NFκB Negatively Regulates Interferon-induced Gene Expression and Anti-influenza Activity*

Interferons (IFNs) are antiviral cytokines that selectively regulate gene expression through several signaling pathways including nuclear factor κB(NFκB). To investigate the specific role of NFκB in IFN signaling, we performed gene expression profiling after IFN treatment of embryonic fibroblasts derived from normal mice or mice with targeted deletion of NFκB p50 and p65 genes. Interestingly, several antiviral and immunomodulatory genes were induced higher by IFN in NFκB knock-out cells. Chromatin immunoprecipitation experiments demonstrated that NFκB was basally bound to the promoters of these genes, while IFN treatment resulted in the recruitment of STAT1 and STAT2 to these promoters. However, in NFκB knock-out cells IFN induced STAT binding as well as the binding of the IFN regulatory factor-1 (IRF1) to the IFN-stimulated gene (ISG) promoters. IRF1 binding closely correlated with enhanced gene induction. Moreover, NFκB suppressed both antiviral and immunomodulatory actions of IFN against influenza virus. Our results identify a novel negative regulatory role of NFκB in IFN-induced gene expression and biological activities and suggest that modulating NFκB activity may provide a new avenue for enhancing the therapeutic effectiveness of IFN.

Type I IFNs 2 (IFN␣, IFN␤, IFN, and IFN) are multifunctional cytokines that are critical in the host defense to infectious agents by modulating innate and adaptive immune responses. IFNs induce their biological effects by regulating the expression of a family of early response genes, called IFN-stimulated genes (ISGs), through JAK-mediated tyrosine-phosphorylation of the STAT factors, STAT1 and STAT2. The phosphorylated STAT proteins dimerize, translocate into the nucleus, and bind to the conserved IFN-stimulus response element (ISRE) within the promoters of ISGs (1). In addition, transcription factors of the IFN regulatory factor (IRF) family have been shown to regulate ISG expression (2). Accumulating evidence indicates that IFNs also activate the nuclear factor B (NFB) transcription factor in a serine/threonine kinase-dependent signaling pathway (3,4).
The mammalian NFB proteins, p50, p52, RelA (p65), RelB, and c-Rel, form homodimers and heterodimers to regulate the expression of genes involved in the immune response, inflammation and cell survival (5). In most cell types, the predominant form of NFB, the p50:p65 heterodimer, is bound to IB inhibitory proteins in the cytoplasm of unstimulated cells. Similar to various other stimuli, IFN␣/␤ activates a phosphatidylinositol 3-kinase/Akt pathway, which results in the dissociation of the inactive cytosolic NFB⅐IB complexes followed by NFB nuclear translocation and DNA binding (3,4). We previously demonstrated that mouse embryonic fibroblasts (MEFs) derived from mice with germ line deletions of the p50 and p65 genes were resistant to IFN-induced NFB activation and were sensitized to the antiviral action of IFN␤ against vesicular stomatitis virus (6). The present study was undertaken to further define the role of NFB in IFN-induced gene expression and the biological actions of IFN. Gene expression profiling identified 35 ISGs whose induction was highly regulated by NFB. A subset of genes that were induced higher by IFN in NFB knock-out (NFB-KO) cells encoded GTP-binding and antigen presentation proteins, which play critical roles in the antiviral and immunomodulatory activities of IFN, respectively. Quantitative RT-PCR demonstrated that these ISGs were induced more rapidly and at significantly lower IFN concentrations in NFB-KO cells relative to wild-type (WT) MEFs. Chromatin immunoprecipitation (ChIP) assays demonstrated that NFB dimers containing p50 and p65 were basally bound to the promoters of these ISGs. IFN induced the binding of STAT1 and STAT2 to these promoters. However, the kinetics of ISG induction in NFB-KO MEFs by IFN correlated with the promoter binding of IRF1. These results suggest that IFN induction of these genes is negatively regulated by NFB. We also found that NFB suppressed not only the direct antiviral action of IFN against influenza virus but also IFN-induced influenza-specific MHC class I antigen presentation. Together, our results suggest that NFB not only regulates the induction of a subset of IFN-induced genes but also the antiviral and immunomodulatory activities of IFN.
RNA Preparation and Microarray Analysis-Total cellular RNA from control and IFN␤-treated (2,500 units/ml for 5 h) WT or NFB-KO MEFs was extracted with TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Approximately 10 g of RNA was submitted to Genome Explorations Inc. (Memphis, TN) for labeling and hybridization to murine U74Av2 GeneChips (Affymetrix Inc.) according to the manufacturer's protocols. Expression values were determined using Affymetrix Microarray Suite (MAS) 5.0 software. All data analysis was performed using GeneSpring software 7.0 (Silicon Genetics, Inc.). The MAS 5.0 gene expression values for each gene were normalized as described previously (6). Fold induction by IFN was calculated in matched pairs of WT and NFB-KO MEFs, respectively. The average of fold induction from three independent sets of GeneChip data for both WT and NFB-KO MEFs was subjected to non-parametric t test. The difference of fold induction for each gene by IFN was calculated by subtracting fold induction in WT MEFs from fold induction in NFB-KO MEFs.
Reverse transcription was performed at 48°C for 45 min, and RT-PCR cycling parameters were as follows: denaturation at 95°C for 2 min, amplification at 94°C for 30 s, and 62°C for 30 s for 35 cycles. The product size was initially monitored by agarose gel electrophoresis and melting curves were analyzed to control for specificity of PCR reactions. The data on IFN-induced genes was normalized to the expression of the housekeeping gene ␤-actin. The relative units were calculated from a standard curve, plotting three different concentrations against the PCR cycle number at the cycle threshold (with a 10-fold increment equivalent to ϳ3.1 cycles).
Immunoblotting-At various times after IFN␤ treatment (1,000 units/ml), MEFs were lysed directly in radioimmunoprecipitation assay buffer (whole cell lysate), or nuclear extracts were prepared with the NE-PER kit (Pierce). Equivalent amounts of protein were subjected to SDS-PAGE. Proteins were transferred to nitrocellulose membranes, immunoblotted for Tap1, Mx1, ␤-actin, or TFIIB and visualized by enhanced chemiluminescence (Pierce).
Antiviral Assay-To determine the cellular sensitivity to the ability of IFN to reduce influenza A virus titer, cell cultures were preincubated overnight with IFN, followed by infection with influenza A/Puerto Rico/ 8/34 (PR8) virus for 1 h at 1 plaque-forming unit per cell. At 24 h  post-infection, the virus yield in the medium was assayed by plaque formation on Madin-Darby canine kidney cells as described previously (9). Influenza-specific Immune Assays-To assay for MHC class I functional expression, 5 ϫ 10 3 WT or NFB MEF cells were co-cultured with 1 ϫ 10 5 4-39 hybridoma cells in individual wells of 96-well plates for 24 h in a 1:1 mixture of DMEM and Complete tumor medium containing 0.1 g/ml D b NP-366 peptide and IFN␤. The 4-39 hybridoma line reacts positively by IL-2 production with the immunodominant D b NP-366 peptide but negatively with D b PA-224, K b PB1-703, K b NS2-114, and D b PB1-F2-62 peptides (10,11). Alternatively, to assay for MHC class I antigen processing and presentation following infection with influenza virus, 5 ϫ 10 3 MEF cells were cultured in individual wells of a 96-well plate in DMEM overnight in the presence or absence of IFN␤. Then MEF cells were infected with PR8 influenza A virus at 1 plaque-forming unit/cell in serum-free DMEM for 1 h, followed by co-culture with 4-39 hybridoma cells (1 ϫ 10 5 ) without or with IFN␤ in DMEM plus Complete tumor medium (1:1) for 24 h. IL-2 production, which in this assay correlates directly to the MHC class I expression on the MEF cell surface, was measured by enzyme-linked immunosorbent assay using purified anti-IL-2 and biotin-anti-IL-2 (Pharmingen) antibodies following the manufacturer's directions.

Identification of ISGs That
Are NFB-dependent-We previously reported that NFB plays a role in regulating ISG expression (6). Subsequently, a functional genomics approach was used to monitor the role of NFB in gene expression changes induced by IFN. For these experiments, we used fibroblasts derived from mice with germ line deletions of the p50 and p65 genes. These MEFs were shown to be resistant to IFN-induced NFB activation (6) and hence can be considered as NFB-KO. In contrast, MEFs expressing p50 and p65 genes were IFNresponsive and considered WT. In three independent experiments, WT and NFB-KO MEFs were treated in the presence or absence of IFN␤ (2,500 units/ml for 5 h), and total RNA was extracted, labeled, and hybridized to murine U74Av2 GeneChips (Affymetrix). Control cultures received no IFN but were treated in parallel. Expression values were determined using Affymetrix Microarray Suite (MAS) 5.0 software, and the data were filtered and analyzed using GeneSpring software 7.0 (Silicon Genetics, Inc.) as described previously (6).
After data processing, a total of 7,412 probe sets were used to compare gene expression between MEFs differing with respect to genotype (NFB-KO and WT) and treatment (control and IFN treatment). Parametric two-way analysis of variance (p Ͻ 0.05, n ϭ 3 for each group) identified 1,375 probe sets whose expression was different between any two of four groups (control WT MEFs, IFN-treated WT MEFs, control NFB-KO MEFs, IFN-treated NFB-KO MEFs). The induction of 35 genes by IFN was significantly (p Ͻ 0.05, non-parametric t test, n ϭ 3 for each group) different between WT and NFB-KO cells. Functional classification of NFB-regulated ISGs using Expression Analysis Systematic Explorer analysis (12) revealed that genes encoding GTP-binding and antigen presentation proteins were significantly overrepresented (variant one-tailed Fisher exact probability test, Expression Analysis System- atic Explorer scores Ͻ 0.01). The NFB-regulated GTP-binding ISGs included the 65-67-kDa guanylate-binding proteins (Gbp1 and Gbp2), the Mx proteins (Mx1 and Mx2), and the 47-kDa GTPase Ifi47 (also called Irg47) (13). GBPs play important roles in resistance to viruses (14 -16), as well as intracellular protozoa and bacteria (13). On the other hand, the NFB-regulated antigen presentation ISGs (Tap1, Tap2, Psmb9/Lmp2, and Psmb8/Lmp7) are involved in degrading intracellular proteins into antigenic peptides and contribute to the transport of these peptides to endoplasmic reticulum where they bind to the assembled MHC class I molecules (17). As shown in Table 1, Gbp1, Ifi47, Mx1, Mx2, Tap1, Psmb9/Lmp2, and Psmb8/Lmp7 were induced by IFN to a greater extent in NFB-KO MEFs than in WT MEFs, while Gbp2 and Tap2 were induced more in WT MEFs than in NFB-KO MEFs. These findings suggested that NFB regulates the expression of a subset of ISGs that play important roles in antiviral and immune responses.
Dose Dependence and Kinetics of NFB-dependent ISGs-To further investigate the induction of ISGs that were negatively regulated by NFB, quantitative real-time PCR assays were performed for Ifi47, Tap1, and Mx1 using RNA from WT and NFB-KO MEFs treated with IFN␤. Consistent with the microarray results, Mx1, Ifi47, and Tap1 ( Fig.  1) were induced by IFN␤ in a dose-dependent manner to higher expression levels in NFB-KO cells than in WT cells. Similarly, we showed previously that Mx1 and Nmi were also induced more in NFB-KO cells than in WT cells by IFN in a dose-dependent manner (6).
We also examined the time course of IFN induction of these genes. Ifi47, Tap1, and Mx1 ( Fig. 2A) were induced more rapidly in NFB-KO cells compared with WT cells. For example in WT MEFs, Mx1 was induced by 2 h of IFN treatment and reached a maximal induction level of ϳ15-fold after 4 h. In sharp contrast, Mx1 was induced ϳ10-fold after 30 min and ϳ220-fold after 4 h of IFN treatment in NFB-KO cells. Moreover, the changes in the mRNA levels were accompanied by changes in the protein levels for Tap1 and Mx1 proteins (Fig. 2B). The effect of NFB on regulation of these ISGs is specific, since the expression profile of another ISG, IRF1, was not regulated by NFB ( Fig. 2A). In addition, the effects of NFB on IFN-induced gene expression did not reflect altered JAK-STAT signaling, since STAT2 activation was similar in WT and NFB-KO MEFs (Fig. 2B). These results suggested that NFB selectively decreased and delayed the transcription of a subset of ISGs.
The Binding of p50 and p65 to the Promoters of NFB-dependent ISGs-To investigate which NFB proteins regulated the transcription of Ifi47, Tap1, and Mx1, we performed ChIP assays. Although STAT proteins were not basally bound to the promoters of Ifi47, Tap1, and Mx1, IFN treatment induced the binding of STAT1 and STAT2 to these promoters in WT and NFB-KO MEFs between 15 and 60 min of IFN treatment (Fig. 3, A-C). This result was expected, since the binding of these STAT proteins is necessary for IFN-induced activation of ISG transcription. In contrast to STAT proteins, the p50 NFB protein was basally bound to the promoter of these ISGs in WT MEFs, while p65 was basally bound to Ifi47 and Tap1 promoters. Since the genes for p50 and p65 were both deleted in NFB-KO mice, neither protein was found bound to the promoter of these ISGs in NFB-KO MEFs. Interestingly, in NFB-KO MEFs IFN treatment resulted in the recruitment of IRF1 to the promoters of all three ISGs, and IRF1 recruitment closely correlated with the rapid and enhanced induction of the genes upon IFN treatment. Although IRF1 itself is an ISG, we found that IRF1 expression was not NFB-regulated; IFN induction of IRF1 was similar in WT and NFB-KO MEFs as determined by microarray analysis (ϳ6-fold) or quantitative real-time PCR (Fig. 2A). Taken together these results demonstrate that NFB directly binds to the promoters of a subset of ISGs and thereby inhibits their induction by IFN, perhaps by inhibiting recruitment of IRF1 to their promoters. p50 was bound to the promoters of all three ISG promoters and correlated with diminished ISG expression. Therefore, we examined the effect of the p50 protein alone on the induction of NFB-regulated ISGs in MEFs derived from p50 knock-out mice. As shown in Fig. 1, Ifi47, Tap1, and Mx1 were induced by IFN␤ to higher expression levels in p50-KO than in either NFB-KO cells or in WT cells. These results suggest that p50 is the predominant negative regulator of these ISGs.
NFB Dampens the Antiviral and Immunomodulatory Activity of IFN against the Influenza Virus-The NFB-regulated ISG Mx1 has selective antiviral activity in mice against the influenza virus (16). Therefore, we investigated whether NFB modulated the anti-influenza activity of IFN. WT and NFB-KO MEFs were infected with the PR8 strain of influenza A in the absence or presence of different concentrations of IFN, and viral replication was measured after 24 h. As shown in Fig. 4, IFN (100 units/ml) results in only a 10-fold reduction in virus titer in WT MEFs. However, a 100-fold lower IFN concentration (1 unit/ml) was sufficient to induce equivalent antiviral effects in NFB-KO as compared with WT MEFs.
The NFB-regulated ISG Tap1 is required for antigen presentation of an immunodominant influenza viral epitope (18). Since IFN also modulates this facet of the host immune response, we next determined the effect of IFN on influenza-specific MHC class I antigen presentation in WT and NFB-KO MEFs. Using a T cell hybridoma specific for the immunodominant D b NP-366 influenza peptide as responder cells, we determined that antigen presentation levels were significantly enhanced by IFN treatment in NFB-KO MEFs as compared with WT cells (Fig.  5). Taken together, these results indicate that NFB dampens several functionally important IFN activities.

DISCUSSION
The purpose of the present study was to characterize the role of NFB in the biological actions of IFN using a functional genomic approach. We show that NFB negatively regulates the induction of a subset of ISGs by IFN and that NFB suppresses the antiviral and immunomodulatory activities of IFN in the context of influenza virus infection. Both IFN and NFB play critical roles in the host defense to various pathogens. However, the role of NFB in the biological functions of IFN is relatively undefined. We previously found that the deficiency of p65 and p50 NFB subunits sensitized MEFs to antiviral action of type I IFNs against vesicular stomatitis virus (6). Indeed, NFB increases the susceptibility of human cells to influenza virus infection (19). Our findings provide a possible molecular mechanism for this observation whereby p50-containing NFB dimers basally inhibit a subset of IFN antiviral genes via binding directly to their promoters.
The role of the JAK-STAT pathway in gene induction and in the generation of many of the biological effects of IFN is firmly established. However, it is now apparent that the activation of the JAK-STAT pathway alone is insufficient to account for all of the biological actions of IFNs. Some of these pathways operate independently of the JAK-STAT pathway, while other pathways may cooperate with STATs to regulate the transcription of target genes. For example, although the role of IRF9 together with the STAT1/STAT2 dimer in the transcriptional regulation of ISGs is well described, other IRF proteins can also act as transcriptional activators and/or repressors of IFN-regulated genes (2). In addition, the p300/CBP co-activators can interact with STAT proteins to regulate ISG transcription (20,21).
We show that, although NFB is basally bound to ISG promoters, the IFN-induced binding of STAT1 and STAT2 to these promoters was unimpeded. Therefore, NFB does not negatively regulate ISG induction through affecting STAT binding to ISG promoters. However, NFB proteins appear to modulate IRF1 promoter binding. IRF1 is also an IFN-induced transcriptional activator of ISG expression (22). We show that in NFB-KO MEFs, but not in WT MEFs, IRF1 is recruited to ISG promoters upon IFN treatment, and this recruitment closely correlates with the more rapid and enhanced ISG expression in NFB-KO cells. Moreover, other co-activators of ISG expression (i.e. p300/CBP) are also bound to ISG promoters in the absence of NFB proteins. 3 Therefore, NFB may negatively regulate ISG expression by directly or indirectly inhibiting the promoter binding of co-activators of ISG expression.
The notion that NFB suppresses transcription is contrary to its classical role of stimulating the transcription of target genes. However, several recent studies indicate that NFB may down-regulate cellular responsiveness to specific cytokines. For example, tumor necrosis factor-mediated JNK signaling is enhanced in NFB-KO mice (23,24). Moreover, we previously showed that the IFN-induced expression of some ISGs (Mx1 and Nmi) is negatively regulated by NFB, while the expression of other ISGs, such as Ifit1 and Isg15, is enhanced by NFB (6). In the present study, we show that NFB has opposing effects on different ISGs that map to a single genetic locus. For example, the induction by IFN of Gbp1 and Gbp2, which are located adjacent to one another on mouse chromosome 6, are negatively and positively regulated by NFB, respectively (Table 1). Similarly, the induction of Tap1 and Tap2, which are adjacent to one another on mouse chromosome 17, is also affected in opposite directions by NFB (Table 1). These results suggest that NFB may act as a transcriptional switch to fine-tune the expression of neighboring ISG family members in a locus. It will be important to determine the cellular consequences of this regulatory pathway with regard to the specific functions of Gbp and Tap family proteins.
The ability to integrate multiple signaling pathways to achieve unique responses is a critical requirement for development and homeostasis in all metazoans (25). In this study, we have characterized the relationship of the classical JAK-STAT pathway to NFB pathway in the transcriptional regulation of Ifi47, Tap1, and Mx1. Moreover, we demonstrate that NFB regulates the antiviral activity of IFN and the promotion of T cell activation. Type I IFNs are broadly used in clinical treatment of viral infections, multiple sclerosis, and cancer. We show that the efficacy of IFN as an antiviral agent against influenza virus is significantly enhanced in NFB-KO MEFs. Interestingly, NFB inhibitors are currently used to treat inflammatory diseases and autoimmune disease. Thus, we propose that the combination of IFN and NFB inhibitors may be used to maximize the therapeutic benefit of IFN in the treatment of human disease while minimizing the dosage and possibly its undesirable side effects.