|Home | About | Journals | Submit | Contact Us | Français|
Arenaviruses perturb innate antiviral defense by blocking induction of type I interferon (IFN) production. Accordingly, the arenavirus nucleoprotein (NP) was shown to block activation and nuclear translocation of interferon regulatory factor 3 (IRF3) in response to virus infection. Here, we sought to identify cellular factors involved in innate antiviral signaling targeted by arenavirus NP. Consistent with previous studies, infection with the prototypic arenavirus lymphocytic choriomeningitis virus (LCMV) prevented phosphorylation of IRF3 in response to infection with Sendai virus, a strong inducer of the retinoic acid-inducible gene I (RIG-I)/mitochondrial antiviral signaling (MAVS) pathway of innate antiviral signaling. Using a combination of coimmunoprecipitation and confocal microscopy, we found that LCMV NP associates with the IκB kinase (IKK)-related kinase IKKε but that, rather unexpectedly, LCMV NP did not bind to the closely related TANK-binding kinase 1 (TBK-1). The NP-IKKε interaction was highly conserved among arenaviruses from different clades. In LCMV-infected cells, IKKε colocalized with NP but not with MAVS located on the outer membrane of mitochondria. LCMV NP bound the kinase domain (KD) of IKKε (IKBKE) and blocked its autocatalytic activity and its ability to phosphorylate IRF3, without undergoing phosphorylation. Together, our data identify IKKε as a novel target of arenavirus NP. Engagement of NP seems to sequester IKKε in an inactive complex. Considering the important functions of IKKε in innate antiviral immunity and other cellular processes, the NP-IKKε interaction likely plays a crucial role in arenavirus-host interaction.
Arenaviruses are a large and diverse family of viruses of relevance as both powerful model systems for experimental virology and clinically important human pathogens (6). The prototypic arenavirus lymphocytic choriomeningitis virus (LCMV) has been instrumental in many landmark studies that provided fundamental concepts in molecular virology, virus-host cell interaction, viral pathogenesis, and viral immunology (38). On the other hand, several arenaviruses have emerged as causative agents of severe hemorrhagic fevers (HF) with high morbidity and significant mortality in humans, posing serious public health problems within their regions of endemicity (12). The arenavirus with the highest impact in human health is Lassa virus (LASV), which causes several hundred thousand infections per year in Western Africa, with thousands of deaths (32). There is currently no vaccine available, and therapeutic options are limited, resulting in 15% to 30% mortality in hospitalized patients. Likewise, in South America, the arenaviruses Junin virus (JUNV), Machupo virus, Guanarito virus, and Sabia virus have emerged as causative agents of severe HF disease (41).
Arenaviruses are enveloped viruses with a bisegmented negative-strand RNA genome and a nonlytic life cycle restricted to the cell cytoplasm (9). The S RNA encodes the viral glycoprotein precursor (GPC) and the nucleoprotein (NP), whereas the L RNA encodes the viral RNA-dependent RNA polymerase L and the matrix protein Z. The arenavirus GPC undergoes proteolytic processing by the cellular protease S1P to yield GP1, which is involved in receptor binding, and GP2, which mediates a pH-dependent fusion event required for arenavirus cell entry (2, 23).
A hallmark of severe arenavirus infection in humans is the inability of the patient's innate and adaptive immune systems to contain the virus, resulting in uncontrolled virus multiplication that often leads to a fatal outcome (12). Thus, pathogenic arenaviruses seem able to subvert the mechanisms of innate pathogen recognition by the infected host (1, 13, 26). Another characteristic feature of arenaviruses is their ability to establish persistent infections in their natural rodent reservoirs and in a wide range of mammalian cells in vitro. Arenavirus persistence is commonly associated with continued replication and expression of viral proteins. Intriguingly, despite widespread viral replication and high viral loads in mice persistently infected with LCMV, only a modest type I interferon (IFN) response is mounted (48), suggesting that the virus escapes innate detection and/or can efficiently counteract mechanisms of innate defense (4). Accordingly, LCMV was shown to be able to interfere with induction of type-I IFNs in the host cell by blocking the activation of transcription factor IRF3 (31), and the NP was identified as the viral component that acted as an IFN antagonist (31). This IFN-counteracting activity is highly conserved among arenaviruses (30), and it was mapped to the C-terminal region of NP (29), which contains a 3′-5′ exonuclease domain (15, 44). Mutations that abrogated the 3′-5′ exonuclease domain also abolished the ability of NP to suppress the induction of type I IFNs (15, 29, 44), indicating a link between the two activities.
The cytosolic pathogen recognition receptors retinoic acid-inducible gene I (RIG-I) and melanoma-differentiation-associated gene 5 (MDA5) have been shown to recognize arenavirus RNA (14, 27, 28, 56). Upon activation, RIG-I and MDA5 associate with the mitochondrial signaling adapter MAVS/IPS-1/Cardif/VISA (20, 34, 51, 55). MAVS consists of an amino-terminal CARD domain, a proline-rich region (PRR) in the middle of the protein, and a C-terminal transmembrane domain that localizes the protein to the outer mitochondrial membrane. Binding of active RIG-I/MDA5 induces aggregation of MAVS to activate innate antiviral responses (18). Aggregation of MAVS leads to the generation of a signaling platform through recruitment of multiple signaling molecules, including the classical IκB kinase (IKK) complex IKKα/IKKβ/NEMO and the nonclassical IKK-related kinases TANK-binding kinase (TBK-1) and IKKε (20, 34, 35, 51, 55). The classical IKK complex IKKα/IKKβ/NEMO is involved in activation of nuclear factor κB (NF-κB), whereas TBK-1 and IKKε activate the interferon regulatory factors (IRF) by direct phosphorylation of IRF3 and IRF7 (17). Upon activation, NF-κB, IRF3, and the activator protein AP1 synergistically activate transcription of beta IFN (IFN-β), which acts in a autocrine and paracrine manner, binding to its cognate type I IFN receptor (IFNAR) and inducing a second wave of type I IFN expression (10). Upon receptor binding, type I IFNs activate the JAK/STAT signaling pathway (50) and induce the expression of hundreds of IFN-stimulated genes (ISGs), establishing an antiviral state within the cell (45, 47).
In the present study, we sought to identify cellular factors targeted by arenavirus NP to disrupt the host innate antiviral signaling. Using a combination of biochemical assays and confocal microscopy, we have demonstrated that the LCMV NP specifically targets IKKε. We found that this NP-IKKε interaction is highly conserved among arenaviruses from different clades. LCMV NP associated with the kinase domain (KD) of IKKε and blocked its activity without undergoing phosphorylation. This newly uncovered NP-IKKε interaction may play an important role in arenavirus-host cell interaction during acute and persistent infection.
Human embryonic kidney cells (HEK293 and HEK293T) and human lung adenocarcinoma epithelial cells (A549) were maintained in Dulbecco's modified Eagle medium (DMEM) (Gibco BRL) supplemented with 10% fetal calf serum (FCS) at 37°C and 5% CO2. To infect cells, we used lymphocytic choriomeningitis virus Armstrong 53b (LCMV) as described in reference 22 and recombinant Sendai virus-green fluorescent protein (SeV-GFP) as described in reference 36 (kindly provided by Laurent Roux, University of Geneva, Geneva, Switzerland). SeV Cantell strain was grown in 10-day-old embryonated eggs, as previously described (21, 31). The viruses were diluted in 10% FCS–DMEM for LCMV or DMEM for SeV and added to the cells for 1 h at 37°C. The inoculum was then removed and replaced by fresh cell medium. For SeV-infected cells, GFP expression was used as a control for infections.
Rat monoclonal antibody (MAb) antihemagglutinin (anti-HA) (high affinity) was from Roche Applied Science (Rotkreuz, Switzerland). Mouse MAb anti-α-tubulin and mouse MAb anti-FLAG (M2) were from Sigma-Aldrich (St. Louis, MO). Mouse IgG1 anti-IKKε was a kind gift from John Hiscott, McGill University, Montreal, Canada. The nucleoprotein of LCMV (LCMV NP) was detected using guinea pig serum anti-LCMV NP or mouse MAb anti-LCMV NP (1.1.3), as previously described (7). Rabbit and goat polyclonal antibodies (pAb) anti-IRF3 were from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit pAb anti-phosphoIRF3 was purchased from Cell Signaling Technology (Danvers, MA). Mouse MAb anti-GFP and mouse IgG2b anti-MAVS were obtained from Clontech (Mountain View, CA) and Alexis Biochemicals (Lausen, Switzerland), respectively. Alexa Fluor 594 goat anti-mouse IgG1, Alexa Fluor 488 goat anti-mouse IgG2b, Alexa Fluor 594 F(ab′)2 fragment of goat anti-rabbit IgG, and Pacific Blue streptavidin were all purchased from Molecular Probes (Eugene, OR), while biotinylated donkey anti-guinea pig IgG was from Jackson ImmunoResearch (Suffolk, United Kingdom). Dynabeads magnetic beads were purchased from Invitrogen.
The expression plasmids for HA-tagged arenavirus (LCMV [wt], Lassa virus [LASV], Whitewater Arroyo virus [WWAV], Latino virus [LATV], and Junin virus [JUNV]) NPs and HA-tagged LCMV Z were described previously (30, 31). The expression plasmids for HA-tagged LCMV NP mutants (LCMV NP D382A, LCMV NP E384A, LCMV NP D459A, LCMV NP H517A, and LCMV NP D522A) were also previously described (29, 39). Expression plasmids for FLAG-tagged IKKε and TBK-1 were kindly provided by Margot Thome Miazza, University of Lausanne, Switzerland. The truncated version of IKKε (IKKε KD) encoding the N-terminal kinase domain (amino acids 1 to 315) was generated by introducing a STOP codon at position 316. The forward primer was T7 universal primer 5′-TAA-TAC-GAC-TCA-CTA-TAG-GG-3′, and the reverse primer was 5′-ATA-TTC-TAG-ATT-AGG-AGA-AGA-CAT-GGA-CGA-C-3′ comprising the STOP codon and XbaI restriction site, allowing further subcloning. Expression plasmids for FLAG-tagged MAVS and RIG-I were kind gifts of Jürg Tschopp, and an expression plasmid for TRAF3 was a kind gift from Pascal Schneider, both of the University of Lausanne, Lausanne, Switzerland. A GFP-IRF3 fusion protein-encoding plasmid and a FLAG-tagged IRF3 expression plasmid were kindly provided by John Hiscott, McGill University, Canada, and Brian Seed, Harvard Medical School, Boston, MA, respectively.
If not indicated otherwise, HEK293 cells were transfected using calcium phosphate as described previously (34). For protein expression analysis, cells were lysed with sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample-loading buffer. The samples were separated by SDS-PAGE, and protein expression was assessed by Western blot analysis. For immunoprecipitation (IP) of HA-tagged NP, cells were lysed at 48 h posttransfection with lysis buffer (50 mM Tris, 280 mM NaCl, 2 mM EGTA, 10% [wt/vol] glycerol, 0.2 mM EDTA, 50 mM NaF, 1% [wt/vol] NP-40) supplemented with protease inhibitor cocktail (Roche, Rotkreuz, Switzerland). For IRF3 immunoprecipitation, cells were lysed with lysis buffer (20 mM Tris, 137 mM NaCl, 2 mM EDTA, 10% [wt/vol] glycerol, 1% [wt/vol] NP-40) complemented with protease and phosphatase inhibitor cocktail (Roche, Rotkreuz, Switzerland). Cell lysates were incubated for 2 h on a rotation wheel at 4°C with protein G-coated magnetic Dynabeads (Invitrogen, Carlsbad, CA) previously incubated with the appropriate antibody to precipitate the target protein. After four wash steps with lysis buffer, magnetic beads were resuspended in SDS-PAGE sample buffer and heated to elute the bound proteins. Eluted proteins were then separated by SDS-PAGE and analyzed by Western blotting.
Activation of the IRF3-dependent promoter reporter plasmid (p55C1B-FF) after SeV infection has been previously described (31). Briefly, 2 × 105 HEK293T cells (12-well plate format, replicates) were cotransfected, in suspension, using calcium phosphate with 0.5 μg of p55C1B-FF and 100 ng of C-terminally HA-tagged wild-type and 3′-5′ exonuclease single-amino-acid LCMV NP pCAGGs expression plasmids, along with 50 ng of an expression plasmid encoding Renilla luciferase (RL) under the control of the simian virus 40 promoter (pSV40-RL) to normalize transfection efficiencies. Empty pCAGGs plasmid was used as a negative control. Twenty-four hours posttransfection, cells were mock or SeV infected (multiplicity of infection [MOI] = 3) for 1 h at room temperature. At 16 to 18 h postinfection (p.i.), luciferase reporter activities and protein expression were analyzed using cell lysates. Luciferase activities were determined using the Promega (Fitchburg, WA) dual-luciferase reporter assay and a Lumicount luminometer. Reporter gene activation was calculated as fold induction (activation) over the noninduced, empty pCAGGs multiple-cloning site (MCS)-transfected control. Protein expression was determined by Western blotting using anti-HA (Sigma) or anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH) (Abcam, Cambridge, United Kingdom) polyclonal antibodies and a horseradish peroxidase (HRP)-conjugated anti-rabbit IgG secondary antibody (GE, Amersham Biosciences, Amersham, United Kingdom). Protein bands were visualized using a chemiluminescence detection kit (SprayGlo) and autoradiography films (Denville Scientific Inc., Metuchen, NJ) according to the manufacturer's instructions.
HEK293T cells (2 × 105) were cotransfected, in suspension, by the use of calcium phosphate with 1 μg of pEGFP-C1-hIRF3 together with 2 μg of the indicated C-terminally HA-tagged pCAGGs LCMV NP expression plasmids, as previously described (21, 31). LCMV Z-HA pCAGGs was used as a negative control. Twenty-four hours posttransfection, cells were infected with SeV (MOI = 3) for 1 h at room temperature and, 16 to 18 h postinfection, cells were fixed with 0.2% (vol/vol) glutaraldehyde containing 2.5% (vol/vol) formaldehyde–1× PBS for 10 min at 4°C and permeabilized using 0.1% (vol/vol) Triton X-100–1× PBS for 10 min at room temperature. After being washed with 1× PBS and blocked overnight at 4°C with 10% (wt/vol) bovine serum albumin (BSA)–1× PBS, cells were incubated with an anti-HA polyclonal antibody (Sigma) diluted in 5% (wt/vol) BSA blocking solution for 1 h at 37°C, washed, and incubated with a Rhodamine-red-conjugated anti-rabbit IgG (H+L; Jackson ImmunoResearch), and 4′,6′-diamidino-2-phenylindole (DAPI; Research Organics, Cleveland, OH) for cellular nucleus staining, in blocking solution for 30 min at 37°C. For the quantification of the nuclear translocation of IRF3, HA-positive cells (indicative of protein expression) were considered. To calculate percentages of cells with nuclear translocation of GFP-IRF3, a total of 50 cells in 3 to 7 nonoverlapping fields were counted under ×20 magnification, in replicate experiments.
HEK293 cells were seeded the day prior to transfection, followed by transfection using calcium phosphate. For IP of FLAG-tagged IKKε and IRF3, lysis buffer containing the following ingredients was used: 50 mM Tris, 150 mM NaCl, 1 mM EGTA, 1 mM β-glycerophosphate, 1 mM EDTA, 1% (wt/vol) Triton X-100, and 1 mM orthovanadate supplemented with protease inhibitor cocktail (Roche, Rotkreuz, Switzerland). For the in vitro kinase assay, the proteins were incubated for 1 h at 30°C with 10 μCi of [γ-32P]ATP per reaction in kinase assay buffer (50 mM Tris, 12 mM MgCl2, 1 mM β-glycerophosphate, 100 μM ATP, 1 mM orthovanadate) in a total volume of 50 μl. SDS-PAGE sample buffer was added to stop the reaction. Proteins were then separated by SDS-PAGE and developed by autoradiography.
Total RNA was purified with an RNeasy minikit (Qiagen, Chatsworth, CA), and cDNA was synthesized using a QuantiTect reverse transcription kit (Qiagen). TaqMan probes specific for IFN-β (Hs01077958_s1), IKKε (IKBKE; Hs01063858_m1), TBK1 (Hs00179410_m1), and GAPDH (Hs99999905_m1) were obtained from Applied Biosystems (Foster City, CA). Real-time PCR (RT-PCR) was performed using a StepOne real-time PCR system (Applied Biosystems), and gene expression levels relative to GAPDH were determined according to the 2−ΔΔCT method (24).
RNA interference (RNAi) was performed using validated small interfering RNAs (siRNAs) for IKKε (IKBKE; SI02655324) and TBK1 (SI02224411) and scrambled siRNA (1027280) as a control from Qiagen (Basel, Switzerland). Briefly, 2 ×106 HEK293 cells were reverse transfected with 15 nM siRNA using a10-cm-diameter dish and Lipofectamine RNAiMAX (Invitrogen, Paisley, United Kingdom) according to the manufacturer's recommendation. Twenty-four hours after transfection cells, were replated, and 48 h after transfection, cells were infected with SeV (MOI = 0.1) or mock infected to be assessed for IFN-β mRNA levels (RT-qPCR) or IRF3 phosphorylation (Western blotting) 24 h postinfection. Depletion of IKKε and TBK1 was confirmed by RT-qPCR using specific probes or Western blot analysis using specific antibodies.
To perform colocalization studies, A549 cells were seeded on glass 8-well LabTek slides and infected with LCMV at an MOI of 1. At 48 h postinfection, cells were fixed with 2% (wt/vol) formaldehyde–PBS for 15 min at room temperature and washed with PBS. Cells were then permeabilized for 30 min at room temperature with 0.1% (wt/vol) saponin, 10% (vol/vol) goat serum, and 100 mM glycine–PBS. Primary and secondary antibodies were diluted in 0.1% (wt/vol) saponin and 1% (vol/vol) goat serum–PBS and incubated overnight at 4°C and 1 h at room temperature, respectively. Before acquisition, LabTek slides were mounted using Mowiol (Sigma). Image acquisition was performed with a Zeiss LSM710 Quasar confocal microscope equipped with a Plan Apochromat lens (63×, 1.2 numerical aperture [NA] objective), 405-nm diode laser, argon lasers (458, 476, 488, and 514 nm), and 561-nm diode-pumped solid-state (DPSS) laser. All images for each data set were acquired the same day with the same microscope settings. Images were first deconvolved using Huygens Essential software (SVI, Hilversum, Netherlands) and then analyzed for colocalization with Imaris 7.2 software (Bitplane, Zürich, Switzerland).
Arenavirus NP can prevent the induction of type I IFNs by blocking activation of IRF3 (30, 31). Specifically, expression of NP prevented the nuclear translocation of IRF3, which is a prerequisite for activation of the IFN-β promoter (30, 31). However, it was unclear at which level arenavirus NP interfered with IRF3 activation and consequent IFN-β expression. To further investigate this issue, we first studied the effect of arenavirus infection on IRF3 phosphorylation. For this, HEK293 cells were infected with the prototypic arenavirus LCMV ARM 53b (henceforth referred to as LCMV), and the expression of IFN-β mRNA was monitored over time by quantitative RT-PCR (RT-qPCR). In parallel, the expression levels of the viral NP were detected by Western blotting. In cells infected at an MOI of 10, we observed a transient induction of IFN-β mRNA at around 8 h p.i., followed by a marked decrease thereafter (Fig. 1A). The peak in IFN-β mRNA correlated with the first detection of NP, whereas decreased levels of IFN-β mRNA at 12 h p.i. correlated with increased NP levels in infected cells (Fig. 1A and andBB).
We next examined the effect of arenavirus infection on phosphorylation of IRF3. HEK293 cells were infected (MOI = 0.1) with LCMV or mock infected, and cells were infected 24 h later with Sendai virus (SeV), a strong inducer of RIG-I/MAVS-induced IRF3 phosphorylation and IFN expression (34), or mock infected. Phosphorylation of IRF3 was detected by Western blotting with an antibody that specifically recognized the phosphorylated form of IRF3 and compared to total IRF3 levels in the cells. In mock-infected cells, infection with SeV induced phosphorylation of IRF3, whereas previous infection with LCMV prevented IRF3 phosphorylation upon SeV infection (Fig. 1C). To address the role of NP in the block of IRF3 phosphorylation in LCMV-infected cells, we transfected HEK293 cells with recombinant LCMV NP using calcium phosphate, a method that avoids induction of a type I IFN response (31, 34). At 24 h posttransfection, cells were infected with SeV and IRF3 phosphorylation was assessed by Western blotting. SeV-induced phosphorylation of IRF3 was inhibited in NP-transfected cells (Fig. 1C).
Sendai virus is a strong inducer of the RIG-I/MAVS signaling pathway (34), and recent studies implicated the RIG-I/MDA5/MAVS pathway in innate detection of arenaviruses (14, 27, 28, 56). In response to viral infection, activated RIG-I helicases associate with MAVS located on the outer mitochondrial membrane, followed by assembly of a signaling complex whose activation results in IRF3 phosphorylation. The observed blockade of IRF3 phosphorylation by LCMV NP (Fig. 1) raised the possibility that NP targets one or several components of the MAVS-associated signaling complex upstream of IRF3. To identify potential targets, we probed possible interactions between LCMV NP and selected components of the RIG-I/MAVS signaling pathway implicated in activation of IRF3, including IRF3 itself, RIG-I, MAVS, the adapter protein TRAF3, and the kinases TBK-1 and IKKε, both implicated in IRF3 phosphorylation. To this end, HEK293 cells were cotransfected using calcium phosphate with an HA-tagged form of recombinant LCMV NP, together with FLAG-tagged recombinant RIG-I, MAVS, TRAF3, TBK-1, and IKKε, as well as GFP-tagged IRF3. After 48 h, cell lysates were prepared and assayed by coimmunoprecipitation (co-IP) using anti-HA magnetic beads. Isolated immunocomplexes were subsequently probed by Western blotting with antibodies to HA to confirm IP of LCMV NP and antibodies to FLAG tag and to GFP to detect binding partners. In these co-IP studies, we detected a strong and specific interaction of LCMV NP with IKKε but not with the closely related TBK-1 (Fig. 2A). IP of NP resulted sometimes in a weak co-IP of TRAF3 (Fig. 2A), but this was not consistently found. RIG-I, MAVS, and IRF3 did not show detectable binding to NP under our experimental conditions (Fig. 2A). To validate the detected interaction of LCMV NP with endogenous IKKε, HEK293 cells were transfected with LCMV NP, followed by infection with SeV. At 24 h p.i., IP of the HA-tagged LCMV NP was performed and the immunocomplex probed for IKKε with a specific antibody. We consistently detected a specific interaction between the viral NP and endogenous IKKε in SeV-infected cells (Fig. 2B). In sum, probing of a selected set of components of the RIG-I/MAVS pathway involved in IRF3 phosphorylation revealed a strong direct or indirect interaction of LCMV NP with IKKε but not with the closely related TBK-1.
We next investigated if the specific interaction of NP with IKKε was conserved among arenaviruses from different phylogenetic clades. For this purpose, we examined the interaction of IKKε with NPs of the Old World arenavirus LASV, endemic in Western Africa, and New World arenaviruses from clades A (White Water Arroyo virus [WWAV]), B (Junin virus [JUNV]), and C (Latino virus [LATV]). HEK293 cells were cotransfected with FLAG-tagged IKKε or TBK-1, together with HA-tagged versions of the viral NPs. Cell lysates were precipitated with anti-HA magnetic beads and immunocomplexes analyzed by Western blotting as described above. All arenavirus NPs tested associated specifically with IKKε but not with TBK-1 (Fig. 3A). To ensure that our co-IP assay conditions allowed detection of specific interactions, we included as a control an HA-tagged form of the LCMV Z protein. Only NP, and not Z, underwent co-IP with IKKε (Fig. 3B). In our experiments, we noticed apparent differences in the binding of some NPs to IKKε. However, this has not been consistently observed and did not correlate with IFN antagonism of the NPs.
The interaction of arenavirus NPs with IKKε may affect the stability of the kinase and thereby prevent IRF3 phosphorylation. To address this possibility, we examined the expression of IKKε in cells infected with LCMV over time. HEK293 cells were either infected with LCMV or mock infected. At different time points, cells were lysed and expression of IKKε and NP was examined by Western blotting. Normalization of IKKε expression levels to uninfected cells revealed no consistent reduction of IKKε expression during 48 h of infection, despite a dramatic increase in NP expression over the course of infection (data not shown). This result indicated that NP expression does not result in rapid degradation of IKKε. We therefore explored the alternative possibility that LCMV infection might affect the ability of IKKε to be recruited to the MAVS signaling platform on mitochondria. To address this possibility, we examined the cellular distribution of NP, IKKε, and MAVS in LCMV-infected cells by confocal microscopy. To this end, we infected A549 cells with LCMV and at 48 h p.i. cells were fixed and specimens examined by confocal laser scanning microscopy. The colocalization of NP, IKKε, and MAVS was assessed as described in Materials and Methods. In LCMV-infected cells, IKKε showed a strong colocalization with NP, but not MAVS, as confirmed by the quantitative image analyses depicted in the corresponding scatter plots (Fig. 4).
The specific ability of arenavirus NP to target IKKε, but not TBK-1, was unexpected, because analysis of embryonic fibroblasts derived from mice deficient in IKKε and TBK-1 suggested a predominant role for TBK-1, rather than IKKε, in IFN induction in response to double-stranded RNA (dsRNA) and virus infection (16, 33, 40). To address the relative contributions of IKKε and TBK-1 to virus-induced IRF3 activation and IFN-β induction in our experimental system, we depleted IKKε and TBK-1 by RNA interference (RNAi) using validated siRNAs (Materials and Methods). Transfection of HEK293 cells with specific siRNAs targeting IKKε or TBK-1 resulted in a marked reduction in the level of the target mRNA (Fig. 5A). Next, we addressed the roles of IKKε and TBK-1 in IFN-β induction upon infection with SeV. For this purpose, we used siRNA treatment to deplete HEK293 cells of IKKε or TBK-1, followed by infection with SeV. We assessed levels of IFN-β mRNA by RT-qPCR as shown in Fig. 1A. Depletion of IKKε or TBK-1 resulted in a significant impairment of IFN-β induction upon infection with SeV (Fig. 5B), suggesting that both kinases are required for an optimal IFN response. In contrast to other systems, such as murine embryonic fibroblasts, where TBK-1 seems to be the preeminent kinase linked to antiviral signaling (16, 33, 40), we consistently found a stronger effect of depletion of IKKε (Fig. 5B), suggesting the existence of cell-type-specific differences. We further observed important quantitative differences in the induction of IFN-β in response to infection with SeV (Fig. 5B) and LCMV (Fig. 1A), with a circa-50-fold-stronger induction of IFN-β transcription upon infection with SeV. We also examined the role of IKKε in virus-induced phosphorylation of IRF3. For this, we depleted HEK293 cells of IKKε, followed by infection with SeV. We assessed IRF3 phosphorylation using a phosphorylation-specific antibody as described above (Fig. 1C). Depletion of IKKε at >80% at the protein level resulted in a marked reduction of IRF3 phosphorylation in SeV-infected cells (Fig. 5C). Together, these data indicate that, in HEK293 cells under our experimental conditions, both TBK-1 and IKKε contributed to optimal IFN-β induction in response to SeV.
Recent studies demonstrated that the C-terminal domain of LCMV NP is required for its ability to act as an IFN antagonist (29). Residues D382 and E384 in the DIEGR motif (29), as well as D459, H517, and D522, which are part of a recently described 3′-5′ exonuclease motif highly conserved within the C terminus of arenavirus NP, were particularly important (15, 29, 44). When expressed in HEK293 cells, all NP mutants showed expression levels similar to the wild-type level (Fig. 6A). To confirm their defect in preventing type I IFN induction via IRF3, HEK293 cells were cotransfected with wild-type NP or each of the mutant NPs, together with the reporter construct p55C1B-FF, driving expression of a Firefly luciferase reporter from an IRF3-regulated promoter and the pSV40-RL plasmid encoding Renilla luciferase under the control of the simian virus 40 promoter to normalize transfection efficiencies. Twenty-four hours posttransfection, cells were infected with SeV, and 16 to 18 h p.i., activation of IRF3 was monitored by detection of luciferase reporter activities in a luminescence assay. Consistent with previous findings (29), all mutants tested were defective in their ability to block IRF3 induction by SeV infection (Fig. 6B). Results were confirmed with a reporter plasmid encoding the monomeric red fluorescent protein under the control of the IFN-β promoter (pIFNβ-mRFP/CAT) (31) (data not shown).
In a complementary approach, we further monitored the ability of our NP mutants to prevent the nuclear translocation of IRF3. To this end, wild-type NP and mutants were cotransfected with a transcriptional fusion of IRF3 protein with green fluorescent protein (GFP-IRF3). Twenty-four hours posttransfection, cells were infected with SeV as described above. At 16 to 18 h postinfection, cells were fixed and permeabilized and intracellular staining for NP was performed using an anti-HA polyclonal antibody (red), along with DAPI staining to visualize nuclei (blue). Nuclear translocation of GFP-IRF3 was observed (green) in NP-expressing cells, and cells showing nuclear accumulation of GFP-IRF3 were scored. Consistent with their inability to block IRF3-dependent activation of the reporter construct p55C1B-FF, none of the 3′-5′ exonuclease mutants significantly prevented nuclear translocation of GFP-IRF3 (Fig. 6C and andDD).
To assess the ability of the NP mutants to bind to IKKε, we performed co-IP. While wild-type LCMV NP strongly interacted with IKKε, all NP mutants tested showed consistently reduced IKKε interaction (Fig. 6E). Gross protein misfolding is highly unlikely as an explanation for the defect of these NP mutants in interacting with IKKε, as many of these NP mutants were previously shown to be active in an LCMV minigenome rescue assay (29). Our data suggest that LCMV NP binding to IKKε involved the C-terminal domain of NP and was affected by mutations in the 3′-5′ exonuclease motif.
The IKK-related kinases IKKε and TBK-1 have a modular structure, consisting of a C-terminal kinase domain, a ubiquitin-like domain (UBL), and an N-terminal coiled-coil domain (17). To define the domain(s) of IKKε involved in binding of the viral NP, we tested the ability of the IKKε kinase domain (KD) to bind NP. FLAG-tagged full-length IKKε and IKKε KD (Fig. 7A) were cotransfected with LCMV NP. As a negative control, we included FLAG-tagged TBK-1. At 48 h posttransfection, cell lysates were prepared and subjected to co-IP. Detection of FLAG-tagged proteins in total cell lysates revealed similar expression levels for full-length IKKε and the IKKε KD (Fig. 7B). We detected a strong association of LCMV NP with the IKKε KD, with signal intensities comparable to those of full-length IKKε. As expected, no interaction between NP and TBK-1 was detected.
A major function of the arenavirus NP is the packaging of the viral RNA into the viral nucleocapsid, a process that is likely mediated by charged residues at the surface of NP (15, 44). To assess a possible role of charged residues on NP in the binding to IKKε, we tested the sensitivity of the NP-IKKε interaction with respect to enhanced ionic strength. For this purpose, we performed co-IP of LCMV NP with IKKε in the presence of increasing salt concentrations in the washing buffer. While the IP of NP was not affected by NaCl concentrations of up to 1,000 mM, the NP-IKKε interaction assessed by co-IP was sensitive to enhanced salt concentrations, with markedly reduced binding at 280 mM NaCl and higher (Fig. 7C). The efficient dissociation of the NP-IKKε complex caused by increased ionic strength suggests an important contribution of ionic bonds to the interaction between NP and IKKε.
The binding of LCMV NP to the kinase domain of IKKε raised the possibility that NP may affect the catalytic activity of IKKε toward IRF3. In addition, or alternatively, the association of NP with IKKε may result in phosphorylation of NP, which may behave like a pseudosubstrate for the kinase. To address these possibilities, we sought to compare the levels of catalytic activity of IKKε, either alone or in complex with NP. In a first step, HEK293 cells were cotransfected with FLAG-tagged IKKε and HA-tagged NP or transfected with FLAG-tagged IKKε alone. At 48 h posttransfection, IKKε/NP-cotransfected cells were lysed and IP with anti-HA magnetic beads was performed. IKKε singly transfected cells were subjected to IP with anti-FLAG magnetic beads. Immunocomplexes were probed for IKKε in Western blotting, revealing efficient IP of IKKε, as well as co-IP of IKKε with NP (Fig. 8A). Immunocomplexes of the IKKε IP and the IKKε/NP co-IP were then mixed with recombinant IRF3 purified from separate, transiently transfected HEK293 cells. In vitro kinase assays using [γ-32P]ATP were performed as described in Materials and Methods. The kinase reaction was stopped by adding SDS-PAGE sample buffer. Products were separated by SDS-PAGE and developed by autoradiography. Consistent with published data (53), IKKε underwent apparent autophosphorylation in anti-FLAG immunocomplexes derived from IKKε singly transfected cells (Fig. 8B). In contrast, no autophosphorylation was detected with the IKKε present in the immunocomplex from the NP/IKKε co-IP (Fig. 8B). IKKε derived from singly transfected cells was able to efficiently phosphorylate IRF3, which was not the case for IKKε in complex with NP (Fig. 8B). The absence of any detectable radiolabeled band in the kinase assay performed on the NP/IKKε co-IP indicates that IKKε in complex with NP was unable to phosphorylate itself, IRF3, or NP and that engagement of the LCMV NP inhibited the kinase activity of IKKε.
In our present study, we sought to identify cellular factors targeted by arenavirus NP to mediate suppression of innate antiviral signaling. The central findings derived from our studies are that (i) LCMV NP prevents activation of IRF3 by blocking phosphorylation of the transcription factor; (ii) LCMV NP specifically targets the IKK-related kinase IKKε but not TBK-1; (iii) the specific binding of NP to IKKε is conserved within the Arenaviridae; (iv) LCMV NP associates with the kinase domain of IKKε involving NP's C-terminal region; and (v) binding of LCMV NP inhibits the kinase activity of IKKε.
Previous studies demonstrated the ability of arenavirus NP to prevent activation and nuclear translocation of IRF3 in response to viral infection (30, 31). Here we demonstrate that, upon infection with SeV, which activates the RIG-I/MAVS pathway, LCMV NP blocks IRF3 phosphorylation required for dimerization and subsequent nuclear translocation of the transcription factor (17). Considering the proposed role of the RIG-I/MDA5/MAVS pathway in innate detection of arenaviruses (14, 27, 28, 56), we probed the interaction of LCMV NP with selected components of the RIG-I/MDA5/MAVS pathway implicated in IRF3 phosphorylation using a co-IP approach. Although by no means comprehensive, our co-IP analysis revealed a strong and specific interaction of LCMV NP with IKKε but not with TBK-1, RIG-I, MAVS, TRAF3, or IRF3 under our experimental conditions. The lack of detectable co-IP of LCMV NP and RIG-I in our hands seems in contradiction to previous reports that demonstrated an interaction of LCMV NP with RIG-I. The reasons for this apparent discrepancy are not clear and may be due to the more stringent washing conditions applied in our co-IP protocol. Of note, the interaction between LCMV NP and RIG-I reported by Zhou et al. was not affected by the D382A mutation that abrogates the ability of NP to suppress type I IFN induction (56) as well as the high-affinity interaction with IKKε (this study). Examination of NPs derived from representative members of the major arenavirus clades revealed evolutionary conservation of the specific interaction of NP with IKKε but not with TBK-1. Confocal microscopy combined with quantitative image analysis revealed that, in LCMV-infected cells, IKKε strongly colocalized with LCMV NP but not with its cellular binding partner MAVS. This suggests that NP somehow sequesters IKKε and prevents its recruitment to the MAVS signaling platform located on the outer mitochondrial membrane. Whether this sequestration is linked to sites of virus replication remains to be determined. Notably, when associated with NP, IKKε seems catalytically inactive, suggesting as rather unlikely a role of IKKε in viral replication via phosphorylation of cellular proteins or viral proteins or both.
The specific ability of arenavirus NP to target IKKε, but not TBK-1, was unexpected and seems different from the pattern seen with other viruses. Targeting of IRF3 phosphorylation as a viral strategy to subvert induction of type I IFN responses was first demonstrated for the phosphoprotein (P) of rabies virus, which was shown to interfere with phosphorylation of IRF3 by TBK-1 (5). More recent studies mapped this activity to a specific domain within the P protein that is dispensable for replication (46). In contrast, we found that the C-terminal region of LCMV NP, which is required for virus RNA replication and transcription, seems involved in binding to IKKε. Notably, the association of NP blocks the catalytic activity of IKKε, as evidenced by a lack of autophosphorylation and the inability of NP-bound IKKε to phosphorylate IRF3 or NP. This is different from the mode of action reported for the paramyxovirus V proteins and Ebola virus VP35, which interfere with both TBK-1- and IKKε-mediated IRF3 phosphorylation by acting as competing pseudosubstrates (25, 43). Moreover, upon phosphorylation by TBK-1 and IKKε, paramyxovirus V proteins are degraded (25), whereas LCMV NP remains stable over time. Therefore, regarding specificity and the molecular mechanism of inhibition of IKKε and TBK-1, arenavirus NPs behave differently from IFN antagonist proteins of Rhabdoviridae, Paramyxoviridae, and Filoviridae.
Arenavirus NP is a versatile protein that fulfills a plethora of functions in virus replication, virion assembly, and interaction with host cell factors. The examination of a set of previously described and well-characterized NP mutants (29) revealed a role of the C-terminal domain of NP in binding to IKKε. The C-terminal domain of NP is involved in binding to the viral matrix Z protein (39) and contains a 3′-5′ exonuclease domain, whose activity was linked to the anti-IFN activity of NP (15, 29, 44). The interaction with Z and the suppression of IFN activation map to partially overlapping regions of NP but also involve distinct functional domains (39). Our present study was limited to NP mutations causing impairment in the 3′-5′ exonuclease function and IFN suppression. All mutants tested showed significantly weaker binding to IKKε, suggesting that overlapping domains of NP are involved in 3′-5′ exonuclease function and IKKε binding. Future comprehensive mutation-function studies would be required to structurally separate the two functions and to distinguish their individual contributions to virus-induced suppression of the induction of type I IFN.
The specific ability of arenavirus NP to target IKKε, but not TBK-1, was unexpected, because existing evidence indicates a more important role for TBK-1 in the induction of type I IFNs in response to virus infection (16, 33, 40). Analysis of murine embryonic fibroblasts derived from mice deficient in IKKε or TBK-1 suggested a predominant role for TBK-1, rather than IKKε, in IFN induction in response to dsRNA and SeV infection (16, 33, 40). However, in macrophages, the IFN response to SeV was normal in TBK-1-deficient cells (40), indicating cell-type-specific differences. To address the relative contributions of TBK-1 and IKKε to IRF3 phosphorylation and IFN-β induction in response to SeV in our system, we performed RNAi. In HEK293 cells under our experimental conditions, we found that both TBK-1 and IKKε were required for optimal induction of IFN-β upon SeV infection. The apparently nonredundant roles of IKKε and TBK-1 suggest that SeV-induced IRF3 phosphorylation in HEK293 cells involves a complex requiring both kinases. It is conceivable that NP targeting IKKε may also affect the function of TBK-1, resulting in a block in IRF3 phosphorylation. However, several lines of evidence suggest that TBK-1 can function in the absence of IKKε (16, 33, 40). Innate detection of RNA viruses, e.g., by Toll-like receptor 3 (TLR-3), results in specific activation of TBK-1 but not IKKε via the signaling adapter TRIF (49). The absence of a direct interaction between arenavirus NP and TBK-1 found here indicates that arenaviruses may use other strategies to evade TBK-1-dependent pathways. One possible strategy is illustrated by the degradation of viral RNA by the 3′-5′ exonuclease activity of LCMV NP that is linked to its function as a type I IFN antagonist (15, 29, 44). By degradation of viral RNA, the 3′-5′ exonuclease of NP may eliminate the “danger signal” preventing recognition by RIG-I helicases. The specificity of NP for IKKε, but not TBK-1, would allow persistent arenaviruses to leave major TBK-1-dependent pathways of innate immunity intact. This strategy may contribute to protecting the natural reservoir host against infection by other pathogens, including DNA viruses, bacteria, fungi, and parasites.
The remarkably conserved specificity of arenavirus NPs for IKKε suggests other nonredundant roles for IKKε in arenavirus-host cell interaction. IKKε has also been implicated in activation of NF-κB in response to phorbol myristate acetate (PMA) and T cell receptor activation but not tumor necrosis factor alpha (TNF-α) and interleukin-1 (IL-1) (42). In breast cancer, IKKε is known to be an oncogene involved in uncontrolled NF-κB activation (3, 11). Interestingly, as noted in the accompanying article, Rodrigo et al. have reported that LCMV NP is able to specifically inhibit activation of NF-κB in response to virus infection with only a mild effect on NF-κB induction by TNF-α (46a). In addition to its role in activation of IRF3 and NF-κB, IKKε but not TBK-1 affects IFN-regulated gene expression by phosphorylation of STAT1 on serine 708 (54). IKKε-mediated phosphorylation of STAT1 affects the quality of the gene expression profile in response to type I and type II IFNs (37, 54). Apart from antiviral signaling, IKKε has been implicated in the regulation of energy metabolism in models of obesity (8) and cell transformation in cancer cells (19). Since arenaviruses are carried in nature by persistent infection of their reservoir rodent species, a major thrust of our current research is to study the role of the arenavirus NP-IKKε interaction in the context of viral persistence. The specificity of NP for IKKε may allow using the virus as a molecular probe to dissect nonredundant functions of IKKε and to elucidate their role for arenavirus-host interaction.
We are especially indebted to the late Jürg Tschopp for stimulating discussions, tools, and reagents. Without his crucial input, ideas, and support, this study would not have been possible. We thank John Hiscott (McGill University Montreal), Laurent Roux (University of Geneva), Brian Seed (Harvard Medical School), and Margot Thome Miazza and Pascal Schneider (both University of Lausanne) for valuable reagents.
This research was supported by a grant from the VonTobel Foundation (to S.K.) and funds from the University of Lausanne (to S.K.). Research at the J.C.D.L.T. laboratory was supported by grants RO1 AI047140, RO1 AI077719, and RO1 AI079665 from NIH/NIAID. Research in the L.M.-S. laboratory was partially funded by NIH grants RO1 AI077719 and HHSN272201000055C.
Published ahead of print 24 April 2012