|Home | About | Journals | Submit | Contact Us | Français|
Since viruses rely on functional cellular machinery for efficient propagation, apoptosis is an important mechanism to fight viral infections. In this study, we sought to determine the mechanism of cell death caused by Ebola virus (EBOV) infection by assaying for multiple stages of apoptosis and hallmarks of necrosis. Our data indicate that EBOV does not induce apoptosis in infected cells but rather leads to a nonapoptotic form of cell death. Ultrastructural analysis confirmed necrotic cell death of EBOV-infected cells. To investigate if EBOV blocks the induction of apoptosis, infected cells were treated with different apoptosis-inducing agents. Surprisingly, EBOV-infected cells remained sensitive to apoptosis induced by external stimuli. Neither receptor- nor mitochondrion-mediated apoptosis signaling was inhibited in EBOV infection. Although double-stranded RNA (dsRNA)-induced activation of protein kinase R (PKR) was blocked in EBOV-infected cells, induction of apoptosis mediated by dsRNA was not suppressed. When EBOV-infected cells were treated with dsRNA-dependent caspase recruiter (dsCARE), an antiviral protein that selectively induces apoptosis in cells containing dsRNA, virus titers were strongly reduced. These data show that the inability of EBOV to block apoptotic pathways may open up new strategies toward the development of antiviral therapeutics.
Ebola viruses are members of the filovirus family and cause severe hemorrhagic fever in humans. Among the five classified Ebola virus species, Zaire ebolavirus (EBOV) is the most pathogenic, with case fatality rates of up to nearly 90%. Due to the high lethality rate and the lack of licensed vaccines and treatments, EBOV is classified as a biosafety level 4 (BSL4) pathogen and a NIAID category A priority pathogen (1). EBOV productively infects a variety of cells and tissues, with monocytes, macrophages, and dendritic cells representing the early target cells of infection (2). Lymphocytes remain uninfected by EBOV, but a decrease in lymphocyte numbers due to bystander apoptosis is observed during EBOV infection (3–12). Increased expression levels of inducers of apoptosis, including the tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL) and Fas ligand (FasL), have been observed during EBOV infection and might play a role in the observed bystander lymphocyte apoptosis (3, 9, 10, 13–15). Only limited and contradictory data are available about apoptosis of EBOV-infected cells (7, 8, 10, 16). Moreover, since molecular biology data on the induction and regulation of cell death in EBOV infection are lacking, it is not known if EBOV is able to control cell death signaling pathways in infected cells.
Cell death is an important host defense mechanism to eliminate infected cells and limit viral replication and propagation. Viruses have evolved multiple mechanisms to interfere with cell death signaling pathways in order to ensure efficient viral replication (17). Three major forms of cell death have been described, each showing different morphological features: apoptosis, necrosis, and autophagy-associated cell death. Necrosis is characterized by the swelling of dying cells and organelles and the lack of nuclear condensation. In contrast, characteristic features of apoptotic cells include cell shrinkage, nuclear fragmentation, and formation of apoptotic bodies (18). Autophagy is characterized by the appearance of double-membrane vesicles (autophagosomes) in the cytoplasm and the loss of organelles (18). Recent evidence suggests that necrosis, which was long believed to be uncontrolled, is also a regulated form of cell death (19, 20). Cross talk between the different cell death pathways on multiple levels creates a dynamic network controlling the induction and execution of cell death (21, 22). Apoptosis is the best-characterized form of programmed cell death and is regulated by a complex signaling network. A hallmark of apoptosis is the activation of cysteine aspartate-specific proteases, the caspases (23). Apoptosis can be induced by members of the TNF family or FasL via the extrinsic pathway. Binding of the ligands to their receptors leads to the formation of the death-inducing signaling complex (DISC) and subsequent activation of initiator caspase 8 and downstream caspases (24, 25). Cellular stresses induce apoptosis by signaling via the mitochondria (intrinsic pathway), characterized by the permeabilization of the mitochondrial membrane and efflux of cytochrome c into the cytoplasm, leading to activation of caspase 9. Induction of apoptosis by the intrinsic pathway hinges on the balance of pro- and antiapoptotic members of the Bcl-2 protein family, which control cytochrome c release from the mitochondria (26). After activation by cleavage, effector caspases process their protein substrates, e.g., DNA repair components like poly(ADP-ribose) polymerase 1 (PARP-1) or structural proteins, resulting in the fragmentation of the cell and the nucleus (23).
An important pathogen-associated molecular pattern that plays a role in the induction of apoptosis is double-stranded RNA (dsRNA). dsRNA is a viral replication intermediate and is recognized by a variety of cellular sensors, leading to induction of antiviral signaling pathways, including activation of dsRNA-dependent protein kinase R (PKR) (27). Activated PKR phosphorylates eukaryotic initiation factor 2α (eIF2α), leading to translational arrest (28–30). PKR has been demonstrated to regulate the induction of apoptosis after viral infection or other cellular stress situations, resulting in activation of caspases 8 and 9 (29, 31). However, the exact mechanism of PKR-mediated apoptosis remains unclear, as multiple cellular signaling pathways have been implied to be influenced by PKR activation (28, 30). EBOV blocks multiple antiviral pathways, including the interferon (IFN) response, RNA interference, and activation of PKR (32).
In this study, we address the question of whether EBOV is able to control apoptotic signaling pathways in infected cells. We show that EBOV does not induce apoptosis in infected cells, whereas necrotic cell death was observed. EBOV-infected cells remain sensitive to the induction of apoptosis by exogenous stimuli, and apoptosis signaling is not inhibited. Although antiviral signaling mediated by PKR is suppressed by EBOV, dsRNA-dependent induction of apoptosis is not blocked. We took advantage of the inability of EBOV to block apoptosis by treating infected cells with dsRNA-dependent caspase recruiter (dsCARE), an antiviral protein that induces apoptosis in cells containing dsRNA (33). Our data show that EBOV replication was inhibited by dsCARE treatment, demonstrating that the targeted induction of apoptosis is a promising approach toward the development of antiviral therapies against EBOV infection.
Vero and HeLa cells were obtained from the European Collection of Cell Culture (ECACC). VeroE6 cells were obtained from the American Type Culture Collection (ATCC). Cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS), penicillin (50 units/ml), and streptomycin (50 mg/ml). Primary human macrophages were generated from Leuko Paks (NY Biologics Inc.) or apheresed peripheral blood mononuclear cells (PBMCs; NIH Clinics Center, Department of Transfusion Medicine) using Ficoll separation (GE Healthcare) and isolation of CD14+ monocytes by magnetic bead selection (Miltenyi Biotech). A total of 1.5 × 106 or 3 × 105 CD14-selected monocytes were seeded into 6-well or 24-well culture plates in RPMI medium without serum and allowed to adhere for 15 min to 1 h. Afterwards, nonattached cells were removed, and RPMI medium containing 5% human AB serum was added. Monocytes were incubated for 6 to 8 days to ensure differentiation into macrophages before infection.
EBOV (Mayinga and Kikwit isolates), recombinant EBOV expressing enhanced green fluorescent protein (EBOV-EGFP) (34, 35), and vesicular stomatitis virus (VSV) were grown in VeroE6 cells. Sendai virus (SeV) strain Cantell was raised in 11-day-old embryonated chicken eggs. Virus titers were determined by a 50% tissue culture infectious dose (TCID50) assay (TCID50/ml) for EBOV, a plaque assay (PFU/ml) for EBOV and VSV, enzyme-linked immunosorbent assay (ELISA) titers (ELISA unit [EU]/ml) for influenza A virus, and a hemagglutination test (hemagglutinin units [HA]/ml) for SeV. All work with EBOV was done under biosafety level 4 conditions at the Institute of Virology, University of Marburg, Germany; the Integrated Research Facility in the Rocky Mountain Laboratories, Division of Intramural Research, NIAID, NIH, Hamilton, MT; the Texas Biomedical Research Institute, San Antonio, TX; or the State Research Center of Virology and Biotechnology, VECTOR, Koltsovo, Novosibirsk Region, Russia.
Cells were plated into 6-well culture plates 1 day prior to infection. For annexin V and fluorochrome inhibitor of caspase (FLICA) staining, detection of caspase cleavage and Bcl-2 proteins, reverse transcription-PCR (RT-PCR), and trypan blue staining, 5 × 104 or 105 Vero or HeLa cells were seeded. For detection of phosphorylated proteins (PKR and eIF2α), 4 × 104 Vero or HeLa cells were seeded. For immunofluorescence analysis, 104 Vero cells or 1.5 × 105 primary human macrophages were seeded into chamber slides (μ-Slide, 8 well; Ibidi) or on coverslips in 6-well plates. Infection of cells was performed in DMEM with 2% fetal bovine serum (FBS) or RPMI with 5% AB serum for primary macrophages. Recombinant human TRAIL (Calbiochem) or camptothecin (CPT) (Sigma-Aldrich) was added at the indicated concentrations to the culture medium to induce apoptosis. Dimethyl sulfoxide (DMSO) was used as a control. For dsRNA treatment, cells were transfected with 1 μg poly(I·C) (pIC) (Sigma-Aldrich), using Lipofectamine LTX reagent (Invitrogen) according to the manufacturer's protocol. As a control, cells were transfected with 1 μg of the empty pCAGGS plasmid.
At the indicated time points, cells grown on coverslips or in chamber slides were fixed and inactivated with 4% paraformaldehyde for 24 h. Cells were permeabilized by using 0.1% Triton X-100 for 10 min. Primary human macrophages were permeabilized with a mixture of acetone-methanol (1:1) for 5 min at −20°C. Primary antibodies used included goat anti-EBOV serum (kindly provided by S. Becker, University of Marburg, Marburg, Germany), mouse anti-EBOV NP antibody, and rabbit anti-SeV antibody (kindly provided by W. J. Neubert, Max Planck Institute, Martinsried, Germany). Secondary antibodies were conjugated with AlexaFluor495 (Molecular Probes), AlexaFluor350 (Invitrogen), or rhodamine (Dianova). 4′,6-Diamidino-2-phenylindole (DAPI) (Sigma-Aldrich) was used for nucleus staining. Coverslips infected with recombinant EBOV (recEBOV) expressing EGFP (34) were fixed and inactivated as described above and then stained with DAPI for visualization of nuclei.
Infected and uninfected cells grown on coverslips were washed twice with PBS, stained with annexin V solution (Annexin V-FLUOS staining kit; Roche Applied Science) or annexin V staining mix (Annexin V-Cy3 apoptosis kit; BioVision) for 20 min at room temperature, and fixed with 4% paraformaldehyde in DMEM overnight at 4°C. Immunofluorescence analysis was performed as described above. For detection of active caspases, infected and uninfected cells grown on glass coverslips were incubated in DMEM containing a fluorochrome inhibitor of caspases (polycaspase FLICA; Immunochemistry Technologies) for 1 h, according to the manufacturer's protocol. Cells were fixed and analyzed as described above.
Monolayers of infected cells were washed with culture medium, collected by using a trypsin-Versene mixture, and centrifuged for 5 min at 10,000 × g. Cell pellets were fixed in 4% paraformaldehyde in Hanks' solution for 24 h to inactivate EBOV. Cells were washed 3 times in Hanks' solution, postfixed in 1% osmium tetraoxide, dehydrated in ethanol and acetone, and embedded in Epon-araldite. Ultrathin sectioning was performed by using a diamond knife, and ultrathin sections were contrasted by uranyl acetate and lead citrate. The samples were examined by using a Jem 1400 transmission electron microscope (JEOL) at 80 kV. Digital photographs were collected by a side-mounted Veleta digital camera (Surface Imaging Systems [SIS]). All reagents were obtained from SPI Supplies.
At the indicated time points, cells were scraped into culture medium to ensure analysis of all cells, including detached apoptotic cells. To analyze caspase cleavage, cells were pelleted and lysed in 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS) buffer (Cell Signaling Technologies [CST]) supplemented with 5 mM dithiothreitol and 1 mM protease inhibitor mix (Complete; Roche) by three freeze-and-thaw cycles. For detection of phosphorylated proteins and PARP-1, whole-cell extracts were prepared in cell extraction buffer (Biosource) complemented with 1 mM protease inhibitor mix (Complete; Roche) and 0.1 μM serine/threonine phosphatase inhibitor (calyculin A; CST). After a short centrifugation step, supernatants were transferred into fresh tubes containing 2× SDS sample buffer (25% glycerol, 2.5% SDS, 125 mM Tris [pH 6.8], 125 mM dithiothreitol, 0.25% bromophenol blue). Inactivation of EBOV was performed by boiling the samples for 10 min. Samples were subjected to Western blot analysis using rabbit polyclonal antibodies to detect human caspases 3 and 9 (CST), PARP-1 (CST), and phospho-PKRThr446 (Epitomics). Mouse monoclonal antibodies were used to detect human caspase 8 (CST), PKR (BD Bioscience), eIF2α (Biosource), phospho-eIF2αSer52 (Biosource), and β-actin (Abcam). Secondary antibodies (Dianova) were conjugated with horseradish peroxidase. Detection was performed by using SuperSignal chemiluminescent substrate (Pierce/Thermo Scientific) according to the manufacturer's protocol.
For quantitative analysis of Bcl-2 protein levels and PARP-1 cleavage, the following primary antibodies were used: mouse anti-Bcl-2 (Sigma-Aldrich), rabbit anti-Bax (Serotec), mouse anti-β-actin (Abcam), and rabbit anti-PARP-1 (CST). Secondary antibodies were conjugated to IRDye800 (Rockland Immunochemicals or LI-COR) or AlexaFluor680 (Molecular Probes). Protein bands were quantified by using the Odyssey imaging system and software (LI-COR) and standardized to β-actin.
At the indicated time points, adherent cells and detached cells in the cell supernatants were collected and combined. Cell solutions were mixed with an equal volume of 0.4% trypan blue solution (Gibco), and live/dead cell numbers were counted by using a TC10 automated cell counter (Bio-Rad).
Total RNA from primary human macrophages was isolated at the indicated time points by using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. Five nanograms of RNA was analyzed by quantitative reverse transcription-PCR (qRT-PCR) using the QuantiFast SYBR green RT-PCR kit (Qiagen). Validated QuantiTect primer assays (Qiagen) for TRAIL, TRAIL receptor 1 (TRAIL-R1), TRAIL-R2, TRAIL-R3, TRAIL-R4, and β-2-microglobulin were used according to the manufacturer's protocol. qRT-PCR was performed with a CFX96 Real-Time PCR cycler (Bio-Rad). β-2-Microglobulin was used as an endogenous control for normalization. Fold expression levels compared to those in noninfected samples collected at 3 h postinfection (p.i.) were quantified by using the ΔΔCT method (Bio-Rad CFX Manager 1.5 software).
Gamma-irradiated supernatants of EBOV-infected primary human macrophages were clarified by centrifugation and used in a Procarta human cytokine assay (Affymetrix) for detection of secreted TRAIL. Analysis was done according to the manufacturer's instructions. Mean fluorescence intensity was measured to calculate the final concentration in pg/ml by using a Bioplex200 system (Bio-Rad) and Bioplex Manager 5 software.
dsCARE and dsCAREΔRBD were constructed as previously described (33). Briefly, PCR was used to generate cDNA fragments flanked by restriction sites for the modified human immunodeficiency virus (HIV) Tat protein transduction domain (PTD) (flanked by BamHI and XhoI), the dsRNA binding domain (RBD) of PKR (flanked by XhoI and KpnI), and the caspase recruitment domain (CARD) of apoptotic protease-activating factor 1 (flanked by KpnI and HindIII). PCR products were ligated into the pRSET B expression vector (pRSET-dsCARE) in the following order: PTD-RBD-CARD.
For generation of dsCAREΔRBD, the RBD and CARD fragments were removed by restriction digestion with XhoI and HindIII from pRSET-dsCARE. PCR was used to generate the CARD flanked by restriction sites for XhoI and HindIII, and the product was ligated with XhoI/HindIII-digested pRSET-dsCARE to generate pRSET-dsCAREΔRBD. For expression of the dsCARE and dsCAREΔRBD proteins, plasmids were transformed into Escherichia coli strain BL21(DE3), and bacteria were grown and harvested as described previously (33), with the exception that lysis was performed at 4°C prior to purification of protein over nickel resin columns (GE HealthCare, USA).
A total of 5 × 104 VeroE6 cells were seeded per well in 96-well plates and infected with recEBOV expressing EGFP (34, 35) at a multiplicity of infection (MOI) of 0.01 or 3. At 1 h p.i., dsCARE was added at the indicated concentrations, and cells were incubated until analysis. Supernatants were collected for virus titer determination at the indicated times. Cells were fixed with 10% buffered formalin solution (Sigma) or 4% paraformaldehyde for 24 h, and green fluorescent protein (GFP) fluorescence was analyzed. The 50% effective concentration (EC50) values were determined with GraphPad Prism 5 software based on determined virus titers.
To analyze qRT-PCR data, Student's t test was performed by using Bio-Rad CFX Manager 1.5 software. To analyze differences in protein levels and trypan blue staining, the Student's t test analysis tool of GraphPad Prism 5 software was used. One-way analysis of variance (ANOVA) was performed by using GraphPad Prism 5 software to analyze changes in PARP-1 cleavage. Statistically significant changes are indicated by asterisks in the figures.
To address the question of whether EBOV induces apoptosis, Vero cells were infected with EBOV, and various apoptosis markers were analyzed. Since SeV is known to induce apoptosis in infected cells (36), it was used as a control for the induction of apoptosis. Translocation of phosphatidylserine shown by annexin V staining was detected in SeV-infected cells but not in EBOV-infected cells (Fig. 1A). Caspase activation during EBOV infection was analyzed by fluorochrome inhibitor of caspase (FLICA) staining of infected cells. FLICA binds to multiple activated caspases and was observed in SeV-infected cells but not in EBOV-infected cells (Fig. 1B). Some activation of caspases was detected by FLICA staining in noninfected cells in EBOV infection at 48 h postinfection (p.i.) (Fig. 1B). Single noninfected apoptotic cells were also observed by annexin V staining in samples infected with EBOV (data not shown). Nuclear fragmentation, a characteristic of late apoptotic cells, was not detected in EBOV-infected cells (Fig. 1A and andB).B). Western blot analysis was used to further investigate caspase activation and late stages of apoptosis. To ensure complete infection of cells and avoid unspecific signals from noninfected cells, cells were infected with EBOV at a multiplicity of infection (MOI) of 5, and the infection rate was determined in parallel by immunofluorescence analysis (data not shown). No caspase 8 cleavage and only weak activation of caspase 3 were observed in EBOV-infected Vero cells. Cleavage of PARP-1, resulting in an 89-kDa fragment, a characteristic of late apoptosis, was minimal in EBOV infection and occurred to the same extent as that observed for uninfected cells (Fig. 1C). In addition, caspases 3, 8, and 9 were not or only inefficiently cleaved in EBOV-infected HeLa cells, whereas cleaved caspases were detected in VSV-infected cells, which were used as a control for the induction of apoptosis (Fig. 1D) (37). These results indicate that EBOV infection does not lead to the induction of apoptosis in infected cells.
Since EBOV-infected cells show a clear cytopathic effect (CPE) late in infection, we analyzed EBOV-infected cells for apoptosis markers up to 6 days p.i. As a control for apoptosis, cells were treated with camptothecin (CPT), which causes DNA damage and activation of mitochondrial apoptotic signaling (38). CPT treatment induced translocation of phosphatidylserine, as detected by annexin V staining (Fig. 2A). Although EBOV infection induced CPE, as reflected by rounding and detachment of cells starting at 3 days p.i. (Fig. 2A, bright-field images), annexin V-positive EBOV-infected cells were not detected (Fig. 2A). To investigate late stages of apoptosis, cell extracts were analyzed for cleavage of PARP-1 (Fig. 2B). PARP-1 is the substrate of a variety of cellular proteases, including caspases, calpains, granzymes, and cathepsins (39). PARP-1 cleavage has been observed for various forms of cell death, leading to the production of different PARP-1 signature fragments that can be specifically associated with apoptosis or necrosis (39). Cleavage of PARP-1 by caspases during late stages of apoptosis generates fragments of 89 and 24 kDa. In contrast, PARP-1 cleavage during necrotic cell death gives rise to specific cleavage products ranging from 40 to 55 kDa depending on the cell type and necrotic stimulus (39, 40). Cleavage of PARP-1 to the 42-kDa fragment is mediated by cathepsins after the release of lysosomal content into the cytoplasm during necrosis (40). Only a weak signal for the apoptotic 89-kDa form of PARP-1 was observed in EBOV-infected cells, whereas higher levels of the necrotic 42-kDa fragment were detected (Fig. 2B). Quantitative Western blot analysis confirmed a minor appearance of the apoptotic product, while the amount of the necrotic cleavage product significantly increased during EBOV infection (Fig. 2C). In contrast, only the apoptotic 89-kDa fragment, but not the necrotic 42-kDa fragment, was detected in CPT-treated cells (Fig. 2B, lanes 1 and 2). These results suggest that EBOV infection leads to the induction of necrotic cell death. In addition to the apoptotic and necrotic PARP-1 fragments, a pattern of cleavage products ranging in size from about 50 to 80 kDa was observed in both infected and noninfected cells. The pattern of these fragments is consistent with that described previously for cathepsin-mediated cleavage of PARP-1 (39, 40) (Fig. 2B, asterisk).
To confirm our findings in primary target cells, human monocyte-derived macrophages were infected with EBOV and analyzed for the induction of cell death. Immunofluorescence analysis showed an initial infection rate of 60 to 70% at 1 day p.i. (Fig. 3A). At 3 days p.i. and later, the infection rate of macrophages was about 90%, and single noninfected cells were still present (Fig. 3A). Although caspase 3 was moderately cleaved in EBOV-infected macrophages (Fig. 3B), only cleavage to the necrotic, but not the apoptotic, PARP-1 fragment was significantly increased (Fig. 3C and andDD).
To further analyze whether EBOV-infected cells undergo necrotic or apoptotic cell death, we examined the morphological changes in infected cells. A characteristic feature of necrotic or autophagic cell death that is not observed during apoptosis is the vacuolization of cells (18). Formation of vacuoles in the cytoplasm of cells infected with recombinant EBOV expressing EGFP was observed late in infection. Due to the expression of EGFP in the cytoplasm, the vacuoles were easily detectable by fluorescence microscopy but were also observed by bright-field microscopy (Fig. 4A) and in cells infected with wild-type EBOV (data not shown). The morphological changes associated with different types of cell death can be observed best by electron microscopy. Vero cells were infected with EBOV and analyzed by transmission electron microscopy (TEM). As a control, cells were infected with influenza A virus. Clear signs of apoptosis were observed in influenza virus-infected cells, including condensation of chromatin with a complete loss of its nuclear structure (Fig. 4B). The cytoplasm kept its moderate electron density and granular appearance, and cellular organelles were not altered. At late times of infection, nuclear fragmentation and the appearance of apoptotic bodies were observed (data not shown). In contrast, replication of EBOV in the cytoplasm of infected cells was not associated with a visible alteration of the cell morphology at initial stages of infection (Fig. 4B). EBOV replication takes place in cytoplasmic inclusions of moderate electron density, which are composed of granular material and contain long tubular viral nucleocapsids (Fig. 4B, insets). At late times of infection, characteristic morphological features of necrosis were observed in the infected cells, including an electron-lucent cytoplasm and swelling of mitochondria and other organelles (Fig. 4B, arrows). The nuclei in EBOV-infected cells remained unaltered and did not show visible changes during infection other than slight swelling in completely destroyed cells (Fig. 4B). None of the analyzed EBOV-infected cells showed signs of chromatin condensation or nuclear fragmentation.
Taken together, our results demonstrate that EBOV does not induce apoptosis in multiple cell types, including primary human macrophages, which represent some of the first target cells during EBOV infection. Although minor activation of caspase 3 was observed, late stages of apoptosis were not detected in EBOV-infected cells. In contrast, our data suggest that EBOV-infected cells undergo necrotic cell death.
Upregulation of inducers of apoptosis, including TRAIL and Fas, has been reported for EBOV-infected patients and experimentally infected animals (3, 9–11, 13–15). However, the effect of apoptosis-inducing ligands on infected cells has not yet been investigated. Primary human macrophages were infected with EBOV, and levels of secreted TRAIL were determined. Increasing levels of TRAIL were detected over the course of infection (Fig. 5A). Upregulation of TRAIL mRNA was also detected (data not shown). Induction of apoptosis by TRAIL has been demonstrated to be mediated by binding of TRAIL to cell surface receptors, inducing activation of caspase 8 and subsequent caspase 3 cleavage (24). In addition, TRAIL decoy receptors lacking the intracellular domain needed for signaling are known to be involved in controlling the induction of apoptosis (25). To investigate if EBOV infection leads to changes in the expression levels of TRAIL receptors (TRAIL-Rs), RNA from primary human macrophages infected with EBOV was analyzed by qRT-PCR. No statistically significant changes in mRNA levels were detected for the death receptor TRAIL-R2 or for the decoy receptors TRAIL-R3 and TRAIL-R4 compared to noninfected cells (Fig. 5B). A slight decrease in the level of death receptor TRAIL-R1 mRNA was observed at 48 h in EBOV-infected cells. To analyze the effect of exogenous TRAIL on EBOV-infected cells, Vero cells were infected with EBOV and treated with TRAIL at 24 h p.i. While EBOV infection did not lead to caspase activation, similar levels of caspase cleavage were observed in EBOV-infected cells treated with TRAIL compared to those in noninfected cells treated with TRAIL, indicating that TRAIL-induced apoptosis is not blocked by EBOV (Fig. 5C).
In addition to receptor-mediated signaling pathways, apoptosis can be induced through the intrinsic mitochondrial pathway. Mitochondrion-induced apoptosis is regulated by the balance of pro- and antiapoptotic members of the Bcl-2 protein family (26). To analyze whether EBOV is able to interfere with intrinsic apoptosis signaling pathways, the effect of CPT on EBOV-infected cells was determined. CPT causes DNA damage and an increased protein expression ratio of Bax (proapoptotic) to Bcl-2 (antiapoptotic) (38). Vero cells infected with EBOV and treated with CPT at 24 h p.i. were analyzed for Bcl-2 and Bax protein levels by using quantitative Western blot analysis. While the Bax-to-Bcl-2 ratio was not changed in untreated EBOV-infected cells, CPT treatment led to a significant decrease in Bcl-2 levels in both noninfected and EBOV-infected cells compared to nontreated control cells (Fig. 6A and andB).B). Similar results were obtained for Bax and Bcl-2 mRNA levels (data not shown). CPT-treated cells with decreased levels of Bcl-2 underwent apoptosis, as shown by caspase 3 and apoptotic PARP-1 cleavage, indicating that CPT-induced apoptosis cannot be prevented by EBOV (Fig. 6C). To compare the numbers of apoptotic cells in CPT-treated EBOV-infected and noninfected samples, we performed trypan blue staining. Treatment with CPT reduced viability of noninfected and EBOV-infected cells to 48% and 51%, respectively (Fig. 6D). Taken together, our data indicate that EBOV infection does not protect cells from apoptosis induced by exogenous stimuli activating different apoptosis signaling pathways.
EBOV VP35 blocks the activation of PKR, a central player in the host defense against viral infection (41, 42). Activation of PKR by dsRNA leads to translational arrest via phosphorylation of the translation initiation factor eIF2α, which may finally result in the induction of apoptosis (28–30). To address the question of whether inhibition of PKR activation in EBOV-infected cells accounts for the observed lack of apoptosis, cells were infected with SeV, EBOV, or both viruses, and activation of PKR and caspases was examined. SeV strain Cantell was used for these studies because it contains defective interfering genomes which trigger dsRNA-mediated antiviral signaling (43). SeV infection of Vero cells induced PKR activation, as shown by using a phospho-specific PKR antibody (Fig. 7A, lane 2). In addition, SeV infection led to cleavage of caspases 3 and 8 (Fig. 7A, lane 2). Neither PKR phosphorylation nor caspase cleavage was detected in EBOV-infected cells (Fig. 7A, lane 3). When cells were infected with EBOV and subsequently superinfected with SeV, PKR phosphorylation was strongly reduced compared to that in SeV-infected samples (Fig. 7A, lanes 2 and 4). However, cleavage of caspases 8 and 3 was induced in EBOV-infected cells superinfected with SeV, demonstrating that induction of apoptosis was not blocked (Fig. 7A). As cleaved caspases are rapidly degraded, Western blot quantification is not a suitable tool to quantify caspase cleavage and was therefore not performed (44, 45).
Since Vero cells lack the IFN genes (46), we confirmed our results using IFN-competent HeLa cells, with a similar outcome. EBOV infection led to the suppression of SeV-induced phosphorylation of PKR and eIF2α. Caspase 8 cleavage was observed in all samples, including noninfected and EBOV-infected cells. However, in contrast to uninfected and EBOV-infected cells, caspase 8 was almost completely cleaved in SeV-infected cells, and complete cleavage of the inactive precursor was observed in cells infected with both viruses (Fig. 7B, lanes 2 and 4).
We also used poly(I·C) (pIC) to test the ability of EBOV to inhibit dsRNA-dependent antiviral signaling. Transfection of noninfected HeLa cells with pIC resulted in activation of PKR and subsequent phosphorylation of eIF2α (Fig. 7C, left, lane 3). In addition, activation of caspases 3, 8, and 9 was observed (Fig. 7C, right, lane 3). Only small amounts of phosphorylated PKR and eIF2α were detected in EBOV-infected cells after pIC transfection, indicating suppression of PKR activity (Fig. 7C, left, lane 6). In contrast, activation of caspase 9 and complete cleavage of the precursor proteins of caspases 3 and 8 were detected after pIC transfection of EBOV-infected cells (Fig. 7C, left, lane 6). These data show that EBOV is not able to prevent dsRNA-induced apoptosis, although PKR activation is suppressed.
Since our data show that EBOV-infected cells are prone to apoptosis-stimulating agents, the question arose as to whether EBOV-infected cells could be targeted by using antivirals that selectively induce apoptosis in virus-infected cells. A promising approach toward the development of pan-antivirals takes advantage of the presence of dsRNA in many virus-infected cells. The recombinant proteins dsRNA-activated caspase oligomerizer (DRACO) and dsCARE have been reported to selectively induce apoptosis in cells containing virus-induced dsRNA, leading to the elimination of virus-infected cells and inhibition of virus spread (33, 47). Three protein domains with well-known functions are fused to form DRACO or dsCARE: (i) a protein transduction domain to allow for easy uptake into cells, (ii) a dsRNA binding domain, and (iii) a caspase recruitment domain (CARD). Binding of DRACO or dsCARE to dsRNA leads to the activation of the CARD, inducing caspase-mediated apoptosis (33, 47). Since the induction of dsRNA-dependent apoptosis was not inhibited by EBOV (Fig. 7C), we investigated the effect of dsCARE on EBOV-infected cells. Cells infected at a low MOI with recombinant EBOV expressing EGFP from an additional transcription unit (34, 35) were treated with various amounts of dsCARE, and infected cells were visualized by EGFP fluorescence on days 1 to 3 p.i. Complete infection was observed for nontreated cells after 3 days, whereas EBOV replication was impaired with increasing dsCARE concentrations (Fig. 8A). A nearly complete block of infection was observed by using 6 μg/ml dsCARE at all analyzed time points (Fig. 8A). Analysis of supernatants from EBOV-infected and dsCARE-treated cells revealed a significant reduction of virus titers. Increased concentrations of dsCARE resulted in decreased virus titers with up to a 2- or 3-log reduction using 6 μg/ml dsCARE (Fig. 8B). Based on the plaque assay results, the calculated EC50 was 1.68 ± 0.06 μg/ml, which is in the same range as the EC50 of 1.5 μg/ml calculated for the inhibition of VSV replication by dsCARE (33).
To ensure that the antiviral effect of dsCARE was based on detection of dsRNA in EBOV-infected cells, we used a dsCARE mutant lacking the dsRNA binding domain (dsCAREΔRBD) as a control. Replication of EBOV-EGFP in cells treated with dsCAREΔRBD was as efficient as in untreated cells, while the number of EGFP-expressing cells was significantly reduced in cells treated with functional dsCARE (Fig. 8C). dsCARE was also effective in cells infected with EBOV at a high MOI (Fig. 8D). As reported previously (33, 47), uninfected cells treated with dsCARE did not show obvious cytotoxic effects. Cell morphology was unchanged (Fig. 8A and data not shown), and nuclei stained with DAPI appeared normal (Fig. 8D). These results indicate that the antiviral effect of dsCARE is based on the detection of dsRNA in EBOV-infected cells.
In this study, we show that EBOV infection does not induce apoptosis. Typical hallmarks of apoptosis, including translocation of annexin V and nuclear condensation, were not detected in EBOV-infected cells (Fig. 1). Slight cleavage of caspase 3 was observed in EBOV-infected primary human macrophages (Fig. 3B), and caspase activation was occasionally detected in EBOV-infected immortalized cells (Fig. 1, ,6,6, and and7).7). Although minor amounts of apoptotic PARP-1 cleavage in EBOV-infected cells were observed, PARP-1 cleavage fragments indicative of necrosis were consistently more prominent and became more abundant during later times of infection, suggesting that EBOV-infected cells undergo necrosis (Fig. 2B and and3C).3C). It is conceivable that noninfected cells account for the observed minor caspase activation. The infection rate of macrophages was about 90% at late times of infection (Fig. 3A). Also, single noninfected Vero cells stained by FLICA or annexin V were observed in the samples infected with EBOV (Fig. 1B and data not shown). It is also possible that the observed weak activation of caspases in EBOV-infected cells plays a role in the induction of regulatory nonapoptotic signaling pathways. Nonapoptotic activation of caspases has been described in cell differentiation and proliferation. Intriguingly, activated caspases are involved in the differentiation of monocytes to macrophages (48, 49). Given the important role that monocytes and macrophages play in EBOV infection, it would be interesting to analyze the impact of EBOV infection on monocyte differentiation.
The TEM data presented here confirm that EBOV-infected cells undergo necrosis (Fig. 4) and are consistent with previous studies in different animal models showing that infected cells in tissues did not exhibit the morphological features of apoptosis (8, 9, 50–53). Although we did not observe double-membrane vesicles in EBOV-infected cells, a hallmark of autophagy, we cannot exclude the possibility that some cells undergo autophagic cell death, since only limited time points were analyzed by TEM. Autophagic cell death involves degradation by the lysosomal machinery. Cathepsins, a group of cellular proteases present mainly in lysosomes, have been implicated in necrotic cleavage of PARP-1 (39, 40). However, it remains to be determined if PARP-1 cleavage is induced during autophagy (54).
To our knowledge, there are no reports on the regulation of apoptosis in EBOV-infected immortalized cells, and data on the fate of infected primary human cells are limited and contradictory. Consistent with our results, Geisbert et al. (8) reported that neither EBOV-infected primary human dermal microvascular endothelial cells nor infected monocytes and macrophages in human peripheral blood mononuclear cell (PBMC) cultures showed visible signs of apoptosis by TEM analysis. In contrast, Gupta et al. (10, 16) detected apoptotic markers, including activation of caspase 3, in the majority of the CD14+ subset (monocytes and macrophages) in EBOV infection of PBMCs. In contrast to our study, monocytes were infected with EBOV and then cocultured with autologous lymphocytes, which suggests that cross signaling between different cell types might play a role in the regulation of apoptosis during EBOV infection. A recent study by Bradfute and colleagues (7) reported increased amounts of apoptotic hepatocytes in EBOV-infected mice and that inhibition of apoptosis delayed the observed liver dysfunction. This suggests a potential role for apoptosis in EBOV pathogenesis, at least in rodents. However, it is not clear if the observed apoptotic cells were actually infected with EBOV (7).
Recently, we reported the accumulation of cytoplasmic vacuoles in cells infected with recombinant Marburg virus (MARV) expressing EGFP, suggesting that nonapoptotic cell death is a general feature of filovirus infection (55). It has been suggested that programmed necrosis plays a role in viral pathogenesis by inducing proinflammatory responses triggering inflammation and the clearance of infected cells (56, 57). In patients, EBOV hemorrhagic fever is associated with a “cytokine storm” late in infection, and the dysregulated cytokine response appears to be associated with fatal outcomes (4, 12). Interestingly, only minimal cellular inflammatory responses occur at the sites of infection (2). With the recent discovery of necrotic signaling pathways, future studies could help determine if necrosis plays a role in EBOV pathogenesis, a process that is still poorly understood. Massive apoptosis of bystander lymphocytes has been observed in fatal cases of EBOV infection, suggesting a role for apoptosis in EBOV pathogenesis (3, 5, 12). However, blocking of lymphocyte apoptosis in an EBOV mouse model did not increase survival (7). Increased expression of apoptosis-inducing ligands such as TRAIL and Fas has been observed during EBOV infection and could play a role in lymphocyte apoptosis (3, 8–11, 14, 15). We observed secretion of TRAIL from EBOV-infected macrophages, yet TRAIL-mediated apoptosis was not inhibited in EBOV-infected Vero cells (Fig. 5C). The regulation of TRAIL expression is not well understood. IFN regulatory factor 3 (IRF3)-induced expression of TRAIL has been observed during SeV infection (58). NF-κB signaling seems to play a role in TRAIL expression, and potential binding sites for NF-AT and AP-1 have been identified in the TRAIL promoter (59). It has been proposed that apoptosis mediated by TRAIL is controlled by differential expression of TRAIL receptors or inhibitors of apoptosis signaling (60). During viral infection, TRAIL-resistant cells can be sensitized to apoptosis via regulation of TRAIL receptor levels (61). Respiratory syncytial virus (RSV), a member of the order Mononegavirales and a close relative of EBOV, has been demonstrated to upregulate TRAIL-R1 and TRAIL-R2, thereby causing targeted killing of infected cells (62). For human immunodeficiency virus (HIV), it was shown that noninfected macrophages remain resistant to TRAIL, whereas downregulation of TRAIL-R3 and TRAIL-R4, the nonfunctional decoy receptors for TRAIL, led to apoptosis only in HIV-infected cells (63). In our study, apoptosis was not induced in noninfected primary macrophages after treatment with up to 10 μg/ml TRAIL (data not shown). These data provide evidence that the main target cells of EBOV are resistant to TRAIL-mediated induction of apoptosis, which might explain why EBOV has not evolved inhibitory mechanisms to block TRAIL signaling in infected cells. It is conceivable that differential expression and surface exposure of TRAIL receptors might determine the sensitivity of EBOV-infected cells to TRAIL-mediated apoptosis. While the expression levels of most TRAIL-R mRNAs in EBOV-infected macrophages did not differ from those in noninfected control cells, a decrease in levels of TRAIL-R1 mRNA was observed (Fig. 5B). Due to BSL-4-related experimental limitations, we were not able to examine the expression of TRAIL receptors on the surface of EBOV-infected cells.
As previously shown, EBOV utilizes VP35 to interfere with host cell-mediated dsRNA responses, including inhibition of retinoic acid-inducible gene I (RIG-I)-mediated IFN induction and suppression of PKR. Binding of dsRNA by VP35 has been shown to be essential for the inhibitory function in IFN signaling, whereas PKR inhibition is independent of dsRNA binding (41, 42). Structural analysis revealed that VP35 not only binds but also masks dsRNA, thereby preventing recognition by RIG-I (64, 65). Prevention of dsRNA sensing by binding of viral proteins to dsRNA has also been described for Lassa virus and might be a key feature of hemorrhagic fever viruses (66). As detection of viral dsRNA is essential for the induction of apoptosis, masking of dsRNA by VP35 might explain the absence of apoptosis in EBOV-infected cells.
Apoptosis induced by SeV infection or transfection of poly(I·C) is not inhibited by EBOV, although PKR activation is still blocked (Fig. 7). This suggests that EBOV is not able to inhibit PKR-independent apoptosis mediated by dsRNA. Induction of apoptosis by dsRNA mediates the formation of dsRNA-DISC complexes independent of ligand-receptor binding, initiating cleavage of caspase 8 and subsequent activation of caspases 3 and 9 (67, 68). RIG-I is involved in the induction of apoptosis in SeV-infected cells, indicating that dsRNA is the recognized pathogen-associated molecular pattern, triggering apoptosis in cells infected with this virus (69). IRF3 also seems to play an important role in dsRNA-mediated apoptosis of SeV-infected cells. Heylbroeck et al. (70) reported that the induction of apoptosis in SeV infection relies on activation and subsequent DNA binding of IRF3, whereas the release of IFN is not essential. In contrast, a recent study by Chattopadhyay et al. (69) showed that IRF3 activation and IRF3-mediated transcription are not required for the induction of apoptosis in SeV infection. Apoptosis was induced via translocation of IRF3 to the mitochondria and interaction with Bax. EBOV VP35 inhibits the RIG-I-mediated induction of type I IFN by blocking the phosphorylation and nuclear translocation of IRF3 (32), but it is not known if it is also able to block the interaction between IRF3 and Bax. However, our data strongly suggest that VP35 does not counteract the induction of dsRNA-mediated apoptosis.
PKR-independent apoptosis triggered by dsRNA relies on signaling through Toll-like receptor 3 (TLR3) and Toll/interleukin-1 receptor domain-containing adapter protein (TRIF) and their assembly into an atypical death complex with caspase 8 (71, 72). To date, our knowledge about TLR signaling in EBOV infection is limited (73–77), and there are no data available on the regulation of TLR signaling by EBOV in the context of apoptosis.
Treatment of EBOV-infected cells with dsCARE successfully inhibited viral spread (Fig. 8). dsCARE binds dsRNA and mediates its antiviral activity through the recruitment of caspases in infected cells, as shown previously for adenovirus and VSV. Cell viability assays revealed that dsCARE does not cause significant cytotoxic effects in uninfected cells (33). In agreement with previously reported data (33), our data indicate that the antiviral activity of dsCARE against EBOV relies on binding to dsRNA, as no effect was observed by using a dsCARE mutant lacking the dsRNA binding domain (dsCAREΔRBD) (Fig. 8C). This implies that dsRNA not only is present during EBOV infection but also can be targeted for antiviral treatment. We were unable to detect dsRNA in EBOV-infected cells by immunofluorescence analysis using the J2 antibody, which is directed against dsRNA (data not shown). The J2 antibody was used in a previous study that reported the presence of dsRNA in cells infected with positive-strand RNA, dsRNA, or DNA viruses but not in cells infected with negative-sense RNA viruses (78). Since there is evidence for the presence of dsRNA in cells infected with negative-sense RNA viruses (79), the lack of detection by immunofluorescence analysis is most likely due to sensitivity limitations.
Since it has been suggested that dsRNA is masked by VP35 in EBOV-infected cells, the question of how dsRNA is recognized by dsCARE arises. One possible explanation is that dsCARE outcompetes and displaces VP35 from dsRNA, leading to the recruitment of caspases by dsCARE CARDs. Another hypothesis is that dsCARE bypasses cellular signaling pathways by directly activating effector caspases, thereby circumventing signaling pathways that are blocked by VP35.
Targeting of dsRNA during viral infection to specifically induce apoptosis in infected cells was previously reported by Rider et al. (47). The broad-spectrum antiviral DRACO was effective against multiple viruses, including DNA, dsRNA, positive-sense RNA, and negative-sense RNA viruses, in cell culture and reduced virus titers in mice infected with influenza virus (47). This emphasizes that dsRNA is a promising target for the development of pan-antivirals, since the presence of dsRNA seems to be a universal feature of virus-infected cells.
In summary, our data show that EBOV neither induces nor actively blocks apoptosis in infected cells but instead induces necrosis. We demonstrate that the inability of EBOV to block apoptosis can be exploited to develop antiviral therapeutics. Furthermore, our data provide evidence for the presence of dsRNA in EBOV-infected cells.
We thank C. F. Basler, Mount Sinai School of Medicine, New York, NY; S. Becker and R. Wagner, Institute of Virology, University of Marburg, Germany; and W. J. Neubert, Max Planck Institute, Martinsried, Germany, for providing reagents. EBOV-EGFP used for the dsCARE studies was kindly provided by J. S. Towner, Special Pathogens Branch, Centers for Disease Control and Prevention, Atlanta, GA, and R. Davey, Texas Biomedical Research Institute, San Antonio, TX. We are grateful to H. Feldmann, F. Feldmann, and E. Haddock for their support and assistance in conducting BSL4 experiments at the Laboratory of Virology, NIAID, NIH, Hamilton, MT. We also thank M. D. Rarick, J. H. Connor, and S. Gummuluru for technical advice and helpful discussions and A. J. Hume for reviewing the manuscript (all Boston University, Boston, MA).
This work was supported by NIH grants U01-AI082954 (to E.M.) and CO6 RR12087 (to J.L.P. and J.A.), start-up funds from Boston University (to E.M.), and funding from the German Research Foundation (grant SFB 535 to E.M.) and the Manchot Foundation (to J.O. and E.M). The production of dsCARE was supported by grants from the National Natural Science Foundation of China (grants NSFC 3092803 and NSFC 81071369 to J.Y. and Z.Y.). Funding was also provided by the Division of Intramural Research, National Institute of Allergy and Infectious Diseases, National Institutes of Health.
Published ahead of print 6 March 2013