Mitochondrial protein mitofusin 2 is required for NLRP3 inflammasome activation after RNA virus infection

Significance

Mitochondria play pivotal roles not only in the energy production, apoptosis, and calcium storages, but also in innate antiviral immunity. Mitochondrial antiviral signaling expressed on the outer membrane of mitochondria is essential for intracellular viral RNA-mediated induction of type I interferon. In addition, damaged mitochondria generate reactive oxygen species required for nod-like receptor family, pyrin domain-containing 3 (NLRP3) inflammasome-dependent inflammatory responses. Here, we demonstrate that mitochondrial membrane potential-dependent association between NLRP3 and mitochondrial outer membrane protein mitofusin 2, a key regulator of mitochondrial fusion, is required for the full activation of the NLRP3 inflammasome after infection with RNA viruses. Our results highlight the importance of mitochondria as a platform of NLRP3 inflammasome activation.

Abstract

Nod-like receptor family, pyrin domain-containing 3 (NLRP3), is involved in the early stages of the inflammatory response by sensing cellular damage or distress due to viral or bacterial infection. Activation of NLRP3 triggers its assembly into a multimolecular protein complex, termed “NLRP3 inflammasome.” This event leads to the activation of the downstream molecule caspase-1 that cleaves the precursor forms of proinflammatory cytokines, such as interleukin 1 beta (IL-1β) and IL-18, and initiates the immune response. Recent studies indicate that the reactive oxygen species produced by mitochondrial respiration is critical for the activation of the NLRP3 inflammasome by monosodium urate, alum, and ATP. However, the precise mechanism by which RNA viruses activate the NLRP3 inflammasome is not well understood. Here, we show that loss of mitochondrial membrane potential [ΔΨ(m)] dramatically reduced IL-1β secretion after infection with influenza, measles, or encephalomyocarditis virus (EMCV). Reduced IL-1β secretion was also observed following overexpression of the mitochondrial inner membrane protein, uncoupling protein-2, which induces mitochondrial proton leakage and dissipates ΔΨ(m). ΔΨ(m) was required for association between the NLRP3 and mitofusin 2, a mediator of mitochondrial fusion, after infection with influenza virus or EMCV. Importantly, the knockdown of mitofusin 2 significantly reduced the secretion of IL-1β after infection with influenza virus or EMCV. Our results provide insight into the roles of mitochondria in NLRP3 inflammasome activation.

Nod-like receptor family, pyrin domain-containing 3 (NLRP3) can be activated by a wide variety of stimuli, such as endogenous danger signals from damaged cells, bacterial components, environmental irritants, and DNA and RNA viruses (1). It forms a multiprotein complex called the NLRP3 inflammasome, which includes an adaptor protein apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) and procaspase-1. The NLRP3 inflammasome-mediated cytokine release requires two signaling pathways (2). The first signal is induced by Toll-like receptors (TLRs), interleukin 1 receptor (IL-1R), or tumor necrosis factor receptor, and leads to the synthesis of inactive NLRP3, pro–IL-1β, and pro–IL-18 in the cytosol. The second signal is triggered by NLRP3 agonists, which induce the activation of caspase-1. Caspase-1 catalyzes the proteolytic processing of pro–IL-1β and pro–IL-18, and their conversion to mature forms, and stimulates their secretion across the plasma membrane (1). These inflammasome-dependent cytokines play a key role in the induction of adaptive immunity and the initiation of tissue healing after influenza virus infection (3–5). Migration of dendritic cells (DCs) to the draining lymph nodes and priming of CD8 T cells during influenza virus infection require IL-1R signaling in respiratory DCs (6). By contrast, chronic activation of the NLRP3 inflammasome has been linked to many inflammatory diseases (7, 8). Therefore, increasing the number of studies dedicated to the investigation of the molecular mechanisms of NLRP3 inflammasome activation will be crucial for improving our understanding of the pathogenesis of infectious and autoinflammatory diseases.

Mitochondria are compartmentalized by two membrane bilayers (outer and inner membranes) and are involved in a wide variety of functions in eukaryotic cells. Within the past decade, novel functions of mitochondria have been discovered demonstrating their crucial role in innate antiviral immunity in vertebrates (9). A direct link between mitochondria and innate immunity was first highlighted with the finding that an adaptor protein, mitochondrial antiviral signaling (MAVS; also known as IPS-1, VISA, or Cardif) (10–13), triggered retinoic acid-inducible gene 1 (RIG-I) and melanoma differentiation-associated protein 5-mediated type I interferon (IFN) induction. In addition to their role in antiviral immunity, mitochondria also function as a platform for the activation of the NLRP3 inflammasome by producing mitochondrial reactive oxygen species (mROS) (14, 15). In this context, NLRP3 agonists trigger the generation of mROS from damaged mitochondria, resulting in the dissociation of thioredoxin (TRX) from TRX-interacting protein, which associates with NLRP3 to facilitate inflammasome formation (16). Furthermore, cytosolic mitochondrial DNA (mtDNA) released from damaged mitochondria has been reported to activate the NLRP3 inflammasome (17) and absent in melanoma 2 inflammasome (15), recently identified as a cytoplasmic DNA sensor for the inflammasome (18–21). Although mitochondria are essential for host-cell defense, the mechanism of their involvement in the activation of the NLRP3 inflammasome is still unclear. In the present study, we demonstrate that the mitofusin 2 (Mfn2) is required for the full activation of the NLRP3 inflammasomes in macrophages.

Results

mROS Action Does Not Correlate with IL-1β Secretion Induced by RNA Virus Infection.

Previous studies demonstrated that mROSs are critical for NLRP3 inflammasome activation when stimulated with monosodium urate (MSU), alum, or ATP (14, 15). To confirm the importance of mROS in RNA virus-induced IL-1β secretion, we first measured the kinetic changes in ROS-producing mitochondria after infection with influenza virus (Fig. S1A). The level of ROS-producing mitochondria in LPS-primed bone marrow-derived macrophages (BMMs) infected with influenza virus reached a peak at 3 h postinfection (p.i.), and decreased to basal levels by 6 h (Fig. S1A). IL-1β secretion was first observed at 9 h p.i. and peaked at 24 h p.i. We did not observe detectable levels of IL-1β at the peak of ROS production. We confirmed that the pattern of ROS kinetics and IL-1β secretion from cells infected with influenza virus was similar to that of encephalomyocarditis virus (EMCV) infection (22). Next, we tested the specific role of mROS by measuring IL-1β secretion from virus-infected cells in the presence or absence of the antioxidant Mito-TEMPO (Santa Cruz Biotechnology), a scavenger specific for mROS, which effectively blocked influenza virus-induced mROS production (Fig. S1B). Mito-TEMPO significantly inhibited the secretion of IL-1β from BMMs induced by ATP, nigericin, or MSU (Fig. S1C), consistent with previous reports (14, 15). By contrast, normal secretion of IL-1β was observed in influenza virus- or EMCV-infected BMMs, or recombinant measles virus lacking the V protein (MVΔV)-infected phorbol 12-myristate 13-acetate–differentiated THP-1 cells in the presence of Mito-TEMPO (Fig. S1 C–E). Taken together, these data suggest that mROSs are not essential for the activation of the NLRP3 inflammasome in the case of RNA virus infection.

RNA Virus-Induced IL-1β Secretion Is Impaired in ρ⁰ Macrophages.

Although RNA virus-induced NLRP3 activation was independent from mROS production, it is well established that mitochondria play a central role in innate immunity against RNA viruses (9, 23, 24). To confirm the importance of mitochondrial functions in RNA virus induced IL-1β secretion, we generated mtDNA-depleted J774A.1 mouse macrophages (ρ⁰ J774A.1) (15). ρ⁰ J774A.1 macrophages expressed diminished levels of mtDNA (Fig. S2A). The production of IL-1β was abrogated in ρ⁰ J774A.1 macrophages primed with LPS and stimulated with ATP, influenza virus, or EMCV (Fig. S2B). We observed a similar extent of infection by influenza virus in J774A.1 and ρ⁰ J774A.1 macrophages and robust production of caspase-1-independent cytokine by ρ⁰ J774A.1 macrophages infected with influenza virus (Fig. S2 C and D). Because mtDNA encodes 13 polypeptides required for mitochondrial respiration and oxidative phosphorylation, we reasoned that ethidium bromide treatment might affect ATP synthesis and cause widespread heterogeneity in the mitochondrial membrane potential [ΔΨ(m)]. We confirmed that ρ⁰ J774A.1 macrophages had a moderate effect on the ΔΨ(m)-sensitive color shift detected by MitoProbe 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide (JC-1), a cationic dye that indicates mitochondrial depolarization through a reduction in its red–green fluorescence ratio (Fig. S2E). These results prompted us to explore the functional role of ΔΨ(m) in the regulation of NLRP3 inflammasome activation.

ΔΨ(m) Is Essential for RNA Virus-Induced NLRP3 Inflammasome Activation.

To examine whether ΔΨ(m) is essential for mitochondria-mediated activation of the NLRP3 inflammasome, we treated BMMs with carbonyl cyanide m-chlorophenyl hydrazone (CCCP), a protonophore that dissipated ΔΨ(m) by increasing membrane permeability to protons without affecting the total number of mitochondria (Fig. 1A, JC-1 green-positive cells) and ratio of influenza virus-infected BMMs (Fig. S3A). Since CCCP treatment might affect the mitochondrial ATP synthesis, we used 0.1 μM of oligomycin B or 1 μM of rotenone to exclude the role of mitochondrial ATP synthesis in NLRP3 inflammasome activation. Although it has been reported that treatment of LPS-primed BMMs with 40 μM of rotenone induces inflammasome activation (14), 1 μM of rotenone did not enhance IL-1β secretion from LPS-primed BMMs (Fig. S4A). CCCP treatment significantly suppressed secretion of IL-1β and active caspase-1 after infection with MVΔV (Fig. S4B), influenza virus, or EMCV (Fig. 1B and Figs. S4 C and D and S5A). By contrast, infection with influenza virus, EMCV, or MVΔV induced normal levels of IL-1β and active caspase-1 in the presence of oligomycin B or rotenone, suggesting that the inhibition of mitochondrial ATP synthesis was not attributable to virus-induced NLRP3 inflammasome activation. Furthermore, loss-of-function induced by depleting ΔΨ(m) was reversible, as the removal of CCCP from treated cells by washing fully restored IL-1β secretion after stimulation with influenza virus, EMCV (Fig. 1C and Figs. S4 E and F and S5C), or ATP (Fig. 1D and Fig. S5 B and D). The recovery of ΔΨ(m) in washed cells was confirmed by the regeneration of large isoforms of optic atrophy 1, a mitochondrial protein that is the target for ΔΨ(m)-dependent proteolysis (Fig. 1E) (25). Treatment of BMMs with CCCP did not affect the expression levels of pro–IL-1β (signal 1) or inflammasome-independent cytokine production after stimulation with LPS (Fig. S3 B and C) (26), suggesting that signal 1 for inflammasome activation is intact in CCCP-treated BMMs.

Fig. 1.

ΔΨ(m) is required for RNA virus-induced IL-1β secretion. (A) BMMs were treated with either DMSO, CCCP (40 μM), oligomycin B (0.1 μM), or rotenone (1 μM), and stained with MitoProbe JC-1. (B) LPS-primed BMMs were stimulated with influenza virus or EMCV in the presence or absence of CCCP (40 μM), oligomycin B (0.1 μM), or rotenone (1 μM). (C) LPS-primed BMMs were infected with influenza virus or EMCV in the presence or absence of CCCP (40 μM). The far right lane (wash) indicates CCCP (40 μM) was washed out 6 h before infection. Cell-free supernatants were collected at 18 h p.i. and analyzed for IL-1β by ELISA. (D) LPS-primed BMMs were stimulated with ATP in the presence or absence of the indicated amounts of CCCP. The far right lane (wash) indicates that CCCP (40 μM) was washed out 6 h before ATP stimulation. Cell-free supernatants were collected and analyzed for IL-1β by ELISA. (E) The processing of OPA-1 in CCCP-treated BMMs was analyzed by immunoblotting to confirm the dissipation of ΔΨ(m). Data are representative of at least three independent experiments, and indicate the mean ± SD. **P < 0.01, ***P < 0.001.

Recently, two papers reported an important role for mtDNA in NLRP3 inflammasome activation (15, 17). Specifically, Nakahira et al. observed that the release of mtDNA into the cytosol depended on the NLRP3 inflammasome and mROS (15). In addition, Shimada et al. demonstrated that oxidized mtDNA binds to NLRP3 and activates the NLRP3 inflammasome (17). Thus, it is possible that the impaired IL-1β release from ρ⁰ J774A.1 macrophages was due to a reduction of mtDNA in ρ⁰ J774A.1 macrophages (Fig. S2A). However, we ruled out this possibility by measuring the levels of mtDNA in CCCP-treated BMMs. The levels of mtDNA were similar in CCCP-treated and untreated BMMs (Fig. S3D). Together, these data suggest that CCCP treatment does not affect signal 1 or mtDNA and that ΔΨ(m) is essential for the activation of the NLRP3 inflammasome (signal 2).

Overexpression of uncoupling protein-2 Inhibits NLRP3 Inflammasome Activation.

We next examined whether the suppression of NLRP3 inflammasome activation by chemical treatment can be reproduced by overexpression of uncoupling protein-2 (UCP-2), which is known to reduce ΔΨ(m) by proton leakage (27, 28). Transfection of human embryonic kidney cell line 293T (HEK293T) cells with full-length UCP-2 (full) but not a truncated mutant, UCP-2 (1–200), which contains only amino acid residues 1 to 200, diminished ΔΨ(m) in transfected cells (Fig. 2A) (26). To assess the role of UCP-2 in NLRP3 inflammasome activation, we chose the NLRP3 reconstitution assay in HEK293T cells, as they are deficient for endogenous NLRP3 inflammasome, but can produce mature IL-1β upon transfection with plasmids encoding NLRP3, ASC, procaspase-1, and pro–IL-1β (29, 30). HEK293T cells were transfected with plasmids encoding NLRP3, ASC, procaspase-1, and pro–IL-1β:UCP-2 (full), or UCP-2 (1–200) at a (microgram:microgram) ratio of 1:10 to reconstitute the NLRP3 inflammasome only in UCP-2 transfected cells. Reconstitution of the NLRP3 inflammasome resulted in IL-1β secretion by HEK293T cells, in which the NLRP3 was absolutely required (Fig. 2B). Inclusion of Myc-tagged full-length UCP-2 [UCP-2 (full)] but not its truncated mutant [UCP-2 (1–200)] inhibited the NLRP3 inflammasome-mediated IL-1β secretion and ASC oligomerization (Fig. 2 B and C and Fig. S6A), without affecting the expression levels of pro–IL-1β and NLRP3 (Fig. 2 D and E and Fig. S6A). Similar results were observed in UCP-2–overexpressing J774A.1 macrophages after infection with influenza virus or EMCV (Fig. 2 F–H and Fig. S6B). These results further highlight the importance of ΔΨ(m) in the activation of the NLRP3 inflammasome.

Fig. 2.

Inhibition of NLRP3 inflammasome-mediated IL-1β secretion by UCP-2. (A) HEK293T cells were transfected with expression plasmid encoding either Myc-tagged UCP-2 (full) or UCP-2 (1–200). Twenty-four hours after transfection, the cells were stained with MitoTracker Red and analyzed by flow cytometry. (B and C) HEK293T cells were transfected with expression plasmids encoding NLRP3, ASC, procaspase-1, pro–IL-1β, and either Myc-tagged UCP-2 (full) or UCP-2 (1–200). pCAEGFP was used as a control. Cell-free supernatants were collected 24 h posttransfection and analyzed for IL-1β by ELISA (B). The presence of Myc-tagged proteins was confirmed by Western blotting analysis with a monoclonal antibody against Myc (C). (D and E) HEK293T cells were transfected with expression plasmids encoding pro–IL-1β, and either EGFP, Myc-tagged UCP-2 (full) or UCP-2 (1–200). Cell lysates were collected 24 h posttransfection and analyzed for pro–IL-1β by ELISA (D) or Western blotting analysis with a monoclonal antibody against mouse IL-1β (E). (F) J774A.1 cells stably expressing EGFP or Myc-tagged UCP-2 were stained with MitoTracker Red and analyzed by flow cytometry. (G) LPS-primed J774A.1 cells stably expressing EGFP or Myc-tagged UCP-2 were infected with influenza virus or EMCV. Cell-free supernatants were collected at 21 h after infection and analyzed for IL-1β by ELISA. (H) Samples from J774A.1 cells stably expressing EGFP or Myc-tagged UCP-2 were analyzed by immunoblot with mouse monoclonal antibody against Mfn2. Data represent the mean ± SD. Similar results were obtained from three separate experiments. ***P < 0.001. IB, immunoblot.

Mitofusin 2 Interacts with NLRP3 and Enhances Inflammasome Activation.

On the basis of a previous study (31), we reasoned that mitochondrial dynamics and fusion could play a major role in NLRP3 inflammasome activation. To gain some mechanistic insight into how intact ΔΨ(m) supports the activation of inflammasomes, we first performed subcellular fractionation and immunoprecipitation studies. After infection of LPS-primed BMMs with influenza virus or EMCV, we detected the NLRP3 in the mitochondrial fraction (Fig. 3A). We next attempted to determine whether NLRP3 associated with mitochondrial outer membrane proteins after infection with influenza virus or EMCV. Mitofusins (Mfn1 and Mfn2) are mitochondrial outer membrane guanosine triphosphatases (GTPases) involved in the regulation of mitochondrial dynamics (32). The mouse embryonic fibroblasts (MEFs) with null mutations in both Mfn1 and Mfn2 completely lack mitochondrial fusion and show a widespread heterogeneity of ΔΨ(m) (33). We therefore examined whether Mfns might guide NLRP3 to the mitochondria after infection with influenza virus. Endogenous NLRP3 was coimmunoprecipitated by antibodies to either Mfn1 or Mfn2 in LPS-primed BMMs infected with influenza virus (Fig. 3B). The NLRP3–Mfn2 association in LPS-primed BMMs was attenuated after dissipation of ΔΨ(m) by CCCP treatment before NLRP3 activation (Fig. 3C). The treatment of BMMs with CCCP at 24 h p.i. did not inhibit the NLRP3–Mfn2 association or IL-1β secretion (Fig. 3 D and E). In addition, the CCCP treatment did not significantly change the expression levels of Mfn2 in BMMs (Fig. 3F), suggesting that ΔΨ(m)-dependent mitochondrial fusion is required, whereas the Mfn2 on the fragmented mitochondria is not sufficient for proper activation of the NLRP3 inflammasome. We confirmed the NLRP3–Mfns associations in BMMs by immunoprecipitation studies in HEK293T cells. In agreement with a previous report (23), MAVS coimmunoprecipitated with Mfn2 in HEK293T cells (Fig. S7A). Similarly, the NLRP3 predominantly coimmunoprecipitated with Mfn2 in HEK293T cells (Fig. S7A). Through a coimmunoprecipitation approach, we mapped the region of Mfn2 that interacted with NLRP3 to a central 4,3 hydrophobic heptad repeat (HR) region 1 (Mfn2-HR1) (Fig. S7 B and C). Importantly, overexpression of the HR region in HEK293T cells significantly inhibited the NLRP3-mediated IL-1β secretion (Fig. S8), suggesting that NLRP3 oligomerization and its recruitment to mitochondria through Mfn2 was required for NLRP3-mediated IL-1β secretion.

Fig. 3.

Endogenous NLRP3 associates with mitofusins after infection with influenza virus or EMCV. (A) LPS-primed BMMs were infected with influenza virus or EMCV for 24 h. Crude mitochondrial fractions were analyzed by immunoblotting with mouse monoclonal antibody against NLRP3. (B) LPS-primed BMMs were infected with influenza virus. Immunoprecipitates from lysates with Mfn1- or Mfn2-specific antibody were analyzed by immunoblotting with mouse monoclonal antibodies against NLRP3 or Mfn2 or rabbit polyclonal antibody against Mfn1. (C) LPS-primed BMMs were infected with influenza virus or EMCV in the presence or absence of CCCP for 18 h. Immunoprecipitates from lysates with Mfn2-specific antibody were analyzed by immunoblotting with mouse monoclonal antibodies against NLRP3 or Mfn2. (D and E) LPS-primed BMMs were infected with influenza virus or EMCV. At 24 h p.i., CCCP was added to culture media for 10 min. (D) Immunoprecipitates from lysates with Mfn2-specific antibody were analyzed by immunoblotting with mouse monoclonal antibodies against NLRP3 or Mfn2. (E) Cell-free supernatants were collected and analyzed for IL-1β by ELISA. (F) Unprimed BMMs were treated with DMSO or CCCP for 30 min. Crude mitochondrial fractions were analyzed by mmunoblotting with mouse monoclonal antibody against Mfn2. Indicated below bands are the signal intensities as normalized to that of the corresponding OPA-1. The value in DMSO-treated BMMs was set to 100%. Data are representative of at least three independent experiments, and indicate the mean ± SD. IB, immunoblot; IP, Immunoprecipitation; N.S., not significant.

To evaluate the role of Mfn2 in IL-1β secretion in response to influenza virus and EMCV infection, we generated J774A.1 cells that stably expressed short hairpin RNA (shRNA) against murine Mfn2. Western blot analysis confirmed knockdown of Mfn2 expression (Fig. 4A). Notably, Mfn2 knockdown J774A.1 cells had a significantly impaired secretion of IL-1β and active caspase-1 after infection with influenza virus and EMCV (Fig. 4B and Fig. S9). The reduction in IL-1β secretion was not due to a general defect in the cytokine response to LPS (Fig. 4 C and D), indicating that Mfn2 was required for optimal activation of the NLRP3 inflammasome after infection with influenza virus and EMCV. Recently, it was reported that the N terminus of NLRP3 interacted with MAVS (31). In our study, the NLRP3–MAVS association was enhanced by transient transfection of HEK293T cells with plasmids encoding Mfn2 (Fig. S10A). In addition, the NLRP3–Mfn2 association was also enhanced by transient transfection of HEK293T cells with plasmids encoding MAVS (Fig. S10B). Collectively, these data indicate that ΔΨ(m)-dependent association between Mfn2 and NLRP3 is required for proper activation of the NLRP3 inflammasome in response to viral infection.

Fig. 4.

Mfn2 is required for IL-1β secretion in response to influenza virus and EMCV. (A) Samples from J774A.1 cells stably expressing shRNA against Mfn2 or EGFP mRNAs were analyzed by immunoblot with mouse monoclonal antibody against Mfn2. Indicated below bands are the signal intensities as normalized to that of the corresponding OPA-1. The value in wild-type J774A.1 cells was set to 100%. (B) LPS-primed J774A.1 cells stably expressing shRNA against Mfn2 or EGFP mRNA were infected with influenza virus or EMCV. Cell-free supernatants were collected at 21 h after infection and analyzed for IL-1β by ELISA. (C and D) J774A.1 cells stably expressing shRNA against Mfn2 or EGFP mRNA were stimulated with LPS for 21 h. Cell-free supernatants were collected and analyzed for pro–IL-1β (C) or IL-6 (D) by ELISA. Data are representative of at least three independent experiments, and indicate the mean ± SD. **P < 0.01.

Discussion

The innate immune system uses pattern-recognition receptors to detect pathogen-associated molecular patterns or damage-associated molecular patterns, and thus regulate the production of type I IFNs and proinflammatory cytokines. Unlike TLRs and RIG-I–like helicases that recognize viral RNA, NLRP3 senses cellular damage or distress because of infection. We previously demonstrated that the influenza virus activates the NLRP3 through its M2 protein, a proton-selective ion channel, whereas EMCV activates the NLRP3 inflammasome by increasing Ca²⁺ levels in the cytosol through the action of virus-encoded viroporin 2B (22, 34). However, the mechanism by which RNA viruses activate the NLRP3 inflammasome is not well understood. Here, we demonstrated that ΔΨ(m), a proton gradient across the mitochondrial inner membrane, is required for NLRP3–Mfn2 association and full activation of the NLRP3 inflammasome. These findings highlight the important physiological role of mitochondria in the activation of the NLRP3 inflammasome in addition to RIG-like receptor-mediated antiviral signal transduction (9, 24).

Although dissipation in ΔΨ(m) decreased the synthesis of mitochondrial ATP, inhibitors of mitochondrial F₁F₀-ATPase, F-type proton-translocating ATPase, or mitochondrial complex I did not suppress IL-1β production, demonstrating that inhibition of ATP synthesis alone is not sufficient to suppress NLRP3 inflammasome activation. Recently, it has become increasingly apparent that mitochondrial dynamics plays an important role in MAVS-mediated type I IFN production and activation of the NLRP3 inflammasome (9, 31, 35). In this context, we previously demonstrated that MEFs with null mutations in both mitofusin genes, Mfn1 and Mfn2, showed a complete lack of mitochondrial fusion, widespread loss of ΔΨ(m), and reduced MAVS-mediated IFNs responses after infection with Sendai virus, vesicular stomatitis virus, and EMCV (26, 36). Subramanian et al. demonstrated that MAVS interacts with the NLRP3 and promotes the NLRP3 inflammasome-mediated IL-1β secretion (31). In this study, we describe a unique mechanism by which the NLRP3 inflammasome is activated in a ΔΨ(m)-dependent manner. Specifically, the ΔΨ(m) was required for inflammasome-dependent IL-1β secretion and NLRP3 association with Mfn2 after infection with influenza virus and EMCV. Mfn2 enhanced the association between NLRP3 and MAVS. Although treatment of BMMs with CCCP did not reduce the expression levels of Mfn2, the knockdown of Mfn2 in J774A.1 cells diminished virus-induced IL-1β secretion, suggesting that Mfn2 on the fragmented mitochondria induced by ΔΨ(m) depolarization is not sufficient, and the association of the NLRP3, Mfn2, and MAVS on ΔΨ(m)-intact fused mitochondria is required for full activation of the NLRP3 inflammasome (Fig. 5).

Fig. 5.

Proposed model of ΔΨ(m)-dependent activation of the NLRP3 inflammasome. Mitochondria are dynamic organelles that undergo continuous cycles of fusion and fission. Mfn1 and Mfn2 are essential for mitochondrial fusion. After infection with RNA viruses, the NLPR3 and MAVS associate with Mfn2 in a ΔΨ(m)-dependent manner. The Mfn2 associates with the NLRP3 through its HR1 to recruit adaptor protein ASC and procaspase-1 and form the NLRP3 inflammasome. The NLRP3 inflammasome activates caspase-1, which cleaves pro–IL-1β to mature forms and stimulates their secretion.

Recently, mitochondrial dysfunctions such as mROS production, decreased ΔΨ(m), and oxidized mtDNA have been implicated in NLRP3-mediated IL-1β responses to nonviral signals (14, 15, 17). These results differ from our findings showing that Mito-TEMPO, which effectively blocked IL-1β production after stimulation with ATP, nigericin, or MSU, did not inhibit influenza virus-, EMCV-, or MVΔV-induced IL-1β secretion. One possible explanation is that ATP, alum, or nigericin induces IL-1β secretion within a few hours poststimulation, while infection with RNA viruses requires 12 h to induce secretion of IL-1β. In addition, CCCP did not induce detectable levels of IL-1β from LPS-primed BMMs, suggesting that mROSs are insufficient for inflammasome activation in the absence of additional factors such as ΔΨ(m). Although mROSs might suppress transcriptional induction of NLRP3 expression, but not the activation of NLRP3 inflammasome (37), the role of mROSs in virus-induced activation of the NLRP3 inflammasome remains to be determined.

Our data reveal a link between ΔΨ(m)-dependent association of the NLRP3 and Mfn2 and regulation of the NLRP3 inflammasome. Because Mfn2 tethers the endoplasmic reticulum and mitochondria to control mitochondrial uptake of Ca²⁺ ions (38), the tethering function of the Mfn2 in the activation of the NLRP3 inflammasome cannot be ruled out. Further studies are required to elucidate the relevance of Mfn2 in the activation of the NLRP3 inflammasome. Better understanding of the mechanism that governs the activation of the NLRP3 inflammasome will facilitate the development of more effective interventions for the treatment of infectious and autoinflammatory diseases.

Methods

Mice.

C57BL/6 mice were purchased from The Jackson Laboratory. All animal experiments were approved by the Animal Committees of the Faculty of Medicine (Kyushu University) and the Institute of Medical Science (The University of Tokyo).

Statistical Analysis.

The analysis of variance test (GraphPad INSTAT) was used for statistical analyses. P values less than 0.05 were considered statistically significant.