Retinoic acid inducible gene-I (RIG-I) plays important roles during innate immune responses to viral infections and as a transducer of cytokine signaling. The mechanisms of RIG-I up-regulation after cytokine stimulation are incompletely characterized. It was previously reported that IFN–γ induces the expression of RIG-I in endothelial cells. In this study, we characterized the mechanism of type I IFN-mediated up-regulation of RIG-I in HeLa cells and found that, in addition to type I IFN, TNF-α, a cytokine that regulates innate immune responses, induced expression of RIG-I. To investigate whether TNF-α- and type I IFN-mediated up-regulations of RIG-I were causally related, we studied the kinetics of these responses. Our results were consistent with a model in which TNF-α functioned upstream of type I IFNs. The ability of TNF-α to up-regulate RIG-I required protein synthesis, expression of functional type I IFNRs, and STAT1 signaling. We also found that IFN-ε was the only IFN isoform expressed constitutively in HeLa cells and that its expression was up-regulated in response to stimulation with TNF-α. The mechanism of up-regulation involved stabilization of IFN-ε mRNA in the absence of transcriptional activation. Silencing the expression of IFN-ε attenuated STAT1 expression and phosphorylation and inhibited RIG-I expression, providing additional support for the participation of IFN-ε upstream of STAT1. Our findings support a sequential mechanism whereby TNF-α leads to stabilization of IFN-ε mRNA, increased IFN-ε synthesis, engagement of type I IFNRs, increased STAT1 expression and phosphorylation, and up-regulation of RIG-I expression. These findings have implications for our understanding of the immune responses that follow cytokine stimulation.
Viral infection leads to the initiation of complex innate immune responses that result from the recognition of viral genetic material by cellular receptors including TLRs (1). Replication of RNA viruses leads to the accumulation of dsRNA molecules that are recognized by TLR3, resulting in the synthesis of type I IFNs and other cytokines (2, 3). IFNs are members of an important family of functionally related cytokines that participate in homeostasis, host defense, and innate immunity (4). They mediate antiviral, antiproliferative (5), antitumor (6), and immunomodulatory responses (7). A number of studies suggest that, in addition to TLR-mediated effects, other mechanisms activate the IFN response during viral infections (8). Yoneyama and coworkers reported that a cytoplasmic RNA helicase, retinoic acid-inducible gene-I (RIG-I,3 also known as Ddx58), binds dsRNA and activates a program of IFN-dependent gene expression (9). In addition, the same group developed RIG-I deficient mice and demonstrated that RIG-I is essential for the production of IFNs in response to RNA viruses owing to its ability to recognize dsRNA generated during viral replication (10, 11). Mice deficient in the genes encoding RIG-I or melanoma differentiation-associated gene 5 (MDA5), a related helicase, are highly susceptible to infection, which led to the conclusion that these proteins are critical for specific antiviral host responses.
The signaling pathways activated by TLR3 and RIG-I differ in the initial adaptor proteins recruited and the motifs involved in RNA recognition, but the mechanisms converge at the end and activate the same protein kinases (12). The relative physiological importance of the two pathways depends on the cell type (12). The RIG-I pathway is critical for responses to viral infection in conventional dendritic cells and in fibroblasts (10). In contrast, plasmacytoid dendritic cells use primarily the TLR system in response to viral infections (10).
RIG-I is a member of the DExH box-containing helicase family; it harbors two caspase recruitment domains (CARD) at the amino-terminal end and an RNA helicase motif at the carboxyl terminus (9). Binding of dsRNA from synthetic and viral origins leads to a conformational change that exposes the N-terminal caspase recruitment domains, a process essential for RIG-I to function as a signal transducer (9, 13). Signaling through RIG-I and its adaptor protein, IFN promoter-stimulator 1 (IPS-1), activates IFN regulatory factor 3 (IRF3) and the host type I IFN response that limits viral infection (14).
Immune responses have common features combined with a degree of site specificity related to the function and anatomical location of an organ (15). The female reproductive tract is immunologically unique because it must provide both tolerance to allogeneic stimuli and protection from diverse pathogens (15). The fact that some infections can be lethal and others can have potentially devastating reproductive sequelae (15) underscores the importance of spatially and temporally regulated immune responses in the female reproductive tract. In this study we used HeLa cervical cancer cells as a model to study immune responses in the cervix, as appropriate organization and regulation of immunity in this organ is essential for healthy female reproductive functions.
We investigated the mechanisms that lead to RIG-I production following the stimulation of HeLa cells with cytokines. We focused our studies on the characterization of downstream events that result from treatment with TNF-α, because this is a key cytokine produced by innate recognition pathways involved in promoting antiviral and inflammatory responses (16). We found that de novo synthesis of IFN-ε is required for TNF-α-mediated STAT1 phosphorylation and that the mechanism involves stabilization of IFN-ε mRNA leading to the up-regulation of RIG-I. These findings, combined with an increasing body of evidence supporting essential functions for RIG-I in the antiviral host response and in adaptive responses (17), point at a key role for IFN-ε and RIG-I as regulators of immunity.
Materials and Methods
Human recombinant TNF-α, polyinosinic:polycytidylic acid (poly(I:C)), and an anti-type I IFNR neutralizing Ab were from Calbiochem. Cycloheximide (CHX) and a pFLAG-CMV expression vector were purchased from Sigma-Aldrich. Lipofectamine 2000, Lipofectamine reagent, FBS, DNase I, Platinum Pfx DNA polymerase, pcDNA3.1/Zeo+, and TRIzol were from Invitrogen Life Technologies. We purchased reporter lysis buffer, the pGL3-basic vector, and a luciferase assay systems kit from Promega. Polyvinylidene fluoride membranes and chemiluminescence detection reagents were from PerkinElmer. Protein content was assessed with a Pierce protein assay kit. Taq polymerase and reagents for cDNA synthesis were from Fermentas. iQ SYBR Green Supermix was from Bio-Rad. ICN Pharmaceuticals provided an anti-actin mouse mAb, and anti-phospho-STAT1 (Tyr701) and anti-STAT1 were purchased from Cell Signaling. Anti-hemagglutinin Abs were from Clontech and a polyclonal anti-IFN-ε Ab was from Abnova. An HRP-conjugated secondary Ab was from BioSource and the anti-RIG-I antiserum was a generous gift from Dr. T. Imaizumi, Hirosaki University, Hirosaki, Japan.
HeLa cells were maintained in a 5% CO2 atmosphere at 37°C in DMEM supplemented with 10% FBS. Cells were serum starved 24 h before TNF-α treatment (10 ng/ml unless otherwise specified). We inhibited protein synthesis by pretreatment of the cells with CHX (1 or 5 μg/ml) 30 min before TNF-α treatment. Blockade of type I IFNR was accomplished by pretreatment of the cells for 10 min with a neutralizing Ab. In studies aimed at characterizing viral responses, we seeded HeLa cells at a density of 3 × 105 per 35-mm dish and then transfected the cells with 10 μg of poly(I:C) using Lipofectamine 2000, following the instructions provided by the manufacturer. The cells were harvested 6 h after transfection.
Conventional RNA interference (RNAi) approaches were avoided in this study owing to the fact that transfection with small interfering RNA (siRNA) activates the IFN system (18, 19). Instead, we used Stealth RNAi (Invitrogen Life Technologies), which has been shown to eliminate nonspecific effects such as the PKR/IFN stress response. The Stealth siRNAs targeting 25-nt sequences were designed using the BLOCK-iT RNAi Designer. The siRNA sequences used were: IFN-ε, 5′-AGAAGUCUUUGAGUCCUCAGCAGUA-3′ (sense) and 5′-UACUGCUGAGGACUCAAAGACUUCU-3′ (antisense); scrambled IFN-ε, 5′-AGAUUCUGAGUUCCUCGACAGAGUA-3′ (sense) and 5′-UACUCUGUCGAGGAACUC AGAAUCU-3′ (antisense).
To assess the efficiency of silencing, we overexpressed IFN-ε in HeLa cells and then transfected scrambled or IFN-ε-specific siRNAs. First, we cloned the cDNA encoding IFN-ε by amplifying cDNA isolated from HeLa cells using Pfx DNA polymerase and specific primers. We introduced a NotI site at the 5′-end of the sense primer and a 5′-SalI site in the antisense primer (both shown in lower case font), as follows: IFN-ε, 5′-GCTTgcggccgcGATGATTATCAAGCACTTCTT-3′ (sense) and 5′-TCTAGAgtcgacCCTCGGGCTTCTAAACTCTGT-3′ (antisense).
The amplified product was inserted into a mammalian expression vector (pFLAG-CMV) and the plasmid DNA (0.5 μg) was then transfected into HeLa cells using Lipofectamine reagent. Six hours later we transfected IFN-ε siRNA (250 picomoles) using Lipofectamine 2000. The cells were incubated for 48 h and then analyzed for the expression of IFN-ε mRNA and protein. To assess the role of silencing endogenous IFN-ε on TNF-α-mediated STAT1 phosphorylation and RIG-I levels, we transfected HeLa cells with IFN-ε siRNAs using Lipofectamine 2000 as described above. Forty-eight hours after transfection we treated the cells with TNF-α for the indicated periods and then harvested the cells.
We subjected DNase I-treated RNA (1 μg) to reverse transcription using a cDNA synthesis kit. A Chromo4 real-time PCR detection system (Bio-Rad) was used for quantitative assessment of RIG-I, IFN-α, -β, -ε, -κ, and -ω and GAPDH. We designed specific primers for quantitative PCR (see below) and used them at a final concentration of 200 nM with single-stranded cDNA as the template. The amplification reactions were performed with iQ SYBR Green Supermix according to the manufacturer’s specifications. Amplification conditions were as follows: 2 min at 50°C followed by 3 min at 95°C and then 40 cycles of 15 s at 95°C, 30 s at 55°C, and 30 s at 72°C. After amplification was complete, a melting curve was generated by heating slowly at 0.1°C per second to 95°C with continuous collection of fluorescence. Melting curves and quantitative analysis of the data were performed using an Opticon monitor, version 3.1.
The sequences of individual primers were: RIG-I, 5′-GCATATTGATGGACGTGGCA-3′ (forward) and 5′-CAGTCATGGCTGCAGTTCTGTC-3′ (reverse); IFN-α, 5′-AGAATCTCTCCTTTCTCCTG-3′ (forward) and 5′-TCTGACAACCTCCCAGGCAC-3′ (reverse); IFN-β, 5′-CCTGTGGCAATTGAATGGGAGGC-3′ (forward) and 5′-CCAGGCACAGTGACTGTACTCCTT-3′ (reverse); IFN-ε, 5′-AGGACACACTCTGGCCATTC-3′ (forward) and 5′-TTGCTTCATGTCGTTCAAGG-3′ (reverse); IFN-κ, 5′-TGAGTTGCCCCAAGAGTTTC-3′ (forward) and 5′-ACAATCTCCCAGGCACAGTC-3′ (reverse); IFN-ω, 5′-TAGCCCTGTTGGATCTCTGG-3′ (forward) and 5′-TGCAGTTGCTGATGAAGTCC-3′ (reverse); GAPDH, 5′-CCACCCATGGCAAATTCCATGGCA-3′ (forward) and 5′-TCTAGACGGCAGGTCAGGTCCACC-3′ (reverse).
Positive controls for amplifications were obtained using cDNA isolated from HeLa cells transfected with individual IFNs. The constructs used for transfection were obtained by amplification of HeLa cell genomic DNA. This was possible owing to the fact that the type I IFNs constitute a family of intronless genes (20). We used Platinum Pfx DNA polymerase for amplification and specific primers harboring a 5′-HindIII site (sense primers, shown in lower case font) or a 5′-XhoI site (antisense primers, shown in lower case font) as follows: IFN-α, 5′-cttaagCTTAATATCTACGATGGCCTCGCCCTT-3′ (sense) and 5′-AGActcgagTTATTCCTTCCTCCTTAATC-3′ (antisense); IFN-β, 5′-cttaagCTTTGTTGTCAACATGACCAACAAG-3′ (sense) and 5′-AGActcgagTCAGTTTCGGAGGTAACCTG-3′ (antisense); IFN-ε, 5′-cttaagCTTGACCTTCACCATGATTATCAAGCAC-3′ (sense) and 5′-AGActcgagCTACCTCGGGCTTCTAAACTC-3′ (antisense); IFN-κ, 5′-cttaagCTTTGCAAAAAAAATGAGCACCAAACCT-3′ (sense) and 5′-AGActcgagTTATTTCCTCCTGAATAGAGC-3′ (antisense); IFN-ω, 5′-cttaagCTTTCATTTCCCAATGGCCCTCCTG-3′ (sense) and 5′-AGActcgagTCAAGATGAGCCCAGGTCTC-3′ (antisense). The products were purified, digested with HindIII and XhoI and then cloned into pcDNA3.1/Zeo+.
After stimulation, cells were washed twice with PBS and lysed in cell lysis buffer (20 mM Tris (pH 7.4), 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100) After one cycle of freezing thawing, the lysates were cleared by centrifugation at 12,000 × g for 2 min at 4°C. We subjected 50 μg of protein extracts to electrophoresis on 8.5% SDS-PAGE gels and then transferred the proteins to polyvinylidene difluoride membranes that were blocked for 60 min at room temperature in 1× TBST buffer (50 mM Tris-HCl (pH 7.5), 250 mM NaCl, and 0.1% Tween 20) containing either 5% nonfat dry milk or 5% BSA. The membranes were incubated for 60 min at room temperature with the primary Ab indicated in each case. After five washes with blocking solution, we added an HRP-conjugated secondary Ab, incubated the membranes for 60 min at room temperature, repeated the washes using TBST, and then visualized the immunoreactive bands using chemiluminescence detection reagents.
Promoter activity assays
We introduced a promoter region of IFN-ε that comprised nt −497 to +296 into pGL3-basic. We used genomic DNA isolated from HeLa cells combined with a sense primer harboring a 5′-XhoI site and an anti-sense primer designed with a 5′-NcoI site (both shown in lower case font) as follows: XhoI-IFN-ε, 5′-GGGctcgagGGCTGTACCAACATATATG-3′ (sense) and NcoI- IFN-ε, 5′-CTTccatggTGGCTTACTTATTCATTGTATGCTT-3′ (antisense).
The cloned IFN-ε promoter region was transfected into HeLa cells and, 24 h later, we treated a portion of the cells with TNF-α (2 h at 10 ng/ml). The remaining cells were stimulated with poly(I:C) (5 h, 25 μg/ml) in the presence of Lipofectamine 2000. We harvested cellular protein using a reporter lysis buffer at 200 μl per well, determined total protein content, and assessed luciferase activity in the extracts.
mRNA stability studies
We stimulated HeLa cells with TNF-α (10 ng/ml) for 2 h and then inhibited RNA synthesis by the addition of actinomycin D (5 μg/ml) for various times as indicated (0–5 h). We isolated the RNA fraction and subjected aliquots to quantitative RT-PCR as described above. We determined the half-life of individual mRNAs using linear regression analysis.
TNF-α induces RIG-I expression in HeLa cells
Several studies have reported that treatment with a variety of cytokines and other stimuli result in the up-regulation of RIG-I. Both the magnitude and nature of these responses depended on the type of stimuli and the cell line tested. IFN-γ (21), IL-1β (22), LPS (23) and dsRNA (24) increased RIG-I expression levels in vascular smooth muscle (21), endothelial cells (24), and fibroblasts (22). It was recently reported that IFN-α and TNF-α treatment enhanced virus-induced expression of IFN-β by activating RIG-I expression in the lung epithelial cell line A549 (2). This system allowed the identification of RIG-I as the central regulator of influenza A virus-induced expression of antiviral cytokines in human lung epithelial cells. Our first objective was to identify a cellular system that allowed an investigation of the molecular mechanisms that result in RIG-I-induction and subsequent signaling in human cancer cells. We initially tested a panel of cancer cells under basal conditions and found that RIG-I was constitutively expressed at low levels in the human cervical cancer cell line HeLa (Fig. 1,A). In addition, we found that HeLa cells responded to cytokine stimulation by up-regulating the expression of RIG-I. Treatment with IFN-β or TNF-α significantly up-regulated the expression of RIG-I after prolonged treatment in a fashion similar to that observed in lung epithelial cells (2). In contrast, we found no significant changes in RIG-I expression in response to other stimuli such as LPS or IL-6 (Fig. 1 A). The levels of RIG-I were not affected after stimulation of a breast cancer (MCF-7) or a kidney embryonic cell line (HEK293) with TNF-α (not shown), although these cell lines have been previously reported to show other responses following treatment with this cytokine (25, 26). These combined results indicated that up-regulation of RIG-I occurred in response to challenge with specific inflammatory stimuli in a cell-specific manner and provided a cellular model to investigate the mechanisms that lead to RIG-I induction during immune responses.
We next investigated the kinetics of TNF-α- and IFN-β-mediated up-regulation of RIG-I mRNA. We found that RIG-I mRNA levels reached maximal expression after 2 h of exposure to IFN-β (Fig. 1,B, open squares). In contrast, the response to TNF-α proceeded with slower kinetics and reached a plateau after 8 h of stimulation (Fig. 1,B, closed circles). The delayed kinetics of TNF-α-mediated effects prompted us to investigate whether protein synthesis was required for the up-regulation of RIG-I mRNA levels in response to TNF-α. We inhibited protein synthesis in HeLa cells by pretreatment with CHX and then stimulated the cells with TNF-α. We found that inhibition of protein synthesis prevented the ability of TNF-α to increase the expression of RIG-I mRNA (Fig. 2). In contrast, stimulation with IFN-β, which presumably acts downstream of TNF-α (Fig. 1,B), resulted in the up-regulation of RIG-I mRNA in a protein synthesis-independent manner (Fig. 2). These results indicated that de novo protein synthesis was required for the ability of TNF-α, but not IFN-β, to induce up-regulation of RIG-I mRNA.
Signaling through type I IFNRs is required for TNF-α–mediated effects
Our next goal was to identify the signaling molecules involved in TNF-α–mediated up-regulation of RIG-I. We hypothesized that IFNs might play an intermediate role owing to observations (Figs. 1 and 2 and Refs. 2 and 9) indicating that RIG-I is up-regulated following exposure to IFNs. IFN leads to JAK-mediated tyrosine phosphorylation of STAT1 and STAT2, a process that results in the dimerization of these transcription factors, migration to the nucleus, and recognition of a conserved IFN stimulus response element within the promoters of target genes (27, 28). It was previously reported that exposure of HeLa and other cells to IFN-α or IFN-β leads to a transient phosphorylation of STAT1 that can be detected within a few minutes (Ref. 29 and our unpublished observations). We used the kinetics of STAT1 phosphorylation as a parameter to investigate whether the IFN pathway contributed to TNF-α-mediated STAT1 phosphorylation in HeLa cells. We found (Fig. 3) that STAT1 phosphorylation proceeded with relatively slow kinetics and became evident only 4 h after exposure to TNF-α. In addition, the ability of TNF-α to increase STAT1 phosphorylation required protein synthesis (data not shown). These observations, combined with our finding that TNF-α-mediated induction of RIG-I mRNA required de novo protein synthesis (Fig. 2), suggested that de novo synthesis of IFN might be required for this response.
Our next objective was to identify the family of IFNs required for TNF-α-induced STAT1 phosphorylation. We focused our attention on the family of type I IFNs, as it was previously reported that the type II IFNR family is primarily expressed in cells of the myeloid lineage. We found evidence supporting the participation of type I IFNR subsequent to TNF-α stimulation when we found that blockade of this receptor subfamily with a specific Ab (anti-IFNR1) almost completely prevented STAT1 phosphorylation (Fig. 4,A). We next investigated whether the expression levels of type I IFNR changed in response to stimulation with TNF-α. We found a similar expression of type I IFNR in unstimulated compared with TNF-α-stimulated HeLa cells (Fig. 4 B). These results are consistent with studies in which TNF-α was shown to have no effect on the expression of type I IFNR in smooth muscle cells (30). Our data suggested that the level of type I IFNR expression was not the limiting factor in determining the degree of STAT1 activation. Instead, the data suggested that TNF-α activated a signaling mechanism that led to the synthesis of new protein, the release of a ligand or ligands of type I IFNR to the extracellular medium, receptor activation, and STAT1 phosphorylation.
IFN-ε is required for TNF-α-induced STAT1 phosphorylation and induction of RIG-I in HeLa cells
To more precisely identify the nature of type I IFN responses in HeLa cells, we surveyed type I IFN expression in resting and stimulated cells using quantitative RT-PCR. We found that HeLa cells constitutively expressed mRNA for IFN-ε but not other previously identified members of the IFN family (Fig. 5,A). Additionally, the expression levels of IFN-ε, a recently discovered type I IFN family member identified in female reproductive organs, increased upon stimulation with TNF-α in a time- and concentration-dependent manner (Fig. 5, A and B). To investigate whether up-regulation of IFN-ε was necessary for TNF-α-induced STAT1 phosphorylation we used Stealth RNAi to silence the expression of IFN-ε. This approach completely suppressed the expression of transfected IFN-ε, as expected (Fig. 6,A). Silencing the expression of endogenous IFN-ε significantly prevented TNF-α-mediated STAT1 expression and phosphorylation and RIG-I expression (Fig. 6,B). The fact that after 8 h of exposure to TNF-α there was a low level of RIG-I expression suggested that IFN-ε-independent mechanisms contribute, in part, to the up-regulation of RIG-I in response to TNF-α. The nature of this mechanism(s) remains to be established. To further investigate the requirement for IFN-ε in TNF-α-mediated stimulation of STAT1 phosphorylation, we treated HeLa cells with a blocking anti-IFN-ε Ab before stimulation with the agonist. We found (Fig. 6 C) that blocking the action of IFN-ε also prevented the ability of TNF-α to induce STAT1 phosphorylation. These findings firmly established that IFN-ε is required for STAT1-mediated responses following stimulation with TNF-α. In addition, our results suggested a previously unrecognized role for IFN-ε as an intermediate in TNF-α-mediated up-regulation of RIG-I in HeLa cells.
TNF-α modulates IFN-ε levels by stabilizing its mRNA
We next investigated whether TNF-α regulates IFN-ε levels at the transcriptional level. We compared responses mediated by TNF-α and poly(I:C), a viral mimetic reported to transcriptionally activate the IFN-β gene (31). We found up-regulation of IFN-ε mRNA levels in HeLa cells exposed to TNF-α or transfected with poly(I:C), as expected (Fig. 7,A). Promoter studies using an IFN-ε luciferase expression construct indicated that viral stimulation activated IFN-ε expression at the transcriptional level (Fig. 7,B). In contrast, TNF-α–mediated up-regulation of IFN-ε was not accompanied by changes in promoter activity (Fig. 7,B). These results were consistent with a mechanism whereby TNF-α controlled the expression of IFN-ε at the posttranscriptional level and prompted us to investigate whether the half-life of IFN-ε mRNA was affected following stimulation with TNF-α. We inhibited RNA synthesis in HeLa cells using actinomycin D and then determined the rate of IFN-ε mRNA decay in the presence and absence of stimulation with TNF-α. We found (Fig. 8) that treatment with TNF-α resulted in robust stabilization of IFN-ε mRNA. These results were consistent with our promoter studies and established that TNF-α increases IFN-ε expression by a mechanism that involves mRNA stabilization in HeLa cells.
Several lines of evidence indicate that intracellular viral infection is detected by RIG-I and that this protein initiates events that activate host-associated antiviral responses (9, 32, 33, 34, 35). The precise mechanism by which RIG-I activates the innate immune system has been the subject of intense research (10, 14, 36, 37). The events that follow the binding of dsRNA by RIG-I include the homodimerization of IFN regulatory factor 3 and the subsequent up-regulation and expression of IFN genes (9). Previous studies demonstrated that the expression of RIG-I and a closely associated helicase, melanoma differentiation-associated gene 5 (MDA5), are transcriptionally regulated by IFN levels (9, 38). These combined results suggested that RIG-I and IFN levels are controlled by a regulatory mechanism involving reciprocal positive feedback participation of both IFN and RIG-I. In our studies, we characterized a cell-specific mechanism that controls RIG-I expression and involves a program of gene activation dependent on signaling mediated by type I IFNs. Our model consisted of stimulation with TNF-α, a cytokine generated primarily by activated macrophages (39) that mediates T cell-dependent immune responses (40), cellular senescence (41), and growth and differentiation (41). TNF-α-mediated effects during viral immune responses require IFN signaling, as demonstrated in studies in which the induction of MX1 was suppressed by the neutralization of type I IFNs with blocking Abs (42). In related work, Jacobsen and coworkers reported increased levels of IFN-β-specific transcripts after TNF treatment (43). In this study, we found that a cervical cancer cell line, but not a breast cancer line (MCF-7) or a kidney embryonic cell line (HEK293), responded to stimulation with TNF-α by up-regulating the expression of RIG-I. This response developed with relatively slow kinetics and required de novo protein synthesis.
Further studies identified IFN-ε as an intermediate required for STAT1 expression and phosphorylation and subsequent induction of RIG-I. Wong and coworkers previously reported the presence of a STAT1 intronic enhancer region that displayed type I and type II IFN-inducible elements (44). This is consistent with our observation that silencing the expression of IFN-ε decreased the ability of HeLa cells to up-regulate STAT1 expression in response to TNF-α. In humans, the type I IFN family of proteins has been classified based on sequence and expression patterns. The “leukocyte” IFNs include IFN-α and -ω (45, 46); IFN-β is a “fibroblast” IFN (47) and IFN-κ is expressed in keratinocytes (48). Despite their rich functional and structural diversity, all type I IFNs bind to type I IFNR, the activation of which initiates a number of phosphorylation events that result in the induction of antiviral genes. In our studies, treatment of HeLa cells with TNF-α did not stimulate production of most of the well-characterized members of the type I IFN family, including IFN-α, -β, -ω, and -κ. These results were in agreement with reports indicating that the levels of IFN-β are not affected by the exposure of HeLa cells to TNF-α (42, 43). In recent work, Hardy and coworkers identified IFN-ε as a new member of the type I IFN family (4). IFN-ε is preferentially expressed in female reproductive organs such as the ovaries and the uterus, but its precise physiological function is unknown. Our studies showed that HeLa cells, which were derived from a cervical cancer specimen, constitutively expressed IFN-ε mRNA. In addition, the expression of IFN-ε mRNA was up-regulated in response to TNF-α in a time- and concentration-dependent manner.
Unlike mechanisms that control the expression of “early response” genes, IFN-ε up-regulation following the stimulation of HeLa cells with TNF occurred with delayed kinetics. The requirement for de novo protein synthesis combined with observations in related systems (43) suggested that the synthesis of IFN-ε was necessary to induce STAT1 phosphorylation and RIG-I synthesis. Stabilization of IFN-ε mRNA, rather than transcriptional activation of the IFN-ε gene, contributed to the observed elevation in mRNA levels. These results are consistent with our observation that the promoter region of IFN-ε (4) does not contain obvious domains that could account for TNF-α-mediated transcriptional activation.
In conclusion, our data are consistent with a mechanism in which IFN-ε plays a key role as the predominant type I IFN involved in signaling events in HeLa cells (Fig. 9). De novo synthesis of IFN-ε seems required to induce STAT1 phosphorylation and increase RIG-I levels. TNF-α increased the levels of IFN-ε by a mechanism that involved mRNA stabilization in the absence of transcriptional activation (Fig. 9). This is in contrast to responses observed after viral challenge that showed a requirement for the transcriptional activation of IFN response genes. We speculate that this mechanism of RIG-I induction may be operative during innate immune responses and that its dysregulation may have pathological consequences.
We thank Drs. Tada-atsu Imaizumi and Hidemi Yoshida (Hirosaki University) for generously providing the anti-RIG-I antiserum and for their helpful advice. We are indebted to Drs. Matthew K. Topham and Xiaoqing Wu for their continued support and valuable suggestions.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Huntsman Cancer Foundation and by a grant from the National Institutes of Health (P01-CA73992).
Abbreviations used in this paper: RIG-I, retinoic acid-inducible gene-I; CHX, cycloheximide; poly(I:C), polyinosinic:polycytidylic acid; RNAi, RNA interference; siRNA, small interfering RNA.