MAVS Regulates Apoptotic Cell Death by Decreasing K48-Linked Ubiquitination of Voltage-Dependent Anion Channel 1

Abstract

The mitochondrial antiviral signaling protein MAVS (IPS-1, VISA, or Cardif) plays an important role in the host defense against viral infection by inducing type I interferon. Recent reports have shown that MAVS is also critical for virus-induced apoptosis. However, the mechanism of MAVS-mediated apoptosis induction remains unclear. Here, we show that MAVS binds to voltage-dependent anion channel 1 (VDAC1) and induces apoptosis by caspase-3 activation, which is independent of its role in innate immunity. MAVS modulates VDAC1 protein stability by decreasing its degradative K48-linked ubiquitination. In addition, MAVS knockout mouse embryonic fibroblasts (MEFs) display reduced VDAC1 expression with a consequent reduction of the vesicular stomatitis virus (VSV)-induced apoptosis response. Notably, the upregulation of VDAC1 triggered by VSV infection is completely abolished in MAVS knockout MEFs. We thus identify VDAC1 as a target of MAVS and describe a novel mechanism of MAVS control of virus-induced apoptotic cell death.

INTRODUCTION

Apoptosis is a fundamental physiological process that plays a crucial role in the development and maintenance of homeostasis in multicellular organisms. It is now well established that mitochondria, as a reservoir for apoptotic proteins, play a critical role in the regulation of apoptosis in mammals. In response to various apoptotic stimuli, several apoptotic factors, such as cytochrome c (Cyto C), are released from the intermembrane space of mitochondria into the cytoplasm to initiate the activation of downstream destructive programs, including the caspase cascade (1–4). Although it remains unclear how these apoptotic initiators cross the outer mitochondrial membrane (OMM) and are released into the cytosol, accumulating evidence indicates that voltage-dependent anion channel 1 (VDAC1) is involved in the release of apoptotic proteins via the OMM (5–12). The VDAC family of proteins includes three isoforms, VDAC1, -2, and -3, all of which are located on the OMM. The release of Cyto C, the interaction of the proapoptotic protein Bax with VDAC1, and the triggering of cell death are all inhibited by anti-VDAC1 antibody. VDAC1 protein has thus been recognized as a key protein in mitochondrion-mediated apoptosis through its involvement in the release of apoptotic proteins located in the intermembrane space and as the proposed target of pro- and antiapoptotic members of the Bcl2 family and hexokinase.

Another evolutionarily conserved defensive weapon multicellular organisms use to eradicate viral infections is the innate immune system. In the case of cytoplasmic infection, binding of viral dsRNA to the helicase domain of RIG-I or MDA-5 induces these proteins to interact with MAVS (also known as IPS-1/VISA/Cardif), a C-terminal caspase recruitment domain (CARD) protein located on the OMM (13–16). MAVS then activates kinases, such as TANK-binding kinase 1 (TBK-1) and IKKε, which phosphorylate interferon (IFN) regulatory factor 3 (IRF-3), finally resulting in the production of cytokines such as type I IFN (IFN-I) (17–20). MAVS has a domain organization similar to that of other tail-anchored membrane proteins, which anchors it to the mitochondria and peroxisomes. MAVS-dependent antiviral signaling occurs from both peroxisomes and mitochondria, but the kinetics of IFN-stimulated genes are different. Peroxisomal MAVS induces the immediate expression of antiviral factors that function to contain a nascent infection. Long-term containment of the infection, however, requires the function of mitochondrial MAVS (21–23). In addition to its well-known function in IFN induction, MAVS is also a proapoptotic molecule that triggers disruption of the mitochondrial membrane potential and cell apoptosis in response to viral infection (21, 24–26). Overexpression of MAVS also activates caspase-3, -8, and -9 (27, 28). Instead, virus-induced apoptosis and caspase activation are delayed and attenuated in cells with reduced level of MAVS. Blocking of IFN-β, NF-κB, or IRF-3 has no effect on MAVS-induced apoptosis (24). These results reveal a new role for MAVS in the regulation of cell death beyond its well-known function of IFN induction in innate antiviral immunity. However, the molecular mechanism by which MAVS triggers cell apoptosis is not yet fully understood.

In this study, we found that only the transmembrane (TM) domain of MAVS is required for its proapoptotic activity via association with VDAC1. As a result, MAVS modulates VDAC1 protein stability through the ubiquitin-proteasome pathway. Functionally, ectopic expression of MAVS leads to VDAC1 upregulation, caspase-3 activation, and apoptosis induction. Furthermore, MAVS-deficient fibroblasts display reduced VDAC1 expression with a consequent reduction of vesicular stomatitis virus (VSV)-induced cell apoptosis and, notably, a completely abolished VSV infection-triggered upregulation of VDAC1. We thus identify VDAC1 as a target of mitochondrion-located MAVS and discover a novel mechanism by which mitochondrion-located MAVS mediates cell apoptosis triggered by viral infection.

MATERIALS AND METHODS

Cell culture and transfection.

Cell lines 293T and MCF-7 were grown in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (FBS; HyClone), 2 mM l-glutamine, 100 U/ml penicillin, and 100 mg/ml streptomycin. Vectors of Flag-tagged MAVS, hemagglutinin (HA)-tagged MAVS, Flag-tagged VDAC1, and Myc-tagged VDAC1 were obtained by cloning the genes into corresponding pcDNA3-based vectors (Invitrogen). Flag-MAVS mutation plasmids were made on the basis of the pcDNA3-Flag-MAVS plasmid with the QuikChange mutagenesis kit (Stratagene). Transient transfections were performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.

Mice and small interfering RNA (siRNA).

MAVS^−/− mice on a 129/Sv/C57BL/6 background were a kind gift from Zhijian J. Chen (University of Texas Southwestern Medical Center). Embryonic fibroblasts were prepared from day 15 embryos of wild-type (WT) and mutant mice, respectively, and cultured in DMEM supplemented with 10% FBS.

For downregulation of VDAC1, nucleotides 606 to 624 of the VDAC1 coding sequence (GGAGACCGCTGTCAATCTT) were chosen as the target for siRNA. Transient knockdown of MAVS was performed by siRNA oligonucleotide (Santa Cruz) transfection by following the manufacturer's instructions. Scrambled siRNA oligonucleotides were used as a control.

Immunofluorescence.

Cells were incubated with 250 nM MitoTracker Deep Red FM (Molecular Probes) for 30 min or BacMam 2.0 CellLight reagents (30 particles per cell; Invitrogen) for 16 h at 37°C prior to fixation. For MAVS or VDAC1 visualization, cells were fixed with 2% paraformaldehyde for 20 min at 25°C and permeabilized for 10 min with 0.2% Triton X-100. Samples were blocked with 2% goat serum in phosphate-buffered saline (PBS) containing 50 mM ammonium chloride for 30 min. Anti-MAVS (1:200; Santa Cruz) and anti-VDAC1 (1:200; Santa Cruz) antibodies were used to detect the MAVS and VDAC1 proteins, respectively. Staining was visualized with secondary antibodies conjugated with Alexa Fluor 488, 594, or 647 (Molecular Probes), and the images were captured with a digital camera under a confocal microscope (Zeiss LSM510).

Luciferase reporter assays.

Cells were transfected with 0.2 μg of the luciferase reporter NF-κB–luc, IFN-β–luc, or IRF-3–luc plus 0.02 μg of the internal control reporter PRL, and luciferase activities were assessed after certain times of culture.

Immunoprecipitation and immunoblot analysis.

Cell lysates were prepared in lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 10 mM sodium fluoride, 10 mg/ml aprotinin, 10 mg/ml leupeptin, 10 mg/ml pepstatin A) containing 1% Nonidet P-40. Soluble proteins were subjected to immunoprecipitation with anti-Flag (M2, Sigma), anti-Myc (Santa Cruz), or anti-mouse IgG (Sigma) antibody. An aliquot of the total lysates (5%, vol/vol) was included as a control. Immunoblot analysis was performed with anti-Myc, anti-HA, horseradish peroxidase (HRP)-conjugated anti-Flag, anti-α-tubulin (Sigma), and anti-VDAC1 (Santa Cruz) antibodies, respectively. The antigen-antibody complexes were visualized by chemiluminescence assay (PerkinElmer).

Protein-binding assays.

Glutathione S-transferase (GST) fusion proteins were expressed in Escherichia coli BL21(DE3) after cloning of the corresponding gene fragments into a pGEX4T-1-based vector (Amersham Biosciences). The GST fusion protein in the clear supernatant was purified by affinity chromatography with a glutathione-Sepharose 4B column. The control GST was also expressed and purified in the same way as GST fusion proteins.

In GST pulldown experiments, cell lysates were incubated for 2 h at 4°C with 5 μg of purified GST or GST fusion proteins bound to glutathione beads. The absorbates were washed with lysis buffer and then subjected to SDS-PAGE and immunoblot analysis. An aliquot of the total lysates (5%, vol/vol) was also loaded as a control. In direct binding assays, immunoprecipitates were separated by SDS-PAGE and then blotted onto nitrocellulose membrane. The membrane was subsequently incubated with purified GST fusion proteins for 2 h at room temperature. The GST fusion proteins binding to nitrocellulose were probed with anti-GST antibody and visualized by chemiluminescence assay (PerkinElmer).

Flow cytometry.

Annexin V staining was performed as indicated, with the annexin V-fluorescein isothiocyanate apoptosis kit (BioVision). In brief, cells were washed with cold PBS and then resuspended in binding buffer. After 15 min of incubation with annexin V dye, cells were washed and propidium iodide was added. Cells were analyzed by flow cytometry. The results were expressed as the mean of three independent experiments.

Viral infection.

Cells were plated in six-well plates at a density of 1 × 10⁶/well and incubated overnight. Viral infection was performed with cells at 80% confluence. Culture media were replaced with serum-free DMEM, and VSV was added to the media at the multiplicity of infection (MOI) indicated. After 2 h of incubation, extracellular virus was removed by washing the cells twice with serum-containing medium. Cells and supernatants were harvested at the indicated times postinfection.

Cross-linking experiment.

Cells (2.5 to 3 mg/ml in PBS, pH 8.3) were incubated with 1,5-difluoro-2,4-dinitrobenzene (DFDNB; Thermo) at 30°C for 30 min. Cross-linking was terminated by the addition of sample buffer and heating at 70°C for 10 min. Cells were then sonicated and subjected to SDS-PAGE and immunoblotting with anti-VDAC1 antibodies.

Statistical analysis.

Analyses were done with the statistical software SAS/STAT. Data analysis over time was undertaken by repeated-measures analysis with SAS/STAT. A P value of <0.01 was considered the threshold for statistical significance.

RESULTS

The TM domain is required for mediation of the proapoptotic activity of MAVS.

Previous reports showed that MAVS-dependent antiviral signaling occurred from both peroxisomes and mitochondria (29); however, little is known about the proapoptotic property of MAVS on peroxisomes. Therefore, three mutant proteins containing, respectively, the Pex13 locational motif (MAVS-Pex), the Fis1 locational motif (MAVS-Mito), and the OMP25 locational motif (MAVS-Mimic), which direct the protein to different cell compartments, were constructed and their effects on cell apoptosis were explored directly. We first transfected MAVS^−/− MEFs with the three mutant constructs and determined their localizations by confocal microscopy. As expected, MAVS-Pex and MAVS-Mito were found primarily on peroxisomes and mitochondria, respectively, while MAVS-Mimic was located on both mitochondria and peroxisomes (Fig. 1). Collectively, the three mutant proteins thereby provided an ideal system to determine the relative roles of mitochondrial and peroxisomal localization in MAVS-mediated apoptosis induction. In concert with previous findings (24, 30), compared to the vector-transfected control cells, ectopic WT MAVS expression led to significant cell death and dose-dependent apoptosis induction, as evidenced by trypan blue staining and annexin V staining, respectively (Fig. 2A and B). Blockage of caspase activity by the pan-caspase inhibitor Z-VAD-FMK resulted in the complete reversal of MAVS-induced cell death (Fig. 3A and B). The MAVS-Pex mutant was completely incapable of inducing cell death and apoptosis, suggesting that mitochondrial localization is essential for MAVS activation of apoptosis. Interestingly, both the MAVS-Mito and MAVS-Mimic mutant proteins failed to induce any apoptotic cell death (Fig. 2A and B). Since all three of the mutant MAVS proteins were expressed well (Fig. 2B), the inability of these constructs to induce apoptosis was not likely to be due to ineffective protein expression. MAVS activity was previously proposed to be linked to its mitochondrial localization and the ability to form oligomers mediated by the TM domain (31, 32), we then generated two MAVS dimerization mutant proteins (A521W and V528W) to see their impact on cell apoptosis. These two mutant proteins, which kept over 60% of their ability to activate IFN signaling (Fig. 3C to E), were completely incapable of inducing cell death and apoptosis (Fig. 3A and B). On the contrary, Y9F mutant MAVS, which failed to activate the IFN-β, IRF-3, and NF-κB promoters (33) (Fig. 3C to E), led to significant cell death and apoptosis (Fig. 3A and B). Taken together, these results suggest that the TM domain is required for mediation of the proapoptotic activity of MAVS and that the MAVS-mediated antiviral response is dispensable for MAVS-mediated apoptosis induction.

Fig 1.

Targeting of MAVS to distinct subcellular compartments by replacement of its TM domain. Micrographs of MAVS^−/− MEFs transfected with mutant MAVS proteins were stained with anti-MAVS antibody. Mitochondria were stained with MitoTracker. Peroxisomes were visualized with BacMam 2.0 CellLight reagents. Each images is representative of at least three independent experiments where >200 cells were examined per condition, and >95% of the cells displayed similar staining.

Fig 2.

The TM domain is required for mediation of the proapoptotic activity of MAVS. (A) 293T cells transfected with expression plasmids encoding Flag-MAVS or mutant forms thereof were determined by trypan blue exclusion cell counting. **, P < 0.01. (B) 293T cells transfected with different amounts of Flag-MAVS or mutant forms thereof were analyzed by flow cytometry with annexin V staining. IB, immunoblotting.

Fig 3.

Dimerization is required for mediation of the proapoptotic activity of MAVS. (A and B) 293T cells were treated with the pan-caspase inhibitor Z-VAD-FMK and then transfected with 0.2 μg of expression plasmids encoding Flag-MAVS or mutant forms thereof. Cell death and apoptosis were determined by trypan blue exclusion cell counting (A) and flow cytometry with annexin V staining (B), respectively. (C to E) 293T cells were transfected with IFN-β–luc (C), NF-κB–luc (D), or IRF-3–luc (E) together with Flag-MAVS or mutant forms thereof. Luciferase activity was measured 24 h later and normalized for transfection efficiency. Cell-based studies were performed at least three independent times with comparable results. The data shown are means ± the standard errors of the means. Student's t test was used for statistical analysis: **, P < 0.01. IB, immunoblotting.

MAVS interacts with VDAC1.

To further dissect the possible molecular mechanism of the proapoptotic function of MAVS, a yeast two-hybrid experiment was conducted that showed that an OMM protein, VDAC1, associated with MAVS (Fig. 4A), implying that MAVS might regulate apoptosis signaling via its interaction with VDAC1. To confirm the yeast two-hybrid analysis, Flag-tagged VDAC1 and HA-tagged MAVS were transfected into 293T cells and a coimmunoprecipitation experiment was performed (Fig. 4B). HA-tagged MAVS was detected by anti-Flag antibody immunoprecipitation from cells cotransfected with Flag-VDAC1 but not from cells cotransfected with a negative control Flag-tagged protein (Fig. 4C). Specially, VDAC1 did not associate with Golgi compartment protein GP73 (Fig. 4D). This observation substantiated the yeast two-hybrid analysis and established an interaction between MAVS and VDAC1. The specificity of the interaction between MAVS and VDAC1 was also confirmed by coimmunoprecipitation analysis with normal serum (IgG). Importantly, endogenous MAVS was found to be specifically coimmunoprecipitated with endogenous VDAC1 but not with VDAC2 and VDAC3 (Fig. 4D).

Fig 4.

VDAC1 interacts with MAVS. (A) AH109 was cotransformed with the plasmids indicated. Positive interaction showed colony formation on synthetic medium lacking tryptophan, leucine, adenine, and histidine. (B) 293T cells were cotransfected with HA-MAVS and Flag-VDAC1 expression plasmids or Flag vector, and anti-Flag or -IgG immunoprecipitates were analyzed by immunoblotting (IB) with anti-HA or anti-Flag antibodies. (C) Lysates from 293T cells were subjected to immunoprecipitation (IP) with anti-GP73 or -IgG, fractionated by SDS-PAGE, and subsequently analyzed by immunoblotting with anti-VDAC1 antibody. (D) Lysates from 293T cells were subjected to immunoprecipitation with anti-MAVS or -IgG, fractionated by SDS-PAGE, and subsequently analyzed by immunoblotting with anti-VDAC1, anti-VDAC2, and anti-VDAC3 antibodies, respectively. (E) Anti-Flag or -IgG immunoprecipitates prepared from cells transfected with Flag-MAVS or Flag vector plasmid were subjected to SDS-PAGE and blotted onto nitrocellulose membrane. The nitrocellulose membrane was incubated with soluble GST-VDAC1 or GST for 2 h and then analyzed with anti-GST or anti-Flag antibody. (F) 293T cells transfected with an expression vector encoding Flag-VDAC1 were stained with anti-Flag antibody and imaged by confocal microscopy. The mitochondria were stained with MitoTracker. The yellow staining in the overlay image indicates colocalization of MAVS and MitoTracker. (G) Micrographs of MEFs stained with anti-MAVS and anti-VDAC1 antibodies. Mitochondria were stained with MitoTracker. Peroxisomes were visualized with BacMam 2.0 CellLight reagents. All of the images in all of the panels are representative of at least three independent experiments where over 200 cells were examined per condition, and >95% of the cells displayed similar staining. The values to the left of panels B to E are molecular sizes in kilodaltons.

To rule out an indirect association mediated by other components in the cell lysates, anti-Flag immunoprecipitates prepared from cells expressing Flag-MAVS were subjected to SDS-PAGE and then blotted onto a nitrocellulose membrane. After incubation with soluble GST-VDAC1 fusion protein, the nitrocellulose membrane was treated with an HRP–anti-GST antibody. The results showed that MAVS bound to VDAC1 directly (Fig. 4E, top). As a control, Flag-MAVS did not bind to GST (Fig. 4E, middle). Therefore, we concluded that MAVS binds to VDAC1. Immunofluorescence analysis of HEK293T cells transfected with Flag-VDAC1 showed that overexpressed VDAC1 localized exclusively on mitochondria (Fig. 4F). Moreover, the staining patterns of MAVS partially overlapped that of VDAC1 on mitochondria in response to VSV infection, indicating that both proteins colocalized in the mitochondrial compartment under physiological conditions (Fig. 4G).

Characterization of the MAVS-VDAC1 interaction.

To further map the VDAC1 binding sites in MAVS, three truncated mutant forms of MAVS lacking the CARD-like domain (residues 10 to 77, ΔCARD), the proline-rich region (residues 103 to 152, ΔPro), or the TM domain (residues 514 to 535, ΔTM) were constructed and overexpressed together with VDAC1. Figure 5A and B show that both HA-tagged VDAC1 and endogenous VDAC1 interacted with WT, ΔCARD, or ΔPro MAVS but not with ΔTM MAVS. Thus, the interaction with VDAC1 was specific for the TM domain of MAVS. To further dissect the interaction of VDAC1 with mitochondrial MAVS or peroxisomal MAVS, three mutations that direct the protein to different cell compartments were overexpressed together with VDAC1. Coimmunoprecipitation analysis showed that VDAC1 bound only WT MAVS and not peroxisomal MAVS (Fig. 5C). Their interaction required the TM domain of MAVS, as VDAC1 failed to bind to MAVS with the TM domain replaced with another mitochondrial location motif (Fig. 5C), indicating that the association of VDAC1 with the TM domain of MAVS may contribute to the function of MAVS in apoptosis regulation.

Fig 5.

Characterization of the MAVS-VDAC1 interaction. (A and B) 293T cells were cotransfected with a Flag-tagged MAVS expression plasmid or Flag-tagged mutant MAVS expression plasmids with (A) or without (B) HA-VDAC1, and anti-Flag immunoprecipitates (IP) were analyzed by immunoblotting (IB) with anti-HA antibody (A), anti-Flag antibody, or anti-VDAC1 antibody (B). (C) 293T cells were cotransfected with Flag-VDAC1 and an HA-tagged MAVS expression plasmid or HA-tagged mutant MAVS expression plasmids, and anti-Flag immunoprecipitates were analyzed by immunoblotting with anti-HA or anti-Flag antibody. (D) 293T cells were cotransfected with Flag-MAVS and Myc-VDAC1 or Myc-mutant VDAC1; anti-Flag immunoprecipitates were analyzed by immunoblotting with anti-Myc or anti-Flag antibody. (E) 293T cells were cotransfected with a Flag-tagged MAVS expression plasmid or Flag-tagged mutant MAVS expression plasmids; anti-Flag immunoprecipitates were analyzed by immunoblotting with anti-VDAC1 or anti-Flag antibody. All coimmunoprecipitation experiments were performed independently two to three times with comparable results. The values to the left of the blots are molecular sizes in kilodaltons.

It has been reported that the N-terminal domain of VDAC1 controls the release of Cyto C and thus apoptosis (34). Therefore, we constructed N-terminally truncated VDAC1 (residues 1 to 25, Δ26) and overexpressed it with MAVS. Figure 5D showed that Flag-tagged MAVS failed to interact with Δ26 VDAC1. Thus, the interaction with MAVS is specific for the N terminus of VDAC1. Interestingly, coimmunoprecipitation assays of overexpressed A521W or V528W with endogenous VDAC1 demonstrated that these two MAVS dimerization mutations resulted in significantly reduced interaction with VDAC1, while the Y9F MAVS mutation did not impair its ability to interact with VDAC1 (Fig. 5E), sustaining the importance of the TM domain of MAVS in mediating the MAVS-VDAC1 interaction.

VDAC1 is required for mediation of the proapoptotic activity of MAVS.

Next, we determined the effect of MAVS-VDAC1 interaction on apoptosis induction. VDAC1-specific siRNA oligonucleotides that reduced VDAC1 expression by about 70% were used (Fig. 6A). As shown in Fig. 6A, ectopic MAVS expression induced significant cell apoptosis, while reduction of the VDAC1 expression level inhibited MAVS-induced cell apoptosis by approximately 70%.

Fig 6.

VDAC1 is required for mediation of the proapoptotic activity of MAVS. (A) 293T cells transfected with Flag-MAVS were transfected with scrambled siRNA or VDAC1 siRNA. Cell apoptosis was measured by annexin V staining 48 h later. (B) WT and MAVS^−/− MEFs were infected with VSV at an MOI of 0.02 for 6 h or treated with ActD for 15 h in the presence or absence of Z-VAD. Cell apoptosis was measured by annexin V staining. (C and D) MAVS^−/− MEFs transfected with Flag-MAVS were transfected with scrambled or VDAC1 siRNA. After 48 h, cells were infected with VSV at an MOI of 0.02 for 6 h (C) or treated with ActD for 15 h (D). Cell apoptosis was measured by annexin V staining. (E) Whole-cell lysates from WT and MAVS^−/− MEFs infected with VSV at the indicated times were analyzed by immunoblotting (IB) with anti-Cyto C antibody, anti-PARP antibody, or anti-caspase-3 antibody. α-Tubulin was used as an equal-loading control (Ctrl). Cell-based studies were performed at least three independent times with comparable results. The data shown are means ± the standard errors of the means. Student's t test was used for statistical analysis: **, P < 0.01.

To further explore the physiological role of MAVS-VDAC1 interaction in mediating virus-induced apoptosis, we extended these findings to investigation of virus-induced apoptosis in MEFs isolated from WT MAVS knockout mice, respectively. Both VSV and actinomycin D (ActD), which can induce mitochondrion-dependent cell apoptosis, were used. As shown in Fig. 6B, VSV- or ActD-triggered apoptosis was significantly higher in WT MEFs than in MAVS^−/− MEFs and Z-VAD-FMK treatment resulted in a complete reversal of VSV- or ActD-induced apoptosis in WT MEFs. Moreover, reexpression of Flag-MAVS in MAVS^−/− MEFs apparently made the cells more sensitive to virus-induced cell apoptosis whereas expression of the Flag vector in MAVS^−/− MEFs retained their resistance. Furthermore, the proapoptotic function of MAVS in response to VSV infection could be abrogated by VDAC1 siRNA knockdown (Fig. 6C). As shown in Fig. 6D, VDAC1 was also required for ActD-induced apoptosis mediated by MAVS. To further delineate the function of MAVS in mitochondrion-mediated apoptosis signaling, Cyto C release, PARP cleavage, and caspase-3 activation were also analyzed by the immunoblotting method. As shown in Fig. 6E, VSV infection induced an increased cytosolic Cyto C level and significant PARP cleavage in WT MEFs, whereas the Cyto C release and PARP cleavage in response to VSV infection were nearly defective in MAVS^−/− MEFs. Caspase-3 was also activated in WT MEFs but not in MAVS^−/− MEFs (Fig. 6E). Taken together, these results indicated that MAVS-induced apoptosis is dependent on VDAC1 and caspase-3 activation.

Binding of MAVS to VDAC1 is required for virus-induced cell apoptosis.

The negative-strand RNA virus VSV was used in this study to further analyze the function of MAVS mutations on the proapoptotic property of MAVS in response to viral infection. MAVS^−/− MEFs transfected with an expression vector encoding Flag-tagged WT or mutant MAVS were infected with VSV for 6 h. Immunoblot analysis of the cell lysates from exponentially growing cells revealed normal expression of MAVS and mutant forms thereof (Fig. 7A, right). As shown in Fig. 7A (left), both ΔCARD and ΔPro MAVS were able to rescue VSV-triggered apoptosis by about 26 and 24% individually, comparable to WT MAVS cell rescue (26%), whereas the introduction of ΔTM MAVS resulted in a nearly complete loss of virus-induced apoptosis (5%). In concert with VSV-induced apoptosis, the MAVS TM domain was also required for ActD-induced apoptosis (Fig. 7B).

Fig 7.

Binding of MAVS to VDAC1 is required for virus-induced cell apoptosis. (A, C, and E) MAVS^−/− MEFs transfected with Flag-MAVS expression plasmid or Flag-tagged mutant MAVS proteins were infected with VSV at an MOI of 0.02. Cell apoptosis was measured by annexin V staining 6 h after VSV infection. (B and D) MAVS^−/− MEFs transfected with Flag-MAVS expression plasmid or Flag-tagged mutant MAVS proteins were treated with ActD for 15 h. Cell apoptosis was measured by annexin V staining. Cell-based studies were performed at least three independent times with comparable results. The data shown are means ± the standard errors of the means. Student's t test was used for statistical analysis: **, P < 0.01. IB, immunoblotting.

To further address the function of MAVS from either mitochondria or peroxisomes, we reexpressed locational mutant MAVS proteins in MAVS^−/− MEFs and then infected them with VSV or treated them with ActD. In concert with previous findings, transient expression of WT MAVS led to significant apoptosis induction by VSV or ActD. In contrast, all three locational mutant MAVS proteins completely failed to mediate apoptosis triggered by VSV or ActD, which is consistent with the results of binding assays (Fig. 7C and D). This proapoptotic effect of MAVS did not require IFN-I-induced MAVS signaling because expression of A521W and V528W dimerization mutant MAVS proteins completely abolished VSV-mediated apoptosis induction, whereas a Y9F mutant protein was able to induce VSV-triggered apoptosis in a way comparable to that of WT MAVS (Fig. 7E).

MAVS stabilizes VDAC1.

To explore the mechanism of the proapoptotic effect of MAVS, we examined the effect of MAVS on endogenous VDAC1 abundance. When expression plasmids encoding MAVS were transfected into MCF-7 cells, remarkably, a striking increase in the abundance of endogenous VDAC1 was found (Fig. 8A). As a control, increasing amounts of Flag-GP73 did not change the endogenous VDAC1 level (Fig. 8B). We next investigated the function of endogenous MAVS in VDAC1 regulation. Expression of MAVS-specific siRNA, but not control siRNA, decreased endogenous VDAC1 abundance in MCF-7 cells (Fig. 8C). Consistently, MAVS^−/− MEFs showed an endogenous VDAC1 protein level lower than that of WT MEFs (Fig. 8D). Reverse transcription-PCR data showed that the abundance of VDAC1 mRNA was not altered by MAVS expression (Fig. 8E), suggesting that MAVS upregulated VDAC1 by posttranscriptional modification. Indeed, the estimated half-life of VDAC1 (about 5 h) is significantly shorter than that of VDAC1 in the presence of MAVS (>24 h; Fig. 8F).

Fig 8.

MAVS stabilizes VDAC1. (A) MCF-7 cells were transfected with plasmids expressing increasing amount of Flag-MAVS. Whole-cell lysates were analyzed by immunoblotting (IB) with anti-Flag or anti-VDAC1 antibody. α-Tubulin was used as an equal-loading control. (B) MCF-7 cells were transfected with plasmids expressing increasing amount of Flag-GP73. Whole-cell lysates were analyzed by immunoblotting with anti-Flag or anti-VDAC1 antibody. α-Tubulin was used as an equal-loading control. (C) MCF-7 cells transfected with small interfering MAVS (siMAVS) or small interfering scrambled oligonucleotides were analyzed by immunoblotting with anti-MAVS or anti-VDAC1 antibody. α-Tubulin was used as an equal-loading control. (D) Whole-cell lysates from WT and MAVS^−/− MEFs were analyzed by immunoblotting with anti-VDAC1 antibodies. α-Tubulin was used as an equal-loading control. (E) MCF-7 cells were transfected with different doses of MAVS expression plasmids. RNA was extracted 24 h later, and the VDAC1 mRNA level was analyzed by quantitative real-time PCR. (F) 293T cells were transfected with the expression vector encoding HA-VDAC1 together with Flag-MAVS or Flag vector. After 24 h, cells were treated with cycloheximide (50 μM). The level of VDAC1 was monitored by immunoblotting with anti-HA antibody. α-Tubulin was used as an equal-loading control. (G) Flag-VDAC1 and Myc-MAVS were cotransfected with plasmids encoding HA-tagged ubiquitin (HA-Ub), HA-tagged K48 ubiquitin (HA-Ub K48) or HA-tagged K63 ubiquitin (HA-Ub K63). Cells were grown in DMEM containing MG132 (20 μM) for 6 h. Anti-Flag immunoprecipitates were analyzed by immunoblotting with anti-HA antibody. Whole-cell lysates were subjected to immunoblotting with anti-Myc and anti-Flag antibodies. α-Tubulin was used as an equal-loading control. (H) WT and MAVS^−/− MEFs were infected with VSV at an MOI of 0.02 for the indicated times. Anti-VDAC1 immunoprecipitates were analyzed by immunoblotting with anti-VDAC1 or anti-K48-Ub antibody. (I and J) MCF-7 cells were transfected with the plasmids indicated. Whole-cell lysates were analyzed by immunoblotting with anti-Flag or anti-Myc antibody. α-Tubulin was used as an equal-loading control. Cell-based studies were performed at least three independent times with comparable results. The values below certain Western blots (WB) are relative levels determined by software-based quantification of the representative experiment shown. The data shown are means ± the standard errors of the means.

To further delineate the mechanism of MAVS-mediated VDAC1 stabilization, 293T cells were cotransfected with plasmids expressing Flag-VDAC1, Myc-MAVS, and HA-ubiquitin in the presence of the proteasome inhibitor MG132. VDAC1 was then immunoprecipitated with anti-Flag antibody and blotted with anti-HA antibody. An immunoprecipitation assay showed that the ubiquitination of VDAC1 occurred mainly with ubiquitin harboring a lysine residue only at position 48 (all others were replaced with arginine) but at a very low level when the lysine residue was only at position 63 (Fig. 8G). Overexpression of MAVS led to a sharp reduction in the K48 ubiquitination level of VDAC1 (Fig. 8G), suggesting that MAVS may have a major role in its proteasome-mediated degradation. Additionally, endogenous VDAC1 was immunoprecipitated with anti-VDAC1 antibody at different times after VSV infection, and K48 ubiquitination was detected with anti-K48-ubiquitin antibody. The decreased intensity of a more slowly migrating smear demonstrated that VDAC1 K48 ubiquitination modification was reduced in WT MEFs in response to VSV infection (Fig. 8H), indicating that VDAC1 was deubiquitinated under physiological conditions of activation. MAVS^−/− MEFs tended to have increased K48-linked ubiquitination at 3 and 6 h after VSV infection (Fig. 8H), which corroborated the initial observation that MAVS modulated VDAC1 protein stability by decreasing the degradative K48-linked ubiquitination of VDAC1 (Fig. 8G). MAVS-Mimic did not affect VDAC1 polyubiquitination modification (Fig. 9A) and MAVS did not affect VDAC2 ubiquitination (Fig. 9B).

Fig 9.

MAVS-VDAC1 interaction is required for inhibition of VDAC1 K48 polyubiquitination. (A and B) Flag-VDAC1, Myc-MAVS, and Myc–MAVS-Mimic (A) or Flag-VDAC2 (B) were cotransfected with plasmids encoding HA-tagged K48 ubiquitin (HA-Ub K48). Cells were grown in DMEM containing MG132 (20 μM) for 6 h. Anti-Flag immunoprecipitates (IP) were analyzed by immunoblotting (IB) with anti-HA antibody. Whole-cell lysates were subjected to immunoblotting with anti-Myc and anti-Flag antibodies. The values to the left of the blots are molecular sizes in kilodaltons. WB, Western blot.

It is of interest to examine whether WT MAVS could affect the protein stability of Δ26 mutant VDAC1. Cotransfection of MAVS significantly increased WT VDAC1 but not Δ26 mutant VDAC1 (Fig. 8I). Notably, overexpression of TM deletion mutant MAVS or dimerization mutant MAVS did not lead to VDAC1 upregulation (Fig. 8I and J), while Y9F mutant MAVS led to significant upregulation of VDAC1. Taken together, these results indicate that the interaction between MAVS and VDAC1 is indispensable for VDAC1 stabilization.

VDAC1 upregulation and oligomerization in response to VSV infection require MAVS.

Previous reports showed that apoptosis induction triggered VDAC1 upregulation and oligomerization (35); however, the molecular mechanism underlying VDAC1 upregulation is elusive. Since MAVS could modulate VDAC1 stability, we then tested whether virus-induced VDAC1 upregulation and oligomerization require MAVS. VDAC1 upregulation was observed in WT MEFs upon VSV infection and ActD treatment but was markedly reduced in MAVS knockout and knockdown cells (Fig. 10A and B). To obtain information on the oligomeric status of VDAC1 under apoptotic conditions, we used a chemical cross-linking assay with the membrane-permeating cross-linker DFDNB. Cells were first infected with VSV for 6 h and then incubated with DFDNB. VDAC1 oligomeric states were then assessed by SDS-PAGE and immunoblotting with anti-VDAC1 antibodies. The formation of VDAC1 homo-oligomers, comprising dimers to higher-molecular-mass complexes, was enhanced severalfold upon VSV infection; in contrast, the VDAC1 oligomeric level was sharply reduced in MAVS knockout and knockdown cells (Fig. 10C and D). Cumulatively, these observations solidify the positive regulatory role of MAVS in virus-mediated apoptosis responses independently of its role in innate immunity and demonstrate that MAVS exerts its proapoptotic effect on virus-induced apoptotic signaling by VDAC1 upregulation.

Fig 10.

VDAC1 upregulation and oligomerization in response to VSV infection require MAVS. (A) WT and MAVS^−/− MEFs infected with VSV for the times indicated were analyzed by immunoblotting (IB) with anti-VDAC1 antibody. α-Tubulin was used as an equal-loading control. (B) MCF-7 cells transfected with small interfering MAVS (siMAVS) or small interfering scrambled oligonucleotides were treated with ActD for 15 h. Whole-cell lysates were analyzed by immunoblotting with anti-VDAC1 antibody. α-Tubulin was used as an equal-loading control (Ctrl). (C and D) Control siRNA (siCtrl)- and MAVS siRNA-transfected 293T cells (C) WT MAVS, and MAVS^−/− MEFs (D) were treated with VSV for 6 h. Cells (2.5 to 3 mg/ml) washed with PBS were incubated with DFDNB (0.1 mM) at 30°C for 30 min and then subjected to SDS-PAGE and immunoblotting with anti-VDAC1 antibodies. The positions of VDAC1 monomers and multimers are indicated. Cell-based studies were performed at least three independent times with comparable results. The values below certain Western blots are relative levels determined by software-based quantification of the representative experiment shown. The data shown are means ± the standard errors of the means. The values to the left of panels C and D are molecular sizes in kilodaltons.

DISCUSSION

Besides its well-described role in IFN-I induction, MAVS may also be involved in cell death signaling during viral infection. For example, overexpression of MAVS activates caspase-3 and induces interleukin-18 processing in HaCaT keratinocytes (22). Transfection of MAVS induces cell death in HEK293T cells, and activation of RIG-I/MDA5 by poly(I·C) triggers apoptosis in a MAVS-dependent manner in human melanoma cells (23). However, the mechanism of MAVS-mediated apoptosis induction remains unclear.

Regulation of the stability of one OMM protein, VDAC1, is a potential mechanism by which MAVS modulates mitochondrial apoptotic signaling. In this study, we characterized VDAC1 as a novel and essential cofactor for MAVS function in apoptotic cell death. Several lines of evidence support this argument. First, VDAC1 was shown to interact directly with MAVS. Mutational analysis revealed that the TM domain in MAVS was required for association with VDAC1. Second, MAVS upregulated VDAC1 protein stability by decreasing its K48-linked ubiquitination level. The half-life of VDAC1 in the presence of MAVS was significantly increased. Third, MAVS-deficient fibroblasts displayed reduced VDAC1 protein and oligomerization levels with a consequent reduction in VSV-induced cell apoptosis. Fourth, the upregulation of VDAC1 triggered by VSV infection was completely abolished in MAVS-deficient fibroblasts. These results suggest that the cellular expression level and the oligomerization state of VDAC1 regulated by MAVS are crucial factors in the process of virus-mediated apoptosis induction.

VDAC1, located on the OMM, is a key player in mitochondrion-mediated apoptosis. Thus, in addition to regulating the metabolic and energetic functions of mitochondria, VDAC1 appears to be a convergence point for a variety of cell survival and cell death signals through its association with various ligands and proteins. The N-terminal region of VDAC1 modulates the accessibility of VDAC1 to Bax, Bcl-x_L, and hexokinase (36). Our observations therefore suggest that the proapoptotic activity of MAVS was mediated through interaction with the N-terminal region of VDAC1, implying that the latter may adopt different conformations, depending on various factors. Eventually, rearrangement and coordination of VDAC1's N-terminal helix with different proteins were shown in this study to be involved in its proapoptotic function.

MAVS contains a C-terminal TM domain that targets the protein not only to mitochondria but also to peroxisomes. This TM domain of MAVS is essential for apoptosis signaling because deletion or replacement of the mitochondrial targeting sequence of MAVS disrupts the apoptosis response. In contrast to mitochondrial MAVS, peroxisomal MAVS failed to associate with VDAC1 and trigger apoptotic cell death, indicating that contact of MAVS with VDAC1 is a prerequisite for the induction of apoptosis by mitochondrial MAVS, highlighting the biological significance of VDAC1-MAVS association as an important apoptosis regulatory control.

VDAC1 concentrations in the OMM are important for the regulation of sensitivity to apoptotic signals, which is important in several human diseases. A previous report showed that apoptosis triggered VDAC1 upregulation (11). However, the molecular mechanism of VDAC1 upregulation remains elusive. Here our results suggest that MAVS modulates VDAC1 protein stability through the ubiquitin-proteasome pathway and the presence of MAVS significantly increases the half-life of VDAC1. MAVS-deficient fibroblasts displayed reduced levels of VDAC1 and oligomerization. As MAVS itself is not an E3 ubiquitin ligase, it would be interesting to identify additional E3 ligases or deubiquitinases involved in this process.

In general, apoptosis-related caspase activity serves two different purposes; one is to eliminate functional proteins by digestion into inactive forms, and the other is to generate new biological activities. In the case of MAVS, both purposes are achieved. First, MAVS induces VDAC1-dependent Cyto C release, caspase activation, and consequent cell apoptosis. Second, caspase-mediated cleavage of MAVS from mitochondria abrogates its VDAC1 binding and apoptosis-induction abilities and thus may act as another negative regulation strategy to inhibit the progression of apoptosis. Therefore, the direct binding of MAVS to VDAC1 and the involvement of VDAC1 in mitochondrial apoptosis raise the possibility that a MAVS-VDAC1 interaction lies at the base of cell survival and apoptosis regulation. These findings also support the central role of MAVS as a platform for the spatiotemporal regulation and integration of multiple innate signaling pathways in mitochondria.