Establishment and Maintenance of the Innate Antiviral Response to West Nile Virus Involves both RIG-I and MDA5 Signaling through IPS-1

Cells and viruses.Low-passage-number primary WT, RIG-I^−/−, and IPS-1^−/− MEFs derived from normal and transgenic C57BL/6 mice, respectively, were kindly provided by Shizuo Akira (16, 19). WT and MDA5^−/− primary MEFs derived from C57BL/6 mice were obtained from Marco Colonna (9). All cell lines were propagated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 2 mM l-glutamine, 1 mM sodium pyruvate, antibiotic-antimycotic solution, and nonessential amino acids (complete DMEM). U-2 OS/NS3/4A cells were propagated in complete DMEM supplemented with 500 mg/ml G418, 1 mg/ml puromycin, and 1 mg/ml tetracycline (42). Working stocks of WNV-TX 2002-HC (TX02) and WNV-NY, which are recent lineage I isolates with identical growth characteristics and antiviral responses in vitro (17), were generated by passaging the virus one time on 293 cells at a low multiplicity of infection (MOI), and aliquots were stored at −80°C. Titers of working viral stocks on U-2 OS/NS34A cells were determined by plaque assay. Viral titers on the various MEF cell lines were determined using a focus-forming assay as previously described with Vector VIP (Vector Laboratories) as a substrate for horseradish peroxidase (8). Titers for WNV were similar for WT and RIG-I^−/− MEFs, which allowed cultures to be comparably infected. The amount of virus added to cultures to achieve the indicated MOI was calculated using the titer of the viral stock on the respective cell line unless otherwise stated. Sendai virus (SenV) Cantrell strain (Charles River) was amplified in chicken embryos.

Expression microarray format and data analysis.Cultures of WT and RIG-I^−/− MEFs in six-well plates were infected with TX02 at an MOI of 0.5 in duplicate. At the indicated time points, total RNA was extracted from cells by use of TRIzol reagent as recommended by the manufacturer (Invitrogen Life Technologies, Inc.). Total RNA samples were amplified using a RiboAmp RNA amplification kit (Arcturus) as described by the manufacturer. The quality of amplified RNA was evaluated by capillary electrophoresis using an Agilent 2100 Bioanalyzer (Agilent Technologies). Microarray format, protocols for probe labeling, and array hybridization are described at http://expression.viromics.washington.edu . Briefly, a single experiment comparing two mRNA samples was done with four replicate mouse (V2) 22K oligonucleotide expression arrays (Agilent Technologies) using the dye label reverse technique. This allows for the calculation of mean ratios between expression levels of each gene in the analyzed sample pair, standard deviation, and P values for each experiment. Spot quantitation, normalization, and application of a platform-specific error model was performed using Agilent's Feature Extractor software, and all data were then entered into a custom-designed database, Expression Array Manager, and then uploaded into Rosetta Resolver System 6.0 (Rosetta Biosoftware) and Spotfire DecisionSite for Functional Genomics 8.1 (Spotfire). Data normalization and the Resolver error model are described on the website http://expression.viromics.washington.edu . This website is also used to publish all primary data in accordance with the proposed MIAME standards (4). Selection of genes for data analysis was based on a change (n-fold) of twofold or greater.

Antibodies.Rabbit anti-human ISG56, rabbit anti-mouse ISG54, rabbit anti-mouse ISG56, and rabbit anti-human ISG15 antibodies were kindly provided by Ganes Sen. Goat anti-SenV, mouse anti-hepatitis C virus (HCV) NS3/4A, and rabbit anti-mouse IRF-3 were purchased from Biodesign, Novocastra Laboratories, and Zymed, respectively. Mouse anti-WNV was obtained from the CDC, and rabbit anti-human IRF-3 serum was kindly provided by Michael David. Goat anti-GAPDH and mouse anti-GAPDH were purchased from Santa Cruz and Abcam, respectively. Peroxidase-conjugated goat anti-rabbit and donkey anti-goat were purchased from Jackson ImmunoResearch. Goat anti-mouse conjugated to peroxidase was obtained from Novocastra. For immunofluorescence assays, donkey anti-goat immunoglobulin G (IgG)-Alexa 594, goat anti-rabbit IgG-Alexa 488, and goat anti-mouse IgG-Alexa 350 were purchased from Molecular Probes, and goat anti-mouse IgG-rhodamine antibody conjugate was purchased from Jackson ImmunoResearch.

Immunoblot analysis.Cells were lysed in radioimmunoprecipitation assay buffer (10 mM Tris, 150 mM NaCl, 0.02% Na-deoxycholate, 1% Triton X-100, 0.1% sodium dodecyl sulfate) containing protease inhibitors (Sigma) and okadaic acid (1 mM) (Sigma). Proteins (20 μg) were resolved on 10% polyacrylamide gels containing sodium dodecyl sulfate. After electrophoresis, proteins were transferred to NitroPure nitrocellulose transfer membranes (Micron Separations Inc.), and blots were blocked overnight at 4°C. Blots were probed with the indicated primary antibodies and appropriate conjugated secondary antibodies. Protein bands were visualized using the ECL+ Western blotting detection reagents (Amersham Biosciences) followed by exposure of the blot to film.

Indirect immunofluorescence analysis (IFA).The indicated cell lines were grown on tissue culture chamber slides and infected with either WNV-NY or SenV (50 hemagglutinating units). At the indicated times postinfection, slides were washed with phosphate-buffered saline (PBS) and fixed with 3% paraformaldehyde for 30 min at room temperature. Cell monolayers were permeabilized with a solution of PBS-0.2% Triton X-100 for 15 min, followed by a 1-h incubation in PBS containing 10% normal goat serum. After being rinsed with PBS, MEFs were incubated for 1 h in the presence of rabbit anti-mouse IRF-3 (1:500), mouse polyclonal anti-WNV antibody (1:750), and goat polyclonal anti-SenV (1:500) in PBS-0.05% Tween 20-3% bovine serum albumin. Cells were washed three times with PBS-0.5% Tween 20 and incubated with donkey anti-goat IgG-Alexa 594 antibody conjugate (1:2,000) for 1 h at room temperature. Cells were washed three times and incubated with goat anti-rabbit IgG-Alexa 488 (1:2,000) and goat anti-mouse IgG-Alexa 350 (1:1,000) for 1 h. Permeabilized U-2 OS/NS3/4A cells were incubated with a rabbit polyclonal anti-human IRF-3 (1:500) and a mouse anti-HCV NS3/4A (1:250) in PBS-0.05% Tween 20-3% bovine serum albumin and washed three times with PBS-0.5% Tween 20. Slides were then incubated with goat anti-rabbit IgG-Alexa 488 antibody conjugate and goat anti-mouse IgG-rhodamine (1:2,000) for 1 h at room temperature. Following incubation with secondary antibodies, cells were washed three times with PBS-0.5% Tween 20 and allowed to dry, and the slides were overlaid with Vectashield solution (Vector Labs). The coverslips were mounted and visualized with a Zeiss Axiovert fluorescence microscope equipped with a digital camera.

RIG-I-mediated regulation of the innate antiviral response to WNV.We have previously demonstrated that ablation of RIG-I delays the induction of a subset of ISGs in response to WNV infection, suggesting that RIG-I signaling is required to trigger the initial antiviral response to WNV (7). In order to examine the global effect of RIG-I ablation on the cell's ability to respond to WNV, gene expression profiles of WNV-infected WT and RIG-I^−/− MEFs were assessed using microarray analysis. Cells were infected at an MOI of 0.5 to allow cell-to-cell spread of the virus within the cultures and thereby more closely approximate an in vivo infection. Total RNA was collected at 20 h postinfection, an early time point at which the antiviral response has been shown to be detectable in WT but not RIG-I-deficient MEFs, and 32 h postinfection, a time point at which a subset of ISGs have previously been shown to be induced in WNV-infected WT and RIG-I^−/− MEFs (7). These two time points were chosen in an attempt to distinguish between antiviral genes with delayed induction profiles in WNV-infected RIG-I^−/− MEFs and genes whose expression is completely dependent on RIG-I signaling. Comparison of the gene expression profiles for WT and RIG-I^−/− MEFs demonstrated that the global response to WNV infection was attenuated at 20 h postinfection in RIG-I-deficient MEFs (Fig. 1). Only 0.3% of genes were differentially regulated in WNV-infected RIG-I^−/− MEFs compared to 3.3% of genes in WT MEFs. In contrast, similar percentages of genes were differentially regulated in RIG-I^−/− and WT MEFs at 32 h postinfection, demonstrating that signaling through RIG-I is essential for stimulation of the initial antiviral response to WNV as well as a broad modulation of the cell's transcriptional profile.

• Open in new tab
• Download powerpoint

FIG. 1.

Microarray analysis of WT and RIG-I^−/− MEFs. Graphic representation of the total number of host cell genes whose expression was differentially modulated in WNV-infected cells compared to mock-infected cultures. Genes were selected based on two criteria: a greater-than-99% probability of being differentially expressed (P < 0.01) and an overall change in expression of twofold or greater. The total number and frequency of genes that fit these criteria are indicated below each bar.

A focused analysis of known immune response genes identified 77 WNV-responsive genes in WT MEFs, all of which exhibited sustained expression throughout the course of the experiment (Fig. 2). Of these genes, 24 exhibited similar kinetics of induction in both WT and RIG-I^−/− MEFs (Fig. 2A; also see Table S1 in the supplemental material), though several genes exhibited a reduced level of induction in RIG-I^−/− MEFs. The induction patterns of the remaining WNV-responsive genes were substantially altered in the absence of RIG-I (Fig. 2B and C; also see Tables S2 and S3 in the supplemental material). Two additional expression profiles were observed for the WNV-responsive genes in RIG-I^−/− cells. The first profile consisted of genes with delayed kinetics of induction (Fig. 2B; also see Table S2 in the supplemental material), similar to the pattern previously reported for ISG54 and ISG56 (7). Expression of these genes was detected only at the 32-h-postinfection time point in RIG-I^−/− MEFs; however, several of these genes, such as the flavivirus resistance gene OAS1b, were expressed at lower levels in RIG-I^−/− MEFs than in WT MEFs. Microarray analysis also identified a novel subset of WNV-responsive genes that were not induced in RIG-I-deficient MEFs at either time point (Fig. 2C; also see Table S3 in the supplemental material).

• Open in new tab
• Download powerpoint

FIG. 2.

Expression profiles of WNV-responsive genes. (A) WNV-responsive genes with similar kinetics of induction in WT and RIG I^−/− MEFs. (B) Genes exhibiting a delayed induction of expression in RIG-I^−/− MEFs compared to WT MEFs. (C) Genes that were not induced in RIG-I^−/− MEFs compared to mock-infected cells at either time point.

IFN-α mediates a second phase of ISG expression in response to WNV infection.Among the WNV-responsive genes detected in WT MEFs were several subtypes of IFN-α (Fig. 2B), the expression of which would be expected to initiate an amplification of the innate antiviral response. In the case of WNV-infected RIG-I^−/− cells, the expression of IFN-α was delayed until approximately 32 h postinfection (Fig. 2B). This delay in IFN-α expression would be predicted to postpone the activation of the IFN amplification loop and the subsequent second round of induction of ISGs. This suggested that the WNV-responsive genes that were not expressed in RIG-I^−/− MEFs by 32 h postinfection (Fig. 2C) might be induced by IFN-α at later times. In order to assess this possibility, the kinetics of induction of ISG15, an IFN-inducible gene that has been previously shown to be induced in response to WNV infection in a variety of cells in vitro and tissues in vivo (8, 18, 41), was examined by Western blot analysis (Fig. 3A). As a control, the expression of ISG54, a protein with an expression profile similar to that of IFN-α in both WT and RIG-I^−/− MEFs, was also monitored. As previously reported, WNV-mediated induction of ISG54 expression was delayed in RIG-I^−/− MEFs compared to WT; however, expression was clearly detectable in both cell lines by 32 h postinfection (7). In comparison to ISG54, ISG15 exhibited an extended lag period prior to the induction of expression in RIG-I-deficient cells.

• Open in new tab
• Download powerpoint

FIG. 3.

Analysis of the induction kinetics of WNV-responsive genes at late times postinfection. (A) Comparison of the kinetics of expression of ISG54 and ISG15 in WT and RIG-I^−/− MEFs infected with WNV. WT and RIG-I^−/− MEFs were infected with WNV and whole-cell lysates were collected at the indicated times postinfection (p.i.). Steady-state levels of ISG54, ISG15, WNV, and GAPDH were examined by Western blotting. (B) WT and RIG-I^−/− MEFs were infected with WNV in the presence of control α-IgG or α-IFN-α/β antibodies (400 U/ml). Whole-cell lysates were assessed for steady-state levels of ISG15, STAT1, and GAPDH. α-, anti.

The timing ISG15 expression suggested that its induction was mediated through the IFN-α amplification loop. To assess the role of IFN-α in WNV-induced ISG expression, WT and RIG-I^−/− MEFs were infected with WNV in the presence of anti-IFN-α/β or control antibodies, and the expression of ISG15 was examined by Western blotting at late times postinfection (Fig. 3B). Anti-IFN-α/β antibodies reduced the level of WNV-induced ISG15 expression in WT MEFs and blocked expression in RIG-I^−/− MEFs, while control antibodies had no effect on the induction of ISG15. To confirm that this observation was not restricted to ISG15, the expression pattern of a second WNV-responsive gene, STAT1α, was also examined. STAT1α was not detected in WNV-infected RIG-I^−/− MEFs by microarray analysis at 32 h postinfection (Fig. 2C). However, Western blot analysis confirmed that STAT1α was induced in RIG-I-deficient cells at later times postinfection (Fig. 3B), confirming that a second round of innate antiviral gene expression is induced during WNV infection. Furthermore, the addition of anti-IFN-α/β antibodies to culture supernatants inhibited the induction of STAT1, demonstrating that induction of a second wave of WNV-responsive gene expression is mediated through IFN.

MDA5 mediates IRF-3 activation in the absence of RIG-I.The fact that cells lacking RIG-I were still able to respond to WNV infection indicated that a second PRR was also involved in detecting viral replication. A likely candidate for a secondary PRR molecule sensor for WNV infection is the RIG-I homolog MDA5. In initial experiments with MEFs derived from MDA5^−/− and littermate control mice, the rate of WNV replication was substantially reduced compared to that previously observed for RIG-I^−/− and control WT MEFs (compare Fig. 3A and 4). The low level of WNV replication in WT and MDA5-deficient cells failed to stimulate a robust antiviral response (Fig. 4). Therefore, MDA5's potential contribution to the stimulation of the antiviral response to WNV infection could not be accurately assessed using these cell lines. In order to overcome this issue, WNV-mediated induction of the innate antiviral response was evaluated for RIG-I^−/− MEFs coinfected with SenV. It has previously been demonstrated that the V proteins of several paramyxoviruses, including SenV, specifically bind to MDA5 and subsequently block its signaling activity (1, 5). Therefore, signaling through the MDA5 pathway can be selectively inhibited by infecting cells with SenV. RIG-I^−/− MEFs were infected with SenV or WNV or coinfected with SenV and WNV, and the innate host response was assessed by monitoring the cellular localization of IRF-3 by IFA (Fig. 5). As previously reported, infection with WNV alone induced IRF-3 nuclear localization (Fig. 5g) (7). In contrast, SenV infection did not induce IRF-3 activation in RIG-I^−/− MEFs (Fig. 5f), confirming that the virus was obstructing signaling through the MDA5 pathway. Similarly, IRF-3 nuclear localization was not detected in cultures coinfected with SenV and WNV (Fig. 5h), demonstrating that SenV blocked the cell's ability to respond to WNV infection. These data indicated that the innate antiviral response to WNV in RIG-I-deficient cells is mediated through MDA5.

• Open in new tab
• Download powerpoint

FIG. 4.

WNV infection of MDA5^−/− and WT MEFs. Whole-cell lysates collected at the indicated times postinfection were analyzed for steady-state levels of ISG54, WNV protein, and GAPDH by Western blot analysis.

• Open in new tab
• Download powerpoint

FIG. 5.

Cellular localization of IRF-3 in RIG-I^−/− MEF coinfected with WNV and SenV. Mock (a, e, and i), SenV (b, f, and j), and WNV (c, g, and k) cultures, along with cultures coinfected with SenV and WNV (d, h, and l), were fixed at 72 h postinfection and probed for WNV protein expression (a to d), IRF-3 (e to h), and SenV protein expression (i to l). α-, anti.

IPS-1 mediates the innate antiviral response to WNV.Signaling through the RIG-I and MDA5 pathways converges at the adaptor molecule IPS-1 (16, 28, 38, 43). If the RIG-I and MDA5 pathways were solely responsible for sensing WNV infection, then IPS-1 would be predicted to play a central role in establishing an antiviral state in response to infection. It has previously been demonstrated that the NS3/4A protease of HCV prevents signaling through both the RIG-I and MDA5 pathways by cleaving IPS-1 (20, 23, 28). Therefore, HCV NS3/4A was used as a tool to investigate the involvement of IPS-1 in WNV-mediated activation of the innate antiviral response. The cellular distribution of IRF-3 was monitored in cultures infected with WNV in the presence or absence of HCV NS3/4A. In control cells, which were not expressing HCV NS3/4A, WNV induced nuclear translocation of IRF-3 as previously reported (Fig. 6A, panel a) (8). Conversely, in the presence of HCV NS3/4A, WNV-mediated translocation of IRF-3 was completely inhibited, suggesting that HCV NS3/4A blocked the cells' ability to respond to WNV infection. Furthermore, HCV NS3/4A expression also prevented induction of the IRF-3 target gene ISG56 in response to WNV infection (Fig. 6B).

• Open in new tab
• Download powerpoint

FIG. 6.

IPS-1 plays a central role in modulating the innate antiviral response to WNV. (A) Induction of IRF-3 nuclear localization by WNV infection. Cellular localizations of IRF-3 (panels a and b) and HCV NS3/4A (panels c and d) protein expression were examined by IFA in U-2 OS/NS3/4A cells propagated in the presence (panels a and c) or absence (panels b and d) of tetracycline. (B) Effect of HCV NS3/4A on WNV-induced expression of IRF-3 target genes. U-2 OS/NS3/4A cells propagated in the presence or absence of tetracycline were infected with WNV (MOI = 3). Induction of ISG56 was assessed by Western analysis of whole-cell lysates harvested at the indicated times postinfection. Blots were stripped and reprobed for HCV NS3/4A to confirm the induction of expression in the absence of tetracycline. The levels of GAPDH expression were also assessed to control for loading. (C) Activation of the innate antiviral response in IPS-1^−/− MEFs. WT and IPS-1^−/− MEFs were infected at an MOI of 5 (based on titer from Vero cells), and whole-cell lysates were collected at 48 h postinfection. Expression of ISG54, ISG56, WNV, and GAPDH was assessed by Western blotting.

The role of IPS-1 in signaling the induction of the innate antiviral response to WNV was also assessed using MEFs recovered from IPS-1-deficient mice. Lysates recovered from WNV-infected WT and IPS-1^−/− MEFs were examined for the expression of ISG56 and a second IRF-3 responsive gene, ISG54 (Fig. 6C). In the absence of IPS-1, WNV-mediated induction of both ISG56 and ISG54 was completely ablated, demonstrating that signaling through IPS-1 is necessary for the establishment of an antiviral state in response to WNV infection.