Influenza A Virus PA Antagonizes Interferon-β by Interacting with Interferon Regulatory Factor 3

The influenza A virus (IAV) can be recognized by retinoic acid-inducible gene I (RIG-I) to activate the type I interferon response and induce antiviral effects. The virus has evolved several strategies to evade the innate immune response, including non-structural protein 1 (NS1) and its polymerase subunits. The mechanism by which NS1 inhibits interferon-β (IFN-β) is well understood, whereas the mechanism by which polymerase acid protein (PA) inhibits IFN-β remains to be elucidated. In this study, we observed that the IAV PA protein could inhibit the production of IFN-β and interferon-stimulated genes induced by Sendai virus through interferon regulatory factor 3 (IRF3), but not through nuclear factor-kappaB (NF-kappaB). In addition, PA inhibited IFN-β induction by RIG-I, melanoma differentiation-associated gene 5, mitochondria antiviral signaling protein, TANK-binding kinase 1, inhibitor of nuclear factor kappa-B kinase-ε (IKKε), and IRF3 overexpression. Furthermore, PA interacted with IRF3 to block its activation. The N-terminal endonuclease activity of PA was responsible for its interaction with IRF3 and inhibition of the IFN-β signaling pathway. In summary, our data reveal the mechanism by which IAV PA inhibits the IFN-β signaling pathway, providing a new mechanism by which the virus antagonizes the antiviral signaling pathway.

The PA subunit has been reported to enhance influenza virus polymerase activity, pathogenicity, transmission, contribute to its capacity to infect a range of hosts (24)(25)(26), and has also been suggested to be involved in the inhibition of IFN-β induction (21); however, this mechanism still remains to be elucidated. In this study, we demonstrated that IAV PA could block IRF3 activation and suppress IFN-β production. The aspartic acid at position 108 in PA is required for this activity.

luciferase assay
To determine the impact of PA on the promoter activities of IFN-β, IRF3, or NF-kappaB, the luciferase reporter plasmids were co-transfected with the indicated plasmids in 293T cells. A cotransfected pRL-TK vector that provided constitutive expression of Renilla luciferase was used as a control (Promega, Madison, WI, USA). After 24 h, cells were infected with 10 hemagglutinating activity units/well of Sendai virus (SEV; Centre of Virus Resource and Information, Wuhan Institute of Virology, Chinese Academy of Sciences) or they were transfected with 200 ng Poly(I:C) (Sigma, St. Louis, MO, USA) for 8 h. Cells were then lysed for measuring the luciferase activity using the dual-luciferase assay system (Promega) according to the manufacturer's instructions.

elisa of iFn-β
To measure IFN-β secretion, A549 cells were transfected with an empty vector or plasmids with Flag-tagged pdm/09 PA. After 24 h, the cells were left uninfected or were infected with SEV for 8 h. The supernatants were harvested for an ELISA using a human IFN-β ELISA kit (Elabscience, Wuhan, China) according to the manufacturer's instructions.

statistical analyses
Statistical analyses for all experiments were performed using Student's t-test or one-way ANOVA (for more than two groups) with GraphPad Prism software (San Diego, CA, USA). The data were representative of at least three separate experiments. The error bars represent the SDs, and p values <0.05 are significant (*p < 0.05, **p < 0.01, ***p < 0.001). resUlTs

Pa antagonizes seV-induced iFn-β Production
To investigate the regulation of influenza virus PA in IFN-β induction, we constructed the expression plasmid of influenza A (H1N1) pdm09 virus (pdm/09) PA and a series of influenza PA expression plasmids from different influenza virus subtypes, including F9PA, H5N1PA, and H7N9PA. Luciferase assays revealed that all subtypes of PA could strongly inhibit SEV-and Poly(I:C)-induced IFN-β promoter activity (Figure 1A), supporting the hypothesis that IAV PA plays a role in inhibition of the IFN-β signaling pathway (21). This hypothesis was further supported by the fact that IFN-β induction by SEV could be further inhibited by pdm/09 PA overexpression, as evidenced by reduced IFN-β transcription and protein levels (Figures 1B,C).
Pdm/09 Pa inhibits the rig-i-like receptor (rlr)-Mediated iFn-β signaling Pathway through irF3, rather than nF-kappaB During influenza virus infection, vRNAs will be recognized by PRRs and trigger the activation of transcription factors, leading to IFN-β induction (10). To describe the molecular mechanism of IFN-β inhibition by PA, a luciferase assay of the IFN-β promoter was performed with several activators, including RIG-I, MDA-5, MAVS, TBK-1, IKKε, and IRF3. PA induced by upstream factors could inhibit IFN-β promoter activation (Figure 2A). The assay indicated that PA may inhibit the IFN-β pathway downstream of these activators.
Because activation of the transcription factors IRF3 and NF-kappaB is essential for IFN-β production, we further investigated whether pdm/09 PA disturbs the IRF3 or NF-kappaB pathway using luciferase assays of the IRF3 (Figure 2B) or NF-kappaB promoter ( Figure 2C). Interestingly, IRF3 promoter activity induced by the above activators was strongly inhibited. However, NF-kappaB promoter activity was not inhibited by pdm/09 PA protein. Therefore, we conclude that PA can interfere in RLR-mediated IFN-β production through the IRF3-dependent signaling pathway.
Pdm/09 Pa interacts with irF3 and suppresses irF3 Phosphorylation, Dimerization, and subsequent nuclear Translocation In the context of viral infection, activation of signaling pathways can induce IRF3 phosphorylation and dimerization, which results in IRF3 accumulation in the nucleus (11,27). We observed that pdm/09 PA overexpression could decrease IRF3 phosphorylation in a dose-dependent manner but could not influence NF-kappaB activation (Figures 3A,B). In addition, pdm/09 PA could also suppress SEV-induced dimerization ( Figure 3C) and nuclear translocation of IRF3 ( Figure 3D). These data indicated that the IAV PA protein inhibits IRF3 activation in response to SEV stimulation. Interestingly, pdm/09 PA and IRF3 showed co-localization in the cytoplasm by confocal microscopy (Figure 3D). Then the cells were lysed for Western blotting using nuclear factor-kappaB (NK-kappaB) and phospho-NF-kappaB antibodies. (c) 293T cells in 6-well plates were transfected with Flag-tagged pdm/09 PA or empty vector. After 24 h, cells were infected with Sendai virus (SEV) or were left uninfected for 8 h and then they were harvested for NATIVE-PAGE using anti-IRF3 antibody or for Western blot using anti-Flag and anti-GAPDH antibodies. (D) 293T cells seeded onto coverslips and placed into 12-well plates were transfected with Flag-tagged pdm/09 PA or empty vector. After 24 h, cells were infected with SEV or were left uninfected for 8 h and were fixed for immunofluorescence assay (IFA), with endogenous IRF3 (red), PA (green), and nuclei (blue) labeled with anti-IRF3 antibody, anti-Flag antibody and DAPI, respectively, using confocal microscopy. (e) 293T cells in 6-well plates were transfected with Flag-tagged PA and HA-tagged IRF3. After 24 h, cells were lysed and precipitated with anti-Flag antibody. The cell lysates and immunoprecipitates (IPs) were analyzed by Western blotting using anti-Flag and anti-HA antibodies. (F) 293T cells were infected with H1N1 pdm/09 ([MOI] = 0.1) for 12 h, and cells were then lysed and precipitated with anti-IRF3 antibody or with normal IgG as a negative control. The cell lysates and IPs were analyzed by Western blotting using anti-IRF3 and anti-PA antibodies.
Since pdm/09 PA antagonizes IRF3 activation, we hypothesized that both proteins interact. A CO-IP experiment showed that pdm/09 PA does co-precipitate with IRF3 in both transfected cells ( Figure 3E) and virus-infected cells (Figure 3F). These data indicate that pdm/09 PA interacts with IRF3 and further blocks IRF3 activation.

The n-Terminal Functional Domain of pdm/09 Pa is responsible for iFn-β suppression
Influenza A virus (IAV) PA consists of two functional domains: a 30-kDa N-terminal fragment (PAN, residues 1-257) and a 53-kDa C-terminal fragment (PAC, residues 258-716) (28). PAC is known as the PB1 binding domain, while PAN possesses endonuclease activity (29). Similar to what was observed for full-length PA, pdm/09 PAN could strongly inhibit SEV-induced IFN-β promoter activation in a dose-dependent manner (Figures 4A,C). Furthermore, the expression levels of IFN-β and ISG-56 were reduced by pdm/09 PA and pdm/09 PAN (Figure 4B), whereas pdm/09 PAC showed no significant effect on IFN-β induction ( Figure 4C). These results indicate that the N-terminus of pdm/09 PA may be responsible for IFN-β suppression.
Because PA interacts with IRF3 and blocks its activation (Figure 3), we further investigated the interaction of PAN with IRF3. Similar to what was observed with pdm/09 PA, PAN was shown to interact with IRF3 in a CO-IP assay ( Figure 4D) and to block SEV-induced translocation of IRF3 into the nucleus ( Figure 4E). Therefore, the N-terminus of pdm/09 PA is responsible for the interaction with IRF3 and suppression of the IFN-β signaling pathway.
The Binding activity of pdm/09 Pa to irF3 is Dependent on asp108 Because the N-terminus of pdm/09 PA contains an endonuclease domain, it is important to investigate the role that the endonuclease activity of pdm/09 PA plays in inhibition of the host IFN-β induction. A previous study showed that a PA-D108A mutant could abolish the endonuclease activity of PA in vitro (30). The D108A mutation impairs the capacity of pdm/09 PA to inhibit IFN-β induction by SEV at the transcriptional ( Figure 5A) and translational levels ( Figure 5B). In addition, ISG-15, ISG-56, and CXCL-10 expression levels in cells transfected with the pdm/09 PA-D108A mutant were only half that of those in cells transfected with pdm/09 PA ( Figure 5B). Furthermore, the mutation abolished the interaction between IRF3 and pdm/09 PA ( Figure 5C). Together, these results indicate that Asp108 of pdm/09 PA contributes to the interaction of pdm/09 PA with IRF3 and inhibition of the IFN-β signaling pathway.

DiscUssiOn
The influenza virus RNA polymerase is a heterotrimeric complex consisting of the PB2, PB1, and PA subunits, all of which are required for vRNA transcription and replication (15)(16)(17). Infection is usually associated with inhibition of the host antiviral response by viral polymerases (31). Upon infection, IAV triggers the activation of the host innate immunity (32), which leads to IFN-β secretion, which mediates an antiviral effect. Hence, IAV employs a variety of strategies to circumvent innate immunity and the host's IFN response. The influenza viral NS1 protein and polymerase proteins could reduce IFN-β synthesis through the inhibition of IFN signaling pathways, thereby circumventing the antiviral effect of host immunity, which is essential for IAV infection (18,19,21,23,(33)(34)(35)(36). However, the mechanism by which PA inhibits IFN-β remains unknown. In our study, we found that all the PA subunits from the IAV strains H1N1, H5N1, and H7N9 could antagonize IFN-β production through IRF3 rather than NF-kappaB.
It is known that upon viral infection, IRF3 is phosphorylated, exposing the IRF association domain at the C-terminus. Subsequently, phosphorylated IRF3 will be dimerized and translocate to the nucleus (37), leading to IFN-β transcription (11,27,(38)(39)(40)(41). In this study, we observed that IRF3 phosphorylation and dimerization were inhibited by pdm/09 PA, resulting in IRF3 accumulation in the cytoplasm. Interestingly, IRF3 and pdm/09 PA were co-localized, and a CO-IP assay further indicated an interaction between pdm/09 PA and IRF3. These results explain the findings that PA can antagonize IFN-β production through IRF3.
The influenza viral PA protein can be digested by trypsin through two domains: the N-terminal domain (from amino acid residues 1-257) and the C-terminal domain (from amino acid residues 277-716). Amino acid residues 257-276 provide a flexible structure to ensure the connection of PA to PB1 (28,30). The N-terminus of PA possesses pleiotropic functions (42), including the host RNA cap-snatching via an endonuclease activity (15,43,44) and proteolysis induction (45,46). Both functions may cause non-specific reductions in host gene expression levels. The capacity of PA to inhibit the IRF-3 signaling pathway, but not the NF-kappaB signaling pathway, suggests that PA inhibits IFN-β via some specific molecular mechanism. To further investigate the precise function of the domains of PA against the host IFN response, we generated plasmids pdm/09 PAN (1-257) and pdm/09 PAC (258-716) and found that PAN activity was similar to that of full-length PA and that pdm/09 PAN could interact with IRF3 and effectively alter the distribution of endogenous IRF3 upon SEV infection.
Furthermore, mutation of PA (PA-D108A mutant) causes loss of endonuclease activity (30), abolishes the interaction of PA with IRF3, and decreases the inhibition of the IFN-β signaling pathway. These data further support our opinion that the N-terminal endonuclease activity may be required for PA to interact with IRF3 and then block its activation.
Regrettably, we failed to generate a recombinant virus carrying the pdm/09 PA-D108A mutation to evaluate the effect of IFN-β inhibition by PA on virulence. The inability to generate the recombinant virus may be due to the essential function of PA endonuclease activity for viral propagation, such as viral genes transcription and the vRNA promoter binding activity (30). Considering that the sequence of the endonuclease domain of PA is highly conserved, we hypothesized that IAV PA could bind to IRF3 through its N-terminal domain and suppress IFN-β induction, although the exact pathogenic role of PA during infection was difficult to be elucidated.
It has been reported that a single amino acid substitution of PB2 (N9D) blocks mitochondrial localization of the PB2 protein and suppresses IFN-β induction (18), while we demonstrated that PA blocks IRF3 activation and IFN-β induction dependent on Asp108. In addition, PA could mediate IFN-β induction through multiple mechanisms (as shown in Figure 5, the pdm/09 PA-D108A mutant is still able to inhibit host IFN-β induction). It is likely that the PA protein impacts the synthesis of host proteins through its helix α4 domain and amino residues 51-57, the flexible loop domain (43,47). The first 85 amino acid residues of PA and amino acid residues from 186 to 247, which are necessary to induce proteolysis of PA, may also contribute to suppression of host IFN-β induction (45). There may also be other mechanisms by which PA inhibits IFN-β that remain unknown.
In summary, we observed that IAV PA can interact with IRF3 and inhibit its activation, thereby blocking the activation of the IFN-β signaling pathway. The interaction requires the N-terminal endonuclease activity of PA. This study reveals a new mechanism by which IAV inhibits the IFN-β signaling pathway.
aUThOr cOnTriBUTiOns CY, ZZ, AZ, and MJ designed the study; CY and ZZ performed experiments; CY, ZZ, SW, XS, and DZ analyzed the data; CY drafted the manuscript; XMS, AZ, and MJ contributed to conduct of the laboratory work and critical review of the manuscript. All authors contributed to read and approved the final version.