RNA virus infection results in expression of type 1 IFNs, especially IFN-α/β, which play a crucial role in host antivirus responses. Type 1 IFNs are induced in a cell type-specific manner through TLR and RIG-I–like receptor pathways, both of which activate IFN regulatory factors (IRFs) and NF-κB transcription factors. Although NF-κB activation and association with the IFN-β promoter after RNA virus infection is well documented, our previous work showed that, surprisingly, NF-κB is not essential for IFN-β gene expression. Thus, the actual function of NF-κB in IFN-β expression and virus replication is not clear. In this study, we found Newcastle disease virus and vesicular stomatitis virus replication is enhanced in mouse embryonic fibroblasts (MEFs) lacking the NF-κB RelA subunit. Increased virus replication was traced to a specific requirement for RelA in early virus-induced IFN-β expression. At these time points, when IFN-β expression is ~100-fold less than peak levels, impaired IFN-β production delayed IFN-induced gene expression, resulting in increased virus replication in RelA−/− MEFs. Importantly, our results show that RelA requirement is crucial only when IRF3 activation is low. Thus, high levels of activated IRF3 expression are sufficient for induction of IFN-β in RelA−/− MEFs, transcriptional synergism with the coactivator CREB-binding protein, and rescue of susceptibility to virus. Together, these findings indicate that NF-κB RelA is not crucial for regulating overall IFN-β production, as previously believed; instead, RelA is specifically required only during a key early phase after virus infection, which substantially impacts the host response to virus infection.
Type I IFNs, IFN-α and IFN-β, are essential for limiting virus replication and promoting clearance by inducing antivirus gene expression and modulating virtually every aspect of innate and adaptive immunity (1, 2). IFN-α/β binds to type I IFN-α receptors (IFNAR1 and IFNAR2) and signal through receptor-bound Janus protein tyrosine kinases and STAT. Activated STAT1/STAT2 associate with IFN regulatory factor (IRF)9 to form IFN-stimulatory gene factor 3, which binds to IFN-stimulated response elements and upregulates IFN-stimulated gene (ISG) expression (3).
IFN-α/β expression can be induced by viruses through endosomal membrane-bound TLRs, including TLR3, TLR7/8, and TLR9 (4, 5). Through MyD88 or TRIF adaptors, TLRs activate the kinases TBK1 and inducible IκB kinase (6–8). These kinases phosphorylate and activate IRF3 and IRF7, which are crucial for inducing IFN-α/β (6, 7). IRF3 is expressed constitutively and contributes to IFN-β expression following activation-induced dimer formation (9); IRF7 expression is induced by virus infection through IFN feed-forward signaling and is essential for optimal IFN-β and IFN-α expression (9–11). The RNA helicases RIG-I and MDA5 are RIG-I–like receptors that recognize the cytoplasmic presence of RNA viruses (12–16). RIG-I–like receptors signal through mitochondrial-bound IPS-1 (also called VISA, MAVS, or Cardif) to activate TBK1/inducible IκB kinase, resulting in IRF3 and IRF7 activation (17–20).
Previous studies documented four transcription factor-binding sites, called positive regulatory domains (PRD-I to PRD-IV), in the IFN-β promoter (21–23). PRD-I/III binds IRF3/IRF7, PRD-II binds NF-κB, and PRD-IV binds ATF-2/c-Jun, which together form the IFN-β enhanceosome, an essential component for virus-induced IFN-β transcription (22, 24). The mammalian NF-κB family contains RelA, cRel, RelB, p50, and p52, which form homodimers or heterodimers (25, 26). NF-κB dimers are retained in the cytoplasm by IκBs, which are subject to IκB kinase (IKK)-mediated phosphorylation under stimulation, resulting in degradation of IκBs and translocation of NF-κB into the nucleus (25, 26). The crucial roles for IRF3 and IRF7 in IFN-β expression were confirmed in mouse knockout studies (9, 10, 27). NF-κB was similarly implicated in IFN-β expression (22, 24, 28–30). Interestingly, RelA association with the IFN-β promoter occurs through specific interchromosomal interactions (31). However, our previous studies showed that Sendai virus and Newcastle disease virus (NDV) infection induced robust IFN-α/β expression in RelA−/−, p50−/−, cRel−/−, p50−/−cRel−/−, or p50−/−RelA−/− mouse embryonic fibroblasts (MEFs) and RelA−/− or p50−/−cRel−/− dendritic cells (DCs), which demonstrated the lack of an essential role for NF-κB in virus-induced IFN-β expression (32). Therefore, the potential role for NF-κB, if any, in IFN-β expression and the host-mediated control of virus replication is unclear. The findings reported in this article demonstrate that NF-κB function is limited to a key early phase after virus infection when IRF3 activation is minimal. Thus, although NF-κB may have little impact on the overall magnitude of IFN-β production, it plays a crucial role in the early phase of type I IFN production and the subsequent expression of ISGs, thereby restraining virus replication.
Materials and Methods
Mice and materials
RelA+/+ and RelA−/− chimera mice were generated as described (33). IKKβ+/− mice were kindly provided by Dr. Zhiwei Li (Moffitt Cancer Center), and IFNAR−/− mice were kindly provided by Dr. Esteban Celis (Moffitt Cancer Center). All mice were maintained under specific pathogen-free conditions, and all experiments using mice were carried out in accordance with institutional guidelines.
Primary RelA+/+, RelA−/−, RelA+/+IFNAR−/−, RelA−/−IFNAR−/−, p50−/−, and cRel−/− MEFs were prepared from day-14.5 embryos, and primary IKKβ+/+ and IKKβ−/− MEFs were prepared from day-12.5 embryos. Biological replicates in the figure legends refer to MEFs isolated from different embryos and infected with virus in independent experiments. MEFs were cultured in DMEM with 10% donor calf serum, penicillin-streptomycin, and glutamine. Wild-type (WT) and IRF3−/− immortalized MEFs were as described (9) and were kindly provided by Dr. Tadatsugu Taniguchi (University of Tokyo, Tokyo, Japan) and Dr. Karen Mossman (McMaster University, Hamilton, Ontario, Canada). A total of 1 × 106 primary MEFs or 0.4 × 106 WT and IRF3−/−-immortalized MEFs were plated in 60-mm cell-culture dishes and cultured overnight before being used experiments. Bone marrow-derived DCs (BMDCs) were cultured as previously described (33).
NDV and vesicular stomatitis virus infection of cells
MEFs and BMDCs were infected with a recombinant variant of NDV that expresses GFP (NDV-GFP) at multiplicity of infection (MOI) of 5 and 1, respectively, as described (32). For MEFs, time points start after the 1-h virus incubation, whereas for BMDCs, time points start from the beginning of the 1-h virus incubation. For vesicular stomatitis virus (VSV) experiments, 1 × 105 MEFs per well were seeded in 12-well plates and infected with VSV. Cell supernatant was harvested 24 h postinfection and titrated on BHK cells by a standard plaque assay.
Chromatin immunoprecipitation, real-time RT-PCR, Western blotting, and native gel Western blotting
Chromatin immunoprecipitation (ChIP), real-time PCR, and Western blotting were performed as described previously (32, 33). For ChIP, the real-time PCR primers were 5′-GCCAGGAGCTTGAATAAAATG-3′ (ChIPIFNbS) and 5′-CTGTCAAAGGCTGCAGTGAG-3′ (ChIPIFNbAS). The primers for IFN-β, IFN-α, IRF7, and β-actin mRNA were as previously described (32). The primers for detecting NDV nucleocapsid (NDVnuc) mRNA were as described (34). Other primers used were 5′-GAAACTTCATTCAAACCCGGCCCA-3′ (oas2-F) and 5′-CCGGAAGCCTTCAGCAATGTCAAA-3′ (oas2-R). Native gel Western blotting was performed as described (35). Briefly, whole-cell extracts were prepared as described (33). A 7.5% native separating gel (5 ml H2O, 2.5 ml 1M Tris-Cl [pH 8.8], 2.5 ml 30% acrylamide-bis, ammonium persulphate, tetramethylethylenediamine) was prerun with native running buffer (25 mM Tris and 192 mM glycine, with and without 1% sodium deoxycholate in the cathode and anode chamber, respectively) for 30 min at 40 mA. Samples were mixed with 2× native sample buffer (125 mM Tris-Cl [pH 6.8], 30% glycerol, 2% sodium deoxycholate) and loaded on the gel. The gel was run at 25 mA for 1 h, followed by standard Western blotting procedures. Rabbit normal IgG and anti-CBP Ab were purchased from Upstate Biotechnology (Lake Placid, NY). Anti-IRF3 Ab was purchased from Zymed (San Francisco, CA). Anti-phosphorylated STAT1, anti-STAT1, anti-RelA, and anti–β-actin Abs were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
Uninfected cells and NDV-infected cells were collected, and flow-cytometric analysis was performed on a FACSCalibur cytometer (BD Biosciences, San Jose, CA); data were acquired using CellQuest software and analyzed using FlowJo software (Tree Star, Ashland, OR).
Transfection and reporter assay
Transfection of HEK 293T cells was performed as described previously (33). RelA and CBP-expressing vectors have been described (33). Superactive (SA) mutant of IRF-3 (S396D) was obtained from InvivoGen (San Diego, CA), cloned into the monokine induced by IFN-γ (MIG) retroviral vector, and confirmed by DNA sequencing. Different amounts of IRF3SA-, 10 ng RelA-, and 500 ng CBP-expressing plasmids in different combinations were transfected with 10 ng IFN-β luciferase vector and 0.2 ng pRL-TK. The total DNA amount was equalized with MIG vector. pRL-TK was used as control and for normalization. Dual-luciferase (Promega, Madison, WI) was measured 48 h later, according to the manufacturer’s suggestions. All transfections were done in duplicate and repeated twice.
Supernatants from cultures of mock-infected or NDV-infected MEFs were collected and frozen until assayed. Mouse IFN-α and IFN-β ELISA kits were purchased from PBL Biomedical Laboratories (Piscataway, NJ), and the assay was carried out according to the manufacturer’s instructions.
Generation of MIG- or IRF3SA-expressing fibroblasts
Immortalized RelA−/− fibroblasts were spin-infected using retroviral supernatant produced by HEK 293T cells transfected with 6 μg MIG- or IRF3SA-expressing MIG retroviral vectors and 4 μg helper vector pCL-Eco. Infected cells were FACS sorted 2 d later based on GFP fluorescence.
All statistical analyses were performed using the Student t test; a p value <0.05 was considered significant.
Increased NDV and VSV replication in MEFs lacking IKKβ or NF-κB RelA subunit
To investigate the role of NF-κB in RNA virus replication, we used the ssRNA virus NDV. NDV-GFP (36) was used to monitor virus replication by flow cytometry. Although NDV can productively infect MEFs, infectious virus is not generated in the absence of exogenous trypsin, allowing examination of virus replication during a single infection cycle. IKKβ plays a key role in activating NF-κB complexes consisting of p50+RelA subunits (37), which are predominant in MEFs. We infected WT and IKKβ−/− MEFs with NDV-GFP and determined GFP levels by flow cytometry 9 h postinfection. In one representative experiment, ~67% of WT MEFs were GFP+, with a mean fluorescence intensity (MFI) of 322.01 (Fig. 1A). Although slightly more IKKβ−/− MEFs were GFP+ (78%), their MFI was substantially greater than that of WT GFP+ MEFs (628.76) (Fig. 1A). This experiment was repeated four times, with IKKβ−/− MEFs showing a statistically significantly greater percentage of GFP+ cells and MFI than WT cells (Fig. 1A). Thus, greater NDV protein synthesis apparently takes place in infected IKKβ−/− MEFs than in WT MEFs. Similar results were obtained at an MOI of 2 or 10 and at 24 h postinfection (data not shown). We then used real-time PCR to detect NDVnuc mRNA levels. NDVnuc mRNA levels in WT MEFs increased dramatically as early as 3 h postinfection and continued to increase until 15 h postinfection (Fig. 1B). NDVnuc mRNA levels were similar between WT and IKKβ−/− MEFs at early time points (0–6 h), suggesting comparable initial NDV infection. However, at later time points, NDVnuc mRNA levels in IKKβ−/− MEFs were substantially greater than those in WT MEFs (Fig. 1B). These results suggest that IKKβ is important for controlling NDV replication. Virtually identical results were obtained using RelA−/− MEFs (Fig. 1C, 1D).
To test whether other NF-κB subunits, such as p50 and cRel, play a similar role in antivirus responses, we infected WT, p50−/−, and cRel−/− MEFs with NDV and assessed NDVnuc mRNA levels by real-time PCR. In contrast to the dramatic and sustained higher NDVnuc mRNA levels in RelA−/− MEFs, the levels were similar between WT and p50−/− MEFs and were slightly higher at 9 h postinfection in cRel−/− MEFs (Fig. 1E). These results suggest that, compared with the important role of RelA, p50 is dispensable and cRel plays a relatively minor role in inhibiting virus replication. Thus, IKKβ-mediated RelA activation is primarily required for limiting NDV replication in MEFs. Therefore, subsequent studies used RelA−/− MEFs.
To ensure that the above results were not NDV specific, we examined another RNA virus (VSV) (38). Importantly, because infectious VSV can be generated following infection of MEFs, we determined the amount of virus produced. WT and RelA−/− MEFs were infected with VSV at an MOI of 0.001 for 24 h, and infectious viral progeny titers in supernatants were measured. RelA−/− MEFs produced ~40-fold higher viral progeny than did WT MEFs (p = 0.0027; Fig. 1F), indicating that much greater VSV replication takes place in RelA−/− MEFs than in WT MEFs.
RelA enhances early IFN-β expression induced by NDV and VSV infection
IFN-α/β inhibit virus replication through induction of ISGs (1, 2). Although a crucial role for NF-κB in IFN-β regulation has been proposed, our previous studies showed that RelA (or p50 and cRel) is not essential for virus-induced IFN-β expression (32). In light of the above findings showing increased virus replication in RelA−/− MEFs, we performed a detailed kinetic examination of IFN-β expression after NDV infection. Interestingly, reduced IFN-β mRNA was noticed in RelA−/− MEFs compared with WT MEFs at early time points (Fig. 2A); IFN-β expression in WT MEFs at these early time points was ~100-fold lower than at the peak response at 12–15 h (Fig. 2B). Interestingly, RelA−/− MEFs typically showed slightly greater IFN-β expression at later time points, possibly as a result of increased virus replication (see later discussion). Consistent with mRNA level, IFN-β protein expression was substantially reduced at early time points (9 h) compared with later time points (18 h) (Fig. 2C). Consistent with the slightly greater NDVnuc mRNA levels in cRel−/− MEFs, we detected moderately impaired IFN-β mRNA expression soon after NDV infection (Fig. 2D).
Expression of IFN-α (non4), IRF-7, and other ISGs, such as 2′-5′ oligoadenylate synthase 2 (oas2), is dependent on initial IFN-β expression. Therefore, expression of these genes can help to determine the functional consequence of reduction in initial IFN-β expression. Importantly, expression of IFN-α (non4), IRF-7, and oas2 mRNA was substantially reduced in RelA−/− MEFs at early time points (Fig. 3A; results of experiment 2 are shown) but not at late time points (Fig. 3B; results of experiment 2 are shown). Similar to IFN-β, IFN-α protein expression was also reduced in RelA−/− MEFs at early but not late time points (Fig. 3C). Similar consequences of the absence of IKKβ were noticed on expression of IFN-β (Supplemental Fig. 1). Therefore, these findings help to explain why NDV replication is increased in RelA−/− MEFs. They may also explain differences in type 1 IFN expression in RelA−/− MEFs at early versus late time points. Thus, initially low expression of IFN-β in RelA−/− MEFs results in reduced ISG expression causing increased virus replication, which activates RIG-I and induces IFN expression (39), leading to greater IFN expression at later time points. Therefore, although RelA is not essential for overall IFN-β expression, it plays a crucial role in early IFN-β expression following virus infection.
To ensure that reduced IFN-β expression and increased NDV replication were due to RelA absence, we re-expressed RelA in immortalized RelA−/− MEFs by retroviral transduction. These studies showed that control MIG retrovirus-transduced RelA−/− MEFs expressed more NDVnuc mRNA at later time points (Fig. 4A) and less IFN-β early postinfection (3-h time point) than did RelA-expressing RelA−/− MEFs (Fig. 4B, 4C). Thus, RelA expression can rescue the phenotype of RelA−/− MEFs.
We next determined whether RelA plays a similar role in IFN-β expression after VSV infection. WT and RelA−/− MEFs were infected with VSV for different periods, and mRNA expression of IFN-β and IFN-α (non4) was determined. Compared with WT MEFs, the expression of IFN-β and IFN-α (non4) in RelA−/− MEFs was dramatically impaired (10-fold lower) at early time points (Fig. 5A). However, similar to NDV infection, RelA−/− MEFs expressed greater IFN-β and IFN-α (non4) than did WT MEFs at later time points (20 h). As shown in Fig. 5B, the expression of ISGs, including IRF7 and oas2, was very low at early time points. Consistent with reduced IFN-β expression, expression of these ISGs was much lower in RelA−/− MEFs than in WT MEFs at 10 h postinfection. Together, these results indicate that RelA is essential for the expression of early IFN-β and ISGs after VSV infection.
Essential role of RelA in early IFN-β expression in conventional DCs
Conventional dendritic cells (cDCs) are professional APCs (40) that play a crucial role in the host antivirus responses. We used mouse BMDCs from WT and RelA−/− fetal liver cell adoptive-transfer mice as cDCs for NDV-infection studies. RelA−/− cDCs expressed similar levels of NDVnuc mRNA as did WT DCs 2 h after NDV infection; however, RelA−/− DCs expressed substantially more NDVnuc mRNA at later time points (Fig. 6A). IFN-β expression in WT cDCs was rapidly enhanced, peaking at 7 h postinfection (Fig. 6B) and decreasing thereafter (data not shown). Compared with WT DCs, IFN-β induction in RelA−/− DCs was reduced early (2 h) after NDV infection (Fig. 6B, 6C). Compared with MEFs (Fig. 2A), NDV induces much faster IFN-β expression in DCs. Thus, the seemingly subtle, but statistically significant, difference in early IFN-β expression between WT and RelA−/− DCs causes greater NDV replication in RelA−/− DCs at late time points of infection. These results suggest that in addition to MEFs, RelA is important for early IFN-β expression and for controlling NDV replication in cDCs.
Absence of RelA does not impact IFN-induced inhibition of virus infection and replication
The decreased expression of ISGs and increased virus replication in RelA−/− MEFs may be the result of impaired IFN-induced responses; a previous study implicated such a link (41). Thus, RelA may be required for IFN-induced responses that inhibit virus replication. To test this possibility, different amounts of IFN-β were used to pretreat WT and RelA−/− MEFs, after which they were infected with NDV. With increasing amounts of IFN-β, NDV replication was similarly inhibited in WT and RelA−/− MEFs, as reflected by a lower percentage of GFP+ cells and lower MFI of GFP+ cells (Fig. 7A). Consistent with reduced GFP+ cells, IFN-β pretreatment inhibited NDVnuc mRNA expression similarly in WT and RelA−/− MEFs (Supplemental Fig. 2A). Furthermore, IFN-β pretreatment increased IFN-α (non4) expression by NDV infection similarly in WT and RelA−/− MEFs (Supplemental Fig. 2B). Next, we determined STAT1 phosphorylation in response to IFN-β treatment. As shown in Fig. 7B, STAT1 phosphorylation was similarly induced in WT and RelA−/− MEFs. Therefore, these results suggest that increased virus replication in RelA−/− MEFs is more likely due to the decrease in IFN-β expression than to impairment in IFN-β–induced antivirus responses.
An additional possibility is that increased virus replication in RelA−/− MEFs involves a pathway distinct from the type 1 IFN pathway. IFN-α/β signal through IFNAR (3). To explore a possible type I IFN-independent role for RelA−/−, we generated RelA−/−IFNAR−/− mice, from which MEFs were obtained. A more severe phenotype in RelA−/−IFNAR−/− MEFs compared with RelA+/+IFNAR−/− MEFs would suggest that RelA may have a type I IFN-independent role. However, NDV replication determined by GFP and NDVnuc expression was similarly increased in IFNAR−/− and RelA−/−IFNAR−/− MEFs (Fig. 7C, 7D). Although not conclusive, these results suggest IFN signaling requirement for RelA control of virus replication.
RelA requirement correlates with IRF3 activation levels after virus infection
The IFN-β promoter has four conserved PRDs (PRDI–IV), which bind to transcription factors NF-κB, IRF3/IRF7, and AP-1 (ATF2/c-Jun) (21, 22, 24). Based on knockout mouse studies, IRF3 plays an especially important role in virus-induced IFN-β expression (9). In light of our findings, we hypothesized a possible role for the kinetics of IRF3 activation in determining RelA requirement for IFN-β expression at early, but not late, time points after virus infection. Interestingly, IRF3 activation determined through dimer formation was undetectable up to 6 h after NDV infection (Fig. 8A). These results suggest that RelA is crucial in the absence of IRF3 activation. Conversely, it is possible that IRF3 is activated and crucial for IFN-β expression, but activated IRF3 levels are too low for detection. To distinguish between these possibilities, we used immortalized IRF3−/− MEFs. As shown in Fig. 8B, IFN-β expression induced by NDV infection was severely impaired in the absence of IRF3 at early and late time points after NDV infection. Correlated with diminished IFN-β expression, NDV replication determined by GFP and NDVnuc expression was much greater in IRF3−/− cells than in WT fibroblasts (Fig. 8C, 8D). Thus, IRF3 is a substantially more important transcription factor than RelA in the overall expression of virus-induced IFN-β. Furthermore, IRF3 is activated at early time points after NDV infection, but activation levels are below detection limits using a dimerization assay.
Our results suggest that RelA requirement for IFN-β expression is evident at time points when IRF3 activation is weak but not when IRF3 activation is strong (Figs. 2A, 8A). To determine whether high levels of activated IRF3 can induce IFN-β expression in the absence of RelA, we expressed a constitutively active mutant of IRF3 (S396D amino acid change; IRF3SA) in RelA−/− MEFs. Remarkably, IRF3SA expression was sufficient for induction of IFN-β expression in RelA−/− MEFs, and it almost completely inhibited NDV infection (Fig. 9A). Thus, strong IRF3 activation can induce IFN-β expression and inhibit virus infection in the absence of RelA.
We then examined the potential mechanism involved in strong IRF3 activation-induced IFN-β expression. The coactivator, CBP, was found to be critically important for IFN-β transcription after virus infection (30, 42). Although RelA is thought to play an especially important role in CBP recruitment, resulting in synergistic enhancement of IFN-β expression (30), our results indicate that RelA is not required for IFN-β expression when IRF3 activation is high. Thus, high levels of activated IRF3 may recruit CBP in the absence of RelA. To test this hypothesis, we performed ChIP analysis to investigate the binding of CBP to the endogenous IFN-β promoter. CBP binding to the IFN-β promoter was below detection limits at 6 h after NDV infection (data not shown). However, at 12 h, when the activated IRF3 level is greatest (Fig. 8A), similar amounts of CBP bound to the promoter in WT and RelA−/− MEFs (Fig. 9B). These results suggest that high levels of activated IRF3 can recruit CBP without requiring RelA.
The above findings suggest that high levels of activated IRF3 are sufficient to recruit CBP in the absence of RelA. We next investigated the role of RelA when activated IRF3 levels are low. We first investigated synergism between IRF3 and CBP in stimulating IFN-β promoter reporter activity. Although CBP increased IFN-β reporter activity in the presence of low levels of IRF3 (1 ng) only slightly, a robust 40-fold increase was noted at higher IRF3 levels (10 ng) (Fig. 9C). Thus, high levels of IRF3 can strongly activate IFN-β promoter activity through synergism with CBP, which is consistent with the above-mentioned findings and the physical interaction between IRF3 and CBP (43, 44). However, coexpression of low levels of RelA greatly increased IFN-β reporter activity in the presence of low levels of IRF3 (1 ng) and CBP (Fig. 9D). In contrast, low levels of RelA, low levels of IRF3SA, or CBP alone did not activate IFN-β promoter activity, whereas RelA+CBP or IRF3+CBP induced it substantially less than did RelA+IRF3+CBP (Fig. 7D). Together with our above findings, these results suggest that high levels of active IRF3 can synergize with CBP and promote IFN-β expression in the absence of RelA. However, low levels of active IRF3 synergize poorly with CBP. Under these conditions, IRF3 and RelA are required for synergism with CBP and proper induction of IFN-β expression. Therefore, these findings together help to explain RelA requirement for IFN-β expression at early (i.e., low IRF3 activation) but not at late time points (i.e., high IRF3 activation) after virus infection.
Classic studies of enhanceosome formation and subsequent IFN-β promoter activation implicated NF-κB as a key mediator of virus-induced IFN-β expression (22, 24, 29, 45, 46). However, although the fact that NF-κB is activated by virus infection and it associates with the IFN-β promoter is beyond dispute, the actual role of NF-κB in IFN-β expression and how that impacts virus replication has remained mostly unexplored. The findings from this study indicate that NF-κB functions in regulating IFN-β expression are limited to an early, but nonetheless crucial, phase of virus infection. Furthermore, the RelA subunit is the main NF-κB component responsible for virus-induced IFN-β expression. We show that this early requirement for RelA is crucial for the timely induction of IFN-β. In the absence of RelA, IFN-β production is delayed, leading to a significant defect in the induction of downstream ISGs, including IRF7. As a consequence, IRF7-dependent secondary antiviral gene transcription is severely defective in cells lacking RelA, even though IFN-β expression eventually achieves or exceeds levels found in WT cells. The crucial importance of this early delay in IFN-β gene induction is highlighted by the increased susceptibility of RelA−/− MEFs to RNA virus replication.
Our results also show that the requirement for NF-κB in IFN-β gene expression is inversely correlated with IRF3 activation, suggesting that NF-κB is especially important for promoting IFN-β expression prior to substantial IRF3 activation. Previous studies identified a crucial role for CBP in virus-induced IFN-β expression (30, 42). Our results, in agreement with those of other investigators (30, 43, 44, 47), show that RelA (or IRF3) alone can synergize with CBP to transactivate the IFN-β promoter. Thus, RelA compensates for low levels of activated IRF3 by synergizing with CBP to transactivate the IFN-β gene. In contrast, IRF3 also synergizes with CBP, and activated IRF3 in RelA−/− MEFs is sufficient to recruit CBP and induce IFN-β expression. Collectively, these observations suggest that RelA is critical for enhancing early IFN-β expression after virus infection when IRF3 activation is weak. Our studies also made possible a direct comparison of RelA and IRF3 function in virus-induced IFN-β expression. Our findings indicate that although RelA function is limited to early time points, IRF3 is crucial at all time points tested. Thus, IRF3 is a substantially more important transcriptional regulator of virus-induced IFN-β than is RelA.
Recent structural studies of the IFN-β enhanceosome revealed a surprising absence of contacts between bound transcription factors, suggesting that different factors may independently associate with DNA and enhance transcription (21). Our functional studies support these findings by showing lack of RelA requirement and IRF3 sufficiency for IFN-β expression (at later time points). Our results also suggest that agents that induce limited IRF3 activation may require IKKβ/NF-κB to synergize with IRF3 and CBP for IFN-β induction. Consistent with this, we found that the TLR4 ligand LPS is a poor inducer of IRF3 activation compared with virus infection (Supplemental Fig. 3A) and that LPS-induced IFN-β is severely impaired in IKKβ−/− MEFs (Supplemental Fig. 3B). Thus, in addition to early stages of virus infection, NF-κB function may be generally important for IFN-β expression by agents that are poor activators of IRF3, including certain TLR ligands (e.g., LPS). Our findings also suggest that agents that induce rapid NF-κB activation may enhance early virus-induced IFN-β expression and, therefore, limit virus replication. Thus, modulators of NF-κB activation may have benefit as antiviral agents. Finally, unregulated IFN-α/β expression contributes to autoimmune diseases, such as systemic lupus erythematosus (2). It will be interesting to determine whether unrestrained NF-κB activation contributes to chronic IFN expression in these diseases.
We thank the Flow Cytometry and Molecular Biology core facilities at the Moffitt Cancer Center. IKKβ+/− mice were kindly provided by Dr. Zhi-Wei Li (Moffitt Cancer Center).
Disclosures The authors have no financial conflicts of interest.
This work was supported by National Institutes of Health Grants R01 AI059715 and DOD BC011057 and by institutional funds from the Moffitt Cancer Center (to A.A.B.). This work was also supported in part by a National Institute of Allergy and Infectious Diseases (NIAID)-funded Center to Investigate Virus Immunity and Antagonism Grant U19AI083025, and by NIAID Grant U19AI083025 (to A.G.-S.).
The online version of this article contains supplemental material.
Abbreviations used in this paper:
- bone marrow-derived dendritic cell
- conventional dendritic cell
- chromatin immunoprecipitation
- dendritic cell
- IFN-α receptor
- IκB kinase
- IFN regulatory factor
- IFN-stimulated gene
- mouse embryonic fibroblast
- mean fluorescence intensity
- monokine induced by IFN-γ
- multiplicity of infection
- Newcastle disease virus
- a recombinant variant of NDV that expresses GFP
- Newcastle disease virus nucleocapsid
- 2′5′ oligoadenylate synthase 2
- positive regulatory domain
- phosphorylated STAT1
- vesicular stomatitis virus
- Received January 15, 2010.
- Accepted May 26, 2010.
- Copyright © 2010 by The American Association of Immunologists, Inc.