LPS is the known component of bacterial pathogens that stimulates a number of proinflammatory factors. However, the mechanism of the induction of these factors by LPS has not been fully elucidated. We show here that LPS induces retinoic acid-inducible gene-I (RIG-I) in vitro and in vivo as a result from autocrine secretion of IFN-β in macrophages. TIR-domain-containing adapter-inducing IFN-β-deficient mouse embryo fibroblast (trif−/−) fail to show expression of RIG-I following LPS stimulation. Interference of RIG-I expression short interfering RNA represses the expression of LPS-induced TNF-α, whereas over-expression of RIG-I leads to the activation of TNF-α promoter and the induction of TNF-α expression. LPS- and IFN-β-induced TNF-α are suppressed in RIG-I-deficient mouse embryo fibroblasts (rig−/−). Thus, RIG-I plays a key role in the expression of TNF-α in macrophages in response to LPS stimulation, mainly for the late phase LPS-induced expression of TNF-α.
Innate immune cells mediate inflammatory responses to microbial pathogens. Bacterial components, such as endotoxin/LPS of Gram-negative bacteria, membrane lipoproteins of Gram-positive bacteria, and nonmethylated CpG-rich DNA of both Gram-negative and Gram-positive bacteria, are known potent initiators of a proinflammatory response. Binding of these bacterial components to cell surface receptors, i.e., TLRs, will initiate several signaling pathways, such as ERK, JNK, and p38, and NFκB pathways (1, 2). In turn, these signaling pathways activate cellular responses, including the production of proinflammatory cytokines, such as TNF-α and IL-1 (1, 2, 3).
Recent studies have shown that the TLR family proteins function as the primary receptors for many, and perhaps most, bacterial components. By far, 13 different TLRs are identified and dramatic progress has been made in defining the specificity of these TLRs for different microbial molecules. For example, the ligands for TLR2, TLR4, and TLR9 are known to be lipoprotein, LPS, and CpG DNA, respectively (4, 5, 6, 7). Although many similarities exist between the signaling pathways of the different TLRs (8), it is clear that stimulation of the different TLRs activates distinct cellular responses (9, 10, 11). The mechanisms used by different TLRs in initiating the downstream signaling pathways are not fully understood. There is still a gap in our knowledge regarding the link between TLR-initiated signaling and the downstream molecules that mediate the induction of proinflammatory factors. Defining the downstream signaling events and molecules would help to achieve a new level of understanding of how the cellular responses are elicited. Determination of the signaling pathways and molecules from ligand recognition to cytokine gene expression will ultimately provide the best means of identifying new therapeutic targets for treatment of inflammation-related diseases.
It is known that dsRNA, derived from viruses, can bind and trigger TLR3, and engagement of this receptor leads to the induction of type I IFN, which is a potent antiviral response (12). Interestingly, gene targeting has shown that TLR3 is not solely responsible for type I IFN production in virus-infected cells (13, 14, 15), indicating the presence of TLR3-independent pathway. Retinoic acid-inducible gene-I (RIG-I)4 is a recently identified protein that plays a key role in innate antiviral response (16). Several recent findings have demonstrated that RIG-I, an unusual RNA helicase protein, mediates the induction of type I IFN through TLR3-independent pathway and acts as a sensor for cytoplasmic viruses (13). RIG-I holds a unique structural feature, which contains both protein-interaction caspase recruit domain (CARD) and a catalytic domain required ATP. Both the ATP-dependent catalytic activity and CARD are required for activation of IRF3 (16). CARD alone is also capable of activating IRF3 and NF-κB, and induction of IFN-β when over-expressed in cells. Although its role in antiviral response is becoming clear, other physiological functions of RIG-I have been under-explored. In our present study, we report that RIG-I plays a key role in the regulation of TNF-α expression via TLR4 signaling pathway in macrophages treated with LPS.
Materials and Methods
Cells and reagents
RAW264.7 and HEK293 cells were cultured in RPMI 1640 and DMEM, respectively, supplemented with 10% FCS in a 5% CO2 incubator. Mouse peritoneal exudated macrophages were obtained by lavage 4 days after the injection of sterile 3% thioglycolate broth (1 ml i.p.). Cells were washed and resuspended in RPMI 1640 medium containing 10% FCS and standard supplements. Primary macrophages were plated in 24-well tissue culture dishes (1 × 106 cells/well). After 2-h incubation to allow for the adherence of macrophages, monolayer was washed three times to remove nonadherent cells and incubated with RPMI 1640 medium containing 10% FCS and standard supplements. The next day, reagents were added at different time points. Wild-type (WT), trif−/−, rig+/−, and rig−/− mouse embryo fibroblasts (MEFs), kindly provided by Dr. Bruce Beutler (Scripps Research Institute, La Jolla, CA) and Dr Zhugang Wang (Shanghai Jiao Tong University School of Medicine, China) (17), were cultured in DMEM supplemented with 10% FBS in a 5% CO2 incubator. LPS was from Sigma-Aldrich (serotype 0111:B4). RAW264.7 cells were transfected using FuGENE6/FuGENEHD (Roche).
Preparation of plasmid constructs
RIG-I and CARD and Helicase expression constructs were made as follows: total RNA from LPS-stimulated mouse macrophage RAW264.7 cells was reverse-transcribed using SuperScript II RT (Invitrogen). The full length of RIG-I cDNA and CARD (1–230aa) and Helicase (206–926aa) were amplified by PCR using specific primers and subcloned into the vector PET30a, PMD-18T, and PMD-18T, respectively. A KpnI/NotI fragment of RIG-I, BamHI/SalI fragment of Helicase, and KpnI/XbaI fragment of CARD were recovered from above vectors and subcloned into the same sites of pcDNA6A/myc/his, pcDNA3-Flag, and pcDNA6A/myc/his, respectively. TNF-α deletion and mutation series are gifts from Jiahuai Han, the Scripps Research Institute, La Jolla, CA (18).
RNAi vector constructs
To construct the short interfering RNA (siRNA)-expressing lentiviral vectors, the siRNA expression cassette was subcloned into FG12 between the XbaI and XhoI sites. The target sequence for siRNA 5′-GCCCATTGAAACCAAGAAATT-3′ was ligated into the BbsI-XhoI of pBS-SKII plasmid. The resulting plasmid was confirmed by DNA sequencing.
Lentiviral vector production
All vesicular steatites virus-G pseudo typed lentiviral vector stocks were produced by calcium phosphate-mediated transient transfect ion of HEK-293T cells. Briefly, HEK-293T cells were cultured in DMEM containing 10% FCS and standard supplements. Cells were cotransfected with appropriate amounts of vector plasmid, the HIV-1 lentiviral packaging constructs preserve, and pMDLg/pRRE, and the vesicular steatites virus-G expression plasmid pHCMVG. Viruses were collected from the culture supernatants on days 2 and 3 post-transfection and concentrated 100–1000-fold by ultracentrifugation and stored at−70°C. The concentrated virus stocks were titrated on HEK-293 T cells based on GFP expression. The RAW264.7 and primary mouse macrophage cells were transduced with concentrated lentiviral vector stocks at a multiplicity of infection of 50–100 in the presence of 4 μg/ml polybrene (Sigma-Aldrich). Cells were incubated overnight at 37°C and treated at the following day after refreshing the medium for recovery.
Luciferase reporter assays
The Dual-Luciferase Reporter Assay system (Promega) was used for luciferase assays. For RNA silencing of gene expression, cells were infected with RIG-siRNA-lentivirus targeting RIG-I and the control virus FG12, supplemented with 4 μg/ml ploybrene followed by cotransfection of pIFN-β-luc or pISRE-luc or pNF-κB-luc or AP-1-luc with the control vector SV40 or pRL-TK. Cells were stimulated with or without LPS for 16 h and then harvested for luciferase assay.
For RT-PCR analysis, cells were first washed twice with PBS, and then total RNA was isolated with Trizol reagent (Invitrogen) and then reverse-transcribed with SuperScript III reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. In tissue TaqMan PCR analysis, GAPDH was used as reference gene in both brain and liver. PCR were performed with following primers: β-actin: forward, 5′-TGCCGCATCCTCTTCCTC-3′; reverse, 5′-CGCCTTCACCGTTCCAGT-3′; RIG-I: forward, 5′-TCGGCGTTGGAGATGCTA-3′; reverse, 5′-TGGAAGAAGGCTTTGAGG-3′; MDA-5: forward, 5′-GCACCTACGCACTTTCC-3′; reverse, 5′-CCGAGCAGACCAGCAT-3′; TNF-α: forward, 5′-TCGTAGCAAACCACCAAG-3′; reverse, 5′-CAATGACTCCAAAGTAGACC-3′; and IFN-β: forward, 5′-GAGTTACACTGCCTTTGCC-3′; reverse, 5′-GATTCACTACCAGTCCCAGA-3′. In tissue Taqman PCR analysis, the following RIG-I primers are used: forward, 5′-CCACCTACATCCTCAGCTACATGA-3′; reverse, 5′-TGGGCCCTTGTTGTTCTTCT-3′. Results were analyzed by electrophoresis on agarose gel.
Cells were lysed with lysis buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% Nonidet p-40, 0.1 mg/ml leupeptin, 1 mM pheylmethylsulfonyl fluoride, and 1 mM sodium orthovanadate). After centrifugation at 12,000g for 20 min at 4°C, supernatant was used for immunoblots using the indicated Abs. RIG-I immuno-serum was made by injecting the rabbit with purified RIG-I protein. IFN-β Ab was obtained from R&D Systems (cat. no. 32400-1; PBL) and TNF-α Ab from Santa Cruz Biotechnology.
RAW264.7 cells (2 × 106/ml) in 12-well plates were treated with LPS (1 μg/ml) for 16 h, and then culture supernatants were harvested. TNF-α protein was measured with commercial ELISA kits (Boster) according to the instructions of the manufacturer.
Construction of adenovirus expression vector
Recombinant adenoviruses expressing RIG-I were prepared using the AdEasy system (Stratagene). In brief, the coding region of RIG-I was cloned into the NotI and PvuI site of the shuttle vector, Shuttle-IRES-hrGFP-1. To produce homologous recombination, Pme I linearized shuttle plasmid containing RIG-I and pAdEasy-1 plasmid were cotransformed into Escherichia coli BJ5183 cells. The resulting plasmid was then transformed into E. coli XL-Gold Ultracompetent cells. After amplification and purification, the recombinant plasmid was linearized with PacI and then transfected into the adenovirus packaging cell line 293A using LipofectAMINE (Invitrogen). Cell lysate was prepared 10 days after transfection and used to infect 293 cells. When most of the cells were killed by the adenovirus infection and detached, the cell lysate was harvested (the process was repeated three times). Control adenovirus expressing GFP (Ad-GFP) was prepared as the same.
In vivo animal model
Mice were treated with LPS by i.p. injection of LPS (Salmonella Abortis Equi, σ, dose: 500 μg/kg) or poly(I:C) (Amersham Biosciences; Little Chalfont, dose: 12 mg/kg), After 2, 6, and 24 h, groups of mice were terminally anesthetized and transcardially perfused with heparinized saline and tissue collected for quantitative PCR analysis. Samples from saline-treated mice were taken at 2 h. Brains from perfused mice were rapidly removed and a thick coronal section (2 mm) taken (−2.7 to −3.7). The dorsal hippocampus was then punched out from this section and kept in RNA-later at 4°C (Qiagen) until further use.
Total RNA was extracted from mouse tissue using RNeasy mini columns (Qiagen) according to the manufacturer’s instructions. Contaminating genomic DNA was degraded during extraction by use of the Qiagen DNase I enzyme. RNA samples were stored at −80°C until assay. Single-stranded cDNA was synthesized from 400 ng total RNA, and the cDNAs for RIG-I (Ddx58, NM_172689) and GAPDH were amplified by TaqMan PCR using specific primers. All procedures were performed under the authority of a U.K. Home Office License in accordance with the U.K. animals (Scientific Procedures) Act 1986, and after local ethical approval by the University of Southampton.
RIG-I is a LPS response gene
LPS, a known initiator of inflammation acts through the activation of LPS receptor TLR4, resulting in the expression of multiple cytokines and TLR response genes (1, 2, 3). We observed that LPS also induced a time- and dose-dependent expression of RIG-I in a macrophage cell line (RAW264.7 cells) (Fig. 1, A and B). The RIG-I homologue MDA-5 was not induced by LPS (Fig. 1,C). Stimulation of RAW264.7 cells not only resulted in increased expression of RIG-I mRNA, but we also found increased protein expression after exposure to LPS. (Fig. 1,D). Then, we tested whether the same response occurred in primary macrophages. Induction of RIG-I by LPS in primary macrophages was similar as compared with RAW264.7. RIG-I induction increased over time and correlated with the induction of IFN-β and TNF-α (Fig. 1 E), although overall expression levels were lower in primary macrophages.
Because viruses and poly(I:C) are able to induce RIG-I (19), therefore, we compared RIG-I expression pattern in macrophages treated with LPS or poly(I:C) using real-time PCR. Both poly(I:C) and LPS can induce RIG-I expression in RAW264.7 cells and primary macrophages, with poly(I:C) being a more potent inducer in RAW264.7 cells. Interestingly, in primary macrophages, LPS was shown to be a more potent inducer of RIG-1 expression as compared with poly(I:C) (Fig. 1,F). Finally, we tested whether LPS can induce RIG-I in vivo. Mice were given LPS (500 μg/kg by i.p. administration) and tissue was collected at different time points to assess RIG-I expression using real-time PCR. As shown in Fig. 1, G and H, induction of RIG-I was shown both in brain and liver after LPS and poly(I:C) challenge. Taken together, RIG-I is a LPS response gene in vitro and in vivo.
Induction of RIG-I by LPS is through autocrine secretion of IFN-β
LPS treatment of macrophages leads to induction and secretion of IFN-β, and autocrine secreted IFN-β leads to the expression of IFN-β response genes (IRGs) in cells (20). We observed that induction of IFN-β appeared earlier than the induction of RIG-I in response to LPS challenge (Fig. 1,A). We, thus, addressed whether the expression of RIG-I was induced by autocrine secreted IFN-β. RAW264.7 cells were treated with LPS in the presence of IFN-β Ab, and it showed that this Ab blocked the induction of RIG-I (Fig. 2,A). RIG-I could also be induced directly by IFN-β in macrophages in a dose- and time-dependent manner (Fig. 2, B and C), and also in 293T cells (Fig. 2,D), indicating that autocrine secreted IFN-β is an inducer of RIG-I expression under LPS challenge. Pretreatment of macrophages with TNF-α Ab did not affect the induction of RIG-I by LPS (Fig. 2,E). IFN-β triggers JAK pathway to induce IRGs through its receptor. We predicted that neutralization of IFN-β and inhibition of JAK activation could block the signaling and suppress the induction of RIG-I by IFN-β, as well as the induction of TNF-α. RAW264.7 cells were pretreated with JAK-I (the inhibitor of JAK) for 30 min or neutralized with IFN-β Ab before LPS stimulation. Addition of JAK inhibitor blocked LPS-induced RIG-I, but only partially blocked the induction and secretion of TNF-α (Fig. 2, F and G). Addition of IFN-β Ab also only blocks the secretion of TNF-α partially (Fig. 2 H).
TIR-domain-containing adapter-inducing IFN-β (TRIF) is a signaling molecule downstream of TLR4, and it mediates the induction of TRIF response genes, including IFN-β, by LPS. It showed that MEF deficient in TRIF did not induce RIG-I after stimulation with LPS (Fig. 2 I). We concluded that the induction of RIG-I by LPS is mediated at least partially by the autocrine secretion IFN-β, resulting in decreased TNF-α production.
RIG-I deficiency in cell suppresses the induction of TNF-α by LPS
RIG-I is able to activate both NF-κB and IRF signaling pathways, which are key transcription factors for the expression of proinflammatory factors (16). We next addressed the role of RIG-I in the regulation of LPS response genes. siRNA technique was used to silence endogenous RIG-I in macrophages. RAW264.7 cells infected with siRNA lentivirus no longer showed induction of RIG-I after stimulation with LPS, whereas control siRNA lentivirus infected cells did not block the induction (Fig. 3,A). We next tested the effect of silencing RIG-I on the expression of TNF-α in LPS-treated RAW264.7 cells and primary macrophages. Cells were transfected with lentivirus carrying RIG-I siRNA, and then treated with LPS for 16 h. Induction of TNF-α was suppressed in LPS-stimulated RAW264.7 cells (Fig. 3,B), as well as in primary macrophages (Fig. 3,C). To confirm this observation, RIG-I-deficient MEF were tested. TNF-α was induced in rig+/+ and rig+/− MEFs, but not in rig−/− MEF, after LPS challenge for 16 h (Fig. 3,D). Autocrine secretion IFN-β was identified to be an inducer of RIG-I (Fig. 2). IFN-β was also able to induce TNF-α in rig+/+ MEF (Fig. 3,E) while in rig−/− MEF it could not (Fig. 3,E). The induction of TNF-α by LPS was reduced in the presence of IFN-β Ab as well as in cells pretreated with JAK inhibitor, although the inhibition was not completed (Fig. 2, F and G). Furthermore, we used a reporter assay to evaluate the role of RIG-I on the LPS-induced TNF-α. RAW 264.7 cell and primary macrophages were infected with siRNA of RIG-I, and then transfected using TNF-α reporter followed by the treatment of LPS. As shown in Fig. 3 F, the activation of TNF-α by LPS failed when primary macrophages were silenced in RIG-I. These results suggested that RIG-I was required for the induction of TNF-α by LPS.
RIG-I mediates later induction of TNF-α
It is known that induction of TNF-α by LPS can be either an early and/or late stage event (21), thus, it is necessary to clarify in which stage RIG-I is required for LPS-induced expression of TNF-α. Primary macrophages were transfected with or without siRNA of RIG-I followed by the treatment of LPS for different time points. At the mRNA level, we observed a biphasic TNF-α induction: the first phase peaking occurred at 2 h, while the second induction started at ∼8 h (Fig. 4,A). The second phase was reduced in RIG-I-silenced macrophages, whereas the early induction was not affected. We further tested this effect at the protein level by ELISA. As shown in Fig. 4,B, LPS induced biphasic secretion of TNF-α at protein level, with similar kinetics compared with the mRNA level. After siRNA treatment, the second phase of TNF-a secretion was reduced by ∼40%, whereas the early phase of secretion remained unchanged (Fig. 4 C). These data suggest that RIG-I mediates the late phase induction of TNF-α in response to LPS.
Over-expression of RIG-I leads to the activation of TNF-α promoters and induction of TNF-α
We next addressed whether RIG-I can directly activate the TNF-α promoter, leading to the induction of TNF-α. An adenovirus vector-expressing RIG-I was constructed, and infection of cells with this vector resulted in expression of ectopic RIG-I (Fig. 5,A). The TNF-α promoter was strongly activated in 293T cells and to a lesser extent also in RAW264.7 cells after over-expression of RIG-I (Fig. 5,B). A series of luciferase reporter constructs containing progressive deletions in the TNF-α 5′ promoter sequence (22) were used to further test the effect of RIG-I on the activation of TNF-α promoter. RIG-I-mediated activation of −615 promoter is comparable to that activated by MKK6. In contrast, deleting the sequence upstream of position −161 strongly reduced RIG-I-mediated activation (Fig. 5 C). The region between −1 and −161 of TNF-α promoter contains several well known cis-elements, such as CRE, κB-3, AP-1, and AP-2 like motifs, which are required for the full induction of TNF-α by MAPK (18). These elements do not appear to be required for RIG-I-mediated activation. According our data, elements on TNF-α promoter between −615 to −161 are essential for RIG-I-mediated activation. An Egr-1 binding sequence was found at −172 in this region, which is a transcription factor induced by LPS. However, mutation of −172 Egr-1 recognized site did not affect the promoter activity, indicating that Egr-1 site does not mediate RIG-I activation of TNF-α promoter. Sequence analysis reveals the presence of κB-1 at −587. This site was deleted and the promoter activity with this deletion was tested. Deletion of −587 not only resulted in the loss of RIG-I-mediated activation of the promoter, it also resulted in loss of the basal activity of the promoter, indicating that κB-1 site is the key element for RIG-I-mediated activation and basal activity of TNF-α promoter.
We next tested whether RIG-I was able to directly drive the expression of TNF-α. RAW264.7 cells were infected with Ad-RIG-I. Expression of TNF-α was determined by RT-PCR. Over-expression of ectopic RIG-I in RAW264.7 cells led to the induction of TNF-α at mRNA (Fig. 5,D) and protein (Fig. 5 E) levels. Although adenovirus itself causes induction of RIG-I and TNF-α, there is an additional induction of TNF-α as a result of ectopic RIG-I expression rather than the adenovirus per se.
It is well known that many cells recognize LPS via TLR4 to activate the inflammatory response, leading to the induction of the expression of various genes. However, the molecular mechanisms by which LPS induce the expression of cytokines, chemokines, and other response genes through TLR signaling are still unknown. We now demonstrate that RIG-I plays a key role in LPS-induced TNF-α in macrophages. The role of RIG-I in RAW264.7 cells and primary macrophages were comparable, though the expression levels of RIG-I in primary cells were lower as compared with the RAW264.7 cell line cells. LPS induces the expression of RIG-I in vitro and in vivo. Administration of LPS to mice results in increased RIG-1 expression level in brain, liver (Fig. 1, G and H), and spleen (data not shown). RIG-I expression also correlates with the induction of IFN-β and TNF-α (Fig. 1,E). Induction of RIG-I is mediated by autocrine secretion of IFN-β, because it is suppressed either by the addition of IFN-β Ab (Fig. 2,A) or JAK pathway inhibitor (Fig. 2,F). TRIF is a downstream molecule that mediates TLR4 and TLR3 signaling to induce IFN-β. TRIF-deficient MEFs are unable to induce RIG-I in response to LPS (Fig. 2 I), which supports the hypothesis that autocrine secreted IFN-β induces RIG-I in LPS-treated macrophages. Although TNF-α is also a secreted factor that is involved in the autocrine induction of gene expression, it is not responsible for the induction of RIG-I (data not shown). However, it enhances that RIG-I-mediated activation of NF-κB and ISRE reporter (data not shown), suggesting that TNF-α is involved in RIG-I-mediated induction of gene expression by different ways. Thus, we conclude that RIG-I is a LPS response gene.
RIG-I belongs to the RNA helicase family with a unique structural feature (16). It confers a CARD motif that has been identified as a domain mediating the interaction with other molecules, and transducing the signals for some crucial cellular processes, such as apoptosis and gene regulation (23, 24). The observation of the induction of RIG-I in response to LPS stimulation in macrophages and in vivo prompted us to address its role in the regulation of LPS-induced genes. Disruption of RIG-I in LPS-treated macrophages by siRNA suppresses the induction of TNF-α as well as IFN-β (Fig. 3, B and C). Results from RIG-I-deficient MEFs are consistent with data from siRNA experiments. LPS induces the expression of TNF-α only in WT and heterozygous rig+/− MEF, but not in homozygous rig−/− MEF (Fig. 3, D and E). Furthermore, over-expression of RIG-I directly activates the promoters of TNF-α (Fig. 5,B) and leads to the expression of TNF-α in macrophage (Fig. 5, D and E). It is interesting that RIG-I-mediated activation of TNF-α promoter is not dependent on the known cis-elements, such as Egr-1, κB like motif, CRE, AP-1, and AP-2 (Fig. 4, B and C), but rather relies on −587 site, a κB-1 like site that has never been addressed before. This site is not only important for RIG-I-mediated TNF-α activation but also for basal activity level of the TNF-α promoter (Fig. 4 C). Therefore, these data strongly demonstrate that RIG-I is required to induce the expression of late phase TNF-α by LPS. It has been demonstrated that RIG-I-mediated activation of NF-κB completely depends on its downstream molecule, VISA/MAVS/IPS-1/Cardif. It is likely that RIG-I-mediated induction of TNF-α also requires the activation of VISA/MAVS/IPS-1/Cardif signaling.
TLR4 is the receptor that mediates LPS signaling through MyD88 and TRIF pathways (4, 25, 26, 27, 28, 29). LPS-induced genes are divided into two classes, MyD88 dependent and TRIF dependent (30). Both pathways activate NF-κB, leading to the expression of proinflammatory cytokines. Induction of inflammatory genes can be mediated by early and delayed activation of NF-κB, mediated by MyD88 and TRIF pathways, respectively. We have shown that RIG-I induction is TRIF dependent, through induction of IFN-β (Fig. 2,I). RIG-I does not affect early induction but only the later phase induction of TNF-α by LPS (Fig. 4), suggesting that RIG-I is responsible for later TRIF-dependent activation of NF-κB and induction of TNF-α.
Though autocrine secreted IFN-β mediates the induction of RIG-I in LPS-treated macrophages (Fig. 2), neutralization of IFN-β with its Ab and inhibition JAK pathway by inhibitor only partially affects the induction of TNF-α by LPS (Fig. 2, F, G, and H). One explanation is that LPS induces the expression of multiple cytokines, through the MyD88-dependent early NF-κB activation pathway, which also are able to induce the expression of RIG-I. For example, IFN-γ could stimulate the expression of RIG-I in MCF-7 cells (31), as well as in monocyte-derived macrophages (32). Therefore, inhibition of JAK either by IFN-β Ab or the inhibitor of JAK could only block JAK pathway-mediated induction of RIG-I, leading to the partial inhibition of the induction of TNF-α. In contrast, MyD88-induced early secretion of TNF-α is independent of RIG-I. Early TNF-a production is induced directly through LPS-TLR4-MyD88-dependent pathway, which is independent of RIG-I. RIG-I is induced by autocrined IFN-β, which results from LPS-TLR4-TRIF pathway and is also an early event after LPS stimulation. Therefore, RIG-I only participate in the late induction of TNF-α.
Our data also support the hypothesis by Marco and his colleagues (33). They proposed that LPS-induced cytokine synthesis is composed of two distinct phases. Induction of gene expression in first phase is a process directly induced by LPS that lasts for several hours. Expression of genes in the second phase occurs independent of LPS and results from the autocrine secreted cytokines. In this study, we show that induction of RIG-I by LPS at 16 h is mediated by autocrine secretion of IFN-β, leading to TNF-α production at the late phase. In this regard, induction of TNF-α by LPS in the later phase is independent of LPS, but dependent on RIG-I. Therefore, RIG-I is essential for LPS-induced TNF-α, and perhaps for many other LPS-induced genes as well, especially those expressed in the late stage. Covert et al. (34) has proposed that later phase induction of TNF-α by LPS is IRF3 dependent. The time window that they looked at is still very early, ∼150–200 min after LPS challenge, which is not the later phase identified by Marco et al. and us.
Because amplification of inflammatory factors usually occurs by so called autoloop cascade, we propose a hypothesis that RIG-I can be a key factor in autoloop cascade for the amplification of proinflammatory factors, and likely an important regulator of the late phase cytokine induction (Fig. 6). The role of RIG-I in this loop is summarized in schematic form (Fig. 5). Although MDA-5 is a homologue of RIG-I and also plays the same role as RIG-I in antiviral signaling (35), it seems not involved in LPS response.
In addition to pathogen stimulated inflammation, many other disease states, such as severe trauma, burning, and surgery, can induce the synthesis of proinflammatory mediators (36, 37, 38). We have observed that serum deprivation could induce the expression of RIG-I in vitro independent of TLR4 and may in turn lead to the induction of proinflammatory genes (data not shown). It is possible that a variety of stressors lead to the induction of RIG-I, and this could be the one of the mechanisms to induce the expression of proinflammatory factors.
LPS stimulates macrophage/dendritic cells to produce various proinflammatory cytokines, chemokines, and type I IFN. The pleiotropic effects of LPS in humans are widely accepted as a cause of endotoxin shock that mainly results from sustained or over-expression of proinflammatory factors. Our data demonstrate that RIG-I is a key factor controlling the expression of TNF-α by LPS, indicating that it could be a target molecule for the therapy of diseases that over-react to the presence of LPS, such as sepsis, septic shock, and systemic inflammatory response syndrome.
In summary, RIG-I plays a key role in late phase expression of TNF-α induced by LPS. Targeting RIG-I could be a key approach to control TLR4-dependent and -independent inflammatory gene expression.
We thank Dr. Bruce Beutler (Scripps Research Institute) and Zhugang Wang (Shanghai Jiao Tong University School of Medicine, China) for supplying MEFs, as well as Jiahuai Han (Scripps Research Institute) for the mutant plasmids. We also thank Jessica L. Teeling and V. Hugh Perry (University of Southampton, U.K.) for revising this manuscript.
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 Grants 30330260 and 30470841 from National Science Foundation, Grant 2006CB503802 from National Basic Research Program, China, and by China Scholarship Programs.
Abbreviations used in this paper: RIG-I, retinoic acid-inducible gene-I; MEF, mouse embryo fibroblast; TRIF, TIR-domain-containing adapter-inducing IFN-β; CARD, caspase recruit domain; WT, wild type; siRNA, short interfering RNA; IRG, IFN-β response gene.