Genome-wide RNAi screen identify modulators of SeV-induced IFNB1 expression

The individually arrayed lentiviral-based short hairpin RNA (shRNA) human library (Mission TRC-Hs1.0 from Sigma-Aldrich) produced in-house was used to knockdown the expression of _∼15,000 human genes by RNA interference (RNAi) combining 3 shRNAs per gene. HEK 293T cells stably expressing the firefly luciferase gene under the control of the IFNB1 promoter was used to monitor type I IFN expression in response to SeV infection (Figure 1A). A robust 100-fold increase in luminescence was obtained upon infection and the cell assay was used to screen the shRNA human library by silencing one gene per well in 96-well plate format (Figure S1). The primary genome-wide screen was performed with control shRNAs on each plate targeting the positive regulator (PR) MAVS gene, which produced a tenfold decrease in luminescence, and the negative regulator (NR) NLR family member X1 (NLRX1) gene [11], which increased the luminescence by approximately twofold (Figure S1A). As control, the transduction of cells with non target sequence (NT) shRNA-expressing lentiviral particles was evaluated on SeV-induced IFNB1 promoter activity three days post-transduction. Similar inductions in luciferase and mRNA levels were detected in control transduced cells compared to non-transduced cells demonstrating that infection of cells with shRNA-expressing lentivirus is not contributing to the SeV-mediated antiviral responses. Following statistical analyses of the data using the strictly standardized mean difference (SSMD) as a cutoff [12], we selected 292 potential PR whose silencing significantly decreased the IFNB1 reporter activity (SSMD≤−1.314, approximately ≥2 SDs from the plate median) and 289 NR whose silencing significantly increased the IFNB1 reporter activity (SSMD≥1.662, approximately ≥2.5 SDs from the plate median). For the 581 gene hits, we produced five independent shRNA-expressing lentiviruses that were tested individually in confirmation and secondary screens (Figure 1B). These secondary assays relied on the activation of the RLR pathway at different landmarks of the signaling cascade, through transfection of polyI∶C (dsRNA mimetic) to active RIG-I directly, overexpression of MAVS leading to its constitutive activation, and overexpression of IRF3(5D), a phosphomimetic mutant of IRF3 constitutively targeted to the nucleus, to activate transcription at the IFNB1 promoter (Figure 1C). The inhibition profile of each shRNA in secondary screens allowed to classify gene hits within the RLR signaling cascade in four functional groups: I - SeV specific, II - cytoplasmic dsRNA sensing, III - MAVS-dependent signaling, IV - nuclear import or transcription factor-dependent process (Figure 1D for validation of secondary assays with control genes). Secondary screening resulted in the identification of 237 gene hits (132 PR and 105 NR) that modulate specifically the reporter activity at the IFNB1 promoter without affecting reporter expression from a nonimmune-related endogenous elongation factor 1 alpha (EF1α, also known as EEF1A1) promoter (Figures 1B and 1C). Further prioritizing studies included the transcriptional modulation of virus-induced endogenous IFNB1 mRNA synthesis with at least two shRNAs, which emphasized 114 gene hits, 80 PR and 34 NR (Figures 1E, 1F and S2). Finally, qRT-PCR validation showed a knockdown efficiency of greater than 40% (median knockdown efficiency of 70%) with two independent shRNAs for 59 gene prioritized gene hits (Figure 1F and see Table S1 for the complete set of primary, secondary and qRT-PCR data). The reliability of the screening approach was confirmed by detection of 10 out of the 114 gene hits (9.6%) previously associated to innate immunity, including the signaling components RIG-I (PR group II), MAVS (PR group III), IRF3 (PR group IV) and the NF-κB subunits REL and RELA (both PR group IV).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Genome-wide gene silencing study of virus-induced innate immune responses and bioinformatics analyses.

(A) Schematic representation of the primary genome-wide screen and secondary screens. HEK 293T cells stably expressing the luciferase gene under the control of the IFNB1 promoter were transduced with arrayed lentiviruses combining three shRNAs per gene (primary screen) or with five individual shRNA-expressing lentiviruses for each gene hit (secondary screens) in a 96-well format. After 72 hours, cells were challenged with SeV virus (primary and confirmation screens) or transfected with polyinosinic∶polycytidylic acid (polyI∶C), MAVS- or IRF3(5D)-expressing plasmids (secondary screens) for 16 hours before measuring IFNB1 promoter-driven luciferase activity. (B) Decision tree of primary screen and summary data of gene hits obtained in secondary and validation screens. Selected gene hits (114) that were confirmed and validated with endogenous IFNB1 screens by qRT-PCR induced a modulation of more than 25% of the IFNB1 promoter activity with at least two independent shRNAs following SeV infection. Prioritized gene hits (59) for which knockdown of the target gene was greater than 40% with two independent shRNAs are also identified. (C) Schematic representation of confirmation and secondary assays for epistasis analysis of gene hits acting on the signaling cascade leading to IFNB1 production. SeV infection (primary and confirmation screens), polyI∶C (dsRNA mimetic), MAVS or IRF3(5D) expressing plasmids transfection (secondary screens) were used to activate innate immune response. A non-specific assay was used to discard gene hits affecting nonimmune-related transcription by measuring transcriptional activity of EF1α constitutive promoter. (D) Heat map indicating modulation of IFNB1 promoter activity following silencing of control genes in confirmation and secondary assays (log₂ scale). The functional profiling data for controls and gene hits allow classification within four functional groups: I - SeV specific, II - cytoplasmic dsRNA sensing, III - MAVS-dependent signaling, IV - nuclear import or transcription factor-dependent process. (E) Functional profiling data of 114 gene hits confirmed with at least two shRNAs in the SeV confirmation screen and further validated using endogenous IFNB1 mRNA quantification by qRT-PCR. Manual clustering was performed to classify each gene hit in one of the four functional groups. (F) qRT-PCR validation data of the endogenous IFNB1 mRNA levels and target gene knockdown efficiency in transduced cells with each lentivirus-expressing shRNA for the 114 gene hits. (G) Enriched Gene Ontology (GO) biological process and molecular function terms (P<0.05) for the 114 genes confirmed by qRT-PCR relative to all genes examined in the genome-wide RNAi screen. (H) Identification of 17 out of 237 gene hits that are induced by more than twofold after SeV infection or IFN-α treatment in our microarray analysis of HEK 293T cells (13 genes) or previously described as ISGs (4 genes in gray) [75]. The gene hits that modulate endogenous IFNB1 mRNA are indicated in bold. (I) Venn diagram representing the overlap between various RNAi screens including: our RLR RNAi screen (290 genes validated in secondary screens, including 53 hits considered as non specific in pEF1α assay), a NOD-like receptor screen (NLR) and nine independent viral replication screens (HIV, influenza, HCV and West Nile virus). Numbers between brackets represent the overlap when only the 237 genes considered as specific in our RLR screen are included.

https://doi.org/10.1371/journal.ppat.1003416.g001

Bioinformatics analysis of gene hits

To create a more comprehensive view of all prioritized gene hits in the regulation of antiviral innate immune responses, we performed a bioinformatics meta-analysis that integrates the data of our functional genomics screen with previous functional and proteomic studies. Gene ontology (GO) molecular functions of the 114 gene hits identified enrichment for 10 statistically significant terms encompassing expected terms: regulation of the NF-κB cascade, response to virus, RLR signaling pathway, regulation of cytokine production and antiviral defense, as well as the interesting functional groups: nuclear pore, proteasome and spliceosome (Figure 1G). Additional microarray analysis was performed to identify genes that were induced by more than twofold following SeV infection or IFN-α treatment. Based on our gene profiling analysis and previously described ISGs [13], 17 out of 237 gene hits were identified as ISGs (Figure 1H). Globally, identified gene hits were significantly enriched in ISGs (12 out of 114 genes or 10.5%) compared to a 2.0% representation of the human genome. The analysis also identified four nuclear pore complex (NPC) genes: nucleoporin 93 kDa (NUP93), RAN member RAS oncogene family (RAN), nucleoporin like 1 (NUPL1) and Exportin 1 (CRM1 homolog, yeast) (XPO1) within the group IV that significantly reduced virus-induced IFNB1 transcription upon individual gene perturbation. This example highlights a central role of XPO1-containing protein complex in innate antiviral response that is targeted by several viruses. Indeed, XPO1 is implicated in the nuclear export of the 60S ribosomal RNA subunit (rRNAs) [13] and specific mRNAs (IFNA1 mRNA [14]), and also contributes to the transport of viral proteins and RNA transcripts [15]. Interestingly, XPO1-dependent export of HIV-1 Rev protein required the DExD box RNA helicase DDX3 [16], which plays a role in type I IFN induction through TBK1/IKBKE (also known as IKKε)-mediated phosphorylation of IRF3 [17], [18].

In order to identify common proteins functioning as sensors of different danger signals and effectors of innate response, and as host antiviral restriction factors targeted by viruses, we compared gene hits of our RLR screen to those of biologically relevant RNAi screens for pathogen replication (HIV [19]–[21], HCV [22], [23], Influenza [24]–[26] and West Nile virus [27]) and for Nucleotide-binding and oligomerization domain (NOD)–like receptor (NLR) response [28] (Figure 1I). These screens were performed with different genome-wide RNAi libraries, cell lines, reporter genes and experimental designs. Nevertheless, the overlap analysis identified 58 putative modulators of virus-induced innate responses that are also associated to virus infection, and 9 common modulators of the RLR- and NLR-mediated recognition and signaling pathway (Supplementary Table 2). The overlapping genes are significantly enriched for the functional groups proteasome and spliceosome, and for the RLR, NLR and cancer pathways. Our comparison also highlights four shared genes between the three types of RNAi screen: RELA/p65, central component of the NF-κB pathway; RIPK2 a serine/threonine protein kinase containing a C-terminal caspase activation and recruitment domain (CARD) acting as a potent activator of NF-κB [29]; KLF6, a zinc finger protein recently identified as a critical transcription factor required for the innate immune response against influenza virus [30]; and CYP2U1, a hydroxylase member of the cytochrome P450 superfamily of enzymes involved in the synthesis of lipids with a potential physiological role in fatty acid signaling processes in brain and thymus [31]. Our genome-wide shRNA screen complements previously available immune and viral replication screens to enrich the global interaction networks of innate immune response, which could identify targets for the development of broad-spectrum anti-infective agents.

Secretion of WNT2B and WNT9B acts in a feedback inhibition of SeV-induced IFNB1 transcription

The RNAi screen identified gene hits encoding WNT ligands that modulate virus-induced innate response. WNT ligands are a family of highly conserved secreted glycoproteins that regulate multiple processes in development and tissue homeostasis. These processes are mediated through canonical WNT/β-catenin (CTNNB1) and non-canonical WNT pathways [32], [33]. In recent years, few studies have reported a role of WNT pathway in the regulation of inflammation. In drosophila, WntD was shown to act as a feedback inhibitor of the NF-κB homolog Dorsal in response to bacterial infection [34]. More recently, Wnt2 and Wnt11 were shown to inhibit bacterial-induced inflammation in intestinal epithelial cells in mouse [35], [36]. We identified two highly prioritized genes encoding WNT2B (also called WNT13, NR group IV) and WNT9B (also called WNT15 and WNT14B, NR group IV), which upon gene silencing significantly increased gene transcription of the pIFNB1-LUC reporter construct and of the endogenous IFNB1 promoter in SeV-infected cells (Figure 2A). The knockdown of both WNT ligands increased pIFNB1-LUC expression induced by transfected polyI∶C and overexpression of MAVS and of IRF3(5D), consistent with a regulatory role of the antiviral response in a feedback inhibition loop (Figure 2A). SeV infection led to a significant increase of the virus-induced representative proteins IFIT1 (ISG56), IFIT2 (ISG54) and DDX58 (RIG-I) at 16 hours post-infection with knockdown of WNT2B or WNT9B by two independent shRNAs (Figure 2B and Figure S3 for knockdown efficiency). The increased pIFNB1-LUC expression upon WNT2B or WNT9B silencing is further illustrated by enhancement of IFIT1 mRNA induction levels, and more importantly by increased levels of IFN-β protein in the supernatant of infected cells (Figures 2C–E). As an example, we observed in WNT9B knockdown cells versus control cells transduced with shRNA NT more than 6,000 pg/mL of secreted IFN-β protein (versus 250 pg/mL), a 650-fold in IFNB1 mRNA (versus 100-fold), a 800-fold in IFIT1 mRNA (versus 250-fold), and a 100-fold increase in NF-κB-driven TNF mRNA levels (versus 50-fold in control shRNA NT cells) (Figures 2E and S4). Co-transfection of shRNA-resistant expression vectors for WNT2B or WNT9B rescued the knockdown by decreased pIFNB1-LUC expression excluding shRNA off-target effects (Figure S5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. WNT2B and WNT9B ligands as novel negative regulators of antiviral innate immunity.

(A) Heat map indicating log₂ effect of silencing WNT2B and WNT9B genes with 3 independent lentivirus-expressing shRNAs (left) and qRT-PCR validation of endogenous IFNB1 mRNA levels and of target gene knockdown efficiency (right). (B) Immunoblot analysis of IFIT1, IFIT2 and DDX58 in WNT2B, WNT9B and WLS knockdown HEK 293T cells (two-independent shRNAs per gene) following infection with SeV for 16 hours. (C) IFNB1 promoter-driven luciferase activity in HEK 293T cells treated as described in (B). (D) Fold induction of IFIT1 mRNA levels in HEK 293T cells treated as described in (B). qRT-PCR determination represents the average mRNA RQ normalized versus ACTIN and HPRT1 mRNA. (E) ELISA quantification of secreted IFN-β protein in supernatant of HEK 293T cells treated as described in (B). (F) Immunoblot analysis of HEK 293T supernatants and lysates at various time points in cells overexpressing WNT2B and WNT9B and infected with SeV or treated with IFN-α. The specific bands are indicated with arrows. (G) Fold induction of CTNNB1-TCF/LEF dependent reporter activity (TOP-flash) with addition of purified WNT3A or WNT9B proteins to HEK 293T cell supernatants for 16 hours. (H) Relative fold induction of IFNB1- or EF1α-driven reporter activity with dose-dependent addition of purified WNT9B protein to cell supernatants of SeV infected HEK 293T cells for 16 hours. (I–J) Fold induction of IFNB1 (I) or IFIT1 (J) mRNA levels following dose-dependent addition of purified WNT3A or WNT9B proteins to cell supernatants of SeV infected HEK 293T cells for 16 hours. qRT-PCR determination represents the average mRNA RQ normalized versus ACTIN and HPRT1 mRNA. P values<0.05 (*), <0.01 (**) or <0.001 (***) are indicated.

https://doi.org/10.1371/journal.ppat.1003416.g002

We then investigated the modulation in gene expression of the two WNT ligands following virus infection. Since WNT2B and WNT9B mRNA levels are barely inducible (less than 2-fold) in all conditions tested (Figure S6), we evaluated the secretion of ectopically expressed WNT2B and WNT9B upon SeV infection and IFN-α treatment (Figure 2F). Our data revealed a secretion of WNT2B that build-up until 8 hours post-infection, while no secretion was observed with IFN-α treatment. The virus-induced secretion of WNT2B was also confirmed in cervical-derived HeLa and hepatoma Huh7 human cells (Figure S7A). In contrast, WNT9B is rapidly secreted with IFN-α treatment, while its secretion is merely initiated 8 hours post SeV infection, a time point correlating with virus-induced IFNB1 expression (Figures 2F and S6). We then confirmed the requirement of virus-induced WNT secretion for the negative regulation of IFNB1 transcription by perturbation of wntless homolog (Drosophila) (WLS) gene encoding a conserved membrane protein dedicated to WNT secretion in signaling cells [37]–[39]. Indeed, we observed that silencing of WLS reduced WNT secretion (Figure S7B), and increased the expression of IFIT1, IFIT2 and DDX58 proteins, IFIT1 mRNA levels and secretion of IFN-β protein induced by SeV in HEK 293T cells (Figures 2B, 2D, 2E and S3 for knockdown efficiency ).

Finally, to confirm a direct role of WNT ligand in the negative regulation of IFNB1 mRNA expression, we incubated SeV-infected cells with purified WNT proteins. As expected, purified WNT9B decreased virus-induced pIFNB1-LUC activity, as well as mRNA levels of IFNB1 and of IFIT1 from endogenous promoters, by up to threefold in dose-response without affecting the constitutive EF1α promoter (Figures 2H–J). In control experiments, purified WNT9B weakly induced the CTNNB1-dependent T-cell factor/lymphoid enhancer factor (TCF/LEF)-promoter driven reporter activity (M50 Super 8×TOP-flash [40]) by less than 3-fold while purified WNT3A protein achieved a 60-fold induction of reporter activity (Figure 2G). In addition, purified WNT3A weakly decreased SeV-induced IFNB1 and IFIT1 transcription and at concentrations (250 nM) well above those required to promote CTNNB1 TCF/LEF-dependent reporter activity (compare Figures 2G with Figures 2I and 2J). Overall, these results strongly support a specific effector role of virus-induced secretion of WNT2B and WNT9B ligands in innate immunity, acting in a feedback inhibition of IRF3- and NF-κB-dependent IFNB1 response to virus infection.

CTNNB1 negatively regulates IFNB1 gene expression

In a canonical WNT pathway, WNT ligands bind to cell-surface receptors of the frizzled family receptor (FZD) and low density lipoprotein receptor-related protein (LRP) 5/6, activate dishevelled (DVL) and ultimately result in stabilization of cytoplasmic CTNNB1 and its nuclear translocation. The process is mediated by the WNT/LRP-signalosome through recruitment at cell membrane of glycogen synthase kinase 3β (GSK3B) and AXIN, disrupting the cytoplasmic GSK3B/AXIN/adenomatous polyposis coli (APC) destruction complex of CTNNB1. WNT-mediated activation of CTNNB1 promotes specific gene expression via interaction with TCF/LEF transcription factors. While SeV infection per see do not induce TCF/LEF transcription (data not shown), we investigated the contribution of CTNNB1 on SeV-mediated IFNB1 and IFIT1 induction to better understand mechanistically the role of WNT2B/WNT9B in regulation of innate immunity. The knockdown of CTNNB1 gene produced the expected enhancement in virus-induced pIFNB1-LUC reporter activity and in IFIT1 protein levels (Figures 3A and 3B). Knockdown of CTNNB1 gene by specific shRNA and siRNAs also enhanced induction of DDX58 protein levels, of mRNA levels of IFNB1, DDX58 and TNF from endogenous promoters and notably increased secretion of IFN-β protein (Figures 3C–G and Figure S3 for knockdown efficiency). Conversely, the ectopic expression of CTNNB1 resulted in diminution of virus-induced IFNB1 transcription in a dose response manner, while promoting the TCF/LEF-dependent reporter activity in control experiments (Figures 3H and 3I). To test the potential additive effect of WNT2B and WNT9B knockdown and to confirm that these WNTs act through CTNNB1 in regulating innate antiviral immunity, we performed double knockdown experiments. We showed an additive effect of the knockdown of WNT2B and WNT9B together in enhancing IFNB1 induction following SeV infection (Figure S8). However, the knockdown of WNT2B or WNT9B in combination with CTNNB1, as expected, do not further increase IFNB1 induction when compared to the knockdown of CTNNB1 alone (Figure S8).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. CTNNB1 is a negative regulator of antiviral innate immunity.

(A) IFNB1 promoter-driven luciferase activity in HEK 293T cells transduced with lentivirus-expressing shRNA NT (control) or 3 shRNAs targeting CTNNB1 for four days and subjected to SeV infection for 16 hours. (B) Immunoblot analysis of CTNNB1 and IFIT1 in HEK 293T cells transduced with 3 lentivirus-expressing shRNAs targeting CTNNB1 and infected with SeV. (C) Immunoblot analysis of CTNNB1, IFIT1 and DDX58 in SeV-infected HEK 293T cells previously transduced with a shRNA or transfected with a pool of four siRNAs targeting CTNNB1 for three days. (D–F) Fold induction of IFNB1 (D), DDX58 (E) and TNF (F) mRNA levels in HEK 293T cells treated as described in (C). qRT-PCR determination represents the average mRNA RQ normalized versus ACTIN and HPRT1 mRNA. (G) ELISA quantification of secreted IFN-β protein in supernatant of HEK 293T cells treated as described in (A). (H–I) Fold induction of IFNB1 promoter-driven luciferase activity (H) and of CTNNB1-TCF/LEF dependent reporter activity (TOP-flash, I) following dose-dependent transfection of CTNNB1-expressing plasmid (50, 100, 150 and 200 ng) for 48 hours in SeV-infected HEK 293T cells.

https://doi.org/10.1371/journal.ppat.1003416.g003

We then confirmed the stabilization of endogenous CTNNB1 upon SeV infection in a kinetics study. Indeed, immunoblot analysis of infected cell extracts demonstrated the progressive increase in activated CTNNB1 (dephosphorylated on Ser37 or Thr41) from 4 to 24 hours post-infection, followed by a slight decrease at 48 hours (Figure 4A). The stabilization of CTNNB1 following SeV infection is also observed following ectopic expression of a FLAG-tag CTNNB1 protein (see Figure S12B). Importantly, the kinetic showed no induction in the basal level of phosphorylated CTNNB1 Ser552 upon SeV infection. This is in contrast to the LRRFIP1-dependent phosphorylation of CTNNB1 at Ser552 upon infection with Listeria monocytogenes, which promotes its transcriptional activation required for IFNB1 production in mouse macrophages [41]. Moreover, the kinetics study of SeV infection performed in CTNNB1 knockdown cells clearly showed that blocking CTNNB1 accumulation upon gene silencing correlated with increased activation of the antiviral response as demonstrated by the protein levels of IFIT1 and DDX58, as well as phosphorylated levels of NFKBIA (IκBα) at serine 32, an indicator of NF-κB activation (Figure 4A). IFIT1 and DDX58 proteins increased from 24 to 48 hours post-infection while phosphorylated NFKBIA increased up to 32 hours post-infection in CTNNB1 knockdown cells. As a control, we demonstrated that this increase in innate antiviral response is not a consequence of increase viral replication, since the levels of SeV hemagglutinin-neuraminidase (HN) viral proteins are unchanged in CTNNB1 knockdown cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. SeV infection induces CTNNB1 stabilization and nuclear translocation.

(A) Immunoblot analysis of dephosphorylated active CTNNB1 at Ser37/Thr41, CTNNB1 phosphorylated at Ser552, NFKBIA (IκBα) phosphorylated at Ser32, IFIT1, DDX58 and SeV protein HN of HEK 293T transduced with lentivirus-expressing shRNA 45 targeting CTNNB1 and shRNA NT (control) for four days and subjected to SeV infection for 0, 4, 8, 24, 32 or 48 hours. (B–D) Fold induction of IFNB1 (B), IFIT1 (C) and TNF (D) mRNA levels in HEK 293T cells treated as described in (A). qRT-PCR determination represents the average mRNA RQ normalized versus ACTIN and HPRT1 mRNA. (E) Confocal analysis of HEK 293T cells using Hoechst, anti-CTNNB1 active form and anti-IRF3 antibodies without virus infection or following 16 hours infection with SeV. Nuclear detection of IRF3 and CTNNB1 is identified by white arrows in infected cells.

https://doi.org/10.1371/journal.ppat.1003416.g004

In addition, qRT-PCR quantification of IFNB1, IFIT1 and TNF mRNA levels showed increase in their promoter activity up to seven-fold in CTNNB1 knockdown cells (Figures 4B, 4C and 4D). Interestingly, a more rapid decline of the IFNB1 mRNA level is observed at 48 hours post-infection (Figure 4B) compared to IFIT1 and TNF mRNA levels, which remained higher at this time point (Figure 4C and 4D). In immunofluorescence experiments, we specifically followed the cytoplasmic CTNNB1 regulated by the GSK3B destruction complex using the antibody specific to the activated/dephosphorylated CTNNB1. Following infection, CTNNB1 is detected in the nucleus of SeV-infected HEK 293T cells, similarly to IRF3 (Figure 4E). Similar results were observed in infected A549 cells together with a nuclear co-staining of CTNNB1 and NF-κB (Figure S9). Altogether, our data strongly suggest that the stabilization and/or accumulation of CTNNB1 in the nucleus upon virus-induced WNT secretion triggered a CTNNB1-dependent negative feedback loop to reduce antiviral innate immune responses.

GSK3 inhibitors block innate response through CTNNB1 stabilization

Since the inhibition of GSK3B catalytic activity results in CTNNB1 cytoplasmic accumulation, we then investigated a role of GSK3 in the negative regulation of IFNB1 induction (Figure 5). The inhibition of GSK3A/B activity with specific inhibitors BIO (5 µM) and BIO-acetoxime (10 µM) was confirmed by lack of autophosphorylation at Tyr279/Tyr216, respectively, using phospho-specific antibodies (Figure 5C). As expected, pharmacological inhibition of GSK3 increased active CTNNB1 levels (dephosphorylated on Ser37 or Thr41) that correlated with the CTNNB1 TCF/LEF-dependent reporter activity (Figures 5B and 5C). More importantly, the inhibition of GSK3 resulted in a drastic reduction of virus-induced IFNB1 transcription as well as expression of IFIT1 and of NFKBIA phosphorylation at serine 32 when compared to untreated SeV-infected cells (Figures 5A, 5C and 5D). As GSK3 phosphorylates multiple substrates other than CTNNB1 affecting different pathways, we further demonstrated that the decrease in virus-dependent induction of pIFNB1-LUC, IFIT1 and phosphorylated NFKBIA by GSK3 inhibitors is specifically abolished in CTNNB1 knockdown cells (Figures 5A and 5D). Parallel transcriptional changes were observed with IFNB1, IFIT1 and TNF mRNA levels (Figure S10A–C). Finally, similar restoration of the antiviral response was observed in GSK3 inhibitor-treated CTNNB1 knockdown A549 cells following SeV infection (Figure S10D). In control experiments, increased CTNNB1 TCF/LEF-dependent reporter activity with pharmacological inhibition of GSK3 was also specifically abrogated by knockdown of CTNNB1 (Figure 5B). To verify if the inhibition of the innate antiviral response observed upon pharmacological stabilization of CTNNB1 was a consequence of its transcriptional activity through the induction of TCF/LEF-dependent genes, we used the CTNNB1-TCF interaction inhibitor PNU74654 [42]. The addition of PNU74654 (6 µM) reduced the GSK3 inhibitor-mediated TOP-flash induction by 70%, without restoring innate antiviral response to SeV infection as determined by pIFNB1-LUC, pISG56-LUC and p2xNF-κB-LUC reporter activities (Figure 5E). This experiment rule out a role of TCF/LEF inducible genes by the canonical WNT3A pathway (involved in embryonic development or cell fates during normal development) as part of the negative regulation of innate antiviral immunity. Altogether, the results strongly suggest a key role of GSK3 catalytic activity in a feedback inhibition of the antiviral innate immune response mediated by a canonical-like WNT2B/WNT9B/CTNNB1 signaling pathway.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Pharmacological inhibition of GSK3 stabilizes an active CTNNB1 pool and negatively regulates antiviral innate immunity.

(A–B) Relative fold induction of IFNB1-driven reporter activity (A) and fold induction of CTNNB1-TCF/LEF dependent reporter activity TOP-flash (B) in HEK 293T cells transduced with lentivirus-expressing shRNA NT (control) or shRNA 45 targeting CTNNB1 for four days and subjected to GSK3 inhibition with BIO (5 µM) or BIO-acetoxime (10 µM) and SeV infection for 16 hours. (C) Immunoblot analysis of autophosphorylated GSK3α/β at Tyr279/216, dephosphorylated active CTNNB1 at Ser37/Thr41 and IFIT1 in HEK 293T cells treated with GSK3 inhibitors BIO (5 µM) or BIO-acetoxime (10 µM) and infected with SeV for 16 hours. (D) Immunoblot analysis of dephosphorylated active CTNNB1 at Ser37/Thr41, IFIT1 and phosphorylated NFKBIA (IκBα) at Ser32 in SeV-infected HEK 293T cells treated as described in (A–B) and treated with BIO (5 µM) or BIO-acetoxime (10 µM). (E) Relative fold induction of TOP-flash, IFNB1, ISG56, NF-κB and EF1α-driven reporter activity in HEK 293T cells subjected to GSK3 inhibition with BIO (5 µM), TCF/LEF interaction inhibition with PNU74654 (6 µM) or both inhibitors prior to SeV infection for 16 hours.

https://doi.org/10.1371/journal.ppat.1003416.g005

WNT/CTNNB1 pathway negatively regulates innate immune response in primary human cells

The regulation of antiviral innate immunity by a WNT/CTNNB1 pathway was further investigated in primary cultures of normal human bronchial epithelial cells (NHBE), human monocyte-derived macrophages (MDM) and human hepatocytes. The negative regulation of SeV-induced innate response was demonstrated in NHBE cells as shown by the knockdown of WNT2B, WNT9B and CTNNB1, which increased IFNB1 mRNA levels by up to five folds (Figure 6A) and TNF mRNA levels by up to six folds (Figure 6B). Significant increased IFNB1 and TNF mRNA levels are also observed in WLS knockdown cells (1.5 to 2 folds). We demonstrated the key contribution of CTNNB1 in immune cells and showed that the treatment of MDM with GSK3 inhibitors led to a robust inhibition in IFNB1 and IFIT1 mRNA levels up to 90%, in accordance with results obtained in non immune cells (Figure 6C). However, TNF mRNA induction was systematically weakly or unaffected by GSK3 inhibitor treatment, demonstrating some differences in regulation of MDM compared to epithelial cells at 4 hours post-infection. We also confirmed in primary cultures of normal human hepatocytes that the pharmacological inhibition of GSK3 resulted in a drastic inhibition of virus-induced IFIT1 transcription that is abrogated in CTNNB1 knockdown cells (Figure 6D). More importantly, we emphasized the contribution of the GSK3/CTNNB1 in regulation of innate immune response following hepatitis C virus (HCV) infection. Indeed, incubation of primary human hepatocytes with the full-length HCV strain JFH1 for 48 hours led to the induction of IFIT1 protein levels (Figure 6E). Using this model, we showed a drastic inhibition of IFIT1 levels when treating hepatocytes with the GSK3 inhibitor BIO, and in accordance with previous data, a restoration of IFIT1 levels in GSK3 inhibitor-treated CTNNB1 knockdown cells (Figure 6E). It is noteworthy that treatment with GSK3 inhibitor did not affect levels of HCV NS3 protease in these primary hepatocytes, confirming that the observed effect on IFIT1 protein is not related to an alteration of HCV infection at early time points. Upon knockdown of CTNNB1 however, the observed increase in IFIT1 protein levels and decrease in NS3 protein of HCV-infected primary hepatocytes prompted us to evaluate the quantitative effect of WNT/CTNNB1 signaling on HCV infection. Indeed, we observed a reduced HCV replication upon knockdown of WNT2B, WNT9B, WLS and CTNNB1 in Huh7 cells infected with the reporter HCV J6/JFH-1(p7-Rluc2A) virus (Figure 6F). Up to 80% inhibition of HCV infection was observed in CTNNB1 knocked down cells, a similar effect than the silencing of Y-box-binding protein-1 (YB1) in a positive control that we previously identified as a cofactor of HCV replication and assembly [43]. Altogether, these results confirm the role of a canonical-like WNT/GSK3/CTNNB1 pathway in a negative regulation of innate immunity upon viral infections of primary human cells including immune cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. WNT/CTNNB1 signaling acts as a negative regulator of the innate immune response in primary human cells.

(A–B) Fold induction of IFNB1 (A) and TNF (B) mRNA levels in NHBE cells transduced with lentivirus-expressing shRNA 77 targeting WNT2B, shRNA 69 targeting WNT9B, shRNA 58 targeting WLS, shRNA 45 targeting CTNNB1 and NT shRNA in control cells for four days and infected with SeV for 16 hours. (C) Fold induction of IFNB1, IFIT1 and TNF mRNA levels in human monocyte-derived macrophages subjected to GSK3 inhibition with BIO (5 µM) and SeV infection for 4 hours. qRT-PCR determination represents the average mRNA RQ normalized versus ACTIN and HPRT1 mRNA. (D) Fold induction of IFIT1 mRNA in primary human hepatocytes transduced with lentivirus-expressing shRNA NT or shRNA 45 targeting CTNNB1 for four days and subjected to GSK3 inhibition with BIO (5 µM) and SeV infection for 5 hours. qRT-PCR determination represents the average mRNA RQ normalized versus ACTIN and HPRT1 mRNA. (E) Immunoblot analysis of dephosphorylated active CTNNB1 at Ser37/Thr41, IFIT1 and HCV NS3 protease in primary human hepatocytes either transduced with lentivirus-expressing shRNA 45 targeting CTNNB1 for four days and/or subjected to GSK3 inhibition with BIO (5 µM) and infected with Hepatitis C virus (HCV) strain JFH1 for 48 hours. (F) Huh7 cells transduced with shRNAs NT, WNT2B 76, WNT2B 77, WNT9B 69, WNT9B 70, WLS 58, WLS 99, CTNNB1 45 or YB1 48 (positive control) were infected with HCV J6/JFH-1(p7-Rluc2A) reporter virus. Following 96 hours of infection, Rluc was measure to assess HCV viral replication and results are represented as relative viral infection with shRNA NT treated cells arbitrarily set to 1. P values<0.05 (*), <0.01 (**) or <0.001 (***) are indicated.

https://doi.org/10.1371/journal.ppat.1003416.g006

Virus-induced CTNNB1 association with IRF3

To elucidate the underlying mechanism by which activation of a canonical-like WNT/CTNNB1 pathway decreases innate antiviral responses by negatively regulating IRF3- and NF-κB- dependent production of IFN-β production, we determined if CTNNB1 physically interacts with IRF3 and NF-κB subunits by co-immunoprecipitation experiments. Immunoprecipitation followed by immunoblot analysis showed that SeV infection or treatment of cells with IFN-α increase the interaction of FLAG-IRF3 with endogenous CTNNB1 (Figure 7). In contrast, the observed interaction of NF-κB RELA/p65 subunit with endogenous CTNNB1 remains unchanged following these stimulations. Ectopic expression of FLAG-tagged IRF3, constitutively active IRF3(5D) and RELA/p65 further demonstrated an interaction with exogenous CTNNB1 in SeV-infected cells (Figure S12A). Reciprocal immunoprecipitation of FLAG-CTNNB1 in SeV-infected cells failed to detect interaction with endogenous IRF3 while confirmed the stable interaction with endogenous p65 (). Importantly, the sole increase in endogenous CTNNB1 by stabilization with a GSK3 inhibitor does not promote the interaction with IRF3 in the absence of viral infection (Figure 7 lane BIO). Altogether, these results suggest a direct role of a virus-induced interaction of CTNNB1 with IRF3 in the negative regulation of innate immunity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. CTNNB1 associates with IRF3 and NF-κB subunit p65.

HEK 293T cells were transfected with FLAG-MCS (control), FLAG-eYFP (control), FLAG-IRF3 or FLAG-p65 expressing plasmids for forty-eight hours. Cell extracts were prepared following 16 hours of treatment with IFN-α, GSK3 inhibitor BIO (5 µM) or SeV infection before being subjected to immunoprecipitation directed against FLAG. Cell extracts and immune complexes were analyzed by Western blotting using anti-FLAG and anti-CTNNB1 antibodies.

https://doi.org/10.1371/journal.ppat.1003416.g007

Ethics statement

Primary human hepatocytes were purified by collagenase perfusion from normal biopsies obtained from healthy peritumoral liver tissues of patients undergoing partial hepatectomy for hepatic tumors [66]. Wedge samples (20–50 g) were flushed of blood and maintained overnight in a preservation solution (Belzer UW; Dupont Pharma) at 4°C. At time of perfusion, a selected vein was cannulated with a tip to allow circulation of the buffers in the whole liver sample. All other surrounding veins were subsequently sealed with Vetbound veterinary glue (3M) to allow good dispersion of the buffers and increase the pressure in the whole liver sample. Perfusion starts with calcium free HEPES buffer for 15 min at 20 ml/min following perfusion with 0.55 mg/ml type 2 collagenase (Worthington Biochemical Corporation) and 1.65 mg/ml CaCl2 for 30 min at 15 ml/min. At the end, liver was opened and cells washed in L-15 media. Viability was assessed by trypan blue exclusion. Cells were seed at a 50 000 cells/cm2 with William's E media supplemented with 10% FBS. All procedures followed were in accordance with the ethical commission of the CHUM, Hôpital St-Luc, Montréal, Canada and with the Helsinki Declaration. Human monocyte-derived macrophages (MDM) were prepared from peripheral blood mononuclear cells (PBMC) purified from whole blood of healthy donors by standard Ficoll (GE Sciences) gradient centrifugation. Monocytes were obtained from PBMC by plastics adherence (>97% CD14+). Following extensive washes to remove lymphocytic non-adherent cells, adherent monocytes were allowed to differentiate into MDM for 5 days in the presence of 50 ng/mL granulocyte-monocyte colony stimulating factor (M-CSF, R&D). The study was approved by the IRCM Research Ethic Review Board and in accordance with the Helsinki Declaration. Samples were collected after written informed consent had been obtained.

Expression vectors

WNT2B, WNT9B, CTNNB1 cDNA were obtained from Open Biosystems. Following PCR-amplification, PCR products were cloned into pcDNA3.1-Hygro-MCS using EcoRV/HindIII [67]. All constructs were verified by nucleotide sequencing. pIFNB1-LUC, pISG56-LUC and p2xNF-κB-LUC luciferase reporter constructs were previously described [68]–[70]. To generate stable 293T cells harboring the pIFNB1-LUC promoter, we replaced the MluI-BamHI fragment encompassing the CMV promoter of pcDNA3.1-hygro(+) with the MluI-BamHI fragment of IFNβ-pGL3 [69], which includes two copy of the IFNB1 promoter (−280/+20) and the firefly luciferase coding sequence. pEF1α-LUC was generated after PCR amplification of the EF1α promoter from pEF/JFH1-Rz/Neo [71] and cloning into pcDNA3.1_fLUC_hygro(+) using MluI and BamHI. HCV genotype 2a strain JFH1 p7(+) [71] and J6/JFH-1(p7-Rluc2A) [72] were previously described.

Cells lines and culture

HEK 293T (human embryonic kidney), Huh7 (human hepatoma) and HeLa (human epithelial carcinoma) cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM, Wisent). A549 (human lung adenocarcinoma epithelial) were cultured in Ham's F-12 medium (Invitrogen). Both media were supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 µg/ml streptomycin and 2 mM glutamine (all from Wisent). Normal human bronchial epithelial cells (NHBE) were obtained from Lonza, cultured in BEGM medium (Lonza) and used between passage 2 and 3. Cell populations of HEK 293T stably harbouring the pIFNB1-LUC and pEF1α-LUC used in the screens were produced after selection with 200 µg/ml of hygromycin B (Wisent). Transient transfections were performed with linear 25 kDa polyethylenimine (PEI) (Polysciences, Inc) at a 3 µg PEI to 1 µg DNA ratio for all plasmids and with lipofectamine 2000 (invitrogen) for polyI∶C. siRNA ON-TARGETplus SMARTpool, Human CTNNB1 (1499) (Thermo Scientific) was transfected with lipofectamine RNAi max (Invitrogen) according to the reverse transfection manufacturer protocol for 72 hours. GSK3 inhibitors BIO (GSK3 inhibitor IX, Sigma) and BIO-acetoxime (GSK3 inhibitor X, EMD Millipore) and inhibitor of TCF/LEF interaction PNU74654 (TOCRIS) were added for 16 hours at the indicated concentrations in 0.5% DMSO.

Lentiviral shRNA library production

MISSION TRC shRNA lentiviral library containing _∼75,000 individual clones representing _∼15,000 genes was purchased from Sigma-Aldrich. Three individual clones per gene were chosen for lentiviral production, covering different regions of the gene sequence. For the genome-wide production of lentiviral particles, HEK293 cells (2×10⁴) were plated one day before the transfection. All transfections were performed using a Biomek FX (Beckman Coulter) enclosed in a class II cabinet according to MISSION Lentiviral Packaging Mix protocol (SHP001). As control the MISSION shRNA NT clone (Sigma SHC002) was included in each 96-well plates. The lentivirus-containing supernatants were collected at 24 and 48 hours post-transfection and pooled. NT shRNA controls and 4% of random samples of each plate were used to measure lentiviral titers for quality control purposes. Titers were determined by limiting dilution assay using HeLa cells. The samples (20 µl) were serially diluted in complete DMEM containing 8 µg/ml of hexadimethrine bromide (Polybrene from Sigma). Dilutions (1∶400 or 1∶10,000) were added to preparation of HeLa cells and media were changed at day 3 and 5 with complete DMEM containing 1 µg/ml of puromycin (Wisent). After four days with the selective agent, cells were stained with 1.25% crystal violet and counted in plaque-forming units (PFU) allowing the lentiviral titer determination.

Genome-wide shRNA screen

Cells were seeded in white 96-well plates at a density of 5,000 293T pIFNB1_LUC and 1,250 293T pEF1α-LUC in 100 µl of complete phenol-red free DMEM containing 4 µg/ml polybrene. Infection with lentivirus encoding shRNA (pool of three shRNAs for the primary screen and five individual shRNAs for secondary screens) were carried out immediately after cell seeding at a multiplicity of infection (MOI) of 5 and incubated for four days at 37°C in an atmosphere of 5% CO₂. Cells were infected with 100 HAU/ml of SeV (Cantell Strain, Charles River Labs) for 16 hours before cell lysis and firefly luminescence reading in 100 mM Tris acetate, 20 mM Mg acetate, 2 mM EGTA, 3.6 mM ATP, 1% Brij 58, 0.7% β-mercaptoethanol and 45 µg/ml luciferine pH 7.9 buffer. All infections were performed using a Biomek FX (Beckman Coulter) enclosed in a class II cabinet.

Large-scale lentiviral shRNA production

293T cells were transfected using PEI with 6 µg pLKO.1-puro encoding shRNA targeting WNT2B (TRCN0000033376 and TRCN0000033377), WNT9B (TRCN0000062069 and TRCN0000062070), WLS (TRCN0000133858 and TRCN0000133999), CTNNB1 (TRCN0000003843, TRCN0000003844 and TRCN0000003845) or shRNA non-target (NT, Sigma), 1.5 µg pMDLg/pRRE, 1.5 µg pRSV-REV and 3 µg pVSVg as previously described [73]. After 48 hours, medium containing lentiviral vector was harvested and cellular debris were removed by filtration (0.45 µm). Titers were determined by limiting dilution assay using HeLa cells, as described above.

High-throughput qRT-PCR assay

For the shRNA validation screen, total RNA was extracted in 96-well plate format using “RNAqueous – 96 automated” kit (Ambion) with a Multimek-96 (Beckman Coulter). RNA concentrations were measured using Ribogreen reagent (Invitrogen) on the Envision plate reader (Perkin Elmer). After normalisation of RNA at 200 ng/well, reverse transcription reactions were performed with a High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Quantitative PCR assays were designed following the Universal Probes Library (UPL) from Roche. The best primer/probe combinations were calculated online with the ProbeFinder assay design software. qPCR reactions were performed in 384-well plates using 2 µl of cDNA samples, 5 µl of the Light Cycler 480 Probes Master (Roche), 1 µM of gene specific primers (IDT) and 2 µM of the matching UPL probe in a total of 10 µl (supplementary Table 3 for the complete list of primers and probes used). Amplification conditions on the Light Cycler 480 (Roche) for the UPL were initial denaturation of 5 minutes at 95°C followed by 45 cycles of 10 seconds at 95°C, 30 seconds at 60°C and 1 second at 72°C. All qPCR were run in duplicates and the relative quantification of mRNAs from the IFNB1 and shRNA targeted genes was determined with the average of Ct (cycle threshold) using the ΔΔCt method. Briefly, the difference in the cycle threshold (ΔCt) value was determined by subtracting the Ct value for the transcripts of interest, from the Ct value of the endogenous control (ACTB: β-actin) and compared with a shRNA NT calibrator: ΔΔCT = ΔCt_Sample−ΔCt_Calibrator. Relative expression (RQ) was calculated using the Sequence Detection System (SDS) 2.2.2 software (Applied Biosystems) and the formula RQ = 2^−ΔΔCT. A second set of primers and probe were designed for genes for which no amplification was obtained in the first round of qPCR (supplementary Table 3). A not determined (nd) note is indicated when no results were available after two rounds of qPCR (supplementary Table 1).

Large-scale RNA extraction and qRT-PCR

Total cellular RNA was extracted with the RNeasy Mini kit (Qiagen). Reverse transcription was performed on 500 ng total cellular RNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems, see (supplementary Table 4 for the list of primers and probes). PCR reactions were performed using 1.5 µl of cDNA samples (15 ng), 5 µl of the Fast TaqMan PCR Master Mix (Applied Biosystems), 10 pmol of each primer (IDT) and 5 pmol of the UPL probe (Roche) in a total volume of 10 µl. The ABI PRISM 7900HT Sequence Detection System (Applied Biosystems) was used to detect the amplification level and was programmed to an initial step of 3 minutes at 95°C, followed by 40 cycles of 5 seconds at 95°C, 30 seconds at 60°C and 1 second at 72°C. All reactions were run in triplicate on biological duplicate and the average values were used for quantification. ACTIN and HPRT1 (hypoxanthine phosphoribosyltransferase 1) were used as endogenous controls. The relative quantification of target genes was determined by using the ΔΔCt method, as described above.

Immunofluorescence analysis

HEK 293T cells were seeded in cover slip-containing 24-well plates. Twenty-four hours later, cells were infected with SeV for 16 hours before being washed twice with PBS, fixed with 4% paraformaldehyde-containing PBS during 20 minutes at room temperature and then permeabilized in 0.2% Triton X-100/PBS during 15 minutes. Blocking was made in PBS with 10% normal goat serum, 5% bovine serum albumin (BSA) and 0.02% sodium azide during 45 minutes at room temperature. Following three rapid washes, cells were labelled with mouse anti-CTNNB1 active (Millipore) and rabbit anti-IRF3 (IBL) primary antibodies diluted in 5% BSA/0.02% sodium azide/PBS during 2 hours. Slides were washed three times in PBS and then labelled with anti-mouse AlexaFluor 488 and anti-rabbit AlexaFluor 546 secondary antibodies (Invitrogen) diluted in 5% BSA/0.02% sodium azide/PBS during 1 hour. Cells were extensively washed and incubated with Hoechst dye (Invitrogen) at a final concentration of 1 µg/mL in PBS. Following three rapid washes, slides were mounted using DABCO (Sigma-Aldrich) as an anti-fading agent. Labelled cells were then examined by laser scanning microscopy using a LSM510 confocal microscope (Zeiss).

Microarray analysis

The microarray studies were performed with HEK 293T cells infected for 16 hours with SeV (100 HAU/ml) or treated for 16 hours with a mixture of IFN-α from human leukocytes (400 U/ml; Sigma) and with untreated cells as control. A total of 10 µg of RNA was reverse transcribed using oligo(dT) 16–18 primers and SuperScript II Reverse Transcriptase (Invitrogen) according to the manufacturer's instructions. Following purification using QIAquick PCR Purification kit (Qiagen), up to 1 µg of purified cDNA was mixed with 5′-Cy3 labeled random nonamers (Trilink Biotechnology) and heated at 95°C for 10 minutes and transferred on ice for 10 minutes. Samples were mixed with 1 mM dNTP and 2 µl of 3′-5′ exo-Klenow fragment (New England Biolabs) and incubated at 37°C for 2 hours. The labeling reaction was stopped using 50 µM EDTA and the DNA precipitated using 0.5 M NaCl and 1 volume isopropanol, washed with 80% ethanol and resuspended in water. Hybridizations were carried out using the Human GE 4×44K v2 Microarrays (Agilent Technologies) containing probes targeting 27,958 Entrez Gene RNAs. Arrays were scanned at 5 µm resolution using a GenePix4000B scanner (Molecular Devices). Data from scanned images were extracted using GenePix 6.1 (Axon) and processed and normalized using ArrayPipe (v2.0). Processed data was used as input for linear modeling using Bioconductor's limma package, which estimates the fold-change between predefined groups by fitting a linear model and using an empirical Bayes method to moderate standard errors of the estimated log-fold changes in expression values from each probe set. P values from the resulting comparison were adjusted for multiple testing according to the method of Benjamini and Hochberg.

Western immunoblot analysis

Cells were washed twice with ice-cold phosphate-buffered saline (PBS; Wisent), harvested and lysed in 10 mM Tris-HCl, 100 mM NaCl, 0.5% Triton X-100, pH 7.6 with EDTA-free Protease Inhibitor Cocktail (Roche). Cell lysates were clarified by centrifugation at 13,000 g for 20 min at 4°C and subjected to sodium dodecyl sulfate-polyacrylamide gel (SDS-PAGE). Western blot analysis was performed using mouse anti-WNT2B (Santa Cruz), anti-ACTIN (Chemicon International), anti-CTNNB1 active (Millipore), anti-FLAG (Sigma), anti-IRF3 (Santa Cruz), anti-NS3 (HCV, Abcam), anti-p65 (Santa Cruz) and rabbit anti-CTNNB1 total (Epitomics), anti-CTNNB1-P-Ser552 (Cell Signaling), anti-IFIT2 (Novus Biologicals), anti-IFIT1 (Novus Biologicals), anti-SeV (HN) [first described by Ilkka Julkunen 74], anti-WNT9B (Aviva systems biology) and anti-NFKBIA-P-Ser32 (Cell Signaling). HRP-conjugated secondary antibodies were from Bio-Rad. The chemiluminescence reaction was performed using the Western Lighting Chemiluminescence Reagent Plus (PerkinElmer).

Luciferase assay

293T cell population stably expressing pIFNB1-LUC were cultured in white opaque 96-well plates with complete phenol red-free DMEM and transfected with 1 ng of Rluc expressing plasmid used to normalized the firefly luciferase readout. For the other fLUC reporter constructs, 293T cells were transfected with 20 ng of M50 Super 8×TOP-flash [40] (Addgene plasmid 12456), pIFNB1-LUC, pISG56-LUC or p2xNF-κB-LUC and 1 ng of Rluc. Cells were lysed directly and luciferase activities were determined by a dual-luciferase reporter assay system (Promega).

SeV, HCV and IFN-α and purified WNT stimulations

Cells were infected with 100 HAU/ml of Sendai virus (Cantell Strain, Charles River Labs) or stimulated with mixture of IFN-α from human leukocytes (400 U/ml; Sigma) for 4 hours in A549 cells and MDM, 5 hours in human primary hepatocytes and 16 hours in 293T, Huh7 and HeLa cells before luminescence reading, immunoblot analysis or qRT-PCR. Primary human hepatocytes were infected for 48 hours with HCV genotype 2a strain JFH1 p7(+) and Huh7 cells were infected for 96 hours with J6/JFH-1(p7-Rluc2A). HCV particles were produced as previously described [43]. Purified WNT9B and WNT3A proteins (R&D systems) were reconstituted at 100 µg/ml in PBS containing 0.1% BSA and added to SeV-infected HEK 293T cell supernatant for 16 hours.

IFN-β ELISA

Secreted IFN-β protein quantification was measure with VeriKine Human IFN-α Serum Sample ELISA Kit (PBL Interferon Source) using 50 µl of HEK 293T cells supernatant previously transduced with lentivirus-expressing shRNAs for three days before infection with SeV for 16 hours.

Statistical analyses

Data are presented as the mean ± standard error of the mean (SEM). Statistical significance for comparison of two means was assessed by an unpaired Student's t test. Analyses were performed using the Prism 5 software (GraphPad). Statistical relevance was evaluated using the following p values: p<0.05 (*), p<0.01 (**) or p<0.001 (***).