Lysine 63-Linked TANK-Binding Kinase 1 Ubiquitination by Mindbomb E3 Ubiquitin Protein Ligase 2 Is Mediated by the Mitochondrial Antiviral Signaling Protein

ABSTRACT

Beta interferon (IFN-β) is involved in a wide range of cellular functions, and its secretion must be tightly controlled to inhibit viral spreading while minimizing cellular damage. Intracellular viral replication triggers cellular signaling cascades leading to the activation of the transcription factors NF-κB and interferon regulatory factor 3 (IRF3) and IRF7 (IRF3/7), which synergistically bind to the IFN-β gene promoter to induce its expression. The mitochondrial antiviral signaling protein (MAVS) is a governing adaptor protein that mediates signaling communications between virus-sensing proteins and transcription factors. The activity of MAVS in the regulation of IFN-β secretion is affected by many cellular factors. However, the mechanism of MAVS-mediated IRF3/7 activation is not completely understood. Here, we identified a highly conserved DLAIS motif at amino acid positions 438 to 442 of MAVS that is indispensable for IRF3/7 activation. Specifically, the L439S and A440R mutations suppress IRF3/7 activation. Pulldown experiments using wild-type and mutant MAVS showed that mindbomb E3 ubiquitin protein ligase 2 (MIB2) binds to the DLAIS motif. Furthermore, the DLAIS motif was found to be critical for MIB2 binding, the ligation of K63-linked ubiquitin to TANK-binding kinase 1, and phosphorylation-mediated IRF3/7 activation. Our results suggest that MIB2 plays a putative role in MAVS-mediated interferon signaling.

IMPORTANCE Mitochondrial antiviral signaling protein (MAVS) mediates signaling from virus-sensing proteins to transcription factors for the induction of beta interferon. However, the mechanism underlying activation of MAVS-mediated interferon regulatory factors 3 and 7 (IRF3/7) is not completely understood. We found a highly conserved DLAIS motif in MAVS that is indispensable for IRF3/7 activation through TANK-binding kinase 1 (TBK1) and identified it as the binding site for mindbomb E3 ubiquitin protein ligase 2 (MIB2). The mutations that targeted the DLAIS motif abolished MIB2 binding, attenuated the K63-linked ubiquitination of TBK1, and decreased the phosphorylation-mediated activation of IRF3/7.

INTRODUCTION

The interferon (IFN) family of cytokines plays important roles in host defense mechanisms against invading viruses (1). IFNs are classified into three major types according to specific receptors that trigger the corresponding cellular signals. Type I IFN mainly comprises 13 alpha IFN (IFN-α) isoforms and a single beta IFN (IFN-β) isoform (2). These proteins bind to the IFN-α receptor (IFNAR) complex, which consists of the IFNAR1 and IFNAR2 proteins (3). IFN-β is the first cytokine released from all cell types in response to viral infection. Secreted IFN-β binds to IFNAR and thereby stimulates the formation of the IFN-stimulated gene factor 3 (ISGF3) protein complex (4). ISGF3 is required for the transcriptional activation of hundreds of IFN-stimulated genes (ISGs) that play multiple roles in antiviral defense. However, ISGs also interfere with normal cellular functions to limit viral spreading (5). Therefore, IFN-β secretion must be tightly regulated to eradicate viruses while minimizing cell damage.

The replication of virus inside host cells produces unique molecular signatures known as pathogen-associated molecular patterns (PAMPs). The detection of PAMPs is mediated by pattern recognition receptors (PRRs), such as Toll-like receptors (TLRs) and retinoic acid-inducible gene-I (RIG-I)-like receptors (RLRs) (6). The recognition of extracellular or endosomal PAMPs is mediated by TLRs, whereas the intracellular detection of PAMPs is mediated predominantly by the cytoplasmic RLRs. Intracellular viral PAMPs bind to the helicase domain of RLRs, thereby enabling the caspase recruitment domain (CARD) of RLRs to interact with the CARD of mitochondrial antiviral signaling protein (MAVS) (7). MAVS is localized at the mitochondrial outer membrane, peroxisomal membrane (8), and mitochondrion-associated endoplasmic reticulum membrane (9). This CARD interaction induces the formation of high-molecular-weight oligomers of MAVS (10), which, in turn, triggers the activation of canonical I-kappa-B kinase complexes (IKKs) and the IKK-related kinases. IKKs then activate NF-κB, and the IKK-related kinases activate interferon regulatory factors 3 and 7 (IRF3/7) (11, 12). TANK-binding kinase 1 (TBK1) and I-kappa-B kinase epsilon (IKKε) belong to IKK-related kinase (13). Among the nine mammalian IRFs identified to date, IRF3 and IRF7 are the major transcription factors for type I interferon (14). IRF3 and IRF7 are closely related homologs with a common mode of activation involving the phosphorylation of a unique serine cluster. The phosphorylation of IRF3/7 is mediated by TBK1/IKKε (13, 15). The synergistic binding of NF-κB and IRF3/7 to the promoter of the IFN-β gene initiates its transcription (16).

MAVS is a key adaptor molecule in the IFN signaling pathway described above. RLR signals converge on MAVS, resulting in the activation of NF-κB and IRF3/7. In contrast to well-established NF-κB activation pathways, it is unclear how the MAVS oligomer leads to IRF3/7 phosphorylation via TBK1/IKKε. An increasing body of evidence indicates that ubiquitination plays a crucial role in the activation of TBK1/IKKε by MAVS. Activated MAVS recruits the E3 ligases, tumor necrosis factor (TNF) receptor-associated factor 2 (TRAF2), TRAF3, and TRAF6 via its TRAF-interacting motif (TIM) (17). TRAF6 is involved in the activation of NF-κB, and TRAF3 plays a role in the activation of IRF3/7 (18, 19). However, TRAF3 knockout does not impair the IFN response (20), suggesting that other E3 ubiquitin ligases can compensate for the loss of TRAF3. We found a putative DLAIS motif in MAVS that is required for the activation of TBK1/IKKε. Wild-type MAVS and motif mutant MAVS were used to screen binding proteins, resulting in the identification of mindbomb E3 ubiquitin protein ligase 2 (MIB2; previously known as skeletrophin). Subsequent studies revealed that MIB2 binding to MAVS via the DLAIS motif stimulates the K63-linked ubiquitination of TBK1, leading to IRF3/7 activation. Our results suggest that MIB2 may be another E3 ligase involved in MAVS-mediated TBK1 activation.

MATERIALS AND METHODS

Bioinformatics analysis.

MAVS sequences of the following 29 mammalian species were downloaded from the NCBI GenBank: Homo sapiens (NP_065797.2), Mus musculus (NP_659137.1), Rattus norvegicus (NP_001005556.1), Sus scrofa (NP_001090898.1), Macaca mulatta (NP_001036131.1), Canis lupus familiaris (NP_001116081.1), Bos taurus (NP_001040085.1), Pan troglodytes (XP_525410.2), Papio anubis (NP_001266458.1), Heterocephalus glaber (XP_004840803.1), Mustela putorius furo (XP_004772903.1), Condylura cristata (XP_004687065.1), Jaculus jaculus (XP_004661501.1), Octodon degus (XP_004634385.1), Sorex araneus (XP_004611046.1), Odobenus rosmarus divergens (XP_004398246.1), Tursiops truncatus (XP_004326011.1), Orcinus orca (XP_004285338.1), Gorilla gorilla (XP_004061802.1), Ovis aries (XP_004014409.1), Felis catus (XP_003983730.1), Saimiri boliviensis (XP_003941135.1), Pan paniscus (XP_003821366.1), Otolemur garnettii (XP_003788260.1), Cavia porcellus (XP_003476457.1), Nomascus leucogenys (XP_004090219.1), Pongo abelii (XP_003779353.1), Callithrix jacchus (XP_002747458.2), and Ceratotherium simum simum (XP_004433848.1). The prediction of the secondary structures of these proteins was performed using Jpred version 3 (21). Multiple-sequence alignment was performed using Clustal Omega (22), and the sequence logos were generated by WebLogo3 (23).

Cells and virus.

The human embryonic kidney 293T and human cervical cancer HeLa cell lines were grown in Dulbecco's modified Eagle's medium (DMEM; Gibco/Life Technologies LLC) supplemented with 10% fetal bovine serum (FBS), 2 mM glutamine (GlutaMAX-I; Gibco), and 1% penicillin-streptomycin (Gibco) and maintained at 37°C in a humidified 5% CO₂ incubator (Thermolyne, Asheville, NC). The Sendai virus (Cantell strain) was cultured in chicken embryo amniotic fluid.

Luciferase assay.

293T cells were seeded into 24-well plates at a density of 10⁵ cells/well for 24 h before transfection. Transfectin (Bio-Rad, Hercules, CA) or calcium phosphate (BD Biosciences, San Jose, CA) was used for transient transfections. The cells were transfected with a pGL3-based firefly luciferase reporter construct (200 ng) and the control Renilla luciferase plasmid pRL-SV40 (Promega, Madison, WI) (2 ng). The amounts of DNA mixtures were adjusted to 300 ng by adding empty pCM vector. For Sendai virus stimulation, the cells were infected with 25 hemagglutinins (HA)/ml Sendai virus at 12 h posttransfection. Twenty-four hours after transfection, the luciferase activity was measured with a luminometer (Victor X4; PerkinElmer, Waltham, MA) using a dual-luciferase assay kit (Promega). The Firefly luciferase values were normalized to the Renilla luciferase values, and the relative luciferase unit (RLU) values were calculated by dividing the normalized values by the negative-control sample value.

Immunoblotting.

The cells were harvested and resuspended in radioimmunoprecipitation assay (RIPA) buffer (Thermo Scientific, Rockford, IL). After incubation at 4°C for 20 min and vortex mixing for 5 min, the lysed cells were centrifuged at 12,000 × g for 5 min to pellet down the cell debris. The supernatants were mixed with SDS sample buffer (Bio-Rad) in a 1:1 ratio and boiled for 5 min at 95°C. The boiled samples were separated by SDS-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane (Bio-Rad). The transferred membrane was blocked in phosphate-buffered saline (PBS) with 0.1% Tween 20 (PBST) and 5% (wt/vol) powdered skim milk for 30 min and then incubated with a primary antibody for 1 h at room temperature. After three washes in PBST for 5 min on a shaker, the membrane was incubated with a horseradish peroxidase-conjugated secondary antibody for 1 h and washed three times in PBST for 5 min. The signals were then detected using an enhanced chemiluminescence system (Thermo Scientific, Rockford, IL) and an ECL detection instrument (ImageQuant LAS 4000; GE Healthcare, Pittsburgh, PA).

GST pulldown assay.

293T cells were seeded in 10-cm-diameter dishes at a density of 10⁷ cells/dish for 24 h prior to transfection. Forty-eight hours after transfection with vectors expressing glutathione S-transferase (GST) or GST fusion proteins, the cells were collected and lysed in NP-40 buffer (50 mM HEPES [pH 7.4], 150 mM NaCl, 1 mM EDTA, and 1% [vol/vol] NP-40) supplemented with Complete Protease Inhibitor Cocktail (Roche Applied Science, Indianapolis, IN). After centrifugation, the supernatants were precleared with protein A/G beads (GE Healthcare) at 4°C for 2 h. The precleared lysates were mixed with a 50% slurry of glutathione-conjugated Sepharose beads (GE Healthcare), and the binding reaction mixture was incubated for 4 h at 4°C. The precipitates were washed extensively with wash buffer (50 mM HEPES [pH 7.4], 150 mM NaCl, and 1 mM EDTA). The proteins bound to glutathione beads were eluted by boiling and then subjected to SDS-PAGE and immunoblotting analysis. An aliquot of the total lysates (5% [vol/vol]) was included in the SDS-PAGE experiments as a control.

Immunoprecipitation.

293T cells were seeded in 10-cm-diameter dishes at a density of 10⁷ cells/dish 24 h before transfection. The cells were transfected with the appropriate plasmid mixture. After 24 h, the cells were harvested and immediately lysed in 1% Triton X-100 lysis buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% Triton X-100, 10% glycerol, 40 mM β-glycerophosphate, 0.1% protease inhibitor cocktail, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1 mM Na₃VO₄, 5 mM NaF, 1 mM dithiothreitol [DTT], and 10 mM N-ethylmaleimide [NEM]). Immunoprecipitation was performed with 1 μg of antibody coupled to protein Sepharose magnetic G beads (GE Healthcare) at 4°C with constant agitation for 1 h. After three washes with washing buffer using a MagRack (GE Healthcare), the samples were denatured in sample buffer and analyzed by immunoblotting.

Mass spectrometry.

293T cells were seeded in 10-cm-diameter dishes at a density of 10⁷ cells/dish 24 h before transfection. Each construct (20 μg) was transfected into five dishes, and the collected lysates were subjected to GST pulldown at 24 h posttransfection. After silver staining (Invitrogen) according to the manufacturer's instructions, the protein bands were excised and analyzed at the Seoul National University Mass Spectrometry facility. The peptides were determined by tandem mass spectrometry and database searches.

Fluorescence-activated cell sorter (FACS) analysis.

293T or 293T MIB2-knockdown cells were seeded in 12-well plates at a density of 2 × 10⁵ cells/well. The next day, green fluorescent protein (GFP)-vesicular stomatitis virus (VSV) was diluted with DMEM (10% FBS) to obtain the appropriate titer. The cells were washed once in PBS followed by the addition of 1 ml of serial dilutions of the virus stock to each well. Twenty-four hours after virus infection, the medium was removed, and the cells were washed once with PBS (1 ml). Next, 0.2 ml of trypsin-EDTA was added to each well, and the plates were then incubated at 37°C for 5 min. This step was followed by the addition of 0.3 ml of chilled DMEM containing 10% FBS to each well to inactivate the trypsin. The cells were pipetted to break up any clumps. An aliquot was collected from each well, and the cells were counted. The remaining cells were centrifuged at 1,400 × g for 5 min, fixed in 500 μl of 1% (wt/vol) formaldehyde (BBC Biochemical), and incubated for 5 min at room temperature. The cells were then centrifuged at 1,400 × g for 5 min, washed twice, and resuspended in 200 μl of sterile PBS containing 1% FBS. The samples were analyzed on a FACSCalibur instrument (BD Biosciences). In general, 10⁴ cells were analyzed per sample. A 488-nm argon ion laser was used, and the green fluorescence signal was collected using a 525/20-nm band-pass filter. The cells were analyzed using FlowJo v10 software (TreeStar). The live cells were gated based on forward scatter and side scatter and subsequently for GFP expression. The data were analyzed as histograms depicting the fluorescence intensity of a single color against the cell number.

Stable knockdown cells.

The short hairpin RNA (shRNA) vectors used to knock down MAVS (TRCN0000438514) and MIB2 (TRCN0000438514) were purchased from the RNAi Consortium (TRC)-Sigma-Aldrich. The shRNA/scramble vector (catalog no. 1864) was acquired from Addgene (Cambridge, MA). Each shRNA vector (3.3 μg) was transfected into 293T cells in 10-cm-diameter dishes with pMD2.G packaging vector (3.3 μg) and pCMV-dR8.7dvpr helper vector (3.3 μg). The transfection was performed using SuperFect (Qiagen). The medium was replaced with 10 ml of DMEM supplemented with 5% FBS. The lentivirus-containing supernatant was collected on days 2 and 3 after transfection by changing the media. The supernatant was centrifuged at 1,000 rpm for 10 min and filtered through a 0.45-μm-pore-size filter. The lentivirus expressing shRNA was infected into 293T cells, and the cells were subjected to puromycin selection 3 days postinfection.

Plasmids, chemicals, and antibodies.

pCMnGFP, pCMcGFP, pCMnFlag, pCMnGST, and pCMBB cloning vectors were constructed for efficient subcloning using common multiple-cloning sites (24). The complete cDNAs of IRF3, IRF7, TBK1, IKKε, and NS3/4a and of the luciferase reporter constructs pGL3-IFNA6, pGL3-IFNB, and pGL3-NF-κB were used as described in detail previously (25). A point mutation was generated by PCR-directed mutagenesis using a QuikChange II site-directed mutagenesis kit (Agilent Technology, Santa Clara, CA).

The expression plasmids for MIB1 (catalog no. 33317), MIB2 (catalog no. 33312), and HA-tagged K63 ubiquitin (catalog no. 17606) were acquired from Addgene (Cambridge, MA). Poly(I·C) and MG132 were purchased from Sigma-Aldrich (St. Louis, MO). BX795 was purchased from InvivoGen (San Diego, CA). The suppliers of the antibodies were as follows: the anti-β-actin (AC-15) primary antibody was purchased from Abcam (Cambridge, MA), and the anti-FLAG (M2) and anti-GST (G1160) primary antibodies were purchased from Sigma-Aldrich. The anti-phospho-IRF7 (Ser471/472, 5184), anti-phospho-IRF3 (Ser396, 4D4G), and anti-phospho-TBK1 (Ser172, D52C2) primary antibodies were purchased from Cell Signaling (Danvers, MA), and the anti-IRF3 (FL-425), anti-IRF7 (F-1), anti-MAVS (E-3), anti-MAVS CARD (H-135), anti-ubiquitin (P4D1), anti-phospho-IκBα (Ser32, B-9), anti-IκBα (H-4), anti-HA (Y-11), and anti-MIB2 (H-102) primary antibodies were purchased from Santa Cruz Biotech (Dallas, TX).

RESULTS

MAVS has two well-defined domains, an N-terminal CARD and a C-terminal transmembrane (TM) domain. The CARD mediates MAVS oligomerization, whereas the TM domain localizes MAVS to the mitochondrial outer membrane, which is crucial for its function. Although these two domains have been extensively studied, little is known about the function of the region between the two domains. We used the term “stem” to describe a region between the CARD and TM domain consisting of amino acids (aa) 100 to 508 (Fig. 1). The most prominent feature of this region is the abundance of proline residues, including five proline dipeptides. Although the leading section of the stem has been described as a proline-rich region, the remaining sequence is also enriched with proline. Another characteristic feature of the stem is the low abundance of secondary structures compared to the CARD or the TM domain.

FIG 1.

Mitochondrial antiviral signaling protein (MAVS) domains and the construction of stem truncations. (Top) MAVS structure. CARD, caspase activation recruitment domain; TM, transmembrane; Stem, the region between the CARD and the TM domain; 2nd structure, predicted secondary structure; open square, beta-sheet; filled square, alpha-helix; Proline frequency, the occurrence of proline residues in a span of five amino acids in each direction; Di-proline position, proline dipeptide locations. (Bottom) Stem truncations inserted between an N-terminal CARD and a C-terminal TM domain. CARD-TM (CS0T) was the backbone for the other constructs. Restriction enzyme sites for MluI and SpeI were inserted for cloning.

To narrow down the identity of the region involved in the activation of IRF3/7, we built constructs containing the CARD, the stem truncation, and the TM domain. First, we built a CARD-TM (CS0T) construct containing only the CARD and the TM domain to use as a backbone for the generation of other constructs. The N terminus of CS0T contained aa 1 to 100 of MAVS, and the C terminus contained aa 509 to 540. The restriction enzyme sites for MluI and SpeI were inserted for subsequent cloning (Fig. 1). The NS3/4a protein of the hepatitis C virus is a protease that cleaves MAVS at the cysteine residue in position 508 to evade host IFN signaling (11). The insertion of the SpeI site in CS0T between aa 508 and 509 destroyed the target site of NS3/4a. As a result, the CS0T-based constructs were resistant to NS3/4a. CS1T was cloned by inserting the full-length stem region into CS0T. CS2T to CS12T contained various stem fragments, as depicted in Fig. 1.

Both IRF3 and IRF7 are activated by TBK1 and IKKε, and IFN-α promoters are specifically stimulated by IRF7 (14). Based on these findings, the activity of TBK1/IKKε is conventionally measured using an IFN-α promoter assay with IRF7 coexpression in 293T cells (17). Additionally, we used NS3/4a to exclude the influence of endogenous MAVS because the CS0T-based constructs were resistant to NS3/4a as described above. We examined the stimulation of the IFN-α6 promoter by coexpressed IRF7 in response to Sendai virus (SV) stimulation with or without NS3/4a. There was no stimulation by SV without IRF7. The promoter was stimulated by SV with IRF7 coexpression, and this stimulation was abrogated by NS3/4a (Fig. 2A). Next, the CS1T resistance to NS3/4a was tested. The promoter stimulation by the overexpression of wild-type MAVS was abolished by NS3/4a but was not altered in the case of CS1T (Fig. 2B). Taking the data together, the activity of TBK1/IKKε shown by CS0T-based constructs can be measured using an IFN-α6 promoter assay with the coexpression of IRF7 and NS3/4a. The IFN-α6 promoter assay was performed using CS0T to CS12T (Fig. 1). CS1T, CS4T, CS8T, and CS9T activated the promoter (Fig. 2C). To evaluate IRF3 phosphorylation, immunoblotting was performed using a phospho-IRF3-specific antibody. The phosphorylation by the constructs matched the promoter stimulation data (Fig. 2D). Based on these data, the stem region from aa 410 to 463, which was common among CS1T, CS4T, CS8T, and CS9T, was critical for IRF3 activation.

FIG 2.

Mitochondrial antiviral signaling protein (MAVS) stem region required for interferon regulatory factor 3 (IRF3)-IRF7 (IRF3/7) activation. (A) The effect of IRF7 and NS3/4a on the stimulation of the alpha interferon 6 (IFN-α6) promoter in response to Sendai virus (SV). The IFN-α6 promoter mixture (200 ng of pGL3-IFN-α6 plus 2 ng of pRL-SV40) was cotransfected with pCM-IRF7 (10 ng) or pCM-NS3/4a (50 ng) into 293T cells seeded in a 24-well plate as indicated. SV was added at 24 h posttransfection, and the luciferase activities were measured 12 h postinfection. The promoter assay data are presented as the means and standard deviations of the results of three independent experiments. RLU, relative luciferase units. (B) Resistance of CS1T to NS3/4a. The IFN-α6 promoter mixture and IRF7-expressing plasmid (10 ng) were cotransfected with wild-type pCM-MAVS (WT) or pCM-CS1T (50 ng) in combination with or without pCM-NS3/4a (20 ng) into 293T cells seeded in a 24-well plate as indicated. The luciferase activities were measured 24 h posttransfection. (C) IFN-α6 promoter assay using CARD-stem truncation-TM domain constructs. Each construct (50 ng) was cotransfected with the promoter mixture (200 ng of pGL3-IFN-α6 plus 2 ng of pRL-SV40 plus 10 ng of pCM-IRF7 plus 20 ng of pCM-NS3/4a) into 293T cells seeded in a 24-well plate as indicated. The luciferase activities were measured 24 h posttransfection. (D) Immunoblot showing IRF3 phosphorylation. Each construct (200 ng) was transfected into stable MAVS-knockdown 293T cells seeded in a six-well plate, and the lysates were subjected to immunoblotting (IB) 24 h posttransfection with the indicated antibodies. pIRF3, IRF3 phosphospecific antibody. MCARD, MAVS-CARD antibody. βActin, beta-actin antibody.

To identify evolutionarily conserved residues between aa 410 and 463, multiple-sequence alignment across mammalian species was performed. Three highly conserved motifs were identified, namely, SKPG (aa 419 to 422), DLAIS (aa 438 to 442), and PEENEY (aa 455 to 460) (Fig. 3A). To test the importance of these conserved motifs for IRF3 activation, a set of mutants was generated using CS1T as a template (Fig. 3A, bottom). The CS1Tm1 contained the K420S and P421R mutations disrupting the SKPG motif; CS1Tm2 contained the L439S and A440R mutations disrupting the DLAIS motif; and CS1Tm3 contained the E458S and N459R mutations disrupting the PEENEY motif. Promoter assays performed using these mutants showed that CS1T, CS1Tm1, and CS1Tm3 activated the IFN-α6 promoter whereas CS1Tm2 did not; all of the constructs activated the NF-κB promoter (Fig. 3B). These results suggest that the conserved DLAIS motif is critical for IRF3 activation. Notably, CS1Tm2 showed the highest NF-κB promoter stimulation despite its inability to stimulate the IFN-α6 promoter. This finding clearly indicates that the activation of IKKs and IRF3 is controlled by different MAVS domains.

FIG 3.

DLAIS motif for the activation of interferon regulatory factor (IRF)3/7. (A) Multiple-sequence alignment of the mammalian MAVS region consisting of residues 400 to 470 and highly conserved motifs. The top logos show the conservation of residues. The shaded boxes indicate highly conserved motifs. The bottom map shows CS1T mutants targeting the conserved motifs. CS1Tm1, the K420S/P421R mutant targeting the SKPG motif; CS1Tm2, the L439S/A440R mutant targeting the DLAIS motif; CS1Tm3, the E458S/N459R mutant targeting the PEENEY motif. (B) Promoter assays using the conserved motif mutants. Each construct (50 ng) was cotransfected with a promoter plasmid mixture (IFN-α6 mixture, 200 ng of pGL3-IFN-α6 plus 2 ng of pRL-SV40 plus 10 ng of pCM-IRF7 plus 20 ng of pCM-NS3/4a; NF-κB mixture, 200 ng of pGL3-NF-κB plus 2 ng of pRL-SV40 plus 20 ng of pCM-NS3/4a) into 293T cells seeded in a 24-well plate. The luciferase activities were measured 24 h posttransfection. (Bottom panels) Immunoblot (IB) showing the expression of the mutants in the cell lysates used for the corresponding promoter assay. RLU, relative luciferase units. The data are presented as the means and standard deviations of the results of three independent experiments. βActin, beta-actin antibody.

The inability of CS1Tm2 to activate TBK1/IKKε was confirmed by examining the phosphorylation of IRF3/7. IRF3 phosphorylation was induced by TBK1, IKKε, and CS1T but not by CS1Tm2 (Fig. 4A). We used the IRF7 K92R mutant lacking DNA binding capacity (26) instead of wild-type IRF7 because the expression level was severely affected by the coexpression of IRF7 and TBK1/IKKε due to the strong induction of IFN1. The assessment of the phosphorylation status of IRF7 K92R showed the same results: phosphorylation was induced by overexpression of TBK1, IKKε, and CS1T but not by overexpression of CS1Tm2 (Fig. 4B). These results indicate that the inability of CS1Tm2 to activate the IFN-α6 promoter is the result of impaired IRF3/7 phosphorylation. Although both TBK1 and IKKε are kinases for IRF3/7, 293T cells express TBK1 only in a resting state. The role of TBK1 in the activation of IRF3/7 by MAVS was assessed using the TBK1 kinase inhibitor BX795. The activation of the IFN-α6 promoter by CS8T (containing stem region aa 410 to 463) was suppressed in the presence of increasing amounts of BX795 in a pattern identical to that obtained for TBK1 (Fig. 4C). The conserved motif mutants were also tested. Promoter stimulation by TBK1, CS1T, CS1Tm1, and CS1Tm3 was inhibited by BX795, whereas CST2m2 did not stimulate the promoter (Fig. 4D). These results indicate that IRF3/7 phosphorylation by MAVS is mediated by TBK1 and that the DLAIS motif is critical for the activation of TBK1.

FIG 4.

TANK-binding kinase 1 (TBK1) ubiquitination via the DLAIS motif. (A and B) IRF3 (A) and IRF7 (B) phosphorylation by the conserved motif mutants. IRF7K92R, IRF7 K92R mutant lacking DNA binding capacity. 293T cells were seeded in six-well plates at a density of 10⁵ cells/well for 24 h before transfection. Each construct (50 ng) was cotransfected with pCM-IRF3 or pCM-IRF7 K92R (100 ng) as indicated, and the cells were subjected to immunoblotting (IB) with the indicated antibodies at 24 h posttransfection. IKKε, I-kappa-B kinase epsilon; pIRF3 and pIRF7, IRF3 and IRF7 phosphospecific antibodies;CS1T, similar to wild-type MAVS except for MluI and SpeI restriction enzyme sites; CS1Tm2, the L439S/A440R mutant. (C and D) Alpha interferon 6 promoter assay using BX795 (TBK1 inhibitor). Each construct (20 ng/well) was cotransfected with a promoter plasmid mixture (200 ng of pGL3-IFN-α6 plus 2 ng of pRL-SV40 plus 10 ng of pCMBB-IRF7 plus 10 ng of pCM-NS3/4a). BX795 was added at the indicated concentrations (C) or at 100 ng/ml (D) at 24 h posttransfection. The luciferase activities were measured 12 h after treatment. RLU, relative luciferase units. The data are presented as the means and standard deviations of the results of three independent experiments. CS1Tm1, the K420S/P421R mutant; CS1Tm3, the E458S/N459R mutant. (E) IκB activation by CS1T and CS1Tm2. 293T cells were seeded in six-well plates at a density of 10⁵ cells/well 24 h before transfection. Each construct (100 ng) was transfected, and the cells were treated with MG132 at 24 h posttransfection. The cells were subjected to immunoblotting (IB) using the indicated antibodies at 6 h after MG132 treatment. pIκB, IκB phosphospecific antibodies; βActin, beta-actin antibody. (F) Posttranslational modifications of TBK1. 293T cells were seeded in 10-cm-diameter dishes at a density of 10⁷ cells/dish 24 h before transfection. Each construct (pCM-Flag-TBK, 10 μg; CS1T and CS1Tm2, 5 μg) was transfected, and the lysates were subjected to immunoprecipitation (IP) at 24 h posttransfection. The precipitates were analyzed by immunoblotting (IB) using the indicated antibodies. WCL, whole-cell lysate. pTBK1, TBK1 phosphospecific antibodies; UBI, ubiquitin antibody.

Because CS1Tm2 could activate the NF-κB promoter (Fig. 3B), changes in IκBα were evaluated to confirm this finding. CS1T and CS1Tm2 induced equal levels of phosphorylation of IκBα; the IκBα levels were reduced in the presence of CS1T and CS1Tm2 and restored in the presence of MG132 (Fig. 4E). Taken together, these findings suggest that NF-κB activity is not controlled by the DLAIS motif. Next, the posttranslational modifications of TBK1 were evaluated. The TBK1 autophosphorylation levels seen with CS1T and CS1Tm2 were comparable. However, TBK1 ubiquitination, which also regulates its activity, was increased only with CS1T overexpression (Fig. 4F). These results suggest that the impaired CS1Tm2-induced phosphorylation of IRF3/7 is associated with a defect in the ubiquitination of TBK1.

GST fusion constructs GST-CS1T and GST-CS1Tm2 were generated to identify a protein that bound to CS1T but not to CS1Tm2. GST pulldown followed by silver staining of the PAGE gel revealed a unique band of approximately 110 kDa in the GST-CS1T lane that was absent in the GST-CS1Tm2 lane (Fig. 5A). Mass spectrometry analysis showed that the band contained MIB2 peptides. MIB2, an E3 ubiquitin ligase belonging to the mindbomb family, was a strong candidate for the missing binding partner because TBK1 ubiquitination was impaired by CS1Tm2 (Fig. 4F). The binding of the MIB1 (homolog of MIB2) and MIB2 was evaluated by a GST pulldown assay. MIB1 bound to neither GST-CS1T nor GST-CS1Tm2, whereas MIB2 bound to only GST-CS1T (Fig. 5B). Next MIB2 binding with the conserved motif mutants was examined. MIB2 bound to GST-CS1T, GST-CS1Tm1, and GST-CS1Tm3 but not to GST-CS1Tm2 (Fig. 5C). These bindings correlated with the promoter assay data (Fig. 3B). Notably, MIB2 showed stronger binding to GST-CS1T than to GST-CS1Tm1 and GST-CS1Tm3, suggesting that, although the DLAIS motif is critical for the physical interaction between MIB2 and MAVS, other motifs also contribute to the binding. To confirm the binding without the influence of endogenous MAVS, the GST pulldown assay was performed in stable MAVS-knockdown 293T cells. MIB2 binding to GST-CS1T was not affected by MAVS knockdown, confirming that the DLAIS motif is required for the MIB2 interaction (Fig. 5D). Next, immunoprecipitation was performed to determine whether viral stimulation induced the endogenous binding between MAVS and MIB2. MIB2 was marginally coprecipitated with MAVS in uninfected cells; however, the coprecipitation was substantially enhanced in response to SV infection (Fig. 5E).

FIG 5.

Physical interaction between mindbomb E3 ubiquitin protein ligase 2 (MIB2) and mitochondrial antiviral signaling protein (MAVS). (A) Silver staining of glutathione S-transferase (GST) pulldown samples from 293T cells expressing empty vector, GST, GST-CS1T, and GST-CS1Tm2. A specific band detected only in the GST-CS1T lane (arrow in the enlarged box) contained MIB2 peptides identified by mass spectrometry. (B to D) GST pulldown assay. Interactions of MIB1 and MIB2 with CS1T or CS1Tm2 (B), MIB2 and the conserved motif mutants (C), and MIB2 and CS1T in MAVS-knockdown cells (D) are indicated. 293T (B and C) or stable MAVS-knockdown 293T (D) (bottom box, MAVS knockdown efficiency) cells were seeded in 10-cm-diameter dishes at a density of 10⁷ cells/dish 24 h before transfection. Each construct (8 μg) was transfected as indicated, and the lysates were subjected to GST pulldown (PD) at 24 h posttransfection. The pulldown samples were subjected to immunoblotting (IB) using the indicated antibodies. WCL, whole-cell lysates; GST-CS1T, GST-tagged CS1T; GST-CS1Tm1, GST-tagged K420S/P421R mutant; GST-CS1Tm2, GST-tagged L439S/A440R mutant; GST-CS1Tm3, GST-tagged E458S/N459R mutant; βActin, beta-actin antibody. (E) Virus-induced binding between endogenous MAVS and MIB2. 293T cells were seeded in 10-cm-diameter dishes at a density of 10⁷ cells/dish 24 h before Sendai virus (SV) infection (25 HA/ml). After 6 h of incubation, 293T cells were subjected to immunoprecipitation (IP) using an anti-MAVS antibody. After IP, IB was performed using the indicated antibodies.

In the experiment described above, the ubiquitination of TBK1 was impaired in the presence of CS1Tm2 compared to CS1T (Fig. 4F). The lysine (K)63-linked ubiquitination of TBK1 is important for the full activation of IRF3/7 (27). Therefore, the K63-linked ubiquitination induced by the overexpression of MIB2 was examined. For a control, a ubiquitin ligase functional mutant (MIB2 MT) was generated by a point mutation targeting the critical cysteine residue in the RING-type 1 domain (C983S) of wild-type MIB2 (MIB2 WT). The K63-linked ubiquitination of TBK1 by SV stimulation was increased in the presence of MIB2 WT, whereas no changes were observed with MIB2 MT (Fig. 6A). This TBK1 ubiquitination was attenuated by MIB2 knockdown (Fig. 6B). We then tested the phosphorylation of IRF3 induced by MAVS and MIB2. CS1T induced IRF3 phosphorylation, which was increased by MIB2 WT but not by MIB2 MT. CS1Tm2 did not induce IRF3 phosphorylation regardless of the presence or absence of MIB2 (Fig. 6C). The assessment of the viral stimulation-induced phosphorylation of IRF3 by MIB2 knockdown showed that IRF3 phosphorylation was attenuated by the knockdown (Fig. 6D). Taken together, these data indicate that MIB2 catalyzes the K63-linked ubiquitination of TBK1 via the MAVS DLAIS motif in response to SV.

FIG 6.

Mindbomb E3 ubiquitin protein ligase 2 (MIB2)-mediated TANK-binding kinase 1 (TBK1) ubiquitination and interferon regulatory factor 3 (IRF3) phosphorylation. (A and B) K63-linked ubiquitination of TBK1 by MIB2. 293T cells were seeded in 10-cm-diameter dishes at a density of 10⁷ cells/dish 24 h before transfection. Each construct (6 μg) was transfected as indicated, and the cells were infected with Sendai virus (SV; 10 HA/ml) at 24 h posttransfection. The cells were subjected to glutathione S-transferase (GST) pulldown (PD) after 6 h of incubation. IB, immunoblotting with the indicated antibody; HA-UBI K63, HA-tagged ubiquitin with all lysine-to-arginine mutations except K63; GST-TBK1, GST-tagged TBK1; WCL, whole-cell lysates; MIB2 WT, wild-type MIB2; MIB2 MT, MIB2 C983S mutant (no ubiquitin ligase activity). (C) Phosphorylation of IRF3 induced by CS1T and MIB2 overexpression. 293T cells were seeded in six-well plates at a density of 10⁵ cells/well 24 h before transfection. Each construct (200 ng) was transfected as indicated, and the cells were subjected to IB at 24 h posttransfection. pIRF3, IRF3 phosphospecific antibody; βActin, beta-actin antibody. (D) Interference with IRF3 phosphorylation by MIB2 knockdown. 293T cells were seeded in six-well plates at a density of 10⁵ cells/well 24 h before transfection. Each shRNA-expressing plasmid (2 μg/well) was transfected, and the cells were stimulated with SV (10 HA/ml) at 2 days posttransfection. The cells were subjected to IB using the indicated antibodies at 2 h after SV stimulation.

The effect of MIB2 overexpression on the activation of IRF7 was evaluated using the IFN-α6 promoter assay. The results showed increasing activation with increasing expression of MIB2 WT, whereas the MIB2 MT did not activate the promoter. Coexpression of NS3/4a abrogated promoter activation induced by MIB2 WT overexpression (Fig. 7A). We generated stable MIB2-knockdown 293T cells. IFN-α6 promoter activation by the SV was reduced in the MIB2-knockdown cells (Fig. 7B). The effect of MIB2 knockdown on viral replication was tested using GFP-expressing vesicular stomatitis virus (GFP-VSV) and FACS analysis. GFP-VSV replication was augmented in the MIB2-knockdown cells (Fig. 7C). These data indicated that MIB2 has a putative function in the activation of IRF3/7 by MAVS.

FIG 7.

Mindbomb E3 ubiquitin protein ligase 2 (MIB2) involvement in interferon signaling. (A) Alpha interferon 6 promoter (IFN-α6) assays with increasing amounts of wild-type MIB2 (MIB2 WT; left panel) or mutant MIB2 (MIB2 C983S mutant [MIB2 MT]; right panel). An IFN-α6 promoter mixture (200 ng of pGL3-IFN-α6 plus 2 ng of pRL-SV40 plus 10 ng of pCMBB-IRF7) and MIB2 WT or MT constructs at the indicated amounts were cotransfected with or without NS3/4a-expressing plasmids (20 ng/ml). The luciferase values were measured at 24 h posttransfection. (Bottom panels) Immunoblots showing the MIB2 WT or MT expression level in the cell lysates used for the promoter assay. RLU, relative luciferase units. The promoter assay data are presented as the means and standard deviations of the results of three independent experiments. βActin, beta-actin antibody. (B) Promoter assays in stable MIB2-knockdown 293T-cells. (Left box) Immunoblot showing the knockdown efficiency of MIB2. The cells were seeded in 24-well plates at a density of 10⁵ cells/well 24 h before transfection. The IFN-α6 promoter mixture was transfected, and Sendai virus (SV; 10 HA/ml) was infected at 24 h posttransfection. (Right panel) The luciferase activities were measured at 12 h postinfection. Control, parental 293T cells; MIB2 KD, stable MIB2-knockdown 293T cells. (C) FACS analysis of GFP-expressing vesicular stomatitis virus (VSV) replication in stable MIB2-knockdown 293T cells. Control or MIB2-knockdown 293 cells were seeded in a 12-well plate at a density of 10⁵ cells/well 24 h before GFP-VSV infection. The cells were infected with the indicated titers, and the infected cells were subjected to FACS analysis at 18 h postinfection. The numbers inside the graph show the percentages of GFP-positive cells. TU, GFP signal transduction unit.

DISCUSSION

In the present study, MIB2 was identified as an E3 ligase that modified TBK1 through K63-linked ubiquitination via MAVS, leading to IRF3/7 phosphorylation in response to viral infection based on the following findings. The MAVS region between aa 410 and 463 was required for IRF3/7 activation. This region contained a highly conserved motif, namely, DLAIS (aa 438 to 442). In the MAVS mutant targeting the DLAIS motif, IRF3/7 activation was lost, whereas NF-κB activation was maintained. The DLAIS mutation impaired TBK1 ubiquitination and IRF3/7 phosphorylation. MIB2 was identified as a binding protein of MAVS and lost binding capacity by the mutation targeting the DLAIS motif. MIB2 overexpression increased IRF3/7 activation via MAVS, whereas MIB2 knockdown decreased the activation in response to SV. MIB2 induced the K63-linked ubiquitination of TBK1 through the MAVS DLAIS motif in response to SV. Finally, MIB2 knockdown decreased the resistance to VSV replication.

MAVS is a key adaptor molecule that receives the PAMP signal from RLRs, amplifies the signal, and activates the downstream transcription factors to induce IFN-β secretion. However, the role of the stem region between the CARD and the TM domain remains unclear, and many cellular proteins bind to this region (28). The most prominent characteristic of the stem region is the abundance of proline residues. Proline has the “α-helix and β-sheet breaker” effect that originates from its bulky side chains, restricted backbone conformation, restrictions on the preceding residue conformation, and inability to function as a hydrogen bond donor (29). These features explain the paucity of the α-helix or β-sheet conformation in the stem region (Fig. 1). In particular, proline dipeptides restrict the ψ and φ angles. Five proline dipeptides are present at aa 103, 109, 119, 312, and 485. The abundance of proline also influences the physical interactions with other proteins. The binding of a proline-rich region does not require the usual energy for the conformational freezing. As a result, the binding is nonspecific but can be very rapid and strong. More than 30 MAVS-binding cellular proteins have been reported (30). Rapid and nonspecific binding derived from an entropy advantage and the large, flat hydrophobic surface of the proline-rich stem explain the abundance of MAVS-binding partners.

The posttranslational modifications of signaling proteins play important roles in the transduction of signals. Phosphorylation and ubiquitination are the major modifications involved in the activation of NF-κB and IRF3/7. TBK1 and IKKε are the main protein kinases for IRF3/7, although they may not contribute equally to the activation through MAVS. IKKε is not involved in the first responses to virus leading to IFN-β secretion in knockout mouse experiments (31). IKKε is a member of the ISG family and functions in the second amplification phase of type I IFN secretion. It was recently shown to bind to a TRAF3-binding site (aa 468 to 540), resulting in the downregulation of the IFN response (17). These reports support our data indicating that TBK1 is specifically involved in IRF3/7 activation through the MAVS DLAIS motif. The ubiquitination of TBK1 in response to RNA viral infection is crucial for the activation of IRF3 via MAVS (32). In particular, the K63-linked ubiquitination of TBK1 is critical for LPS- and RLR-induced IFN production (27). These reports underscore the role of E3 ligases in the signal transduction from MAVS to TBK1. TRAFs are well-studied E3 ligases involved in MAVS activity. MAVS associates with TRAF2, TRAF3, and TRAF6, leading to the activation of both IKKs and TBK1/IKKε (33). The MAVS proline-rich region (aa 100 to 150) contains two TRAF-interacting motifs, and the binding of TRAF2 and TRAF6 leads to the activation of NF-κB. Although an E3 ligase for TBK1 ubiquitination has not been completely elucidated, TRAF3 is a strong candidate involved in IRF3/7 activation through MAVS. TRAF3 was recently shown to bind to the MAVS PEENEY motif (aa 455 to 460), resulting in the activation of IFN antiviral response genes (17). However, we showed that, despite the high conservation of the PEENEY motif among mammalian species (Fig. 3A), the E458S and N459R mutations targeting the PEENEY motif did not have an effect on IRF3/7 activation (Fig. 3B). Furthermore, TRAF3 knockout does not impair the activation of IFN signaling (20), suggesting that another E3 ubiquitin ligase compensates for the loss of TRAF3. In the present study, we showed that MIB2 binds to MAVS (Fig. 5B and C), and the binding data correlated well with the promoter assay data (Fig. 3B). Therefore, MIB2 may be another E3 ligase involved in the activation of IRF3/7 through MAVS.

MIB2 was first cloned and described as an actin-binding, cytoskeleton-related protein whose expression is silenced in a number of melanoma cell lines (34). The protein contains multiple ankyrin repeats and RING finger domains that function as an E3 ubiquitin ligase. MIB2 modifies the intracellular region of Jagged-2, a Notch signaling ligand, by ubiquitination (35). MIB2 expression is primarily regulated by Snail, a transcriptional repressor, and activator protein-2, which stimulates the MIB2 promoter (36). In Drosophila, MIB2 regulates muscle development and maintains the integrity of fully differentiated muscles by preventing their apoptotic degeneration (37). MIB2 has a homolog, MIB1, that also functions as an E3 ubiquitin ligase and plays critical roles in Delta-Notch signaling and mouse neuronal development (38). However, in contrast to MIB1 results, MIB2-knockout mice show no apparent phenotype (39), and the role of MIB2 in mammals had remained unclear. The first report on the involvement of MIB2 in cytokine signaling characterized it as an activator of NF-κB and the mitogen-activated protein kinase promoter. A spliced form of MIB2 activated these promoters in a screen of cDNA libraries from 293T cells (40). Recently, the roles of MIB1 and MIB2 in the innate immune response to RNA viruses have been established by dynamic protein interactome network analysis (41). Here, we showed that MIB2 bound directly to MAVS via the DLAIS motif and modified TBK1 with K63-linked ubiquitin, leading to IRF3/7 phosphorylation and activation. MIB1 also catalyzes the formation of K63 ubiquitin chain on TBK1 (41). In contrast to MIB2, however, MIB1 did not bind to MAVS (Fig. 5B). Therefore, the activation mechanism of MIB1 might be different from that of MIB2. In addition to MIBs, there are a number of E3 ligases involved in the TBK1 ubiquitination: TRAF2, TRAF3, tumor necrosis factor alpha-induced protein 3 (TNFAIP3), and X-linked inhibitor of apoptosis (XIAP). The redundancy of E3 ligases suggests the complex upstream signals for TBK1 ubiquitination. MIB2 is substantially expressed in heart, brain, liver, kidney, skin, and small intestine; MIB1 is expressed in testis, skin, and small intestine (42). These expression patterns indicate that different E3 ligases control the TBK1 ubiquitination in the context of cell types or tissues.

We hypothesized the model of MAVS-mediated IRF3/7 activation on the basis of the literature and our experimental results. Based on the features associated with proline abundance, the stem structure may make an efficient binding platform for many interacting proteins. The recognition of viral PAMPs by the helicase domain of RLRs releases the CARD, allowing it to bind to the CARD of MAVS. This CARD interaction induces the formation of a MAVS oligomer. MAVS CARD aggregation has a tendency to propagate in a manner similar to that seen with prion proteins (10). This finding suggests that MAVS functions as an amplification platform for PAMP signaling (43). CARD aggregation may extend the stem region to make the TIM and the DLAIS motif accessible to E3 ligases simultaneously. TRAF2 and TRAF6 bind the TIM and facilitate the phosphorylation of IκBα, which targets it for proteasomal degradation, thereby liberating NF-κB. MIB2 binds to the DLAIS motif, resulting in the K63-linked ubiquitination of TBK1, and the ubiquitin-modified TBK1 can then phosphorylate and activate IRF3/7.