cell protein to get transported into the nucleus, as AnkG does not contain a predicted nuclear localization.
The amino-terminal fragment AnkG1–69 contains one or more regions necessary for inhibition of apoptosis and for binding to p32, whereas AnkG70–338 neither binds to p32 nor inhibits apoptosis (20). If the change in intracellular localization depends on binding to p32, the intracellular localization of AnkG70–338 should not change after staurosporine treatment. Thus, we analyzed the intracellular localization of AnkG1–69 and AnkG70–338 after cellular perturbation. As shown in Fig. 2C, AnkG1–69 was mainly localized in the nucleus under healthy and apoptotic conditions. The nuclear localization of GFP-AnkG1–69 under healthy conditions might be due to its small size, 34 kDa. GFP-AnkG1–69 can freely migrate into the nucleus and is most likely actively retained within the nucleus. In contrast, AnkG70–338 was mainly localized in the cytoplasm, and this localization did not change after treatment with staurosporine. These results led to the hypothesis that trafficking of AnkG might depend on binding to p32 and that nuclear localization might be important for AnkG-mediated apoptosis inhibition. However, to prove the first hypothesis, it was necessary to generate an AnkG mutant unable to bind to p32.
An arginine-rich region within AnkG is required for binding to p32.To narrow down the region within AnkG required for binding to p32, we generated different AnkG truncations. We expressed HA-tagged AnkG truncations and GFP-p32 in HEK293 cells, precipitated proteins from the cell lysates with an anti-GFP antibody, and evaluated the coimmunoprecipitation of the different AnkG truncations by immunoblot analysis. As shown in Fig. 3A, HA-AnkG, HA-AnkGΔAnk, HA-AnkG1–157, HA-AnkG1–91, and HA-AnkG1–69, but not HA-AnkF, HA-AnkG70–338, HA-AnkG50–338, or HA-AnkG29–338, coprecipitated with GFP-p32. Thus, the first 28 aa of AnkG are most likely required for binding to p32. This N-terminal part contains seven arginine residues. Because p32 was shown to bind to arginine-rich regions (26, 27), we generated point mutations within this arginine-rich N-terminal part, replacing arginine with serine. To analyze binding of the AnkG mutants to p32, coimmunoprecipitation was performed. While HA-AnkG and HA-AnkGR23S coprecipitated with GFP-p32, HA-AnkGR22/23S did not (Fig. 3B). Hence, we identified the p32-binding region within AnkG and generated an AnkG mutant unable to bind to p32.
Identification of the p32 binding site. (A and B) HEK293 cells were cotransfected with plasmids encoding GFP-tagged p32 and the indicated HA-tagged AnkG mutants (HA-tagged AnkF was used as a negative control). The proteins were precipitated from the cell lysates with an anti-GFP antibody. Immunoblot analysis was used to detect p32 (anti-GFP) and Ank proteins (anti-HA) in the lysates (pre-IP) and precipitates (IP). (C) Representative immunofluorescence micrographs show HeLa cells expressing GFP-tagged AnkGR22/23S (green). The cells were fixed, permeabilized, and stained with antitubulin antibody (red) and DAPI (blue). (D) CHO-FcR cells expressing GFP-tagged AnkGR22/23S were incubated with 2 μM staurosporine. At the indicated time points, cells were fixed and the localization of AnkG was analyzed in at least 100 transfected cells per sample from four independent experiments using confocal microscopy. n.s., not significant.
AnkG intracellular localization and trafficking depend on p32 binding.Next, we analyzed the intracellular localization of the AnkG mutant AnkGR22/23S by immunofluorescence. As shown in Fig. 3C, ectopically expressed GFP-AnkGR22/23S colocalized with α-tubulin, suggesting that intracellular localization of AnkG is dependent on p32 binding. To analyze whether intracellular trafficking of AnkG also depends on p32 binding, we treated GFP-AnkGR22/23S-expressing cells for different time periods with staurosporine and analyzed the intracellular localization of AnkGR22/23S by immunofluorescence. The majority of GFP-AnkGR22/23S colocalized with α-tubulin, with less localized to the nucleus. Importantly, the intracellular localization of GFP-AnkGR22/23S was not changed by treatment with staurosporine (Fig. 3D). Taken together, the results show that nuclear localization and trafficking of AnkG depend on its binding to p32.
AnkG has to migrate into the nucleus to inhibit apoptosis.Having demonstrated that AnkG traffics into the nucleus after apoptosis induction, our goal was to determine whether this nuclear localization is essential for the antiapoptotic activity of AnkG. Consequently, we expressed a chimera comprising GFP-AnkG fused to the nuclear export signal of the HIV-1 Rev protein (GFP-NES-AnkG). This nuclear export signal has been used successfully to prevent nuclear import of the Golgi vesicle tethering protein p115 (28). GFP-NES-AnkG was excluded from the nucleus (Fig. 4A) but still bound to p32 (Fig. 4B). As shown in Fig. 4C, GFP-NES-AnkG was present exclusively in the cytoplasm and did not migrate into the nucleus after apoptosis induction (Fig. 4C). Thus, GFP-NES-AnkG can be used to analyze whether AnkG nuclear localization is required for apoptosis inhibition. Therefore, we ectopically produced GFP, GFP-AnkG, and GFP-NES-AnkG transiently in CHO cells and treated the cells with staurosporine to induce cell death. Nuclear fragmentation was visualized by DAPI staining and counted to measure apoptosis. Whereas 35% of cells expressing GFP had fragmented nuclei, this number was reduced to 20% in cells expressing GFP-AnkG (Fig. 4D). Importantly, 40% of cells expressing GFP-NES-AnkG had fragmented nuclei, demonstrating that NES-AnkG does not inhibit apoptosis. Thus, the antiapoptotic activity of AnkG strictly requires its translocation into the nucleus.
AnkG has to migrate into the nucleus to inhibit staurosporine-induced apoptosis. (A) Representative immunofluorescence micrograph showing CHO-FcR cells expressing GFP-tagged NES-AnkG (green). The cells were treated with Mitotracker (red), followed by fixation, permeabilization, and staining of the nuclei with DAPI (blue). (B) HEK293 cells were cotransfected with plasmids encoding the indicated GFP-tagged AnkG mutants or GFP as a negative control. The proteins were precipitated from the cell lysates with an anti-GFP antibody. Immunoblot analysis was used to detect endogenous p32 (anti-p32) or AnkG (anti-AnkG) in the lysates (pre-IP) and precipitates (IP). (C) CHO-FcR cells expressing GFP-tagged NES-AnkG were incubated with 2 μM staurosporine. At the indicated time points, cells were fixed and the localization of AnkG was analyzed in at least 100 transfected cells per sample from three independent experiments using confocal microscopy. n.s., not significant. (D) CHO-FcR cells expressing GFP, GFP-AnkG, or GFP-NES-AnkG were treated with staurosporine for 4 h. The cells were fixed and permeabilized, and the nuclei were stained with DAPI. The nuclear morphology of at least 100 GFP-expressing cells was scored in four independent experiments. n.s., not significant. *, P < 0.02.
Neither AnkGR22/23S nor NES-AnkG prevents pathogen-induced apoptosis.Next, we asked whether AnkG delivered into the host cell by the T4SS also depends on nuclear localization to exert its antiapoptotic activity. Because C. burnetii harbors several antiapoptotic effector proteins, the construction of an ankG deletion mutant complemented or not with nes-ankG might not provide an answer to this question. Instead, we employed a gain-of-function analysis using a Legionella pneumophila ΔflaA strain to determine whether translocation of different AnkG mutants could prevent apoptosis. The L. pneumophila ΔflaA strain caused rapid apoptosis in mouse bone marrow-derived dendritic cells (DCs) and thus could not replicate in these cells (21). These pathogen-induced incidents were blocked by adding AnkG to the repertoire of L. pneumophila effector proteins (20). Therefore, this model can be used to analyze whether AnkG has to bind to p32 or whether AnkG has to migrate into the nucleus to inhibit pathogen-induced apoptosis. We infected DCs with the L. pneumophila ΔflaA strain containing either the empty vector (pJV400), AnkG (pJV400-AnkG), NES-AnkG (pJV400-NES-AnkG), or AnkGR22/23S (pJV400-AnkGR22/23S). At 2 h and 10 h postinfection, the cells were lysed and bacterial CFU were counted. As shown in Fig. 5A, bacterial uptake was not affected by the addition of AnkG or any of the AnkG mutants. At 10 h postinfection, only 12% of the initial inoculum of L. pneumophila ΔflaA strain containing vector alone was recovered, suggesting that these bacteria induce apoptosis in their host cells and thus are not able to survive and replicate (Fig. 5B). In contrast, nearly 50% of the L. pneumophila ΔflaA strain encoding AnkG were recovered, suggesting that AnkG delivered into the host cell by the L. pneumophila T4SS is able to disrupt pathogen-induced apoptosis in DCs, in agreement with a previous report (20). Neither the L. pneumophila ΔflaA strain encoding NES-AnkG nor the L. pneumophila ΔflaA strain encoding AnkGR22/23S seemed to inhibit pathogen-induced apoptosis in DCs, as demonstrated by recovery rates of less than 10%. These results support our previous findings and suggest that AnkG depends on binding to p32 for proper localization and trafficking into the nucleus and that the nuclear localization is essential for inhibition of host cell apoptosis.
Neither AnkGR22/23S nor NES-AnkG can prevent pathogen-induced apoptosis. (A and B) Dendritic cells were infected with Legionella pneumophila ΔflaA strains containing the indicated plasmids. The data shown are from one representative experiment of three experiments with similar results. Shown are the bacterial uptake 2 h after infection (A) and the relative number of intracellular bacteria 10 h after infection compared to the 2-h value (B). *, P = 0.001. (C) Representative immunofluorescence micrograph showing CHO-FcR cells expressing GFP-NLS-AnkGR22/23S (green) incubated with Mitotracker (red), followed by fixation, permeabilization, and staining of the nuclei with DAPI (blue). (D) CHO-FcR cells expressing GFP, GFP-AnkG, or GFP-NLS-AnkGR22/23S were treated with 2 μM staurosporine for 4 h. After treatment, the cells were fixed and permeabilized and the nuclei were stained with DAPI. The nuclear morphology was scored of at least 100 GFP-expressing cells in three independent experiments. *, P < 0.01.
The intracellular trafficking, but not the antiapoptotic activity of AnkG, depends on binding to p32.The previous experiments did not clarify whether the binding to p32 is also necessary for AnkG-mediated antiapoptotic activity. To address this question, we constructed a chimera by fusing the simian virus 40 (SV40) large T antigen nuclear localization signal (29) to the amino terminus of AnkGR22/23S (GFP-NLS-AnkGR22/23S). Ectopic expression of this construct displays nuclear localization (Fig. 5C). Next, we ectopically produced GFP, GFP-AnkG, and GFP-NLS-AnkGR22/23S transiently in CHO cells and treated the cells with staurosporine to induce cell death. Nuclear fragmentation was visualized by DAPI staining and counted to measure apoptosis. Whereas 35% of cells expressing GFP had fragmented nuclei, this number was reduced to 23% in cells expressing GFP-AnkG (Fig. 5D). Importantly, 22% of cells expressing GFP-NLS-AnkGR22/23S had fragmented nuclei. Thus, AnkG depends on binding to p32 for proper localization and trafficking but not for antiapoptotic activity.
DISCUSSION
The elimination of infected cells via apoptosis is an evolutionarily conserved defense mechanism (30). So it is not surprising that many intracellular pathogens have developed mechanisms to counter apoptosis induction by their host cells (16). Several intracellular pathogens inject effector proteins into the host cell to prevent premature host cell death. However, their molecular mechanisms of action are distinct (31). In this study, we analyzed the antiapoptotic activity of AnkG. We showed that the effector protein AnkG localizes in association with the host cell mitochondria in unstressed cells (Fig. 2A). This is in contrast to a report showing that mCherry-AnkG colocalized with microtubules (32). The difference in localization of AnkG cannot be explained by the cell line used, because both studies used HeLa cells. The only other difference is the tag used. However, we have not detected any tag-dependent differences in the intracellular localization of AnkG so far. GFP-, HA- and myc-tagged AnkG proteins all displayed the same intracellular localization in HeLa and CHO-FcR cells (data not shown). Interestingly, ectopically expressed GFP-AnkGR22/23S colocalized with tubulin (Fig. 3C) and thus displayed the same intracellular localization as reported for mCherry-AnkG (32). This localization is surprising, as one would predict that AnkG unable to bind p32 would display cytoplasmic localization. Furthermore, after staurosporine treatment, GFP-NES-AnkG displays partial colocalization with tubulin (data not shown). Therefore, it can be speculated that microtubule association might play a role in AnkG activity under certain cellular conditions.
There are several antiapoptotic type III or type IV secretion system effector proteins that target the host cell mitochondria, the central organelle of the intrinsic apoptotic pathway. Such targeting of the mitochondria by bacterial proteins seems to be evolutionarily conserved, as plant pathogens also target the mitochondria to suppress the hypersensitive response, a form of programmed cell death (33). As shown in Fig. 2A, AnkG only partially colocalized with mitochondria, suggesting that this effector protein is not transported into the mitochondria, as has been shown for Ats1 and PorB. Ats1 from Anaplasma phagocytophilum uses the mitochondrial import machinery to get transported into the mitochondria (34), while the meningococcal PorB associates with a porin located in the outer mitochondrial membrane (35). Importantly, the antiapoptotic activity of Ats1 correlates with mitochondrial import (34). AnkG, in clear contrast, has to get transported into the nucleus to act antiapoptotically (Fig. 4D). Interestingly, this transport into the nucleus, which depends on the ability of AnkG to bind to p32, happens only under apoptotic or stress conditions (Fig. 2B and 3D). This leads to the hypothesis that AnkG primarily targets the mitochondria to sense host cell apoptotic stress and then hitchhikes to the nucleus, the organelle of activity. As a consequence, it can be concluded that the activity of AnkG is adjusted by a host cell stress sensor which regulates the transport process.
For intracellular trafficking of AnkG from the mitochondria to the nucleus, and thus, for activity control, binding to p32 is essential. This is in agreement with a report that proposed that p32 is involved in bridging a signaling pathway that extends from the mitochondria to the cell nucleus (23). However, once AnkG is within the nucleus, binding to p32 is not needed for antiapoptotic activity (Fig. 5D). This result suggests that AnkG must instead interfere with a nuclear function to prevent host cell death. There are several effector proteins known to target the host cell nucleus. The Chlamydia trachomatis effector protein NUE is a histone methyltransferase targeting histones (36). AnkA from A. phagocytophilum mediates epigenetic changes at the CYBB promoter (37), leading to a global downregulation of host defense genes (38). How AnkG modulates nuclear function has yet to be determined, but the activity of AnkG is clearly regulated by host cell stress signaling and p32-dependent trafficking. This is, to our knowledge, the first example that an antiapoptotic effector protein is regulated by host cell protein-mediated trafficking.
The question of how effector proteins are regulated has only rarely been investigated. There are several avenues of regulation possible: (i) regulation by time point and dosage of translocation, (ii) regulation by modulation through other effector proteins, or (iii) regulation by host cell-dependent modification (phosphorylation, lipidation, sumoylation, etc.). Examples already exist for the last scenario. It was shown that intracellular localization and thereby the function of Legionella pneumophila effector proteins containing a CAAX motif are affected by lipidation through the host cell farnesyltransferase and class I geranylgeranyltransferase (39). The Helicobacter pylori T4SS effector protein CagA is phosphorylated by the host cell tyrosine kinases Src and Abl. Phosphorylated CagA can then modulate various signaling cascades associated with cell polarity, cell proliferation, actin-cytoskeletal rearrangements, cell elongation, disruption of tight and adherence junctions, proinflammatory responses, and apoptosis inhibition (40). In this study, we have identified a fourth possibility to regulate the activity of effector proteins: regulation by stress sensing and intracellular trafficking. In our opinion, more knowledge about host cell requirements for regulation of effector proteins is needed. This knowledge not only will help to understand microbial pathogenesis better but also will allow us to develop new strategies for therapy. The first steps down this avenue have already been taken. An exemplified study showed that identifying host cell signaling pathways required for bacterial survival might help to control infection (41).
ACKNOWLEDGMENTS
This work was supported by the Deutsche Forschungsgemeinschaft (DFG) through the Collaborative Research Initiative 796 (grant SFB796 to A.L. and C.B.) and through the Priority Programme SPP1580 (to A.L.) as well as by the ERA-NET PathoGenoMics 3rd call (to A.L.).
We thank Christian Bogdan for his valuable comments on the manuscript.
FOOTNOTES
- Received 26 September 2013.
- Returned for modification 8 October 2013.
- Accepted 7 April 2014.
- Accepted manuscript posted online 14 April 2014.
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
