Influenza virus-induced AP-1-dependent gene expression requires activation of the JNK signaling pathway.

Influenza A virus infection of cells results in the induction of a variety of antiviral cytokines, including those that are regulated by transcription factors of the activating protein-1 (AP-1) family. Here we show that influenza virus infection induces AP-1-dependent gene expression in productively infected cells but not in cells that do not support viral replication. Among the AP-1 factors identified to bind to their cognate DNA element during viral infections of Madin-Darby canine kidney and U937 cells are those that are regulated via phosphorylation by JNKs. Accordingly, we observed that induction of AP-1-dependent gene expression correlates with a strong activation of JNK in permissive cells, which appears to be caused by viral RNA accumulation during replication. Blockade of JNK signaling at several levels of the cascade by transient expression of dominant negative kinase mutants and inhibitory proteins resulted in inhibition of virus-induced JNK activation, reduced AP-1 activity, and impaired transactivation of the IFN-beta promoter. Virus yields from transfected and infected cells in which JNK signaling was inhibited were higher compared with the levels from control cells. Therefore, we conclude that virus-induced activation of JNK and AP-1 is part of the innate antiviral response of the cell.


Here we show that influenza virus infection induces AP-1-dependent gene expression in productively infected cells but not in cells that do not support viral replication. Among the AP-1 factors identified to bind to their cognate DNA element during viral infections of Madin-Darby canine kidney and U937 cells are those that are regulated via phosphorylation by JNKs. Accordingly, we observed that induction of AP-1-dependent gene expression correlates with a strong activation of JNK in permissive cells, which appears to be caused by viral RNA accumulation during replication. Blockade of JNK signaling at several levels of the cascade by transient expression of dominant negative kinase mutants and inhibitory proteins resulted in inhibition of virus-induced JNK activation, reduced
AP-1 activity, and impaired transactivation of the IFN-␤ promoter. Virus yields from transfected and infected cells in which JNK signaling was inhibited were higher compared with the levels from control cells. Therefore, we conclude that virus-induced activation of JNK and AP-1 is part of the innate antiviral response of the cell.
Influenza A virus is the prototype of the Orthomyxoviridae that are characterized by a segmented negative strand RNA genome. The function of the 10 different virus-encoded proteins has been well elucidated (reviewed in Ref. 1). Transcription and amplification of the viral genome is catalyzed by the RNA-dependent RNA polymerase (RDRP) 1 complex, consisting of three proteins, PB2, PB1, and PA, which are associated with the ribonucleoprotein complex. Genomic influenza virus RNAs carry conserved nucleotide sequences at their 5Ј-and 3Ј-ends, which are in part complementary and serve as promoters for synthesis of both complementary RNA intermediates and novel viral RNA (vRNA) (1).
Because of the relatively small coding capacity of the viral genome, it is not surprising that influenza A viruses extensively manipulate and exploit host cell functions to support viral replication. On the other hand, the host cell has developed defense mechanisms against viral infections, which include an effective expression of antiviral cytokines that may require activation of cellular signaling pathways. Although only little is known about the intracellular signaling cascades that are activated by influenza virus, there are numerous reports on downstream target genes, such as interleukins (IL), tumor necrosis factor-␣, and interferons (IFN), which are immediately transcribed in response to infection (for a review, see Ref. 2).
The best studied cell systems with respect to influenza virusinduced cellular gene expression are macrophages and monocytes. These cells are susceptible to influenza virus infection and express a set of cytokines such as tumor necrosis factor-␣, IL-1, IL-6, and IFN-␣/␤ upon infection (3)(4)(5). Other virus-induced genes include monocyte chemoattractant protein-1 in monocytes (6) or IL-8 in human lung epithelial cells (7,8) or rat kidney cells (9). Several influenza virus-induced genes are dependent on activation of the AP-1 family of transcription factors, including the IFN-␤ gene that encodes for one of the most potent antiviral cytokines. This transcription factor family consists of homodimers and heterodimers of Jun (c-Jun, JunB, JunD) Fos (c-Fos, FosB, Fra-1, Fra-2), or the activating transcription factor (ATF-2, ATF-3) proteins, which bind to a common DNA site, the AP-1 site (reviewed in Ref. 10). The regulation of AP-1 activity occurs at two major levels. First, the abundance of the proteins is commonly regulated by controlling the transcription of AP-1 factor genes (11). The regulation of transcription is best understood for c-jun and c-fos genes that are rapidly up-regulated upon cell stimulation (11). At a second level, the activity of different AP-1 factors is posttranscriptionally modulated by phosphorylation events. This includes AP-1 factors c-Jun and the constitutively expressed JunD and ATF-2, which are phosphorylated by kinases of the JNK family, resulting in an in-creased transactivation potential of the factors (reviewed in Refs. 10 and 12).
JNKs, also known as stress-activated protein kinases (SAPKs) belong to the family of mitogen-activated protein kinases (MAPKs) that are regulated by dual phosphorylation of threonine and tyrosine residues in a TXY motif (reviewed in Refs. 13 and 14). This activation is mediated by dual specificity kinases termed MAPK kinases (MKKs). For the JNK subgroup of MAPKs, at least two different MKKs have been identified as specific activators, MKK4/SEK (for SAPK/ERK kinase) (15) and MKK7 (16,17). It was recently shown that MKK4 and MKK7 act synergistically in JNK activation, whereby MKK4 had a preference for the tyrosine residue and MKK7 for the threonine residue within the TXY motif (18). This suggests that the full activation of JNK may require phosphorylation by two different MKKs. Since MKK4 or MKK7 are preferentially activated by different upstream activators (19 -21), there is the potential for integrating the effects of different extracellular signals. Specificity within the different signaling cascades is accomplished by scaffold proteins, such as the JNK-interacting protein-1 (JIP-1), which has been shown to form a complex with JNK, MKK7, and the MKK7 activator mixed lineage kinase (22).
In this report, we show that influenza virus infection of cells induces JNK activation and AP-1-dependent gene expression in a MKK4/SEK-and MKK7-dependent manner. This activation selectively occurs in cells permissive for virus replication and appears to be caused by accumulation of virus-specific RNA. Our findings link influenza virus replication to the activation of an important signaling pathway in the cell, which seems to be required for the antiviral response to infection.

EXPERIMENTAL PROCEDURES
Viruses, Cells, and Viral Infections-Avian influenza virus A/Bratislava/79 (H7N7) (FPV), human influenza virus A/Asia/57 (H2N2) (Asia), and the reassortant virus WSN-HK (H1N2), which contains seven genes of human influenza A/WSN/33 (H1N1) and the neuraminidase gene from human influenza virus A/HK/8/68 (H3N2), were used for infection of different cell lines. Madin-Darby canine kidney (MDCK) cells were grown in minimal essential medium; the human cervical carcinoma cell line HeLa and human embryonic kidney HEK293 cells were grown in Dulbecco's modified Eagle's medium; U937 promonocytic cells and A3.01 human T lymphoma cells were grown in RPMI 1640 medium. All growth media contained 10% heat-inactivated fetal calf serum and antibiotics. Cells were washed with phosphatebuffered saline and infected with various viruses at the indicated multiplicity of infection (MOI) in phosphate-buffered saline/BA (phosphatebuffered saline containing 0.2% bovine serum albumin, 1 mM MgCl 2 , 0.9 mM CaCl 2 , 100 units/ml penicillin, 0.1 mg/ml streptomycin) for 30 min at 37°C. The inoculum was aspirated, and cells were incubated with RPMI 1640/BA (U937 and A3.01 cells), minimal essential medium/BA (MDCK), or Dulbecco's modified Eagle's medium/BA (HEK293 and HeLa cells) (medium containing 0.2% bovine serum albumin and antibiotics). The amount of infectious virus grown in MDCK cells was determined by plaque assays.
DNA Constructs and Cloning-Expression vectors for glutathione S-transferase (GST) fusion proteins pEBG-SAPK␤, pEBG-SEK1 and pGEXKG-c-Jun-(1-135) were a kind gift of J. Kyriakis and L. Zon (Charlestown, MA). Generation of pEBG-SEK1(KϾR) and pEBG-SAPK␤(KKϾRR) was described previously (23). The cDNAs of a hemagglutinin epitope (HA)-tagged version of SAPK␤, the transactivation domain deletion mutant of c-Jun, TAM67, and the JNK binding domain of JIP-1 (22) were cloned into the multiple cloning site of pRSPA (24). Expression vectors pCS3 MKK7wt, MKK7(S3E), and MKK7(KϾM) were a kind gift of P. M. Holland (Fred Hutchinson Cancer Research Center, Seattle, WA) and were described earlier (25). The 5ϫ AP-1 luciferase construct contains five copies of a conserved 9-base pair AP-1 motif, also known as the TRE (12-O-tetradecanoylphorbol-13-acetateresponsive element) (26), in front of a luciferase reporter gene. The empty pHMG expression vector and pHMG expression plasmids for influenza virus NP, PB1, PB2, and PA were kindly provided by J. Pavlovic (Institute of Medical Virology, Zü rich, Switzerland), and constructs that allow RNA polymerase I-driven synthesis of influenza vRNA-like templates carrying the antisense reading frame for a GFP (pPol I GFP) or a luciferase (pPol I Luc) reporter gene were kindly provided by P. Palese (Department of Microbiology, Mount Sinai School of Medicine, New York). The IFN-␤ promoter/enhanceosome luciferase construct was a generous gift of J. Hiscott (Lady Davis Institute for Medical Research, McGill University, Montreal, Canada) (27). The double-stranded RNA (dsRNA) analog poly(IC) and the p38-specific inhibitor SB202190 were purchased from Sigma or Calbiochem, respectively.
Transient Transfections and Reporter Gene Assays-Transfection of HEK293 cells was performed by a calcium phosphate coprecipitation method according to a modified Stratagene transfection protocol. A3.01 and U937 cells were transfected with DMRIE-C (Life Technologies), and MDCK and HeLa cells were transfected with FUGENE (Roche Molecular Biochemicals) or LipofectAMINE 2000 (Life Technologies) according to the manufacturer's instructions. Unless otherwise indicated, cells were harvested 24 h after transfection into 100 l of lysis buffer (50 mM Na-MES, pH 7.8, 50 mM Tris-HCl, pH 7.8, 10 mM dithiothreitol, 2% Triton X-100). The crude cell lysates were cleared by centrifugation, and 50 l of precleared cell extracts were added to 50 l of luciferase assay buffer (125 mM Na-MES, pH 7.8, 125 mM Tris-HCl, pH 7.8, 25 mM magnesium acetate, 2 mg/ml ATP). Immediately after injection of 50 l of 1 mM D-luciferin (AppliChem) into each sample, the luminescence was measured for 5 s in a luminometer (Berthold). As a transfection control, ␤-galactosidase activity of cotransfected Rous sarcoma virus-␤-galactosidase vector was analyzed, and luciferase activities were normalized on the basis of protein content. Mean and S.D. values of at least three independent experiments performed in duplicate or triplicate are shown in Figs. 1, 6, and 8, A and B.
Immune Complex Kinase Assays and Western Blots-Cells were lysed in Triton lysis buffer (20 mM Tris/HCl, pH 7.4, 137 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EDTA, 50 mM sodium glycerophosphate, 20 mM sodium pyrophosphate, 5 g/ml aprotinin, 5 g/ml leupeptin, 1 mM sodium vanadate, 5 mM benzamidine) on ice for 10 -20 min. Cell lysates were then centrifuged, and supernatants were incubated with a specific antiserum and protein A-agarose (Roche Molecular Biochemicals) for 2 h at 4°C to precipitate the endogenous kinase. Endogenous JNK1 was precipitated with an anti-JNK1 antiserum (Santa Cruz Biotechnology). Transfected HA-or GST-tagged SAPK␤, GST-tagged SEK/MKK4, or Myc-tagged MKK7 were detected with either a monoclonal antibody against the HA epitope (12CA5), an antiserum against GST (23), or the monoclonal anti-Myc epitope antibody (9E10). Immune complexes were used for in vitro kinase assays as previously described (23). Briefly, immunoprecipitated kinases were washed twice, both in Triton lysis buffer supplemented with 500 mM NaCl and in kinase buffer (10 mM MgCl 2 , 25 mM ␤Ϫglycerophosphate, 25 mM HEPES pH 7.5, 5 mM benzamidine, 0.5 mM dithiothreitol, and 1 mM sodium vanadate). The assays were performed in kinase buffer supplemented with 5 Ci of [␥-32 P]ATP, 0.1 mM ATP, and recombinant GST-c-Jun-(1-135) or GST-SAPK␣ as substrates for JNK/SAPK or MKK4 and -7, respectively, at 30°C for 15 min. Proteins were separated by SDS-polyacrylamide gel electrophoresis and blotted onto polyvinylidene difluoride membranes. Phosphorylated substrates were detected by a BAS 2000 Bio Imaging Analyzer (Fuji) and by autoradiography. Equal loading of immunoprecipitated kinases was analyzed by Western blotting using the appropriate anti-kinase antiserum or anti-tag antibody in a standard enhanced chemiluminescence reaction (Amersham Pharmacia Biotech). To detect activated ATF-2, an anti-phospho-ATF-2 antiserum was used (New England Biolabs). Equal ATF-2 loading in these assays was controlled with a pan-ATF-2 antiserum (Santa Cruz Biotechnology).

Influenza Virus Induces AP-1-dependent Gene Expression in
Productively Infected Cells-To selectively analyze the regulation of virus-induced AP-1-dependent gene expression without the involvement of other regulatory transcription factors, we used a promoter reporter gene construct containing five copies of an AP-1-binding DNA element in front of a luciferase reporter gene (26). Productive infection of MDCK cells with influenza A virus resulted in a virus titer-dependent transactivation of the AP-1 dependent promoter as early as 4 h postinfection (p.i.) (Fig. 1A). The early onset of luciferase expression suggests that the effects are directly induced by the virus and not by an autocrine event. Essentially the same results were obtained in the promonocytic U937 cell line, which is also permissive for influenza virus replication (Fig. 1B). In cells that only allow an abortive virus replication, such as T lymphocytes (A3.01) (29) or HeLa cells (30), the AP-1-dependent promoter activity remained completely silent (Fig. 1B), indicating that productive replication is required to induce AP-1-dependent transcription.
AP-1 Factors Are Constitutively Bound to Their Cognate DNA Element in MDCK and U937 Cells-A first regulatory step in AP-1 activation is transcriptional up-regulation of different AP-1 components upon cell stimulation (10). To analyze whether influenza A virus infection results in an up-regulation of AP-1 factors, EMSAs from nuclear lysates of infected and noninfected MDCK cells were performed. Although there was a slight up-regulation of AP-1 factors 1-2 h after infection ( Fig.  2A), the major portion of AP-1 factors was already bound to DNA in uninfected cells. Competition with the nonlabeled AP-1 DNA probe showed that the protein-DNA complexes observed were AP-1-specific ( Fig. 2A). To analyze which AP-1 factors are constituents of the complex, supershift assays using specific antibodies against certain AP-1 proteins were performed. The shifted protein-DNA complexes in the presence of the antibodies indicated that c-Jun, c-Fos, and JunD were components of the AP-1 complexes. Further, with regard to these three fac-tors, the composition of the AP-1 complexes did not differ 4 or 8 h postinfection (p.i.) (Fig. 2A). The same results were obtained with U937 cells, albeit in these cells more AP-1 specific complexes were observed (Fig. 2B). Furthermore, ATF-2 was identified as another AP-1 binding factor (Fig. 2C), while transcription factors JunB, Fra-1, and CREB did not bind to the AP-1-specific DNA probes in both cell lines ( Fig. 2C and data not shown). Thus, c-Jun, c-Fos, JunD, and ATF-2 were identified as AP-1 binding factors in uninfected and infected cells and may contribute to the observed induction of AP-1-dependent transcription.
ATF-2 Is Phosphorylated in Response to Virus Infection-The transcriptional activity of AP-1 complexes is potentiated upon phosphorylation of AP-1 factors (10). One of these factors is ATF-2, which plays a critical role in the virus-induced expression of the antiviral cytokine IFN-␤ (31, 32). ATF-2 was indeed phosphorylated upon infection with different influenza A virus strains in permissive 293 (Fig. 3A), COS7 (Fig. 3B), and U937 cells (data not shown). Two different MAPKs were reported to phosphorylate ATF-2, p38 and JNK/SAPK (33,34). Although p38 is activated upon influenza virus infection (data not shown), ATF-2 phosphorylation was not decreased or abolished in cells treated with effective concentrations of the p38-specific inhibitor SB202190 (Fig. 3B). This indicates that the JNK/ SAPK subgroup of MAPKs may be primarily responsible for the observed phosphorylation event.
Productive Infection of Cells with Different Influenza A Viruses Results in a Strong Activation of JNK-Since the data shown in Fig. 3B suggest that JNK causes virus-induced phosphorylation of ATF-2, we analyzed whether infection with different influenza A viruses results in JNK activation. Cells were infected and JNK activity was assessed in immune complex kinase assays at different time points postinfection. A strong activation of JNK was observed upon productive infection of MDCK, U937, or HEK293 cells. Activation kinetics did not differ between the virus isolates but were dependent on the cell line chosen. For example, infection of MDCK cells with the influenza A virus strain Asia (Fig. 4A), the FPV strain (Fig. 4B) or the reassortant virus WSN-HK (data not shown) resulted in  (Fig. 4E). The same results were observed in Asia-infected A3.01 T lymphocytes (data not shown). This indicates that the potential of influenza A viruses to replicate in a certain cell type correlates with both the ability to induce JNK activity (Fig. 4) and AP-1-dependent gene expression (Fig.  1). In support of this assumption, we observed that inhibition of virus replication by the anti-influenza virus drug amantadine (35) results in a 50% decrease in JNK activity 8 h postinfection (Fig. 5A)  tagged versions of either dominant negative or wild-type forms of SEK/MKK4 (Fig. 5B) or MKK7 (Fig. 5C). Cells were infected for 8 h, and transfected MKKs were specifically precipitated from cell lysates with the respective anti-tag antibody. Activities of SEK/MKK4 or MKK7 were then assessed in immune complex kinase assays using recombinant GST-tagged SAPK␣, which is the rat isoform of JNK2, as a substrate. Both wild-type forms of SEK/MKK4 (Fig. 5B, lane 4) and MKK7 (Fig. 5C, lane  4) were activated upon infection, which was not the case in the control samples with the dominant negative kinases (Fig. 5, B  and C, lane 2). To analyze whether the observed phosphorylation of GST-SAPK␣ was on the functional relevant sites, we performed a coupled assay by adding in recombinant GST-c-Jun-(1-135) as a substrate for GST-SAPK␣ to the sample. Indeed, GST-c-Jun-(1-135) was phosphorylated selectively in the samples corresponding to the infected cells (Fig. 5D, lanes  2 and 4), indicating that virus-induced activation of both MKK4 and -7 resulted in a functional phosphorylation of GST-SAPK␣. For a further proof that SEK/MKK4 and MKK7 are upstream of JNK upon virus infection, the activity of the kinase was assessed in the presence of dominant negative forms of both MKKs. Fig. 5E shows that expression of both dominant negative MKK4/SEK (lane 3) and dominant negative MKK7 (lane 4) results in an inhibition of virus-induced activation of coexpressed HA-tagged JNK/SAPK. A concentration-dependent inhibition was not only observed by expression of SEK(KϾR) but also of the JNK-binding domain of JIP-1 (Fig. 5F). Overexpressed JIP-1 inhibits JNK by competitive binding and retention in the cytoplasm (22). On the other hand, transient expression of wild-type MKK7 acts synergistically with virus infection to induce JNK/SAPK activity (Fig. 5G, lane 4). This indicates that the amounts of cellular MKK7 are limited and that the signal transmission is greatly enhanced if a surplus of the kinase is provided. Expression of an active mutant of MKK7 (MKK7(S3E)) resulted in a strong JNK activation already in the absence of virus, which is again significantly enhanced after infection (Fig. 5G, lanes 5 and 6), presumably due to the additional action of virus-activated MKK4/SEK. Taken together, virus infection induces the activity of both MKK4/SEK and MKK7, which act synergistically to activate JNK/SAPK.  (Fig. 6A). As observed in the kinase assays (Fig. 5G), expression of MKK7wt and active MKK7(S3E) enhances the effect of virus infection on AP-1-dependent gene expression (Fig. 6B), whereas expression of the inhibitory mutant SEK(KϾR) or JIP-1 results in an inhibition, both in U937 and MDCK cells (Fig. 6C). Thus, . Cell lysates were subjected to SDS-polyacrylamide gel electrophoresis, blotted onto nitrocellulose membranes, and incubated with an antibody specific for phosphorylated ATF-2. Equal loading of ATF-2 was confirmed using a pan-ATF-2 antiserum. ATF-2 phosphorylation was also detected in lysates of virus-infected U937 cells with an earlier onset consistent with the kinetics of JNK activation in these cells (data not shown). A and B), HEK293 (C and D), HeLa (E), and U937 cells (F) were infected with Asia (A, C, and E) or FPV (B, D, and E) at a MOI of 5. Cells were harvested at the indicated times postinfection, and JNK1 was immunoprecipitated from cell lysates. Immune complexes were subjected to in vitro kinase assays using purified GST-c-Jun-(1-135) as a substrate. Proteins were separated by SDS-polyacrylamide gel electrophoresis, and labeled substrate bands were detected by exposure to x-ray films. Equal loading of JNK1 proteins was confirmed by Western blot. influenza virus-induced AP-1 dependent gene expression is regulated by MKK4/SEK and MKK7-mediated activation of JNK/SAPK and c-Jun.

Accumulation of RNA Synthesized by the Viral Polymerase Complex Results in Activation of JNK-A remaining question
is which viral components are responsible for activation of JNK in the host cell. Recently, we identified influenza A virus hemagglutinin, nucleoprotein, and matrix protein as viral transactivators toward the transcription factor NF-B (36). However, expression of these proteins resulted in neither the activation of JNK (data not shown) nor the activation of AP-1dependent promoters (36). These studies were extended to the viral NS1 and NEP/NS2 proteins, which also did not induce JNK activity when overexpressed (data not shown). Another component of the virus is the single-stranded vRNA, which is transcribed and replicated by the viral RDRP complex consisting of the PB2, PB1, and PA proteins (1). The vRNA is transcribed into mRNA and amplified via a complementary RNA replication intermediate. To assess whether dsRNA has the capacity to induce JNK signaling, we treated MDCK cells with the dsRNA analog, poly(IC). Indeed, poly(IC) stimulation induced a time-and concentration-dependent activation of JNK (Fig. 7A), which was not inhibited by actinomycin D or cyclo-heximide (data not shown). This indicates that activation occurs directly and does not involve new RNA or protein synthesis. To analyze whether accumulation of influenza-like RNAs gives rise to JNK activation, we used a plasmid-based replication system for vRNA-like RNA polymerase I (Pol I) transcripts. Briefly, cells were transfected with expression plasmids for the RDRP constituents PB2, PB1, PA, and NP and for a vRNA-like Pol I transcript carrying the conserved 5Ј and 3Ј promoter structures of influenza vRNA segments flanking an antisense reporter gene. Such constructs have been successfully used to provide vRNA-like templates for the RDRP in plasmid-derived reverse genetics approaches (37). In these studies, it was demonstrated that all three polymerase genes as well as the NP gene have to be cotransfected to allow transcription and amplification of the reporter gene template (37). Using GFP as a reporter gene (pPol I GFP) ϳ5-10% of transfected cells showed green fluorescence after cotransfection of plasmids expressing PB2, PB1, PA, and NP, while no green cells were detectable if either the plasmids for NP or the template vRNA were replaced by empty vectors (data not shown). This indicates that only 5-10% of transfected cells carry all of the plasmids required for transcription and amplification of the template vRNA. Nevertheless, we observed a significant acti- vation of GST-SAPK␤ in the samples cotransfected with all essential replicase genes and either pPol I GFP or pPol I Luc plasmids expressing template vRNAs (Fig. 7B, lanes 3 and 6).
No SAPK␤ activation was detectable if either NP (Fig. 7B, lane  2) or the template RNAs were omitted (Fig. 7B, lane 4) or replaced by empty pPol I vectors (Fig. 7B, lanes 5 and 7). These findings demonstrate that JNK/SAPK is activated during influenza virus-specific transcription and replication of an influenza-like RNA template by the viral RDRP complex, suggesting that vRNA accumulation is the trigger for activation of the JNK signaling pathway.
The JNK Signaling Pathway Is Critical for Virus-induced Activation of the IFN-␤ Promoter-IFN-␤ is one of the most potent antiviral cytokines. The IFN-␤ promoter/enhanceosome was shown to bind c-Jun/ATF-2 heterodimers, which are essential for the virus-induced transcription of the cytokine gene (31). To finally analyze whether virus-induced IFN-␤ expression is dependent on the JNK signaling pathway, we cotransfected dominant negative mutants of JNK/SAPK (SAPK␤(KKϾRR)) or c-Jun (TAM67) with an IFN-␤ promoter luciferase construct. Both, treatment with poly(IC) and infection with influenza A virus, resulted in a strong transactivation of the IFN-␤ promotor, which was efficiently blocked in the presence of dominant negative JNK/SAPK or c-Jun (Fig. 8, A  and B). These data demonstrate that JNK/SAPK is essential for RNA-or influenza virus-induced transcription of the cytokine gene. Since the JNK pathway appears to regulate the expression of an efficient antiviral cytokine, inhibition of the pathway should increase the ability of the virus to replicate in cells. Indeed, higher virus yields were obtained in infected MDCK expressing dominant negative mutants of either MKK7, JNK/SAPK, or c-Jun (Fig. 8C). Taken together, these data suggest that activation of the JNK signaling pathway and AP-1-dependent gene expression upon influenza virus infection are part of the innate immune response of the cell, e.g. by triggering expression of antiviral cytokines such as IFN-␤. DISCUSSION The elucidation of intracellular signaling pathways that are activated by influenza A virus infection is important for the understanding of both viral replication strategies and host defense mechanisms. Here we show that productive infection of cells results in a strong activation of the JNK subgroup of MAPK signaling pathways. Further, expression of dominant interfering mutants revealed that the JNK signaling pathway is a major player in transduction of virus-induced signals leading to AP-1-dependent expression of IFN-␤. Thus, virus-induced activation of JNK and AP-1 appears to be part of the innate antiviral response of the host cell.
AP-1 family transcription factors are regulated both on a transcriptional and posttranslational level (10). Our results demonstrate that transcriptional up-regulation of AP-1 factors only plays a minor role for influenza virus-induced AP-1-dependent gene expression in MDCK and U937 cells. In both uninfected and infected cells, the major portion of AP-1 components were constitutively bound to AP-1 sites, and the composition of the complexes was not significantly changed during infection (Fig. 2). This is consistent with the appearance of constitutively expressed JunD and ATF-2 in the complex. The weak up-regulation 1-2 h after infection (Fig. 2) may be attributed to c-Jun and c-Fos that were also identified in the complexes. It is still an open question which virus-induced signaling events are responsible for the observed up-regulation of these factors. Recently, it was shown that pharmacological inhibition of the mitogenic Raf/MEK/ERK signaling cascade at the level of MEK results in a block of 12-O-tetradecanoylphorbol-13-acetate-induced c-Jun and c-Fos mRNA synthesis (38). Using the same inhibitor, U0126, in EMSA experiments, we could not interfere with influenza virus-induced up-regulation of AP-1 factors, 2 although we observed an activation of the Raf/MEK/ERK cascade early after infection (39). Given that a variety of signaling pathways are known to be involved in regulation of AP-1 factor expression (40), inhibition of MEK by U0126 may not be sufficient to result in detectable differences in the EMSA.
In an analysis of the viral features that give rise to the strong JNK activation, we have now identified accumulation of virusspecific RNA as a major trigger. This finding is consistent with two other reports published while this manuscript was in preparation (41,42). In these studies, it was shown that treatment with the dsRNA analog poly(IC) as well as infection with vesicular stomatitis virus (41) or encephalomyocarditis virus (42) results in JNK activation. However, poly(IC) might have its limitations to represent RNA of replicating single-stranded RNA viruses. In our study, we have reconstituted the influenza virus replicase complex to mimic ongoing viral genome transcription and replication in the absence of infection. This is the first demonstration that RNA accumulation within a virus-like setting causes activation of the JNK signaling cascade. This finding further indicates that viral RNA still has signaling 2 C. Ehrhardt and S. Ludwig, unpublished results. capacity, although it is packaged with NP to form RNP complexes. It will now be important to identify the direct cellular signaling sensors for viral RNA. The only cellular signaling enzyme known so far to be directly regulated by dsRNA is the 68-kDa double-stranded RNA-dependent kinase (43), which inhibits protein synthesis by phosphorylation of the ␣-subunit of the eukaryotic initiation factor 2 (44). It was recently shown that virus-induced activation of IB kinase and NF-B is dependent on RNA-dependent kinase activity; however, JNK was activated independently of RNA-dependent kinase in these studies (41). Our experiments indicate that influenza A virusinduced AP-1 activation is mediated by JNK and its upstream activators MKK4/SEK and MKK7. A variety of kinases that are direct upstream activators of MKK7 and MKK4/SEK are known (45) and may be involved in virus-induced JNK activation, but none of these kinases had so far been shown to be activated by dsRNA. Thus, the search is open for a dsRNAregulated enzyme upstream of MKK4/SEK and MKK7.
Usually, 5-6 h after infection with influenza A virus, an efficient host-cell protein synthesis shut off is initiated (1). The induction of JNK activity 2-4 h postinfection and the amount of luciferase that is accumulated as early as 4 h postinfection show that the viral induction of both AP-1-dependent gene expression and transactivation of the IFN-␤ promoter occur before the protein-synthesis block. The late virus-induced activity of JNK at 8 -10 h postinfection may only poorly contribute to gene expression events. Although this late activation phase is not counteracted by the virus, it may not positively interfere with virus replication, since cells in which JNK signaling is impaired allowed a more efficient virus propagation (Fig. 8C).
Another interesting observation is the inability of nonpermissive cells, such as T lymphocytes or HeLa cells, to support virus-induced JNK1 and AP-1 activation. These cells are readily infected; however, they undergo an abortive infection at some point during the viral life cycle. In HeLa cells, synthesis of viral proteins was found to be normal, but there appears to be a defect during budding of novel virus particles (46), presumably due to a misincorporation of the HA into the membrane (47). In T lymphocytes, the point at which viral replication is impaired is less well understood; however, there is also a significant amount of viral proteins synthesized without formation of new virus particles (29). It will now be of great interest to elucidate why there is a lack of JNK activation despite the fact that virus-specific RNA is still synthesized in these cells.
In summary, our results demonstrate for the first time that productive influenza A virus infection results in the activation of a member of the MAPK signaling cascades, which are important mediators of intracellular signaling. Virus-induced activation of AP-1-dependent gene expression occurs early during the viral life cycle, is mediated by viral RNA-induced activation of the JNK signaling cascade, and appears to be part of the innate immune response to infection.