Interferons (IFNs) are secreted cytokines produced by cells following virus infection that generate an antiviral state in neighboring cells (Pestka, 1997; Stark et al., 1998; Barber, 2001). In addition to their anti-viral functions IFNs also have multiple and diverse biological activities, including protection against bacterial and parasitic infections, antitumor activity by direct growth suppression and by generating an anti-tumor immune response and modulation of differentiation (Fisher and Grant, 1985; Pestka, 1997; Stark et al., 1998; Barber, 2001). Five classes of IFNs are currently recognized, namely, IFN-α, IFN-β, IFN-γ, IFN-τ (bovine) and IFN-ω, which have distinct biological activities, modes of induction and amino acid sequences (Stark et al., 1998). Type I and type II IFNs are classified based on interactions with their respective type I or II cell surface receptors (Pestka, 1997). IFN-α (12 subtypes), IFN-β, IFN-τ, and IFN-ω interact with type I IFN receptors and IFN-γ binds and signals through type II IFN receptors (Pestka, 1997). The binding of IFNs with their cognate receptors results in activation of IFN-stimulated genes (ISGs) that classically involves Janus tyrosine kinases (JAKs) and signal transducers and activators (STATs) (Stark et al., 1998). Binding of Type I IFNs to cognate receptors results in sequential phosphorylation of Tyk2, JAK1, STAT1, and STAT2 and the STAT1/STAT2 heterodimer translocates into the nucleus and interacts with the IRF family transcription factor p48, thus generating the interferon stimulatory gene factor-3 (ISGF3) transcription factor complex that binds to IFN-stimulated regulatory elements (ISRE) in target promoters to drive transcription (Stark et al., 1998).
IFNs protect against viral infection by inducing expression of ISGs, such as GTPase Mx protein, 2′-5′-oligoadenylate synthase (OAS) and dsRNA-dependent protein kinase (PKR) (Stark et al., 1998). The GTPase Mx protein confers antiviral activity against RNA viruses by interfering with viral transcription or cytoplasmic virus multiplication (Staeheli et al., 1993). Activation of OAS by dsRNA promotes the synthesis of 2′-5′ oligoadenylates (2–5A) resulting in activation of RNase L which causes nonspecific degradation of single-stranded RNAs, limiting replication of single-stranded RNA viruses or viruses generating single-stranded RNAs as replication intermediates (Castelli et al., 1997). The activation of PKR by dsRNA mediates phosphorylation of several cellular proteins, including eukaryotic translation initiation factor eIF-2α (Meurs et al., 1990; Der and Lau, 1995). Phosphorylation of eIF-2α leads to stabilization of the complex between eIF-2α and eIF-2β thereby preventing further translation initiation events. In this context, PKR activation by dsRNA blocks the further synthesis of both cellular and viral proteins, contributing to growth suppression.
Classically, the induction of Type I IFNs by viral infection requires Toll-like receptors (TLRs) that recognize pathogen-associated molecular patterns (Meylan and Tschopp, 2006). TLR3 and TLR8 (mouse TLR7) function as signaling receptors for extracellular double-stranded RNA (dsRNA) and viral single-stranded RNA, respectively (Alexopoulou et al., 2001; Heil et al., 2004). However, recent seminal studies have identified a family of related molecules, including melanoma differentiation associated gene-5 (mda-5) and retinoic acid inducible gene-I (RIG-I), that induce type I IFN and function as a first line of defense against viral infection in a TLR-independent manner (Kang et al., 2002; Andrejeva et al., 2004; Yoneyama et al., 2004; Breiman et al., 2005; Berghall et al., 2006; Gitlin et al., 2006; Johnson and Gale, 2006; Meylan and Tschopp, 2006). mda-5 and RIG-I have a unique structure consisting of a caspase recruitment domain (CARD) in the N-terminus and a DExH RNA helicase domain in the C-terminus (Kang et al., 2002; Yoneyama et al., 2004; Johnson and Gale, 2006). Binding of dsRNA with the helicase domain of RIG-I or mda-5 activates the ATPase activity leading to a conformational change and subsequent recruitment of a CARD-containing adapter protein MAVS/VISA/IPS-1/Cardiff that relays the signal to the kinases TBK1 and IKK-i, which phosphorylates IFN-regulatory factor-3 (IRF-3) and IRF-7, transcription factors essential for the expression of type-I IFNs (Kawai et al., 2005; Meylan et al., 2005; Seth et al., 2005; Xu et al., 2005; Kumar et al., 2006; Lin et al., 2006b). By controlling the induction of IFN-β and the anti-viral ISGs, RIG-I and mda-5 play essential roles in innate immunity.
RIG-I and mda-5 both induce type I IFNs. However, these molecules are also ISGs and are also induced by viral infection and by dsRNA (Kang et al., 2004; Yoneyama et al., 2004). Being IFN-inducible genes, they act in a positive feedback loop thus amplifying the effects of IFNs. Type I IFNs have profound growth-suppressing properties and the CARD domain present in these molecules supports their involvement in caspase recruitment and apoptosis. Indeed, we have demonstrated that overexpression of mda-5 alone induces apoptosis in a variety of cells (Kang et al., 2002; Lin et al., 2006a). These findings suggest that in addition to the anti-viral effects, RIG-I and mda-5 might also mediate the growth suppression properties of IFNs.
Type I IFN regulation of mda-5 expression has been studied in detail (Kang et al., 2002, 2004). However, although it is known that RIG-I is inducible by type I IFNs, little is known about the molecular mechanism of this regulation. In the present study, we cloned the promoter region of the RIG-I gene and identified the essential elements regulating its expression. We demonstrate that RIG-I expression is downregulated in cancer cells when compared to normal cells and its expression is primarily regulated by interferon regulatory factor-1 (IRF-1). A positive correlation was evident between expression of IRF-1 and RIG-I. In addition to its anti-viral effects, IRF-1 also has tumor suppressor properties (Romeo et al., 2002). In these contexts, we hypothesize that RIG-I may also function as a tumor suppressor and its overexpression might be exploited as a potential strategy for cancer gene therapy.
Materials and Methods
Cell lines and culture conditions
Normal P69 human immortalized prostate epithelial cells and DU-145, PC-3 and LNCaP prostate carcinoma cells were cultured as described (Lebedeva et al., 2003). Normal HBL-100 human immortal mammary epithelial cells and MCF-7, T47D, and MDA-MB-231 human breast cancer cells were cultured as described (Su et al., 1998). Normal FM516-SV human immortal melanocytes and WM35, HO-1 and MeWo melanoma cells were cultured as described (Lebedeva et al., 2002). Primary human fetal astrocytes (PHFA) and T98G, U87MG and U251MG human malignant glioma cell lines were cultured as described (Su et al., 2003). L9 rat glioma cells were obtained and grown as suggested by the American Type Culture Collection. All cultures were maintained at 37°C in a humidified 5% CO2/95% air incubator. Human cells were treated with IFN-β (1,000 U/ml) or with poly(I).poly(C) (50 µg/ml) for various analyses (Kang et al., 2002).
5′ rapid amplification of cDNA ends (RACE)
5′ RACE was performed using GeneRacer kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. The sequences of the primers used were GSP (complementary to 218- to 241-bp of the published RIG-I cDNA sequence): 5′ GTAGCTCAGGATGTAGGTAGGGTC 3′ and Nested GSP (complementary to 203- to 223-bp of the published RIG-I cDNA sequence): 5′ AGGGTCCAGGGTCTTCCGGAT 3′. The PCR products were cloned into a TA-cloning vector (PCR2.1; Invitrogen) and sequenced.
Cloning of the RIG-I promoter
Human genomic DNA obtained from primary human fetal astrocytes was used as a template to clone the RIG-I promoter by PCR. The genomic sequence upstream of the RIG-I cDNA sequence was obtained from the human genome database. The primers used were sense: 5′ GAGATGAGGTTTCACCATGT 3′ and antisense: 5′ AAGGGAAAATCGAAAGTGCAAC 3′. The cycling parameters were: 94°C for 2 min; 35 cycles of 94°C for 45 sec, 58°C for 1 min and 72°C for 3 min, with a final extension of 72°C for 10 min. The PCR product was cloned into PCR2.1 vector and sequenced.
Construction of plasmids, transient transfection and luciferase assay
The full-length RIG-I promoter (−1902 to −1) was cloned into the KpnI and XhoI sites of pGL3-basic Vector (Promega, Madison, WI) to generate p-1902. The serial deletion mutants of p-1902 were constructed using the Erase-a-Base system (Promega) according to the manufacturer's instructions. Cells were seeded at 2 × 105/35 mm tissue culture plate and ∼24 h later transfected with Lipofectamine-2000 transfection reagent (Invitrogen) and 6 µg of plasmid DNA per plate that included 5 µg of luciferase expression construct and 1 µg of β-galactosidase-expression plasmid (pSV-β-gal; Promega) according to the manufacturer's instructions. Luciferase assays were performed 48 h post-transfection using a Luciferase Reporter Gene Assay kit (Promega) according to the manufacturer's protocol. The β-galactosidase activity was determined using the Galacto-Light Plus kit (Tropix). Luciferase activity was normalized by β-galactosidase activity and the data from triplicate determinations were expressed as mean ± SD.
RNA extraction and Northern blot analysis
Total cellular RNA was isolated by the guanidinium/phenol extraction method and Northern blotting was performed as described (Su et al., 2001). Fifteen micrograms of RNA were denatured and electrophoresed in 1.2% agarose gels with 3% formaldehyde, transferred to nylon membranes and hybridized with 32P-labeled cDNA probes as described previously (Su et al., 2001). Following hybridization, the filters were washed and exposed for autoradiography. The cDNA probes were full-length human RIG-I and GAPDH.
Western blotting was performed as previously described (Su et al., 2001). Briefly, cells were harvested in RIPA buffer (1× PBS, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) containing protease inhibitor cocktail (Roche, Mannheim, Germany), 1 mM Na3VO4 and 50 mM NaF and centrifuged at 12,000 rpm for 10 min at 4°C. The supernatant was used as total cell lysate. Thirty micrograms of total cell lysate were used for SDS–PAGE and transferred to a nitrocellulose membrane. The primary antibodies used were anti-IRF-1 and anti-actin (Santa Cruz Biotechnology, Santa Cruz, CA).
Electrophoretic mobility shift assay (EMSA)
Nuclear extracts were prepared from 2 to 5 × 108 cells as described (Su et al., 2000, 2001). The sequences of the probes were as follows: wild-type IRF-1, 5′-GTTGCACTTTCGATTTTCCCTT-3′, mutant IRF-1, 5′- GTTGCACTAACGACTTTCCCTT-3′. The double-stranded oligonucleotides were labeled with [γ-32P]ATP (Amersham, Piscataway, NJ) and T4 polynucleotide kinase. The labeled probes were then incubated with nuclear extract at room temperature for 30 min. The reaction mixture consisted of 32P-labeled deoxyoligonucleotides (≥50,000 cpm), 2 µg of poly(dI–dC) and 10 µg of nuclear protein extract with 10 mM HEPES pH 7.5, 50 mM KCl, 5 mM MgCl2, 0.5 mM EDTA, 1 mM DTT and 12.5% glycerol. After incubation for 30 min at room temperature, the reaction mixtures were electrophoresed on a 5% non-denaturing polyacrylamide gel with 0.5× TBE (160 V for 3 h). The gel was dried and autoradiographed. Nuclear extracts were also incubated with a 10- and 100-fold molar excess of cold IRF-1 competitor together with the 32P-labeled probe. Supershift assays were performed by incubating nuclear extracts with anti-IRF-1 or anti-actin antibody (2 or 10 µg) together with 32P-labeled probe for 30 min at room temperature. The reaction mixtures were then electrophoresed and processed as described above.
Quantitative PCR (Q-PCR)
Q-PCR was performed using a kit from Stratagene based on SYBR Green I DNA-binding dye. The primers for RIG-I: sense, 5′ TGCAAGCAGAGGCCGGCATGAC 3′ and antisense, 5′ CCCAACTTTCAATGGCTTCATAAAG 3′ and for GAPDH: sense, 5′ GCGTCTTCACCACCATGGAGAA 3′ and antisense, 5′ GAGTCCTTCCACGATACCAAAG 3′.
Chromatin immunoprecipitation (ChIP) assays
ChIP assays were performed using a commercially available kit from Active Motif (Carlsbad, CA). Primary human fetal astrocytes (PHFA) (5 × 107) were treated or untreated with IFN-β and cells were fixed with formaldehyde for 10 min at RT. Cells were harvested and the nuclei were isolated using a dounce homogenizer. The nuclear pellet containing chromatin was sheared with Enzyme Shearing cocktail solution, the chromatin was pre-cleared with protein G beads and anti-IRF-1 antibody was added to the pre-cleared chromatin. The DNA-protein complex was precipitated using protein G beads, washed thoroughly and DNA was eluted from the beads. The eluted DNA was treated with RNase A and proteinase K, purified and used as template for PCR using primers, sense 5′ CTGTGTCTAAACGTTAGCGCTA 3′ and antisense 5′ ACTAAAGGGAAAATCGAAAGTGC 3′. The PCR products were analyzed by agarose gel electrophoresis.
DNA affinity purification assay (DAPA)
DAPA was performed using µMACS Streptavidin Kit from Miltenyi Biotec (Bergisch, Uladbach, Germany). PHFA (2 × 107) were treated or untreated with IFN-β and nuclear extract was prepared from the cells. Five micrograms of double-stranded biotinylated oligo containing IRF-1 binding site (5′ GTTGCACTTTCGATTTTCCCTTTAGTTATT 3′) was incubated with 500 µg of nuclear extract for 30 min at RT and with µMACS Streptavidin Microbeads for 30 min at 4°C. The binding reaction was applied to an equilibrated µ column in the magnetic field of µMACS separator. The column was rinsed and the target molecules bound to the biotinylated probe were eluted. The eluate was used directly for SDS–PAGE and Western blot analysis using anti-IRF-1 antibody. Biotinylated GAPDH probe was used as control.
Immunofluorescence staining was performed in frozen sections of rat brains with L9 (rat glioma) tumor implants. Sections were permeabilized with 0.1% Triton X-100 in PBS for 30 min. Sections were then blocked for 1 h at room temperature with 2% goat serum and 1% BSA in PBS and incubated with anti-RIG-I antibody (1:50) (DDX58 antibody, Abgent) overnight at 4°C. Sections were then rinsed three times for 5 min each in PBS, and incubated with Alexa488-conjugated anti-rabbit IgG (Molecular Probes, Carlsbad, CA) for 1 h at room temperature. After three rinses for 5 min each in PBS, excess PBS was removed and sections were mounted in VectaShield fluorescence mounting medium containing 4,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA). A fluorescence microscope analyzed the images.
All the experiments were performed at least three times. The results are expressed as mean ± SD. Statistical comparisons were made using an unpaired two-tailed Student's t-test. A P < 0.05 was considered as significant.
RIG-I is an IFN-β-inducible gene
To determine IFN induction of the RIG-I gene, we employed normal cells from four different lineages, namely prostate, breast, melanocytes and astrocytes and their malignant counterparts. P69 is an SV40 TAg-immortalized human prostate epithelial cell line and DU-145, PC-3, and LNCaP are prostate carcinoma cells. HBL-100 is a spontaneously immortalized human breast epithelial cell line and MCF7, T47D, and MDA-MB-231 are breast carcinoma cells. FM516-SV (henceforth termed FM516) is an SV40 TAg-immortalized human melanocyte and WM35, HO-1 and MeWo are human melanoma cells. PHFA are primary human fetal astrocytes and U251MG, U87MG, and T98G are human malignant glioma cells. These cells were treated with IFN-β (1000 U/ml) for 6–48 h and RIG-I mRNA expression was analyzed by Northern blot analysis. Under basal condition, RIG-I mRNA expression could not be detected in any of the cancer cell lines (Fig. 1A). On the contrary, under basal conditions RIG-I mRNA expression could be detected in all four normal cells (P69, HBL100, FM516 and PHFA), although RIG-I mRNA level in HBL100 cells was much lower than that in the other three normal cells. This may reflect the instability and later stage of progression of these cells, which over time in culture develop both transformed and tumorigenic (in nude mice) properties. In all the cell lines studied, IFN-β treatment markedly induced RIG-I mRNA expression at 6 and 12 h post-treatment (Fig. 1A). The expression gradually decreased by 24–48 h post-treatment. However, RIG-I mRNA could be detected even at 48 h after IFN-β treatment and did not return to the basal level.
To confirm that RIG-I mRNA expression is higher in normal than in cancer cells we performed quantitative PCR (Q-PCR) using mRNA from P69, FM516 and PHFA and their tumorigenic equivalents, DU-145, MeWo and T98G cells, respectively. RIG-I mRNA was 2–3.5 fold higher in normal cells than in the cancer cells (Fig. 1B). To compare normal and cancer cell expression of the RIG-I protein in an in vivo situation we used a syngeneic rat glioma model in which L9 rat glioma cells were implanted into rat brains. Brain samples were harvested from these animals and immunocytochemistry was performed in tissue sections containing both the normal brain tissue as well as the implanted glioma cells. DAPI staining clearly demarcated the boundary between the normal brain and the implanted tumor (Fig. 1C, right part). While the normal brain contained dispersed cells, as revealed by sparse nuclear staining by DAPI, the tumor region contained a dense arrangement of DAPI-stained nuclei. RIG-I expression was abundantly detected in normal brain, while in the malignant counterpart it was hardly detectable (Fig. 1C, left part). These results bolster the finding that RIG-I is expressed at higher levels in normal cells when compared to cancer cells.
Cloning and characterization of the RIG-I promoter
To obtain insights into the process of IFN-β-induction of the RIG-I gene, we cloned the RIG-I promoter. Initially we determined the transcription start site by identifying the 5′ end of RIG-I cDNA using 5′ rapid amplification of cDNA end (RACE) method with primers complementary to 218- to 241-bp and 203- to 223-bp of the published RIG-I cDNA sequence (GenBank accession# AF038963). Sequence analysis of multiple clones representing the 5′ RACE products confirmed that the 5′-end of the RIG-I cDNA conforms to the 5′-end of the published RIG-I cDNA sequence. Human genome database mining allowed us to identify the upstream genomic sequence and we cloned an ∼2-kb fragment of genomic DNA just upstream of the RIG-I cDNA sequence by PCR using human genomic DNA as a template. This fragment (−1,902 to −1, when the transcription start site is considered as +1) was cloned into the pGL3-basic vector (Promega) upstream of the luciferase gene generating the construct p-1902.
The 16 cell lines, in which RIG-I expression was analyzed, were transfected with p-1902 and a β-galactosidase expression plasmid, treated with IFN-β (1,000 U/ml) or poly(I).poly(C) (50 µg/ml) for 6 and 24 h and luciferase and β-galactosidase activities were quantified. As anticipated from the mRNA expression studies, under basal conditions the relative luciferase activity of p-1902 in normal cells (P69, HBL100, FM516, and PHFA) was ∼2- to 3-fold higher than in cancer cells (Fig. 2). Treatment with IFN-β or with poly(I).poly(C) significantly increased p-1902 activity in all the cells with at least a twofold or greater induction, both at 6 and 24 h after treatment. These findings indicate that IFN-β and dsRNA induction of RIG-I expression occurs at the transcriptional level.
To identify the essential element(s) conferring IFN-β induction, a series of deletion mutants of p-1902 were generated (Fig. 3A). By exonuclease digestion p-1141, p-362, p-166, and p-87 were constructed and these plasmids were transfected into P69, HBL100, FM516, and PHFA and cancer counterparts, namely, DU-145, T47D, HO-1, and T98G, respectively. The basal luciferase activity did not alter significantly between p-1902 and p-166, although the activity of the p-362 construct was lower than that of p-1902, p-1141, and p-166 in some cell lines (such as HO-1) indicating the presence of an inhibitory element between –1141 and –362 (Fig. 3B). Treatment of p-1902, p-1141, p-362 and p-166 transfected cells with IFN-β resulted in a significant induction in luciferase activity, although the induction of p-362 by IFN-β was attenuated in some cell lines (such as DU-145, HBL100, FM516, and HO-1) indicating that a putative inhibitory element between −1141 and −362 also interferes with IFN-β induction. Deletion of p-166 to p-87 resulted in a complete loss of luciferase activity including IFN-β-inducibility, indicating that the region −166 to −1 contains the important elements controlling both basal activity and IFN-β-responsiveness of the promoter.
Sequence analysis identified the TATA box at position −144 of the promoter sequence (Fig. 3A). Another TATA box was also identified at position −162. Which of these TATA boxes are functional remains to be determined. The position of the TATA boxes explains why deletion from −166 to −87 resulted in a complete loss of promoter activity. Sequence analysis did not identify any ISRE sequence, conferring IFN-β-responsiveness, between −166 and −1. However, a putative interferon regulatory factor-1 (IRF-1) binding site was detected between −17 and −4 (Fig. 3A, underlined). The canonical IRF-1 binding site is “G(A)AAA G/C T/C GAAA G/C T/C” (Tanaka et al., 1993). The putative IRF-1 binding site in the RIG-I promoter is “ACTTTCGATTTTCC”.
IRF-1 regulates the basal activity and IFN-β and dsRNA-responsiveness of the RIG-I promoter
To decipher the role of IRF-1 in regulating RIG-I promoter activity, the IRF-1 site in p-1902 was either mutated (ACTTTCGATTTTCC to ACTAACGACTTTCC; IRF mut) or deleted (IRF del). Of interest, the promoter activity in all the 16 cell lines was markedly reduced in the IRF mut or almost completely abolished in the IRF del construct when compared to the activity of p-1902 (IRF wt) (Fig. 4). Additionally, in all the cell lines, while p-1902 (IRF wt) activity could be significantly induced by IFN-β (1,000 U/ml) or poly(I).poly(C) (50 µg/ml) treatment for 24 h, both the IRF mut and the IRF del completely lost their responsiveness to IFN-β or poly(I).poly(C). These findings together with data from the serial deletion analysis studies suggest that both the TATA box and IRF-1 binding site play a crucial role in regulating RIG-I promoter activity. Both of these sites are necessary for maintaining the basal activity of the promoter since loss of either the TATA box (in p-87) or the IRF-1 binding site (in IRF mut or IRF del) abolished RIG-I promoter activity. Additionally, the IRF-1 binding site also conferred IFN-β- and dsRNA-mediated augmentation of RIG-I promoter activity.
To strengthen our findings, P69, HBL100, FM516, and PHFA and their tumorigenic counterparts were co-transfected with p-1902 and either an IRF-1 expression plasmid or an empty vector (pcDNA3.1). Overexpression of IRF-1 significantly increased the activity of p-1902 in all of the cell lines, further confirming the involvement of IRF-1 in regulating RIG-I promoter activity (Fig. 5A). To corroborate these findings, we depleted either IRF-1 or p48, a component of ISGF3 complex, by siRNA in normal cells and their malignant counterparts, and then analyzed IFN-β-induction of the RIG-I promoter. As shown in the Figure 5B, both IRF-1 and p48 siRNA effectively inhibited corresponding protein synthesis in FM516 cells. IRF-1 siRNA, but not p48 siRNA, completely inhibited IFN-β-induction of the RIG-I promoter in FM516 and MeWo cells (Fig. 5C) indicating that IRF-1 plays an essential role in regulating RIG-I promoter activity. Similar results were also obtained in P69, HBL100, and PHFA cells and their tumorigenic counterparts (data not shown).
The binding of IRF-1 to its putative site in the RIG-I promoter was confirmed by electrophoretic mobility shift assay (EMSA) using a radiolabeled probe corresponding to the putative IRF-1 binding site in the RIG-I promoter and nuclear extracts from PHFA and T98G cells. A single shifted band was detected when the probe was incubated with the nuclear extract (Fig. 6A, lane 2), while no band was detected in the absence of nuclear extract (Fig. 6A, lane 1). The intensity of the band was significantly increased upon treatment with IFN-β or poly(I).poly(C) (Fig. 6A, lanes 3 and 4, respectively). The band was competed by increasing concentrations of cold probe (Fig. 6A, lanes 5 and 6) and supershifted by an anti-IRF-1 antibody (Fig. 6A, lane 7), but not by an anti-actin antibody (Fig. 6A, lane 8). A radiolabeled probe with a mutated IRF-1 binding sequence did not generate any shifted band (Fig. 6A, lane 9). These findings confirm that IRF-1 binds to the RIG-I promoter and regulates its activity. IRF-1 DNA binding is markedly increased by IFN-β or dsRNA treatment further supporting the central role of IRF-1 in mediating regulation of RIG-I expression by these agents.
We performed Chromatin Immunopricipitation (ChIP) assays and DNA Affinity Purification Assays (DAPA) to reinforce the EMSA findings. PHFA cells, either untreated or treated with IFN-β (1,000 U) for 12 or 24 h, were subjected to ChIP assay using anti-IRF-1 or anti-p48 antibodies and primers flanking the IRF-1 binding site in the RIG-I promoter resulting in a PCR fragment of 184-bp. As shown in Figure 6B, anti-IRF-1 antibody precipitated the IRF-1 binding element in the RIG-I promoter under basal conditions. The intensity of the DNA band increased markedly upon IFN-β treatment (Fig. 6B, top part, lanes 2 and 3). Anti-p48 antibody failed to precipitate any DNA fragment (Fig. 6B, second part). In DAPA, nuclear extracts of untreated or IFN-β-treated PHFA cells were incubated with a biotinylated IRF-1 binding element in the RIG-I promoter and the proteins bound to the biotinylated DNA were purified by streptavidin beads and subjected to SDS–PAGE and Western blot analysis using anti-IRF-1 antibody (Fig. 6C). Under basal conditions, a faint amount of IRF-1 bound to the RIG-I promoter (Fig. 6C, lane 1). However, this binding increased significantly upon IFN-β treatment (Fig. 6C, lanes 2 and 3). A biotinylated GAPDH probe did not bind to IRF-1 confirming the specificity of this assay (Fig. 6C, lanes 4–6). In total, these findings provide further confirmation that IRF-1, but not the ISGF3 complex, binds to the RIG-I promoter.
IRF-1 expression is elevated in normal cells and is induced by IFN-β
We observed that RIG-I expression is higher in normal cells, induced by IFN-β and controlled by IRF-1. To establish a connection among these findings, we analyzed the expression level of IRF-1 and its IFN-β-inducibility in the 16 cell lines by Western blot analysis. The basal IRF-1 level in normal cells was significantly higher than that in cancer cells (Fig. 7). In most cancer cells, except DU-145, PC-3 and MCF7, IRF-1 expression was not detected. Upon IFN-β treatment IRF-1 was induced and the induction kinetics were similar to that of RIG-I. The maximum induction of IRF-1 was detected between 6 and 12 h with a gradual decrease from 24 to 48 h.
RIG-I was first cloned as a retinoic acid (RA)-inducible gene (Sun, 1997). However, the role of RIG-I in RA signaling is not known. RIG-I is also an ISG that plays an essential role in generation of type I IFNs in response to viral infection (Yoneyama et al., 2004). Mouse fibroblasts and dendritic cells from RIG-I−/− mice fail to generate IFN-β upon RNA virus infection when compared to their wild-type counterparts (Kato et al., 2006). The importance of RIG-I in innate immune response mandates a detailed study of its expression regulation. We now identify a minimum region comprising −166 to −1 of the RIG-I promoter that confers basal activity as well as IFN-β- and dsRNA-mediated induction of the RIG-I promoter. Additionally, our studies confirm that a single transcription factor, IRF-1, primarily regulates all aspects of activity of the RIG-I minimal promoter. This promoter-luciferase system is amenable for high-throughput library screening to identify small molecules specifically inducing RIG-I expression and with potential to be used for early intervention and treatment of deadly viral infections for which no vaccinations are currently available.
The induction of type I IFN-stimulated genes is primarily regulated by the ISGF-3 transcription factor complex binding to an ISRE sequence (Stark et al., 1998). However, IRF-1 also plays a pivotal role in regulating expression of ISGs. IRF-1 was originally identified as a DNA-binding protein of the positive regulatory domain I (PRDI) element within the human IFN-β promoter (Fujita et al., 1988; Miyamoto et al., 1988). While regulating the expression of type I IFN, IRF-1 itself is an ISG (Miyamoto et al., 1988). The consensus IRF-1 binding site, known as IRF-E, is “G(A)AAA G/C T/C GAAA G/C T/C” which is present in the regulatory region of a variety of genes regulating diverse functions (Tanaka et al., 1993). IRF-1 regulates genes involved in anti-viral response, such as IFN-α/β, 2′, 5′ OAS and PKR; anti-bacterial response, such as iNOS and gp91phox; anti-proliferative response, such as lysyl oxidase and p21WAF1/CIP1/mda-6; apoptosis, such as caspases 1 and 7; immune response, such as IL-12 and IL-15; and inflammation, such as cyclooxygenase-2 (Sato et al., 2001; Kroger et al., 2002; Romeo et al., 2002). Thus IRF-1 has potent anti-proliferative, apoptotic and immunomodulatory properties. In addition to mediating IFN-induced gene regulation, IRF-1 can function independent of IFN signaling, such as mediating N-ras-induced upregulaiton of p21WAF1/CIP1/mda-6 (Passioura et al., 2005). The promoters of the ISGs, such as 2′, 5′ OAS, and PKR, contain an ISRE sequence in addition to IRF-E thus having a dual type I IFN-mediated regulation of gene expression (Benech et al., 1987; Tanaka and Samuel, 1994). It has been documented that p48 of ISGF-3 and IRF-1 do not perform redundant functions in the cell, but rather complement one another in both type I and II IFN responses (Kimura et al., 1996). In contrast, analysis of the RIG-I promoter sequence did not reveal any ISRE elements stressing the importance of IRF-1 in regulating RIG-I expression. Additionally, blocking p48 expression, a component of the ISGF3 complex, using an siRNA strategy did not affect IFN-β-induction of the RIG-I promoter, whereas blocking IRF-1 using a similar siRNA strategy effectively inhibited induction of RIG-I promoter activity by IFN-β (Fig. 5C). IRF-1 is also induced by retinoic acid (RA) and thus might also mediate RA-induction of RIG-I (Percario et al., 1999).
We document that both the TATA box and the IRF-E are required for regulation of the basal expression of the RIG-I gene. Deletion of either of these elements abolishes the activity of the promoter. We observed that under basal condition the expression of IRF-1 and RIG-I was detected in normal cells of diverse lineages, while the expression, predominantly that of RIG-I, was decreased or lost in cancer cells. Among the IRF family, IRF-1 functions as a tumor suppressor. IRF-1 reverses IRF-2-induced transformation of NIH3T3 cells and tumor formation in nude mice (Harada et al., 1993). IRF-1 inhibits anchorage-independent growth of cells transformed by various oncogenes and overexpression of IRF-1 inhibits tumor formation in syngeneic mice (Tanaka et al., 1994; Yim et al., 1997; Kirchhoff and Hauser, 1999; Kroger et al., 2001; Kroger et al., 2003). Although there is no spontaneous tumor development in IRF-1−/− mice, IRF-1/p53 double-deficient mice have increased tumor incidence and enhanced multiplicity and alteration of tumor spectrum, when compared to p53−/− mice alone (Nozawa et al., 1999). IRF-1 maps to chromosome 5q31.1, which is frequently deleted in patients with leukemia or myelodysplasia (Willman et al., 1993). The loss of IRF-1 allele has also been documented in esophageal and gastric cancers (Ogasawara et al., 1996; Tamura et al., 1996). IRF-1 expression is lost in human breast cancer when compared to adjacent normal tissue suggesting its involvement in mammary carcinogenesis (Doherty et al., 2001).
Our preliminary studies reveal that similar to mda-5, overexpression of RIG-I also induces apoptosis in diverse cancer cells. Regulation of the RIG-I gene by IRF-1 and the expression profile and growth suppressing properties of RIG-I indicate that RIG-I might be a tumor suppressor gene. Based on these considerations, in-depth analysis of RIG-I in the context of cancer will be of value. Expression analysis in patient-derived samples of various cancers will prove informative and should be performed. Additionally, performing this type of analysis with IRF-1 is supported by our immunohistochemistry observations that in sections of rat brain containing a transplanted L9 rat glioma a clear demarcation can be made between normal cells (elevated level of IRF-1 expression) versus the L9 tumor (barely detectable IRF-1 expression) (Fig. 1C). The consequences of overexpression of RIG-I also need to be analyzed in detail. Although most RIG-I-/- embryos were non-viable, dying at embryonic days 12.5–14.0, a few mice were born alive (Kato et al., 2006). Nevertheless, these mice showed growth retardation and died within 3 weeks after birth. Histological analysis of embryonic day 12.5 embryos revealed massive liver degeneration because of uncontrolled apoptosis (Kato et al., 2006). These findings indicate that RIG-I regulates diverse functions during embryogenesis. Generation of a conditional RIG-I knockout mouse that allows switching off this gene at any time point during development will greatly facilitate evaluation of RIG-I in the context of tumorigenesis and growth regulation.
In summary, we now demonstrate that a single transcription factor IRF-1 regulates basal expression as well as IFN-β and dsRNA-mediated induction of RIG-I. In this context, experimental manipulation of IRF-1 will facilitate modulation of cellular responses against viral infection mediated through RIG-I regulation. Our studies also elucidate a potential role of RIG-I in controlling tumorigenesis, uncovering novel avenues for further analysis of this crucial and functionally important molecule.
The present study was supported in part by National Institutes of Health grants R01 GM068448 and R01 CA035675, the Samuel Waxman Cancer Research Foundation and the Chernow Endowment. P.B.F. is the Michael and Stella Chernow Urological Cancer Research Scientist and a SWCRF Investigator.