Encephalomyocarditis Virus Disrupts Stress Granules, the Critical Platform for Triggering Antiviral Innate Immune Responses

Abstract

In response to stress, cells induce ribonucleoprotein aggregates, termed stress granules (SGs). SGs are transient loci containing translation-stalled mRNA, which is eventually degraded or recycled for translation. Infection of some viruses, including influenza A virus with a deletion of nonstructural protein 1 (IAVΔNS1), induces SG-like protein aggregates. Previously, we showed that IAVΔNS1-induced SGs are required for efficient induction of type I interferon (IFN). Here, we investigated SG formation by different viruses using green fluorescent protein (GFP)-tagged Ras-Gap SH3 domain binding protein 1 (GFP-G3BP1) as an SG probe. HeLa cells stably expressing GFP-G3BP1 were infected with different viruses, and GFP fluorescence was monitored live with time-lapse microscopy. SG formations by different viruses was classified into 4 different patterns: no SG formation, stable SG formation, transient SG formation, and alternate SG formation. We focused on encephalomyocarditis virus (EMCV) infection, which exhibited transient SG formation. We found that EMCV disrupts SGs by cleavage of G3BP1 at late stages of infection (>8 h) through a mechanism similar to that used by poliovirus. Expression of a G3BP1 mutant that is resistant to the cleavage conferred persistent formation of SGs as well as an enhanced induction of IFN and other cytokines at late stages of infection. Additionally, knockdown of endogenous G3BP1 blocked SG formation with an attenuated induction of IFN and potentiated viral replication. Taken together, our findings suggest a critical role of SGs as an antiviral platform and shed light on one of the mechanisms by which a virus interferes with host stress and subsequent antiviral responses.

INTRODUCTION

In eukaryotic cells, viral infections induce several responses. Cellular pathogen recognition receptors such as RIG-I-like receptors (RLRs) and Toll-like receptors recognize specific pathogen-associated molecular patterns and activate the transcription of hundreds of genes, including interferons (IFNs), inflammatory cytokines, and antiviral proteins. Secreted IFNs, in turn, activate a secondary JAK-STAT signaling cascade, which culminates in the activation of various interferon-stimulated genes (ISGs) (1, 2). A representative ISG, protein kinase RNA activated (PKR), acts as an antiviral protein by inducing the blockade of viral translation (3–5). PKR is also known to be associated with the cellular stress responses. Virus infection results in the accumulation of double-stranded RNA (dsRNA), thereby activating PKR and phosphorylation of the α subunit of eukaryotic initiation factor 2 (eIF2α), leading to the formation of stress granules (SGs) (6, 7). Several studies have reported the interaction between viruses and SGs, especially the effects of specific types of viruses on the fate of SG formation and how viruses modulate stress granule assembly (8–11). Recently, we reported that RLR recruitment to SGs during SG formation is critical for RLR-mediated signaling and that nonstructural protein 1 of influenza A virus (IAV) blocks RLR signaling by inhibiting SGs and the antiviral response (12). Accumulating evidence suggests that viruses have evolved strategies to prevent SG formation. These results suggest that virus-induced SGs potentially serve as platforms for antiviral activity; however, the underlying molecular mechanism still remains to be elucidated.

In the present study, we aim to delineate the physiological impact of stress granule formation and its viral modulation. We employed an enhanced green fluorescent protein (EGFP)-tagged stress granule marker, Ras-Gap SH3 domain binding protein 1 (G3BP1), to probe the subcellular distribution of virus-induced SGs (13, 14). This system allows us to monitor SGs in an individual virus-infected cell. Infection with RNA and DNA viruses displayed three distinct patterns: stable, transient, and alternate formation of SGs. We focused on encephalomyocarditis virus (EMCV), which exhibited transient formation of SGs. We show that EMCV disrupts SGs through G3BP1 cleavage. Furthermore, we found that EMCV-induced SGs are required for efficient activation of IFN and cytokine genes. We propose a new antiviral concept highlighting the potential cross talk of virus-induced stress responses and activation of the IFN signaling cascade. This may provide new insight into understanding the mechanism by which antiviral genes are regulated.

MATERIALS AND METHODS

Plasmid constructs.

The stress granule marker constructs pEGFP-C1-G3BP1 (NCBI RefSeq accession no. NM_005754) was a kind gift from Jamal Tazi (Institute de Génétique Moléculaire de Montpellier, France). The pEGFP-C1-G3BP1 Q325E mutant construct was generated by site-directed mutagenesis with a KOD-Plus mutagenesis kit (Toyobo, Japan) using primers containing the desired mutation according to manufacturer's instructions and were completely sequenced by using an ABI Prism DNA sequencer to verify the presence of the mutation. This plasmid contained a single-point amino acid substitution at position 325 (from glutamine to glutamate), which is resistant to cleavage by 3C^PRO of poliovirus (PolioV) (15). Expression vectors for EMCV pFirefly-leader and pFirefly-3C proteases were described previously (16).

Viruses.

PolioV (Mahoney strain), vesicular stomatitis virus (VSV) (Indiana strain), EMCV, adenoviruses (type 5), Sindbis virus (SINV), and Theiler's murine encephalomyelitis virus (TMEV) (GDVII strain) were prepared by infecting BHK cells at a multiplicity of infection (MOI) of 1. Cell culture medium was collected after confirming cytopathic effects following infection. Medium containing newly produced viruses was centrifuged at 1,500 rpm for 5 min to pellet the cell debris, and supernatants containing viruses were collected and stored at −80°C. The viral titer was assessed by a plaque assay on L929 cells, as previously described (17). Newcastle disease virus (NDV) (Miyadera strain), Sendai virus (SeV) (Cantell), and influenza A virus with a deletion of the NS1 gene (IAVΔNS1) (strain A/Puerto Rico 8/34) (18, 19) were propagated in the allantoic cavities of embryonated chicken eggs, and stocks were then stored at −80°C.

Generation of stable HeLa cells and general cell culture conditions.

Cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (Nacalai Tesque, Japan) and penicillin-streptomycin (100 U/ml and 100 μg/ml, respectively; Nacalai Tesque, Japan). To generate HeLa cells stably expressing the EGFP-G3BP1 wild type (wt) and the EGFP-G3BP1 Q325E mutant, pEGFP-C1-G3BP1 and pEGFP-C1-G3BP1 Q325E mutant expression constructs were linearized by using the restriction enzyme ApaLI (TaKaRa, Japan). The linearized plasmids were then transfected into HeLa cells by using FuGENE6 (Promega, USA) according to manufacturer's recommendations. Transformants were selected by including 1 mg/ml of G418 in the culture medium. Individual colonies were isolated and characterized.

Live-cell imaging and immunofluorescence microscopy.

For the live-cell imaging analysis, HeLa cells stably expressing EGFP-G3BP1 (HeLa/G-G3BP cells) were seeded into a 12-well plate and incubated at 37°C. After 24 h, cells were washed with DMEM (10% fetal bovine serum and 1% penicillin-streptomycin) for several rounds. Cells were then infected with various types of RNA and DNA viruses. After 1 h of infection, virus was removed and replaced with 1.0 ml of DMEM imaging medium (4,500 mg/liter d-glucose and l-glutamine, 25 mM HEPES buffer, no sodium pyruvate, and phenol red; Invitrogen). Imaging was immediately initiated every 10 min. Live cells were maintained on the microscope stage at 37°C with 5% carbon dioxide in a humidity-controlled chamber. Images were taken by using Biophotonics-ImageJ software. All imaging was performed by using a Leica CTR 6500 instrument.

For the immunofluorescence analysis, cells were seeded into either a 12-well plate or an 8-well chamber slide and incubated at 37°C. After 24 h, cells were subjected to various treatments, such as plasmid transfection or virus infection. Cells were then rinsed in phosphate-buffered saline (PBS) several times, fixed with 4% paraformaldehyde solution for 10 min at room temperature, washed with PBS for two additional rounds, permeabilized with acetone-methanol (1:1) for 1 min, and blocked with phosphate-buffered saline containing 0.1% Tween 20 (PBST) and containing bovine serum albumin (BSA) (5.0 mg/ml) for 1 h at 4°C. Cells were then incubated with primary antibody followed by fluorophore-conjugated secondary antibodies (Invitrogen) for 1 h at 4°C. Cells were washed with PBST extensively and mounted. All images were obtained by using a Leica CTR 6500 instrument.

siRNA-directed gene silencing.

The small interfering RNA (siRNA) universal negative control and siRNAs targeting the stress granule marker G3BP1 (50 nM) and the dsRNA protein kinase PKR were purchased from Invitrogen and transfected by using either Lipofectamine 2000 (Invitrogen) or RNAiMax (Invitrogen) according to the manufacturer's recommendations. The sequences of siRNAs are as follows: sense sequence 5′-CGG AUU AGC GAC AAA UUU AUU-3′ and antisense sequence 5′-UAA AUU UGU CGC UAA UCC GUU-3′ for RIG-I, sense sequence 5′-UUU ACU UCA CGC UCC GCC UUC UCG U-3′ and antisense sequence 5′-ACG AGA AGG CGG AGCGUGAAGUAA A-3′ for PKR#1, sense sequence 5′-AUG UCA GGA AGG UCA AAU CUG GGU G-3′ and antisense sequence 5′-CAC CCA GAU UUG ACC UUC CUG ACA U-3′ for PKR#2 (#n, designated siRNA number), and sense sequence 5′-UAA UUU CCC ACC ACU GUU AAU GCG C-3′ and antisense sequence 5′-GCGCAUUAACAGUGGUGGGAAAUUA-3′ for G3BP1. At 48 h posttransfection, cells were subjected to viral infection or other treatments. A specific antibody for G3BP1 (Santa Cruz) was used to monitor the knockdown efficiency.

RNA analysis.

RNA was harvested from cells by using TRIzol (Invitrogen) according to the manufacturer's instructions. Contaminating DNA was then eliminated by using recombinant DNase I (10 units/μl; Roche) according to the manufacturer's protocol. Treated samples were purified by phenol-chloroform extraction. Five hundred nanograms of purified RNA was used as a template to synthesize cDNA by using a High Capacity cDNA reverse transcription kit (Applied Biosystems), as specified by the manufacturer, with the following cycles: 25°C for 10 s, 37°C for 2 h, and 85°C for 10 s. The concentration of cDNA was quantified by the use of a spectrophotometer, and the final concentration was adjusted to 1 μg/μl. cDNA samples were then subjected to either standard PCR or real-time quantitative PCR (RT-qPCR) analysis with specific probes from the TaqMan gene expression assay (Applied Biosystems). Quantification of EMCV RNA was performed by using SYBR master mix (Applied Biosystems) with specific primers targeting the EMCV capsid coding region. Standard PCR was performed with cDNA samples together with a master mix containing 1× PCR buffer, 2.5 mM each deoxynucleoside triphosphates (dNTPs), 0.2 units of Ex Taq polymerase, and 1.0 μM both forward and reverse primers. PCR buffer, dNTPs, and Ex Taq polymerase were purchased from TaKaRa, Japan. Primers were all customized and purchased from Invitrogen. PCR was performed with a 50-μl reaction mixture with an initial annealing temperature of 56°C to 60°C. PCR products were analyzed by agarose gel electrophoresis.

Western blotting.

Cells were collected in ice-cold PBS by using a scraper. Cells were collected by centrifugation and lysed by NP-40 buffer (50 mM Tris [pH 8.0], 150 mM NaCl, 1% [vol/vol] NP-40, 1 nM vanadate, 1 mM leupeptin, and phenylmethanesulfonyl fluoride), followed by centrifugation at 15,000 rpm for 10 min and ultracentrifugation at 100,000 rpm for 5 min. The supernatant was mixed with an equal volume of 2× SDS buffer, boiled for 5 min, separated by SDS-PAGE (30 μg/lane), and transferred onto a nitrocellulose membrane. The membranes were incubated in blocking buffer (PBS, 5% [wt/vol] dry milk powder) for 30 min at room temperature, followed by incubation with primary antibody diluted in blocking buffer at 4°C overnight. Membranes were washed extensively with TBST (Tris-buffered saline, 0.1% Tween 20), followed by incubation with a conjugated secondary antibody for 1 h at room temperature. The proteins were visualized by using alkaline phosphatase buffer containing 5'-bromo-4-chloro-3'-indolylphosphate (BCIP)-Nitro Blue Tetrazolium (NBT) (Promega) color development substrate (100 mM Tris-HCl [pH 9.0], 150 mM NaCl, 1 mM MgCl₂, 66 μl of NBT [50 mg/ml], and 33 μl of BCIP [50 mg/ml]).

Antibodies.

The antibodies used in this study include mouse monoclonal anti-green fluorescent protein (GFP) (1:1,000 dilution) (MBL), goat polyclonal anti-G3BP1 (1:500) (catalog no. sc-70283; Santa Cruz), mouse monoclonal anti-G3BP1 (1:1,000) (catalog no. sc-365338; Santa Cruz), rabbit polyclonal anti-PKR (1:1,000) (catalog no. sc-709; Santa Cruz), rabbit polyclonal anti-TIA-1/R (1:1,000) (catalog no. sc-48371; Santa Cruz), goat polyclonal anti-TIAR (1:1,000) (catalog no. sc-1749; Santa Cruz), rabbit polyclonal anti-HuR (1:1,000) (catalog no. sc-365816; Santa Cruz), and propidium iodide (PI) (1:2,000 in PBST) (Miltenyi Biotec). The RIG-I antibodies were generated by immunizing a rabbit with a synthetic peptide corresponding to amino acids 793 to 807 of RIG-I and MDA5. Mouse monoclonal anti-poly(A) binding protein (PABP) (1:1,000) (catalog number ab6125; Abcam), rabbit monoclonal antiactin (1:5,000) (Poly6221; BioLegend), mouse anti-FLAG (1:1,000) (Sigma-Aldrich), and rabbit monoclonal anti-phospho-PKR pT446 (1:1,000) (Epitomics Inc.) antibodies were also used. Anti-EMCV polyclonal antibody was obtained by immunizing a rabbit with purified EMCV virions. Anti-MDA5 polyclonal antibody was obtained by immunizing a rat with recombinant MDA5 (produced in insect cells), which was preactivated with RNA ligands.

Quantification of the distribution pattern of virus-induced SGs.

SG formation was quantified visually by using eyesight counting. The total number of cells displaying each unique distribution pattern in each location was recorded, and the percentage of each pattern was calculated. As for the fixed cells, 10 pictures were taken randomly at different locations. Cells displaying SG foci were quantified manually. Graphs display the average percentages of replicates (at least 20 times).

RESULTS

Characterization of HeLa cells stably expressing an SG marker, G3BP1.

To monitor SGs in living cells, we generated HeLa/G-G3BP cells (Fig. 1). Constitutive aggregation of intrinsic SG components has been reported to lead to a severe stall in protein synthesis and eventual apoptosis (14, 20). All the HeLa/G-G3BP clones displayed uniform and high-level GFP expression, and their growth rate was comparable to that of parental cells (our unpublished observations). It has been well documented that G3BP1 accumulates in SG foci in response to arsenite treatment (oxidative stress) and virus infection (12, 13). HeLa/G-G3BP clone 12 was treated with arsenite or infected with Newcastle disease virus (NDV) or influenza A virus (IAV) with an NS1 deletion (IAVΔNS1), and GFP localization was then examined by confocal microscopy. As shown in Fig. 1A, a speckle-like localization of GFP was induced by these stimuli. Other clones also exhibited similar speckle formation after arsenite treatment or NDV infection (Fig. 1B and C). We confirmed that other SG components, TIA-1, TIAR, HuR, and eIF3, colocalized with the GFP speckles (our unpublished observations). These results indicate that EGFP-G3BP1 acts as a suitable probe for virus-induced SGs. However, since transient overexpression of G3BP1 results in SG formation without external stress (13), we tested if the HeLa/G-G3BP clones would exhibit a normal antiviral response. As shown in Fig. 1D, all clones exhibited induction of IFN-β mRNA comparable to that exhibited by parental cells. We chose clone 18 for further analyses.

Fig 1.

Characterization of HeLa/G-G3BP cells. (A) HeLa/G-G3BP1 clone 12 was mock treated or stimulated, as indicated. DAPI, 4′,6-diamidino-2-phenylindole. (B and C) Cells were fixed and examined for GFP fluorescence. Four independent HeLa/G-G3BP cell clones were stimulated by arsenite (B) or by infection with NDV (C), and the percentage of GFP speckle-positive cells was determined. (D) Parental HeLa cells and HeLa/G-G3BP1 clones were infected with NDV for 12 h, and the IFN-β gene expression level was determined by RT-qPCR. (Error bars indicate standard deviations of duplicates [n = 2].)

G3BP1 exhibits three redistribution patterns after infection with both RNA and DNA viruses.

To examine the dynamics of cytoplasmic SGs induced by viral infection, the cells were infected with different viruses, as shown in Fig. 2, and monitored live for distribution of GFP fluorescence (representative results are shown in Movies S1 to S9 in the supplemental material). Cells infected with SeV, IAV, VSV, and TMEV did not show SG formation (8). Other viruses induced SGs, typically forming a large number of small granules at around 5 h postinfection and gradually fusing to each other. SG formation was quantified (Fig. 2A to K) and classified into three predominant patterns, stable formation (Fig. 2L), transient formation (Fig. 2M), and alternating formation (Fig. 2N), within a single cell. NDV, IAVΔNS1, and adenovirus type 5 displayed stable formation of SGs (see Movies S1 to S3 in the supplemental material). Whereas SINV, EMCV, and PolioV induced foci at around 5 to 6 h postinfection, the foci disappeared thereafter (transient formation) (Fig. 2D to F; see also Movies S4 to S6 in the supplemental material). Interestingly, adenovirus type 5 with an E1A deletion exhibited multiple rounds of formation and disappearance of SGs (alternate formation) in the majority of cells (Fig. 2I; see also Movie S7 in the supplemental material). A similar oscillation of SGs in cells infected with hepatitis C virus (HCV) and treated with IFN was reported previously (21). Collectively, these live-cell-imaging analyses demonstrated that viral infections trigger host stress responses; however, different viruses induce distinct response patterns, presumably through specific underlying mechanisms. The observed SG formation patterns are unlikely to be due to G3BP1 overexpression, because wt HeLa cells exhibited transient SG formation upon EMCV infection when endogenous G3BP1 was used as a marker (Fig. 2O).

Fig 2.

Three major forms of virus-induced stress granule distribution patterns in HeLa/G-G3BP cells infected with different viruses. (A to K) HeLa/G-G3BP cells were infected with NDV (A), IAV (B), IAVΔNS1 (C), EMCV (D), SINV (E), PolioV (F), SeV (G), VSV (H), adenovirus type 5 with an E1A deletion (Adeno5ΔE1A) (I), wild-type adenovirus type 5 (Adeno5WT) (J), and TMEV (K) for approximately 9 to 12 h, and SG formation was monitored and quantified as described in Materials and Methods. (Error bars indicate standard deviations of triplicates [n = 3].) N.D., not detectable; ∗∗, P < 0.005; ∗, P < 0.05. (L to N) Representative cell images taken at the indicated times after infection for stable (NDV) (L), transient (SINV) (M), and alternating (Adeno5ΔE1A) (N) SG formation. (O) Wild-type HeLa cells were mock infected or infected for 4 or 12 h and fixed to examine the localization of endogenous G3BP1 by immunostaining.

EMCV infection results in the cleavage of G3BP1.

We focused on the mechanism of transient formation of SGs by EMCV because PolioV has been reported to inhibit SG formation by cleavage of G3BP1 (15). We examined if EGFP-G3BP1 is cleaved by EMCV by Western blotting. The EGFP-G3BP1 fusion protein is detected as a polypeptide of 96 kDa, and EMCV infection resulted in the appearance of an 80-kDa GFP-containing protein at 6 h postinfection, and nearly complete cleavage of EGFP-G3BP1 occurred at 10 h postinfection (Fig. 3A). Because the fusion protein contains an EGFP moiety at the N terminus of G3BP, the cleavage of G3BP1 is likely to occur at the C-terminal region of G3BP1. We verified the cleavage site by using an antibody detecting the C-terminal epitope of G3BP1 (see Fig. S1A and S1B in the supplemental material). Because the mapped cleavage site was close to that of PolioV and cleavage by PolioV is prevented by an amino acid substitution within G3BP1 (Q325E) (15), we therefore examined this mutant for cleavage by EMCV (Fig. 3B). We found that the G3BP1 Q325E mutant was resistant to cleavage by EMCV, suggesting a common cleavage mechanism. To examine whether the disruption of SGs by EMCV is due solely to cleavage of G3BP1, we examined other SG components, such as PABP, TIA-1/R, HuR, and PKR, which are also essential for SG formation. Figure 3C shows that the levels of SG components, with the exception of G3BP1, did not change upon EMCV infection and that G3BP1 cleavage coincided with the detection of EMCV proteins. Expression of EMCV 3C protease but not leader protein by transfection was sufficient to reproduce G3BP1 cleavage at Q325 (Fig. 3D), strongly suggesting that the cleavage is mediated by 3C protease. We next examined SG formation of HeLa/G-G3BPQ325E cells. In sharp contrast to cells expressing wild-type G3BP1 (see Movie S6 in the supplemental material), HeLa/G-G3BPQ325E cells exhibited stable formation of SGs, as judged by single-cell imaging (Fig. 4A and B; see also Movie S8 in the supplemental material) and quantification (Fig. 4C). These results suggest that EMCV disrupts SGs by cleavage of G3BP1 through a mechanism similar to that of PolioV.

Fig 3.

EMCV infection results in the cleavage of G3BP1. (A) Immunoblotting (IB) showing the kinetics of G3BP1 cleavage in EMCV-infected HeLa/G-G3BP1 cells. N.C., negative control. (B) HeLa cells stably expressing FLAG-G3BP1 Q325E protein were infected with EMCV, and the G3BP1 Q325E protein level was monitored by immunoblotting. (C) Western blot analysis of HeLa/G-G3BP1 cells infected with EMCV. Lysates were prepared at the indicated time points after infection and subjected to immunoblotting with the indicated antibodies. FL, full-length; n.s., not significant. (D, left) HeLa cells were transiently transfected with an empty vector or the expression vector for leader or 3C and analyzed for endogenous G3BP1 by Western blotting. (Right) HeLa/G-G3BP1 and HeLa/G-G3BP1Q325E cells were transiently transfected with an empty vector or the expression vector for leader or 3C and analyzed by Western blotting using anti-GFP. MW, molecular weight (in thousands); p.h.i., hour postinfection; cp, cleavage fragment.

Fig 4.

HeLa/G-G3BPQ325E cells display stable formation of SGs induced by EMCV infection. (A and B) Both HeLa/G-G3BP1 (A) and HeLa/G-G3BP1Q325E (B) cells were infected with EMCV. GFP fluorescence images of these cells taken every 40 min are shown. (C) Quantitative analysis of SG formation pattern of HeLa/G-G3BP1Q325E cells infected with EMCV. (Error bars indicate standard deviations of triplicates [n = 3].) N.D., not detectable; ∗∗, P < 0.005.

G3BP1 negatively regulates EMCV replication.

To examine the impact of SG disruption on EMCV replication, we infected both HeLa/G-G3BP and HeLa/G-G3BPQ325E cells with EMCV and analyzed viral replication by RT-qPCR (Fig. 5A). The EMCV RNA level recovered from HeLa/G-G3BP cells was 6-fold higher than that recovered from HeLa/G-G3BPQ325E cells. Similarly, a significantly lower viral yield was observed for cells expressing G3BP1 Q325E, suggesting that SG formation is critical for suppressing EMCV replication. To further confirm the involvement of G3BP1, we depleted endogenous G3BP1 by siRNA-mediated knockdown (Fig. 5B) and examined its effect on EMCV replication. G3BP1 knockdown caused increased EMCV replication, as judged by the approximately 5-fold augmentation of viral RNA and viral yield (Fig. 5B). These results suggest that G3BP1 is involved in the negative regulation of EMCV.

Fig 5.

Cleavage or knockdown of G3BP1 results in enhanced EMCV replication. (A) HeLa/G-G3BP1 and HeLa/G-G3BP1Q325E cells were infected with EMCV. (Top) Total RNA was harvested at 12 h postinfection, and EMCV RNA was quantified by qPCR. (Bottom) The culture supernatant was subjected to plaque titration. (B) HeLa cells were transfected with either control siRNA or siRNA that targeted G3BP1. (Top) After 48 h, G3BP1 was detected by Western blotting (left) or by staining using anti-G3BP1 antibody (right). (Bottom, left) To investigate the effect of G3BP1 knockdown on viral replication, the cells were infected with EMCV for 12 h, total RNA was extracted, and EMCV RNA was quantified by qPCR. (Right) The culture supernatant was analyzed to determine viral titers.

G3BP1 is critical for EMCV-induced interferon and cytokine gene activation.

Based on the above-described findings, we next asked how G3BP1 exerts its antiviral role. The type I interferon system constitutes major innate antiviral responses; therefore, we examined EMCV-induced IFN-β gene activation in HeLa/G-G3BP and HeLa/G-G3BPQ325E cells (Fig. 6). In HeLa/G-G3BP cells, IFN-β mRNA accumulated at 4 h postinfection, followed by a gradual decrease. However, IFN-β mRNA levels persisted in HeLa/G-G3BPQ325E cells after 8 h postinfection (Fig. 6B). In agreement with these results, the amount of IFN-β protein released into the culture medium at 24 h was significantly augmented by the Q325E mutation (Fig. 6A). A similar enhancement of cytokine mRNA was observed for CXCL10, interleukin-6 (IL-6), and RANTES (Fig. 6C to E). We investigated gene activation at early time points between 0 and 4 h and observed similar activation kinetics between HeLa/G-G3BP and HeLa/G-G3BPQ325E cells (Fig. 7), suggesting that the reduced gene activation of HeLa/G-G3BP cells is due to G3BP1 cleavage. The Q325E mutation did not affect IFN-β gene induction in the case of IAVΔNS1, which did not cause G3BP1 cleavage (Fig. 6F). Next, we examined the effects of depletion of endogenous G3BP1 on cytokine gene activation. As expected, knockdown of endogenous G3BP1 attenuated the expression of IFN-β and other cytokine genes (Fig. 8A to D). These results strongly suggest that G3BP cleavage leads to attenuation of antiviral cytokine induction.

Fig 6.

Inhibition of G3BP1 in EMCV-infected cells results in sustained cytokine/chemokine mRNA accumulation. HeLa/G-G3BP1 and HeLa/G-G3BP1Q325E cells were infected with EMCV. (A) Culture supernatant was subjected to an enzyme-linked immunosorbent assay for human IFN-β (hIFN-β). (B to E) Total RNA was harvested at the indicated time points. mRNA levels for IFN-β (B), CXCL10 (C), IL-6 (D), and RANTES (E) were determined by RT-qPCR. (F, left) Both HeLa/G-G3BP and HeLa/G-G3BPQ325E cells were infected with IAVΔNS1, and the IFN-β mRNA level was quantified as described above. (Right) Lysates of IAVΔNS1-infected HeLa/G-G3BP1 cells were examined for cleavage of G3BP1 by Western blotting. Data depicted are representative of two independent experiments. (Error bars indicate standard deviations of duplicates.) ∗∗, P < 0.005; ∗, P < 0.05.

Fig 7.

IFN production and cytokine gene activation in HeLa/G-G3BP and HeLa/G-G3BPQ325E cells at the early phase. HeLa/G-G3BP1 and HeLa/G-G3BP1Q325E cells were mock treated or infected with EMCV for the indicated times. Total RNA was extracted, and mRNA levels for IFN-β (A), CXCL10 (B), IL-6 (C), and RANTES (D) were quantified by RT-qPCR.

Fig 8.

Knockdown of G3BP1 attenuates EMCV-induced cytokine/chemokine gene activation. HeLa cells were transfected with either control siRNA or siRNA that targeted G3BP1. After 48 h of incubation, cells were infected with EMCV for 12 h, and total RNA was collected as indicated. mRNAs levels for IFN-β (A), RANTES (B), CXCL10 (C), and IL-6 (D) were determined by RT-qPCR. Data are representative of two independent experiments. (Error bars indicate standard deviations of duplicates [n = 2].) ∗, P < 0.05.

It has been well documented that MDA5 senses EMCV infection (22–25) and that virus- and oxidative stress-induced SGs recruit RIG-I, MDA5, and LGP2 (12). Therefore, we hypothesized that EMCV-induced SG regulates IFN-β gene activation by facilitating MDA5 activation. We examined MDA5 localization in EMCV-infected HeLa cells by immunostaining. MDA5 displayed relocalization to speckle-like granules upon EMCV infection (Fig. 9A). The speckles also contained endogenous G3BP1 (Fig. 9A) and TIAR (Fig. 9B). Interestingly, PI, a dye that binds to dsDNA and dsRNA, stains cytoplasmic speckles found only in virus-infected cells, and the dsRNA speckles colocalized with G3BP1 and TIAR. These observations suggest that EMCV infection induces SGs, which recruit SG components, MDA5, and EMCV dsRNA.

Fig 9.

EMCV infection recruits MDA5 into SGs. HeLa cells were mock treated or infected with EMCV (MOI of 10) and fixed. The cells were stained for MDA5, G3BP1, and PI (A) or MDA5, TIAR, and PI (B).

PKR is essential for SG formation and IFN induction in EMCV infection.

Various types of viruses were shown to induce SG formation through PKR activation (26–28). We therefore examined whether EMCV induces SG formation in a PKR-dependent manner. Endogenous PKR expression was efficiently downregulated by siRNA (Fig. 10A). Under these conditions, SG formation by EMCV was decreased significantly (Fig. 10A). We next asked whether cleavage of G3BP1 results in PKR dephosphorylation. Immunoblot analyses showed that PKR was autophosphorylated at 4 h postinfection; however, at 12 h, when G3BP1 cleavage was nearly complete, PKR phosphorylation was undetectable (Fig. 10B, lane 3), suggesting that G3BP1 cleavage resulted in PKR dephosphorylation. Finally, we examined whether the final outcome of signaling, IFN-β gene expression, was dependent on PKR. In PKR knockdown cells, the induction of IFN-β mRNA by EMCV was significantly decreased compared to that in control cells (Fig. 10C). We further confirmed previous reports that IFN induction by poly(I·C) or IAVΔNS1 infection was PKR dependent (12). From the data presented above, we concluded that the loss of PKR impaired EMCV-induced SG formation, leading to a reduction of IFN-β gene activation.

Fig 10.

Involvement of PKR in EMCV-induced SG and IFN-β gene activation. (A) Knockdown of PKR expression results in reduced SGs. (Left) HeLa cells transfected with siRNA targeting PKR for 48 h were examined for PKR expression by Western blotting. (Middle) The cells were infected with EMCV for 6 h and stained for endogenous G3BP1. (Right) SG-containing cells were quantified. (B) HeLa cells infected with EMCV for 0, 4, and 12 h were analyzed for G3BP1, phospho-PKR, EMCV proteins, and actin by immunoblotting. (C) HeLa cells transfected with siRNA targeting PKR for 48 h were mock treated, transfected with poly(I·C), or infected with IAVΔNS1 or EMCV. After 12 h, IFN mRNA was quantified by RT-qPCR. ∗∗, P < 0.005; ∗, P < 0.05.

DISCUSSION

Viral infection causes stress in host cells, resulting in SG formation. To date, both pro- and antiviral roles have been described for virus-induced SGs (28–30), and this issue remains controversial.

In this study, we demonstrated that SGs are potentially involved in mediating virus-triggered IFN responses. It was reported previously that PolioV 3C protease cleaves G3BP1 at residue Q325, resulting in the disruption of SGs (15). This observation indicates not only that G3BP1 is a component of SGs but also that its inactivation by cleavage causes the disruption of SGs. Here, we show that EMCV shares G3BP cleavage activity with specificity identical to that of PolioV 3C, requiring intact Q325. Interestingly, coxsackievirus also disrupts SGs (31) by a similar mechanism (G. Fung, C. S. Ng, J. Zhang, J. Shi, J. Wong, P. Piesik, L. Han, F. Chu, J. Jagdeo, E. Jan, T. Fujita, and H. Luo, unpublished observation), suggesting that this strategy is shared by some picornaviruses to evade immune responses. At the early phase of EMCV infection, cleavage of G3BP1 was not evident. However, at 4 h postinfection, cleavage was detectable, and at 10 h, cleavage reached completion, suggesting that the accumulation of 3C is necessary for the disruption. We observed that stable expression of the G3BP1 Q325E mutant blocked the disassembly of SGs as well as enhanced IFN-β production at a late phase of infection. Furthermore, knockdown experiments showed that G3BP1 is necessary for efficient activation of the IFN-β gene, particularly in the later stages of infection. Although it was reported previously that PolioV 3C cleaves RIG-I and MDA5 (32) and that EMCV cleaves RIG-I (33), we did not observe these cleavages, even under conditions in which G3BP1 was cleaved by EMCV or PolioV (Fig. 11). Taken together, we conclude that G3BP1 is a physiological regulator of IFN-β gene induction through the formation of SGs, which recruit the RNA sensor MDA5. In addition, the persistent activation of the IFN-β gene at late time points is likely due to the increase of the local concentrations of both MDA5 and its ligands within the condensed granules.

Fig 11.

RIG-I was not cleaved after EMCV or PolioV infection. (A) HeLa cells were either mock treated or infected with EMCV for the indicated times. RIG-I was detected by Western blotting. (B) HeLa cells were mock treated or infected with PolioV for 9 h. G3BP1 (left) and RIG-I (right) were examined by Western blotting. CTD, C-terminal domain; shRIG-I, short-hairpin RIG-I.

Collectively, the data presented above strongly suggest that 3C protease of EMCV acts as a critical factor for evading host IFN production to ensure efficient replication. It was demonstrated previously that PKR plays a critical role in dsRNA- or IAVΔNS1-induced SG formation and subsequent IFN-β gene activation (12). Our observation that PKR is required for efficient IFN gene activation by EMCV suggests that PKR is responsible for initiating SG formation (Fig. 10).

Considering that the assembly of SGs is a part of the antiviral response of the host, it is plausible that viruses evolve strategies to block it. Indeed, IAV, SeV, and TMEV do not induce SG (Fig. 2), and it was reported previously that leader RNA, NS1, and leader protein are responsible for inhibition, respectively (8, 12, 34). Although TMEV belongs to the Picornaviridae, its mechanism of SG inhibition appeared to be distinct from those of EMCV and PolioV. TMEV and mengovirus inhibit SG by the action of leader protein (8, 31). We found that 3C but not the leader protein of EMCV inhibits SG formation (Fig. 12). It is tempting to speculate that leader proteins of TMEV and mengovirus inhibit IFN production (35, 36) through the blockade of SG formation, where RLR and viral RNA efficiently interact, as one of the mechanisms. Interestingly, although the leader protein of EMCV did not affect SGs, it inhibited IFN gene activation (Fig. 12), suggesting that leaders of different cardioviruses are functionally equivalent (37, 38) but with distinct modes of action. Therefore, these viruses encode multiple inhibitory proteins to efficiently manipulate host immune responses. EMCV and SINV induced SGs at early time points after infection, but SG formation was disrupted later. A similar phenomenon was reported previously for West Nile and dengue viruses by monitoring TIA-1/R as an SG marker (29). In the case of EMCV and PolioV, G3BP1 cleavage by viral 3C protease is responsible for the disassembly of SGs. Therefore, active mechanisms for the disruption of SGs by SINV, West Nile virus, and dengue virus have been suggested, although the underlying mechanisms remain to be determined. In addition to transient formation of SGs, some viruses exhibited alternating formations of SGs; SGs were formed at an early stage and then disappeared and re-formed at a later stage. This alternating pattern is also dependent on the cell lines used (our unpublished observations), suggesting that the pattern of SG formation is determined by a dynamic balance between the host antiviral response and the viral inhibitory mechanism (21). Such a host mechanism could be a therapeutic target to enhance host defense against viruses.

Fig 12.

EMCV 3C but not leader inhibits SG. (A) HeLa/G-G3BP1 and HeLa/G-G3BP1Q325E cells were transiently transfected with an empty vector or the expression vector for leader or 3C for 48 h. Cells were treated with 0.5 mM sodium arsenite for 30 min, fixed, and stained for TIAR, an SG marker. (B) HeLa cells were transiently transfected with and empty vector or the expression vector for leader or 3C (0 μg, 2 μg, and 4 μg) for 48 h. Cells were mock treated or transfected with long poly(I·C) (2 μg/μl) for 12 h. Total RNA was collected, and the mRNA level for IFN-β was determined by RT-qPCR. Data are representative of three independent experiments. (Error bars indicate standard deviations of duplicates [n = 3].)

Here, we provide evidence that EMCV-induced SGs are involved in regulating IFN-β gene expression. Thus, virus-induced SGs might play dual roles: (i) suppressing viral replication through an inhibition of viral protein synthesis and (ii) serving as a platform to facilitate IFN-β production.