NF- (cid:1) B-dependent Induction of Tumor Necrosis Factor-related Apoptosis-inducing Ligand (TRAIL) and Fas/FasL Is Crucial for Efficient Influenza Virus Propagation*

Activation of the transcription factor NF- (cid:1) B is a hallmark of infections by viral pathogens including influenza viruses. Because gene expression of many proinflammatory and antiviral cytokines is controlled by this factor,

The IB kinase (IKK)/NF-B 1 signaling module is activated by a variety of extracellular agents, including proinflammatory cytokines and pathogenic invaders. The NF-B/IB family of transcription factors promotes the expression of well over 150 different genes, such as cytokine or chemokine genes or genes encoding for adhesion molecules or anti-and proapoptotic proteins (1). The canonical mechanism of NF-B activation in-cludes activation of IKK, which phosphorylates the inhibitor of NF-B, IB, and targets the protein for subsequent degradation (2,3). This leads to the release and migration of the transcriptionally active NF-B factors, such as p65 or p50, to the nucleus (4,5). The IKK complex consists of at least three isozymes of IKK, IKK1/IKK␣, IKK2/IKK␤, and NF-B essential modulator (NEMO)/IKK␥. The most important isozyme for NF-B activation via the degradation of IB is IKK2 (6), whereas IKK1 seems to primarily phosphorylate other factors of the NF-B/IKK family, namely p100/p52 (7,8). The NF-B essential modulator acts as a scaffolding protein for the large IKK complex (9) that was reported to contain still other kinases such as IKK-⑀, TANK-binding kinase-1 (TBK-1), mitogen-activated protein kinase/extracellular signal-regulated kinase kinase kinase 1 (MEKK1) (10), NF-B-inducing kinase (NIK) (11,12), and the double-stranded RNA (dsRNA)-activated protein kinase (PKR) (13,14).
NF-B co-regulates one of the most important antiviral gene expression events, the transcriptional induction of IFN␤ (15). IFN␤ is one of the first antiviral cytokines to be expressed upon virus infection, initiating an autoamplification loop to cause an efficient and strong type I IFN response (16). The IFN␤ enhanceosome, which mediates the inducible expression of IFN␤, carries binding sites for transcription factors of three families, namely the AP-1 family members and c-Jun NH 2 -terminal kinase targets c-Jun and ATF-2, the NF-B factors p50 and p65, and the interferon-regulatory factors (IRFs) (17,18). Besides NF-B a main determinant of a strong virus-and dsRNAinduced IFN␤ response is IRF-3. Recently the IKK family members IKK-⑀ and TBK-1 were identified as IRF-3 kinases (19,20). Both kinases appear to act downstream of the GTPase Rac1 and the p21-activated kinase PAK1 in response to Sendai virus and influenza virus infection (21).
NF-B is activated by multiple families of viruses, including HIV-1, HTLV-1, hepatitis B and C viruses, Epstein-Barr virus, vesicular stomatitis virus, and influenza viruses (22). Although for some of these viruses, e.g. retroviruses or oncogenic viruses, activation of this transcription factor may support viral replication, it is a common view that in general NF-B acts in an antiviral fashion upon infection with RNA viruses, such as vesicular stomatitis virus, Sendai virus, or influenza virus (23,24). RNA virus infections commonly result in activation of an innate antiviral response that includes expression of type I interferons. In addition to IFN␤, also other genes involved in the induction of inflammation and immune responses are reg-ulated by NF-B, such as IL-6, TNF-␣, or IL-12. Accordingly, vesicular stomatitis virus-induced expression of IFN␤, IL-6, and IL-12 is impaired in cells deficient for the NF-B activator IB kinase 2 (IKK2) (23).
Another mode of NF-B interference with virus propagation is through its capability to regulate apoptosis, a morphologically and biochemically defined form of cell death (25) that has been demonstrated to play a role in a variety of diseases including virus infections (26). Apoptosis is mainly regarded to be a host cell defense against virus infections because many viruses express antiapoptotic proteins to prevent this cellular response. NF-B is mainly regarded as a survival factor by up-regulating genes encoding for antiapoptotic proteins, such as Bcl-X L , A20, or cellular inhibitors of apoptosis proteins (27,28). However, NF-B was also reported to act proapoptotically under certain conditions, e.g. by up-regulating the death-inducing CD95 ligand and its receptor (27,28). For example, Dengue virus-infected HepG2 hepatocytes undergo apoptosis in an NF-B-dependent manner (29), while NF-B activation induced by the hepatitis C virus core protein protects cells from apoptosis in the same cell line (30). Thus, the concept of a context-dependent regulation of apoptosis by NF-B has emerged (28).
A common viral inducer of NF-B-dependent responses appears to be dsRNA. Most RNA viruses produce dsRNA-like replication intermediates representing a shared molecular pattern that may be sensed by the cell as an alert signal (31).
Activation of IB kinase and NF-B by influenza virus or influenza viral products is well documented (32)(33)(34)(35)(36); however, the function of this signaling module in the virus-infected cell is not yet as clear. The knowledge so far is restricted to data obtained with a recombinant virus termed delNS1 with a deletion of the complete coding region of the viral non-structural protein 1 (NS1) (37). Infection of cells with influenza delNS1 virus results in enhanced NF-B activation and IFN␤ production (24) as well as enhanced apoptosis (38), suggesting that the NS1 protein is at least a partial antagonist of these presumed antiviral responses. Nevertheless, the consequences of influenza virus-induced NF-B activation for the outcome of an infection have never been directly addressed yet.

EXPERIMENTAL PROCEDURES
Viruses, Cell Lines, and Viral Infections-Avian influenza virus A/Bratislava/79 (H7N7; fowl plague virus (FPV)) and human influenza viruses A/Puerto-Rico/8/34 (H1N1; PR8), A/Asia/57 (H2N2; Asia), and A/Victoria/3/75 (H3N2; Victoria) were taken from the virus strain collection of the Institute of Virology, Giessen, Germany, and were used for the infection of different cell lines. Madin-Darby canine kidney (MDCK) cells and African green monkey kidney cells (Vero) were grown in minimal essential medium (MEM) supplemented with 10% heat-inactivated fetal bovine serum and antibiotics. A549 human lung carcinoma cells were grown in Ham's F-12 medium supplemented with 10% heatinactivated fetal bovine serum and antibiotics. Primary human endothelial cells were obtained from Clonetics and cultured as described previously (39). A549, MDCK, and Vero cell lines stably expressing transdominant IB␣ mutant as well as constitutively active and dominant negative IKK2 were generated with a retroviral transduction approach using the pCFG5-IEGZ retroviral vector system (40) and amphotrophic Phoenix producer cell lines essentially as described by Denk et al. (41). For infection cells were washed with PBS incubated with virus at the indicated multiplicities of infection (MOI) diluted in PBS/BA (PBS containing 0.2% bovine serum albumin, 1 mM MgCl 2 , 0.9 mM CaCl 2 , 100 units/ml penicillin, and 0.1 mg/ml streptomycin) for 1 h at room temperature. The inoculum was aspirated, and cells were incubated with either MEM or Ham's F-12 containing 0.2% bovine serum albumin and antibiotics. 9 or 24 h post-infection (p.i.) supernatants were collected to assess the number of infectious particles (plaque titers) in the samples. Briefly, MDCK cells grown 90% confluent in 6-well dishes were washed with PBS and infected with serial dilutions of the supernatants in PBS/BA for 1 h at 37°C. The inoculum was aspirated, and cells were incubated with 2 ml of MEM/BA (medium containing 0.2% bovine serum albumin and antibiotics) supplemented with 0.6% agar (Oxoid), 0.3% DEAE-dextran (Amersham Biosciences), and 1.5% NaHCO 3 at 37°C, 5% CO 2 for 2-3 days. Virus plaques were visualized by staining with neutral red.
Inhibitors, Antibodies, and Reagents-Caspase inhibitor Z-VAD-FMK or inhibitor control Z-FA-FMK (Alexis Biochemicals, Grü nberg, Germany) was supplied ready-to-use at a concentration of 2 mM in Me 2 SO. Me 2 SO was used as solvent control at a final concentration of 2%, representing the highest inhibitor concentration. The mouse antipoly(ADP-ribose) polymerase (anti-PARP) monoclonal antibody was purchased from Transduction Laboratories. Antibodies for TNF-related apoptosis-inducing ligand (TRAIL) or extracellular signal-regulated kinase 2 (ERK2) Western blots were purchased from Santa Cruz Biotechnology (sc-8440 or sc-154, respectively). mAbs against TRAIL, TRAILreceptor 1 (R1) and -R2, CD95, and CD95L for FACS, as well as recombinant FLAG-or His-tagged human TRAIL and recombinant FasL/CD95L, are available from Alexis Biochemicals. Soluble TRAILreceptor 2 (TRAIL-R2-Fc), soluble TNF-␣ receptor (sTNF-R2-Fc), and CD95-Fc were produced as fusion proteins with human IgG-Fc at the Division for Apoptosis Regulation, Deutsches Krebsforschungszentrum (DKFZ), Heidelberg, Germany (42). A receptor-neutralizing mAb against the type I IFN receptor was purchased from PBL Biomedical Laboratories. A goat antiserum against the influenza virus NP was a kind gift of Robert G. Webster, Memphis, TN.
Flow Cytometry Analysis-TRAIL was detected by an intracellular staining procedure. A549 cell lines were infected with FPV at an MOI of 5 for 8 h in the presence of 2 M monensin to avoid protein secretion. Cells were fixed with 4% paraformaldehyde at 4°C for 20 min and subsequently washed twice in permeabilization buffer (0.1% saponin, 1% fetal calf serum, PBS). After incubation with a mAb against TRAIL or isotype control antibodies (BD Biosciences) cells were stained with biotin-SP-conjugated goat anti-mouse IgG (Dianova) and streptavidin-Cy-chrome (BD Biosciences). In another set of assays CD95, CD95L, and TRAIL detection was performed according to the same protocol but without monensin/saponin treatment. Fluorescence was determined in the FL3-channel using a FACSCalibur cytometer (BD Biosciences). All FACS analyses were repeated at least twice and revealed essentially similar results.
RNase Protection Assay-A549 cell lines were mock-infected or infected with FPV at a MOI of 1. 24 h p.i. cells were lysed, and RNA was isolated using the TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Purified RNA was subjected to the RiboQuant multiprobe RNase protection assay hAPO-3 (Pharmingen, BD Biosciences) according to the manufacturer's instructions.
Influenza A Virus Replication Rescue Assay-A549 cell lines were infected with FPV at a MOI of 1. After 1 h at 37°C, the inoculum was removed. 6 h after mock infection, 100, 50, 10, 5, or 1 ng/ml recombinant human TRAIL or FasL was added to the medium. 3 h after the addition of TRAIL or CD95L (9 h p.i.) supernatant was collected to assess the number of infectious particles (plaque titers) in the samples.
Analysis of the Apoptotic Phosphatidylserine Switch-MDCK cell lines were infected with different influenza A viruses at a MOI of 1. 24 h p.i. cells were washed with PBS, detached from the culture plates with 0.25% trypsin, and spun down. Cell pellets were washed twice with PBS. Annexin V-Alexa568 (Roche Applied Science) staining to detect phosphatidylserine exposure at the outer leaflet of the cell membrane as an early apoptotic marker was performed according to the manufacturer's instructions. Apoptotic cells were detected and quantified by FACS analysis.

RESULTS
To analyze the roles of IKK and NF-B activation during an influenza virus infection we established host cell lines that stably express a dominant negative mutant of IKK2 (IKK KD) or a non-degradable phosphorylation site mutant of IB␣ (IB␣mut) (41) that both prevent NF-B activation. In addition an active mutant of IKK2 (IKK EE) was stably expressed in cells resulting in a slightly enhanced NF-B activity in stimulated or infected cells (Fig. 1, B and C). In contrast expression of IKK KD or IB␣mut in these cell lines resulted in an efficient reduction of NF-B activation. TNF-␣-induced IB␣ degradation was impaired (Fig. 1A), and influenza virus-induced transcription from a NF-B-responsive promoter element (Fig. 1B) was reduced by about 80 -90% in the presence of the mutants. High expression of IKK EE appears to not be well tolerated in a cell line because a strong down-regulation of the transgene is commonly observed after a few passages in all cell types analyzed so far (data not shown). Nevertheless a slight constitutive activity and an enhanced NF-B-dependent transcriptional activity upon stimulation or virus infection were detected in the IKK EE cell lines (Fig. 1, B and C). This was also consistently observed when the artificial NF-B reporter gene construct was replaced by a reporter gene plasmid harboring the IFN␤ pro-moter/enhanceosome (Fig. 1C). In line with the data in Fig. 1B the dominant negative mutants also conferred inhibition of transcriptional activation (Fig. 1C), although less pronounced because of the involvement of other virus-induced transcription factors controlling IFN␤ expression. These findings support earlier reports suggesting that NF-B regulates IFN␤ expression as an antiviral function (23).
However, when we examined the efficacy of influenza virus production in these different cell lines, we surprisingly observed exactly the opposite. Virus propagation was impaired upon NF-B inhibition and slightly enhanced in cells expressing the active form of IKK EE (Fig. 1, D and E). The effects were similar in MDCK epithelial cells and in Vero cells (Fig.  1D) and could also be observed in primary human endothelial cells or by infection with influenza viruses of other subtypes (e.g. H1N1, H2N2, and H3N2) (data not shown). Although the levels of reduction and enhancement of virus propagation were not very pronounced, which may in part be due to inefficient i. to perform luciferase assays. Each bar represents the average and standard deviation of three independent transfections. C, IFN␤ promoter/enhanceosome luciferase-reporter gene assay. Experiments were performed as described in B replacing the 3ϫNF-B promoter element with a luciferase-reporter plasmid driven by the IFN␤ promoter/enhanceosome, which carries the AP-1, IRF, and NF-B binding sites responsible for inducible IFN␤ expression. D, MDCK and type I interferon-deficient Vero cell lines were infected with FPV (MOI ϭ 1). Supernatants were collected 9 h p.i., and virus yields were determined in plaque assays on MDCK cells. Mean virus titers from vector-bearing cell lines were 2.25 ϫ 10 7 plaque-forming units/ml (Vero) or 7.08 ϫ 10 7 plaque-forming units/ml (MDCK). Data are shown as the percentage of virus titers compared with the vector control. Each bar represents the average and standard deviation of three independent experiments. E, experiments were performed as described in D using the different A549 cell lines infected with FPV for 12 and 24 h, respectively. Data are shown as absolute plaque-forming units (pfu) in a logarithmic scale. Each bar represents the average and standard deviation of three independent experiments. action of the transgenes in cells from another species, it should be stressed that similar effects were consistently observed in each experiment within a set of individual experiments. Vero cells do not express type I interferons (46), excluding a prominent involvement of these cytokines in the observed effects. Inhibition or enhancement of virus propagation correlated with the efficacy of transgene action (data not shown) and was strongest in A549 lung epithelial cells, where the differences in virus titers between the different cell lines were around 10-fold (Fig. 1E). These different efficacies of virus production in the different cell lines were not due to a delay or shift in replication kinetics because we observed the same pattern at every time point throughout a 48-h period of virus growth regardless of whether virus in the supernatant or cell-associated virus was examined (data not shown). These results differ from studies obtained upon infections of the respective NF-B-defective CRL cell lines with Borna disease virus, another nucleus-replicating negative-strand RNA virus. Here, inhibition of NF-B had no pronounced effect on replication; however, activation of the pathway by expression of IKK EE resulted in a strongly impaired virus replication because of a massive induction of type I interferons. 2 This suggests a crucial function of NF-B specifically for influenza virus replication, which appears to occur independently of a type I interferon response.
Because NF-B is a regulator of both pro-and antiapoptotic genes (28), we examined viral induction of apoptosis in the different NF-B cell lines. To monitor an early apoptotic event we chose the switch of phosphatidylserine to the outer leaflet of cells, a process that can be examined by annexin V stainings in FACS analysis ( Fig. 2A). Compared with the vector control a higher number of cells were found to be annexin V-positive in infected IKK EE cell lines, whereas fewer cells were positive when NF-B was inhibited ( Fig. 2A). Another early hallmark of apoptosis induction is caspase activation. This could be monitored by proteolytic cleavage of caspase substrates, such as PARP. Influenza virus infection resulted in a significant cleavage of PARP in vector cell lines that was further enhanced in IKK EE-expressing cells (Fig. 2B). In contrast, virus-induced PARP cleavage was significantly reduced in infected cells expressing transdominant negative mutants of IKK or IB␣ (Fig. 2B). Essentially the same patterns of early apoptosis induction that matched with viral replication efficacies were also observed with other virus isolates, such as human isolates of the H1N1, H2N2, and H3N2 subtypes (data not shown). Thus, we face a general phenomenon in which NF-B acts in a proviral and proapoptotic rather than an antiapoptotic fashion in the context of an influenza virus infection. These findings are consistent with our earlier report demonstrating that caspase activation is required for efficient virus propagation (47). Furthermore, the present data links viral NF-B activation to this phenomenon.
Taken together, the data suggested that during influenza virus infection NF-B up-regulates one or more proapoptotic factors, which results in enhanced virus production. Therefore, we examined virus-induced gene expression of a variety of apoptosis regulators, such as caspase 8, CD95/Fas, death receptor 3 (DR3), TRAIL or TNF-R1 in vector, and IKK EE-or IKK KD-expressing cell lines using RNase protection assays (Fig. 3A). One of the genes that was strongly up-regulated upon virus infection in vector-transfected or IKK EE-expressing cell lines but completely absent in cell lines expressing IKK KD encodes TRAIL. TRAIL was expressed in increasing amounts over time during an influenza virus infection (Fig. 3B), and inhibition of NF-B signaling by IKK KD or IB␣mut resulted in a complete block of virus-induced TRAIL synthesis (Fig. 3C).
Another gene that was found to be up-regulated upon infection in a NF-B-dependent manner was the death receptor Fas/CD95 gene (Fig. 3A). Subsequent FACS analysis revealed that this is also true on the protein level; both Fas/CD95 and its ligand FasL/CD95L were induced by influenza virus infection on the protein level, and the expression of both genes was regulated by NF-B (Fig. 3D). In contrast, the TRAIL-R1 and -R2 are not regulated by influenza virus infection but are constitutively expressed both in A549 cells and in primary human endothelial cells (data not shown).
To further test whether the apoptosis inducers TRAIL and the Fas/FasL system might be involved in NF-B-dependent enhancement of influenza virus replication, we determined virus titers in the absence or presence of TRAIL-R2-Fc or CD95-R-Fc, efficient inhibitors of TRAIL or CD95L, respectively (42,48). Both agents not only resulted in a dose-dependent decrease of virus production if applied individually (Fig. 4, A and C) but also acted synergistically to reduce virus titers by more than 1 order of magnitude (Fig. 4C). This indicates that both TRAIL and CD95L are proviral factors. In contrast, presence of soluble TNF-R2-Fc (48), a receptor for another proapoptotic and virus-induced NF-B-dependent cytokine, did not lead to a significant reduction of virus titers (Fig. 4A), although the same concentration of the reagent efficiently blocked TNF-␣-induced IB␣ degradation (data not shown). Consistent with the finding that TRAIL and FasL support virus propagation, 2 O. Planz and S. Ludwig, unpublished data.

FIG. 2. NF-B acts proapoptotically in the context of an influenza virus infection.
A, annexin V staining of A549 for detection of a phosphatidylserine switch as an early marker of apoptosis. A549 cell lines were infected with FPV (MOI ϭ 1) and stained 24 h p.i. with annexin V. Subsequently, samples were analyzed by FACS. B, immunoblot for PARP cleavage as an early apoptotic marker. A549 cell lines were mock-infected or FPV-infected (MOI ϭ 1) and were lysed 24 h p.i. in Triton X-100 lysis buffer. After SDS-PAGE and blotting, membranes were subjected to Western blot using an anti-PARP mAb detecting both the cleaved and the uncleaved form of the protein. ERK blots served as a loading control. stimulation of infected cells with low concentrations of recombinant human TRAIL or FasL enhanced virus production in A549 cells and rescued the phenotype of NF-B deficiency in the different cell lines (Fig. 4, B and D). Because IFN␤ is also regulated by NF-B and because type I interferons are apoptosis inducers, IFN␣/␤ might still play a supportive role in the observed phenomena. However, when type I IFN receptorblocking antibodies were added to the cell culture medium, we observed slightly enhanced rather than reduced virus titers (Fig. 4E). No difference was observed in IB␣mut-expressing cell lines, indicative of a block of IFN synthesis in cells with impaired NF-B signaling. These findings demonstrate that small amounts of type I interferons are still induced in influenza A virus-infected wild type cells but not in IB␣-expressing cell lines. However, these small amounts of IFN␣/␤ confer a slight antiviral rather than a proviral effect (Fig. 4E). Thus, the mechanism by which NF-B acts in a proviral manner in influenza virus-infected cells includes gene regulation of apoptosis inducers such as TRAIL and CD95L. The action of these proviral apoptosis inducers appears to be dominant over a potential antiviral effect conferred by IFN␤ or TNF-␣, which both are only expressed in very little amounts (Fig. 4E and data not shown). DISCUSSION Although the role of NF-B in influenza virus-infected cells has never been studied in detail, it was a common thought that both NF-B activation and apoptosis induction are antiviral responses to influenza virus infections. According to the data presented here NF-B indeed regulates expression of antivirally acting cytokines such as IFN␤; however, the effect that is finally dominant for the outcome of virus propagation is the regulation of provirally acting factors such as TRAIL, Fas, and FasL.
The inhibiting or enhancing effects of the different NF-B mutants on virus propagation in the stably transduced cell lines might be considered relatively small with regard to a fast replicating virus such as influenza virus. However, several points have to be taken into account when judging the significance of the effects. First of all the dominant negative transgenes do not fully block NF-B activation resulting in a certain degree of leakiness. Thus, a maximum of 80 -90% inhibition of NF-B activity correlates quite well with an up to 10-fold inhibition of virus replication. Furthermore, NF-B acts in both anti-and proviral gene expression. This means that the proviral activity observed here is diluted to a certain extent by antiviral gene expression events, e.g. control of IFN␤ expression (Fig. 4E). This is not the case for the provirally acting NF-B-controlled gene products such as TRAIL and FasL, and consequently the combined blockade of these factors by the receptor-Fc fusion proteins results in an inhibition of virus propagation by more than 1 log (Fig. 4C). It is likely that there are still other NF-B-controlled factors acting in the same way and further contributing to the effects. Although earlier reports demonstrated NF-B-dependent Fas and FasL expression (49 -51) and coexpression of both proteins on the surface of influenza virus-infected cells (52), little is known about the virus-induced transcriptional regulation of TRAIL so far. Nevertheless, recent studies have shown that expression of TRAIL can be induced in a NF-B-dependent manner in Jurkat T-cells (53). We now show here that this is also true for a viral inducer in epithelial cells. Our data are supported by findings from a recent transcriptional profiling approach identifying the TRAIL gene among 84 of 13,000 genes that were deregulated in response to infection with a human influenza virus (54). Furthermore, TRAIL and/or TRAIL-R1/2 have been shown to be up-regulated during infections with several other viruses (48,(55)(56)(57), and it has been proposed that this is an antiviral response to selectively kill infected cells or cells of the immune system. However, our data clearly indicate that in the case of influenza virus infection NF-B-dependent expression of TRAIL and the Fas/FasL system and subsequent induction of early apoptotic processes are proviral events. Thus, although the cell may aim to fully execute the apoptotic program to prevent spread of infection, the virus appears to take advantage of the early apoptotic events such as caspase activation. This is fully consistent with our earlier observation that activity of caspases, in particular caspase 3, is required for efficient virus propagation in infected cells that do not yet exhibit morphological changes connected with late apoptosis, such as membrane blebbing or DNA condensation (47). With respect to the step in the viral life cycle that is affected we observed that caspase activity can promote nucleocytoplasmic export of the viral ribonucleoprotein complexes, most likely After infection TRAIL-R2-Fc (10 g/ml) or TNF-R2-Fc (10 g/ml) as control was added to the medium. After 9 and 24 h p.i. supernatants were collected to determine virus titers. Mean virus titers from untreated cells were 1.5 ϫ 10 5 plaque-forming units/ml (9 h) or 3.6 ϫ 10 7 plaqueforming units/ml (24 h). B, A549 cell lines were infected with FPV (MOI ϭ 1). 6 h p.i. recombinant FLAG-tagged human TRAIL (huTRAIL) (10 ng/ml) was added to the infection medium. 9 h p.i. supernatants were collected and assayed to determine viral titers. pfu, plaque-forming unit. C, A549 cells were infected with FPV (MOI ϭ 1). The infection medium was supplemented with different concentrations of TRAIL-R2-Fc or CD95-Fc or both. 24 h p.i. supernatants were collected and plaque-titrated. D, A549 cell lines were infected with FPV (MOI ϭ 1). 6 h p.i. recombinant CD95 (50 ng/ml) was added to the infection medium. 9 h p.i. supernatants were collected and assayed to determine viral titers. E, A549 cell lines were infected with FPV or PR8 (both with a MOI of 1). Type I IFN receptor-neutralizing antibodies (IFNR-Ab) (20 g/ml) were added to the infection medium. 24 h p.i. supernatants were collected and titrated on MDCK cells. Each bar represents the average and standard deviation of three independent experiments. WT, wild type. through facilitating the passive transport process (47). Strikingly, this is consistent with an earlier finding, where overexpression of the antiviral protein Bcl-2 leads to reduced influenza virus titers and retention of the viral ribonucleoprotein in the nucleus (58).
It therefore seems that influenza virus has acquired the capability of taking advantage of the protection machinery of the host cells, thereby supporting viral replication. It is more efficient for a viral invader to take advantage of existing cellular activities rather than actively induce such processes in the host cell. In line with this assumption the virus needs mechanisms to keep a balance between limitation of the antiviral response and maintenance of sufficient signaling strength to support virus growth. Such a balance control may be provided by the viral NS1 protein, which keeps activities of certain transcription factors in a tolerated limit (24,45,59), thereby preventing an overflow of an antiviral response but still allowing some proviral proteins to be produced. Although NS1 may prevent dsRNA-induced activation of IRF-3, which mainly controls the type I IFN response, the protein will not prevent induction of NF-B by viral proteins (35). In support of such a balanced response we observed that recombinant TRAIL or recombinant CD95L only enhanced virus propagation if cells were stimulated late in infection in concentrations up to 10 -20 ng/ml or 50 ng/ml, respectively (Fig. 4, B and D). Earlier stimulation or higher concentrations of TRAIL or CD95L resulted in a loss of the supportive effects (data not shown). This late requirement of TRAIL correlates well with the expression kinetics of the protein (Fig. 3C) and the observed late requirement for caspase activity in the viral replication cycle (47). Furthermore, the recent finding that influenza A virus strains express a proapoptotic protein, PB1-F2, late in the virus life cycle is consistent with a requirement of apoptosis induction for efficient virus growth (60).
In summary, we have identified NF-B and some of its proapoptotic target genes as crucial cellular factors for influenza virus replication. This surprising dependence of influenza virus on NF-B activity may also have implications for future antiviral strategies.