Potential conflict of interest: Nothing to report.
During a viral infection, binding of viral double-stranded RNAs (dsRNAs) to the cytosolic RNA helicase RIG-1 leads to recruitment of the mitochondria-associated Cardif protein, involved in activation of the IRF3-phosphorylating IKKε/TBK1 kinases, interferon (IFN) induction, and development of the innate immune response. The hepatitis C virus (HCV) NS3/4A protease cleaves Cardif and abrogates both IKKε/TBK1 activation and IFN induction. By using an HCV replicon model, we previously showed that ectopic overexpression of IKKε can inhibit HCV expression. Here, analysis of the IKKε transcriptome profile in these HCV replicon cells showed induction of several genes associated with the antiviral action of IFN. Interestingly, IKKε still inhibits HCV expression in the presence of neutralizing antibodies to IFN receptors or in the presence of a dominant negative STAT1α mutant. This suggests that good IKKε expression levels are important for rapid activation of the cellular antiviral response in HCV-infected cells, in addition to provoking IFN induction. To determine the physiological importance of IKKε in HCV infection, we then analyzed its expression levels in liver biopsy specimens from HCV-infected patients. This analysis also included genes of the IFN induction pathway (RIG-I, MDA5, LGP2, Cardif, TBK1), and three IKKε-induced genes (IFN-β, CCL3, and ISG15). The results show significant inhibition of expression of IKKε and of the RNA helicases RIG-I/MDA5/LGP2 in the HCV-infected patients, whereas expression of TBK1 and Cardif was not significantly altered. In conclusion, given the antiviral potential of IKKε and of the RNA helicases, these in vivo data strongly support an important role for these genes in the control of HCV infection.(HEPATOLOGY 2006;44:1635–1647.)
Hepatitis C virus (HCV) infection leads to the development of chronic hepatitis in 60% to 90% of infected individuals, cirrhosis in 0.5% to 30% of cases, and hepatocellular carcinoma at a rate of 1% to 3% per year.1 The current approved treatment for HCV infection is pegylated interferon alpha (IFN-α) in combination with ribavirin. This leads to clearance of the virus in 50% to 80% of cases, depending on the infecting HCV genotype. In particular, HCV of genotype 1 is the most resistant to IFN treatment.2 HCV can interfere with the response of cells to IFN, through its viral proteins targeting either the IFN-activated JAK/STAT signaling pathway3–5 or through other processes, leading to inhibition of action of the IFN-induced antiviral proteins.6 In addition to this, HCV interferes with IFN induction, and more generally, with the induction of the innate immune response. The innate immune response is triggered in response to a variety of pathogens, such as bacteria and viruses, and is essential for a rapid limitation in the spread or action of these pathogens. IFN induction can take place after interaction of extracellular nucleic acids to members of the Toll-like receptor family7–10 and after viral intrusion in the cytoplasm through the interaction of viral double-stranded RNAs (dsRNAs) with specific RNA helicases, such as RIG-I11 or MDA-5.12 For instance, the interaction of dsRNA with RIG-I provokes a change in its conformation that allows its N terminus CARD to trigger IFN induction11, 13, 14 through association with the recently identified IPS-1/MAVS/VISA/Cardif protein.15–18 The particularity of this latter adapter protein, referred to here as Cardif, is to localize to the mitochondrial membrane.16 Cardif, in turn, recruits MAPKs, the IKKαβγ complex and the TBK1/IKKε kinases, thus provoking activation of nuclear factor kappaB (NF-κB) and IRF319, 20 and, ultimately, IFN induction.19, 20
The genome of the hepatitis C virus presents structured dsRNA regions, such as its 5′NTR and 3′NTR, which can trigger IFN-β induction soon after introduction in the cellular host.21 However, HCV interferes with IFN induction through the action of its NS3/4A protease, which can cleave Cardif.18, 22, 23 In addition, the NS3/4A protease can cleave the TRIF adapter, which links the IFN-inducing kinases to the dsRNA-activated TLR3,24 thus emphasizing the importance of these two IFN signaling pathways and the necessity for the virus to inhibit them to favor its propagation.
The replication of HCV in the host may thus depend on a balance between its ability to abrogate both IFN induction and action and the ability of the cells to mount an efficient antiviral response. We have recently shown that the TBK1/IKKε kinases are not directly affected by the NS3/4A protease and that overexpression of IKKε in HCV replicon cells results in the inhibition of the HCV expression by 80%.25
In addition to phosphorylating IRF3 and inducing IFN, IKKε is also known to induce genes through the NF-κB and c/EBPδ transcription factors.26–28 Therefore, to have exhaustive information on the involvement of IKKε in an antiviral action against HCV, we have analyzed its transcriptome profile after provoking its overexpression by transfection in cells expressing an HCV replicon. This revealed several genes involved in translational control, genes involved in apoptosis, and genes of the ISGylation pathway as well as genes from the chemokine family. Several of these genes are known to be induced directly through IRF3 and NF-κB, without the need for IFN induction, and, indeed, we could demonstrate that the antiviral response triggered by IKKε can develop in absence of IFN action. This shows that IKKε has the potential to provoke an immediate innate immune response against a viral intrusion. Furthermore, analysis of liver biopsies indicated a downregulation of the IKKε expression levels in the HCV-infected patients who were nonresponders to IFN treatment. Importantly, expression levels of RIG-I and MDA5, two RNA helicases responsible for IKKε activation, through Cardif, were found to be inhibited in all HCV-infected patients. This indicates that HCV may preferentially replicate and propagate from cells in which the IFN-inducing pathway is severely impaired.
The pcDNA3/AMP/zeo/flag (IKKεwt) and pcDNA3/AMP/zeo/flag (IKKεK38A) plasmids were described previously.19, 25 The pcDNA1/AMP and the pISRE-Luc were obtained from Invitrogen (Carlsbad, CA) and Stratagene (La Jolla, CA), respectively. The pRC/CMV flag STAT1α Y701F (Addgene plasmid 8702)29 and pRC/CMV flag STAT1α (Addgene plasmid 8691)30 were obtained from J Darnell through Addgene.
For neutralization of IFN-α binding, we used a purified preparation of anti-IFNAR1 monoclonal antibody 64G12 as described.31, 32 For the detection of total levels and phosphorylated levels of Stat1, we used polyclonal antibodies anti-Stat1, anti phosphoSTAT1 Ser727, and anti phosphoSTAT1 Tyr701 (Cell Signaling Technology, Danvers, MA). For the detection of flag STAT1α Y701F, flag STAT1α, and flag (IKKεwt), we used anti-flag monoclonal antibody (anti-FLAG M2; Sigma, St. Louis, MO). For individual detection of IKKε, we used mouse anti-IKKε antibodies from BD BioSciences (Franklin Lakes, NJ).23 Protein loading control was performed with monoclonal anti-actin antibodies (Sigma).
The Huh-7 and Huh-7 carrying the full-length HCV replicon 5.1 of genotype 1b were cultured as described.25
Transfection and Reporter Assay.
Cells were transfected using lipofectamine 2000 (Invitrogen), and the reporter assays were performed as described.25
Real-Time RT-PCR Analysis.
Total cellular RNA was extracted by acid-guanidinium thiocyanate-phenol chloroform using RNAble (Eurobio, Les Ulis, France) and according to the manufacturer's instructions. The sequence of the primers used for the reverse transcription polymerase chain reaction (RT-PCR) of GAPDH, IFN-β, IKKε, ISG15, ISG56, MDA-5, NOXA, and RIG-I were designed using the software LC Probe design (Roche, Basel, Switzerland) and choosing primers on each side of an intron (Supplementary Table 1; available at the HEPATOLOGY website (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). The sets of HCV primers33 were designed as described.33 (Supplementary Table 1). The following sets of primers were obtained commercially from the Qiagen (Valencia, CA) company (Human CCL3, gi: 48949814; Human CEBPδ, gi: 288772795; Human HECT E3, gi: 7705930; Human ISG 1-8U, gi: 11995467; Human KLF4, gi: 31560548; Human RANTES, gi: 22538813; Human IFI-6-16, gi: 13259551). For each RT, the GAPDH antisense primer was included in each tube, for subsequent normalization. In the case of IKKε, IFN-β, and HCV, the RT was performed on 1 μg total RNA using each specific antisense primer and the rTth polymerase (Applied Biosystems, Foster City, CA) as described.33 For the following genes: CCL3, CEBPδ, HECT E3, ISG15, ISG56, ISG 1-8U, IFI-6-16, KLF4, MDA-5, NOXA, RANTES, RIG-I, the RT was performed using the MuLV reverse transcriptase and oligo dT primers (Applied Biosystems). The cDNAs were then purified with the high pure PCR product purification kit (Roche Applied Science, Indianapolis, IN) in a final volume of 50 μL. Quantitative PCR was performed on a light cycler apparatus using SYBR Green, with 2 to 5 μL of the purified cDNA in a 10-μL reaction mixture containing 2 μL Light Cycler FastStart Plus DNA master SYBR Green Kit (Roche) and 0.5 μmol/L of each primer, as described.33 Standard curves were established using 10-fold serial dilutions of: (1) plasmids containing the entire cDNA of the gene of interest (pcDNA3/AMP/zeo/flag (IKKεwt); (2) plasmids containing amplicons (GAPDH, IFN-β, ISG15, ISG56); (3) HCV synthetic plus strand RNAs; and (4) purified PCR products (CCL3, CEBPδ, HECT E3, ISG 1-8U, IFI-6-16, KLF4, MDA-5, NOXA, RANTES, RIG-I). The measured amounts of each mRNA were normalized to the amounts of GAPDH mRNA. The amounts of RNA were expressed as number of copies/μg total RNA.
Quantitative PCR of Liver Biopsies.
Before RNA extraction, liver tissues were homogenized in 1 mL Ultraspec (Biotex, Houston, TX), and total RNA was obtained following the Ultraspec protocol. Two micrograms of the RNA preparations were submitted to DNAse (1 U/μL, Gibco-BRL, Grand Island, NY) in a total volume of 20 μL, for 15 minutes at room temperature, followed by inactivation at 65°C for 10 minutes in the presence of 2.5 μmol/L EDTA and 1 minute at 90°C. The reverse transcription was performed in 42 μL using 2.4 U Mu-MLV reverse transcriptase (Gibco-BRL) and 400 ng random primers (Roche) for 60 minutes at 37°C followed by 1 minute at 94°C. Quantitative PCR was performed as described above using 2 μL of the reverse transcription reaction and the primers for RIG-I, MDA5, LGP2, Cardif, TBK1, IKKε, CCL3, IFN-β, and ISG15 (Supplementary Table 1). For internal control genes, we used primers for three genes (TRIM44, HMBS, and BC002942) that have been determined to be the most stable and expressed at low levels in liver and are therefore the most representative genes to allow normalization of the samples34 (Agnès Marchiato and Pascal Pineau, personal communication). For each liver biopsy, the arithmetic mean of the three Ct (Cycle treshhold) obtained for the internal controls was deduced from the Ct of the gene of interest to give a ΔCt value. These were then transformed to relative quantities using the formula xΔCt, where x represents the efficiency of PCR amplification in the Light Cycler apparatus (in our experiments, x = 1.9, representing an efficiency of 90%).
Huh7 and Huh7 HCV replicon cells were seeded at 2 × 106 cells and at 2.2 × 106 cells/100-mm plates, respectively, and transfected after 24 hours using lipofectamine 2000 (Invitrogen). At 24 hours after transfection, the cells were scraped in their culture medium, pelleted by centrifugation, washed twice in phosphate-buffered saline, and the final pellet was either stored at −80°C before further processing or processed immediately. The cell pellets were resuspended in 400 μL lysis buffer (20 mmol/L HEPES,35 1% Triton X-100, 100 mmol/L NaCl, 1% aprotinin, 1 mmol/L PMSF) containing 1 mmol/L sodium orthovanadate (Na3VO4), 10 mmol/L β-glycerophosphate, and 50 mmol/L sodium fluoride (NaF) as phosphatase inhibitors as described in Tanabe et al.36 After 20 minutes of incubation on ice, the cell extracts were centrifuged at 12,000g for 20 minutes at 4°C, transferred to other tubes, and stored at −80°C. The immunoblot analysis was performed as described.25
DNA Microarray Protocol.
Total cellular RNAs were extracted by acid-guanidinium thiocyanate-phenol chloroform using RNAble (Eurobio) and further purified with an RNAeasy kit (Qiagen).The quality of RNAs was monitored on Agilent RNA Nano LabChips (Agilent Technologies, Santa Clara, CA). The reverse transcription (on 5 μg total RNA using OligodT primers), the in vitro transcription of the cDNA in presence of biotin, and the DNA fragmentation were performed according to Affymetrix standard protocols. Fragmented, biotin-labeled cRNA samples were hybridized on GeneChip® Human Genome U133A 2.0, containing probe sets representing 18,400 transcripts and variants including 14,500 well-characterized human genes, using Affymetrix standard protocols. For each sample, arrays were performed in triplicate, to minimize experimental variations. For each array, the cell intensity files (*.CEL) were generated with GeneChipOperating Software. Data analysis was performed using SPlus ArrayAnalyser software (Insightful). Data processing was done with the RMA method.37, 38 Statistical analysis to compare replicates arrays was done using the Local Pool error test.39 The P values (the probability that the variability in a gene behavior observed between classes could occur by chance) were adjusted using Benjamini-Hochberg algorithm.40 Upregulated genes were distributed into clusters of cell functions using NetAffyx Gene Ontology Mining Tool software.
Microarray Analysis of Genes Induced by IKKε in the Huh-7 Cells Expressing the HCV Replicon.
Total RNAs were isolated 48 hours after transfection of Huh-7 cells expressing a full-length HCV replicon with either an IKKε expressing vector or an empty vector, used as control. HCV expression was inhibited in the IKKε-transfected cells as previously reported (not shown25). After purification, the mRNAs were reverse-transcribed using oligo-dT as primers and labeled with biotin after reamplification from the resulting cDNAs. They were then used as probes to hybridize 14,500 gene-expressing chips (Affymetrix; Affymetrix, Santa Clara, CA). This revealed a specific induction of 43 genes at or above twofold (base log2) and a limited number of 24 downregulated genes (Table 1) out of a list of 231 genes retrieved by the program (Supplementary Table 2). Specific induction was confirmed by qRT-PCR on 12 of the induced genes (CCL3, ISG15, MDA-5, KLF4, ISG6-16, ISG56, ISG 1-8U, HECT E3 Ubiquitin ligase, NOXA, RANTES, RIG-I, and C/EBPδ) (Fig. 1).
Table 1. Transcriptome Profile of the Top 50 Induced and the Total (24) Down-regulated Genes by IKKε in the Huh7 HCV Replicon Cells Using Affymetrix U133 2.0 Chips
Fold Change (log2)
Mean values IKKe
Mean values CONT
NOTE. The accession number of the genes is given on the left, followed by the gene name. The numbers represent the fold change and the mean values (of triplicate samples) for IKKε-transfected and control HCV replicon cells. The full transcriptome profile (231 genes) is shown in Supplementary Material.
Actin related protein 2/3 complex, subunit 1A, 41kDa
Protein phosphatase, EF hand calcium-binding domain
Protein inhibitor of activated STAT, 2
Kallmann syndrome 1 sequence
Doublecortin and CaM kinase-like 2
Zinc finger protein 659
Sushi, von Willebrand factor type A
Ankyrin repeat and SOCS box-containing 9
Transducin (beta)-like 1X-linked
HIV-1 Rev binding protein
Histone 1, H4c
Histone 1, H1a
Ribosomal protein L37a
HIR histone cell cycle regulation defective homolog A
Zinc finger protein 236
Ectopic Expression of IKKε in the HCV Expressing Cells Induces a Majority of Genes Linked to the IFN Antiviral Action.
IKKε is functional in cells that have been subjected to lipopolysaccharide (LPS),41 PMA,42 dsRNA treatment, or viral infection43 and is known to activate the IRF3, c/EBPδ, and NF-κB transcription factors.19, 20, 26, 44
Accordingly, we observed that the IKKε transcriptome profile in the HCV replicon cells contains the genes MDA-5, ISG6-16, MX1, RIG-I, ISG9-27, ISG1-8D, ISG20, UbcH8, and the complement factor B, all genes that have been previously reported to be induced in response to a viral infection or dsRNA treatment. Similarly, it contains the genes IRF9, ISG58, PLSCR1, RIG-G, and STAT1, which are known to be induced by RA, LPS, or PMA (Table 2). More specifically, several genes of the IKKε transcriptome have been previously reported to be directly induced by IRF3. Those are: ISG15, ISG1-8, ISG56, ISG54, CXCL10, Viperin, NOXA, RANTES CXCL11, and USP18.45–48 Induction of haptoglobin and antichymotrypsin ACT (SERPIN A3) by IKKε can be presumably attributed to c/EBPδ activation.49, 50
Table 2. List of Highly Induced Genes in the Huh7 Replicon Cells Upon IKKϵ Overexpression and Relation with Known Inducers
Fold Change log2
Previously Described Inducers
NOTE. Huh-7cells expressing a full length HCV replicon were plated in 60 mm at 7×105 cells/dish and transfected 24h after seeding with 3μg of an IKKε expressing vector or an empty vector, used as control. Each sample was performed in triplicate. At 48 hours after transfection, total RNAs were extracted with RNABle and processed for the micro-array analysis. This list shows 52 entries corresponding to genes which have been elevated around 2-fold or greater. Note that 9 genes are represented twice, which reduces the real number of highly induced genes to 4 3. The real change values are given next to those expressed in log2. The accession number of the genes is given on the left, followed by their usual name. Whenever possible, the genes were ascribed to their capacity to be induced by IRF3, c/EBP, NF-κB, three transcription factors known to be activated by IKKε, and more generally by virus, dsRNA or other stimuli of interest, using data retrieved from the literature. References are numbered on the right and given below:
(1)- C. P. Elco et al, J Virol79 3920 (2005); (2)-M. Grove, M. Plumb, Mol Cell Biol13, 5276 (1993); (3) T. Nakaya et al., Biochem Biophys Res Comm283, 1150 (2001); (4)- D. C. Kang et al., Oncogene23, 1789 (2004); (5)- Z. Y. Chen, et al, Exp Cell Res281, 19 (2002); (6)- P. M. Martensen, J. Justesen, J Interferon Cytokine Res24, 1 (2004); (7)- N. Grandvaux et al., J Virol76, 5532 (2002); (8) H. Song, et al, Oncogene23, 8301 (2004); (9)- X. Weihua, et al, Biochim Biophys Acta,1492, 163 (2000); (10)- S. Matikainen et al., Cell Growth Differ8, 687 (1997); (11)- A. Dastur, et al, J Biol Chem281, 4334 (2006); (12)- M. Aebi et al., Mol Cell Biol9, 5062 (1989); (13)- R. Lin, et al, Mol Cell Biol19, 959 (1999); (14)-T. Imaizumi et al., Endothelium11, 169 (2004); (15)- J. B. Andersen, et al, Eur J Biochem271, 628 (2004); (16)- T. Milosavljevic, et al, Gen Physiol Biophys22, 181 (2003); (17)- T. Niikura, et al, Blood Cells Mol Dis23, 337 (1997); (18)- K. W. Zhao et al., Blood104, 3731 (2004); (19)- Z. Liu, et al, Mol Cancer Res3, 21 (2005); (20)- G. S. Sivko, J. W. DeWille, J Cell Biochem93, 830 (2004); (21)-M. Yu et al., Proc Natl Acad Sci U S A94, 7406 (Jul 8, 1997); (22)- N. Kalsheker, et al., Biochem Soc Trans30, 93 (2002); (23)- L. H. Wong et al., J Biol Chem277, 19408 (2002); (24)- L. Espert et al., Oncogene23, 4636 (2004); (25)- C. Zhao, et al, Proc Natl Acad Sci U S A102, 10200 (2005); (26)- A. Marson, et al., J Biol Chem279, 28781 (2004); (27)- S. Tu naru et al., Nat Med9, 352 (2003); (28)- J. Xu, et al, Mol Endocrinol19, 527 (2005); (29)- J. Vandermeer, et al, Arch Otolaryngol Head Neck Surg130, 1374 (2004); (30)- O. Malakhova, et al, J Biol Chem277, 14703 (2002); (31)- T. K. Leung, et al, Biochem J267, 125 (1990); (32)- P. S. Lagali, et al, Biochem Biophys Res Commun293, 356 (2002); (33) K. J. Helbig, et al, Hepatology42, 702 (2005); (34)- D. K. St Clair, et al, Methods Enzymol349, 306 (2002).
Some of the genes of the IKKε transcriptome may be induced as secondary events through some transcription factors induced by IKKε. This might be the case for the KLF4 transcription factor, which is induced in the liver in response to IFNγ51 and therefore could be induced here via STAT1, which is strongly induced by IKKε. Another example is that of the OAS genes.52 Although those genes normally respond to the ISGF3 complex, resulting from the association of IRF9 with STAT1 and STAT2, here they may have been induced through IRF9 alone, because IRF9 overexpression has been previously noted to provoke the induction of several type I IFN-inducible genes.53 The rest of the genes strongly induced by IKKε were previously associated with induction by STAT3 (MPK5, c/EBPδ) and by NF-κB, Sp1, HNF-1β, IRF1, or stress (Nicotinamide N-Methylase, ISG20, HSP70B, Vanin, MnSOD). With regards to NF-κB, IKKε was recently shown to phosphorylate p65 at two of its serine residues (468 and 536) and to control its nuclear import.28
In summary, the transcriptome profile of IKKε confirms the induction of already described genes and allows depiction of new IKKε-induced genes. In particular, these data point out that IRF3-induced genes and genes related to the antiviral action of IFN represent an important fraction of all the genes strongly induced by IKKε (Table 2).
The Protein Kinase IKKε Inhibits HCV Expression in the Absence of IFN Induction.
IKKε can play an important role in IFN induction, via its ability to activate IRF3 and NF-κB. As many as 21 of the IKKε-induced genes in the HCV replicon cells are IFN-inducible genes (see Table 2); therefore IKKε may inhibit HCV expression through IFN induction and subsequent induction of a panel of genes with antiviral activity. Indeed, inhibition of HCV expression by IKKε was equivalent to that provoked by a concentration of IFN-α of between 1 and 10 U/mL (Fig. 2A). However, analysis of the microarray data indicates that several of the IKKε-induced genes, with known antiviral potential, may have been directly induced by IKKε, for instance, through IRF3 or NF-κB. To determine whether the antiviral effect of IKKε requires IFN induction, we then examined its effect on HCV expression in the presence of anti-IFNAR1 antibodies known to compete specifically for the binding of type I IFNs to the IFNAR chain of the IFN receptor.31, 32 Analysis of STAT1 phosphorylation in response to IFNα or IFNγ allowed to determine that the IFN receptor was functional in the HCV replicon cells as in the parental Huh-7 cells (Fig. 2B). The HCV replicon cells were then transfected with IKKε wt or the catalytically inactive mutant IKKε K38A and further incubated in presence of anti-IFNAR1 antibodies at a concentration sufficient to inhibit induction of an ISRE promoter-dependent luciferase reporter gene by 10 U/mL IFN-α (Supplementary Fig. 1). Both IKKε wt and IKKε K38A were expressed at similar levels, after transfection, as controlled by qRT-PCR (4-4.5 × 106 in the absence and 2 to 2.5 × 106 RNA copies in the presence of anti-receptor antibodies). In parallel, the Huh7 HCV Replicon cells were also incubated in presence of IFN, in the absence or presence of the anti-IFNAR antibodies. The results show that, although the antiviral action of IFN-α (90% inhibition of HCV expression) was significantly abrogated in the presence of anti-IFNAR1 (residual 27% of inhibition), the antiviral action of IKKε (86% in absence of antibodies) was only partially abrogated, with still 63% of inhibition of HCV expression (Fig. 2C).
As another readout to determine whether the IKKε-mediated ability to inhibit HCV expression requires IFN action, we have analyzed its effect in the presence of STAT1αY701F, a dominant negative form of STAT1α. Expression of this mutant prevents formation of any ISGF3 transcription factor complex (STAT1/STAT2/IRF9) that would occur on IFN activation of the JAK/STAT signaling pathway and induction of the IFN-stimulated genes.54 Transfection of the Huh7 HCV Replicon cells with IKKεwt again resulted in strong inhibition of HCV expression (81%), as shown in Fig. 2A,C. The HCV expression remained strongly inhibited when IKKεwt was coexpressed with STAT1αY701F, with an increase in inhibition (88%). This further demonstrates that the antiviral activity of IKKε was independent of IFN production and action (Fig. 2D). In addition, we noticed that, although expression of the STAT1αY701F mutant alone had only a partial negative effect on HCV expression, similar to that of the IKKεK38A mutant (35%), expression of a wild-type form of STAT1α proved to inhibit HCV expression to levels comparable to that of IKKε (Fig. 2D). This STAT1-mediated mechanism of HCV inhibition might be related to the ability of STAT1 to provoke apoptosis, for instance, through interaction with p5355 (see Discussion). With regard to the IFN-independent antiviral activity of IKKε, overexpression of STAT1α (wt or Y701F) was reported to attenuate the IFN-α signaling, presumably by blocking IFN-α activated STAT2 translocation from IFNAR2 to IFNAR1.56 Therefore, if the anti-HCV effect of IKKε had depended on IFN production, it would have been attenuated on STAT1α overexpression. The fact that IKKε still inhibits strongly HCV expression in presence of STAT1α confirms its independence, at least in part, from the necessity to produce IFN to generate an antiviral effect.
Therefore, these results show that the ability of IKKε to inhibit HCV can develop in the absence of IFN induction, probably through one or several of the genes identified by the microarray analysis.
The Expression of the Protein Kinase IKKε and of Some Other Components of the IFN Induction Pathway Are Downregulated in HCV-Infected Patients.
We then analysed the in vivo expression levels of IKKε, to determine whether this protein kinase has the possibility to play an antiviral role in natural HCV infection. For this, we performed qRT-PCR on RNAs extracted from liver biopsies from nine HCV-infected patients (6 men, 3 women, aged 29-73 years) and eight HCV-infected patients who were nonresponders to IFN/ribavirin treatment (5 men, 3 women, aged 46-64 years) (Table 3). For controls, we used biopsies collected by laparectomy from tumor-free livers in 12 patients (8 men and 4 women aged 45-68 years); one patient had cholelithiasis, another hydatid cyst, and the remaining (10 patients) suffered from metastatic gastrointestinal cancer. To have a general picture of the expression of the IFN-inducing pathway, we also included in this study the analysis of the expression levels of the three RNA helicases: RIG-I, MDA5, and LGP2, the RIG-I/MDA5 adapter Cardif, TBK1, the second IRF3-phosphorylating kinase and three genes, downstream of the IFN-inducing pathway: IFN-β, CCL3, and ISG15.
Table 3. Clinical Data on Control and HCV-Infected Patients
Viral Load (IU/mL)
Time After Treatment
The ALT and AST normal values are <25 IU/mL and < 29 IU/mL respectively.
5,00E + 06
1 + 1 + 1 + 0
3,45E + 05
1 + 1 + 1 + 0
2,36E + 05
3 + 0 + 3 + 0
2,02E + 05
1 + 3 + 4 + 0
2,00E + 06
0 + 1 + 1 + 0
2,00E + 05
1,00E + 06
1 + 1 + 1 + 0
1 + 1 + 3 + 1
5,46E + 07
1 + 1 + 1 + 0
1,12E + 05
3 + 4 + 3 + 3
2,00E + 08
2,00E + 07
1 + 1 + 3 + 1
1,20E + 07
1 + 1 + 3 + 1
2,40E + 04
1 + 1 + 3 + 0
5,91E + 05
3 + 3 + 3 + 1
9,36E + 06
1 + 1 + 3 + 0
4,00E + 05
1 + 1 + 3 + 1
We found that the relative expression levels of the components of the IFN-inducing pathway vary with the following order : Cardif < RIG-I < LGP2, TBK1 < MDA5 < IKKε, ISG15 < IFN-β, CCL3, with Cardif and CCL3 having, respectively, the lowest and highest values (Fig. 3). Expression of CCL3 was increased in the HCV-infected patients, and this increase was statistically significant in the patient non-responders to IFN treatment as compared with the controls. CCL3 is a known marker of inflammation in hepatocytes,57 and this increase therefore may reflect the inflammation state of the liver, associated with HCV infection. The IKKε protein kinase also can be induced in inflammatory states.41, 42 However, in contrast to the situation with CCL3, the IKKε expression levels did not increase in the HCV-infected patients and, in fact, even decreased with a significant decrease in the non responder patients as compared with the controls. This inhibition, thus, is indicative of some inhibitory effect on the IKKε expression attributable to the viral infection.
The expression levels of the three RNA helicases RIG-I, MDA5, and LGP2 were all downregulated in the nonresponder HCV-infected patients, and inhibition of RIG-I was already significant in the HCV-infected cohort. LGP2 is an RNA helicase related to RIG-I and MDA5, but devoid of CARD domain which has been reported to negatively control the activity of the two other RNA helicases through heterodimerization.13 The expression levels of Cardif, TBK1, and IFN-β were not found to change significantly on HCV infection, whereas that of ISG15 was strongly inhibited. Altogether, the in vivo data collected from the analysis of the liver biopsies from HCV-infected patients show that the RNA helicases/IKKε pathway is particularly sensitive to the infection. A downregulation in the IKKε expression levels therefore may be detrimental for efficient induction of genes, such as ISG15. The decreased expression of IKKε, of the RNA helicases and of ISG15 cannot be attributed to HCV-associated cell death because all HCV patients presented moderate liver damage (Table 3). Coupled with the low expression levels of Cardif in the liver samples and the sensitivity of this protein to the HCV NS3/4A protease-mediated cleavage, these data confirm the importance of IKKε in the establishment of the antiviral state against HCV infection, because the inhibition of its expression by the infection can alter its efficacy to induce antiviral genes. Given the fact that IKKε was recently shown to strongly colocalize with Cardif,23 these in vivo data confirm the in vitro observations that showed the importance of IKKε for the establishment of an ongoing antiviral state against HCV infection. The activity of this kinase in liver cells may be important to contribute to an equilibrium between the HCV infection and the protection of the cells and to limit the propagation of the virus.
We previously showed that overexpression of the IKKε protein kinase in Huh7 cells expressing a full-length HCV replicon can inhibit the expression of this replicon. Here, we have established the transcriptome profile induced by IKKε in these HCV replicon cells. We showed that IKKε can induce expression of several genes that can play a role in antiviral action as well as in inflammatory responses and that its antiviral action still takes place in the presence of antibodies directed against IFNAR receptors, or in the presence of dominant negative STAT1α mutant. This indicates that an increase of IKKε expression levels may give the cells the possibility of resisting HCV infection.
The transcriptome profile induced by IKKε in the HCV replicon cells contains most of the genes involved in protein conjugation by the ubiquitin-like protein ISG15. In addition to ISG15, a known IRF3-induced gene,58 the transcriptome contains the E2-(UbcH8) ligase and E3-CEB1/HERC5 ligase and the USP18 protease. Moreover, the IKKε transcriptome also contains the RIG-I, ISG56, ISG54, STAT1, and MxA genes, which can be ISG15 targets.59 ISG15 forms covalent interaction with proteins but instead of targeting them for proteasomal degradation, like ubiquitin, it provokes posttranslational modifications that play a role on half-life, cellular localization, and protein activity. ISG15 conjugaison of proteins (or ISGylation) requires the activity of the E1-(Ube1L), E2-(UbcH8), and E3-CEB1/HERC5 conjugating enzymes,59–61 whereas the deconjugation process is performed by the specialized ubiquitin protease USP18. The importance of USP18 in innate immunity to viral infection was demonstrated using USP18-deficient mice, which were resistant to lymphocytic choriomeningitis virus or vesicular stomatitis virus.62 Although the exact antiviral activity of the ISG15 pathway remains to be determined, its induction by IKKε may play an important role in the antiviral action of IKKε against HCV.
The transcriptome induced by IKKε in the HCV replicon cells also revealed the expression of two apoptosis-mediating proteins of the BH3-only family: NOXA and the recently described apoL6.63 They both belong to the pro-apoptotic group of the Bcl2 intracellular protein family,64 and direct induction of NOXA by IRF3 has been recently reported.48 Their induction may be of interest in view of the close association of IKKε with the mitochondria-associated Cardif23 and a possible role for Cardif in the control of apoptosis.16
The protein kinase IKKε is strongly related to the other IRF3-phosphorylating kinase, TBK1. Studies with TBK1 and IKKε murine-deficient cells showed that TBK1 plays a major role in IFN-β induction in response to LPS, dsRNA (delivered intracytoplasmically), and virus infection. However, use of the IKKε/TBK1 doubly deficient cells revealed a complete abolition of IFN-β induction, and IKKε has been defined as an IRF3-phosphorylating kinase that could compensate, at least in part, for TBK1 deficiency in the TBK1−/− cells.65 Both kinases can be activated by Cardif; however, the mechanism of their activation is unclear, in particular their mode of interaction with Cardif. In one report, TBK1 was found to associate with Cardif and IKKε was not examined17; in another study, neither of these kinases was found to associate with Cardif but the data were presented only for TBK1.15 In contrast, more recently, a strong association of Cardif with IKKε and no interaction with TBK1 was presented.18 In accord with this latter result, recent confocal microscopy analysis demonstrated a tight colocalization for IKKε with the mitochondrial protein Cardif, whereas TBK1 was associated with other vesicles.23 The Cardif/IKKε association could represent an important threat for HCV, which can be alleviated through the NS3/4A-mediated cleavage of Cardif.18, 22, 23
We showed that IKKε overexpression provoked inhibition of HCV expression. Based on these results, one might expect to find increased hepatic expression of IKKε in chronic hepatitis C as an adaptive response to halt HCV replication through the action of one or several of the antiviral molecules induced by IKKε. However, we found that the IKKε expression levels were downregulated in liver biopsies from HCV-infected patients, with an even more pronounced inhibition in the liver biopsies from patients who have been resistant to IFN treatment.
A decrease in the IKKε levels may create a situation favorable for HCV persistence. In addition to this, inhibition of the expression levels of the two RNA helicases RIG-I and MDA5 in the non responders HCV-infected patients, indicates that HCV infection can disrupt sufficiently the RNA helicase/IKKε chain to favor its propagation. RIG-I and MDA5 belong to the IFN-induced genes but even in case that induction of these genes might occur during IFN therapy, this is probably not sufficient to resist the infection because most HCV-infected patients are resistant to IFN/ribavirin treatment. A decrease in the IKKε levels may be a critical factor to explain HCV persistence and viral resistance to IFN because, as shown, IKKε mediates the induction of essential antiviral genes, including the two helicases, ISG15, enzymes, and targets of ISGylation, STAT1 among others.
The determination of the antiviral genes that are specifically induced by IKKε to provoke HCV inhibition obviously requires transcriptome profile analysis of IKKε-induced genes in HCV-infected TBK1−/− MEFs. This is at present not possible because the cellular models for HCV infection currently are restricted to human hepatoma–derived cell lines. However, we noticed that the following genes induced by IKKε in the HCV replicon cells: Mx1, OASL, apolipoprotein L, MDA5, ISG56, IP-10, Usp18, ISG15, CXCL11, CXCL-10, have been recently reported to be specifically induced by TBK1.66 These genes therefore cannot be considered as specifically induced by IKKε, in normal conditions of activation of the innate immune response. Similarly, we found that the anti-HCV activity of IKKε persisted in presence of dominant negative STAT1 (Fig. 2D), which allowed ruling out of any STAT1-induced genes, such as KLF4, as candidate for a specific IKKε-induced antiviral response. However, STAT1 itself belongs to the IKKε gene profile in the HCV replicon cells and, interestingly, we observed that individual overexpression of STAT1wt resulted in a strong inhibition of HCV expression, similar to that provoked by IKKε (Fig. 2D). STAT1 has been reported to provoke apoptosis.55 NOXA, an another important apoptosis-inducible gene,67 was also highly induced by IKKε in the HCV replicon cells. Whether the antiviral activity of IKKε.against HCV is mediated through apoptosis remains to be determined.
Inhibition of IKKε expression in the HCV infected livers is intriguing because this gene can be induced through NF-κB, which is known to activate on HCV infection, for instance through the expression of the core and the NS5A proteins.68 Moreover, IKKε can autoactivate and sustain its own induction, through C/EBPδ activation. One possibility is that IKKε activation may depend on its localization at the mitochondria in the vicinity of Cardif.23 This information points to the importance of pursuing the generation of efficient and well-tolerated drugs that could abrogate the cleavage of Cardif during HCV infection.
We thank Agnès Marchiato and Pascal Pineau (Unité de Recherche Organisation Nucléaire et Oncogenèse; Pasteur Institute) for primers allowing the normalization of the liver biopsies data.