Epidermal growth factor receptor signaling impairs the antiviral activity of interferon-alpha

Authors

  • Joachim Lupberger,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    Search for more papers by this author
    • These authors contributed equally.

  • François H.T. Duong,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    3. Department of Biomedicine, Hepatology Laboratory, University of Basel, Basel, Switzerland
    Search for more papers by this author
    • These authors contributed equally.

  • Isabel Fofana,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    Search for more papers by this author
  • Laetitia Zona,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    Search for more papers by this author
  • Fei Xiao,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    Search for more papers by this author
  • Christine Thumann,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    Search for more papers by this author
  • Sarah C. Durand,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    Search for more papers by this author
  • Patrick Pessaux,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    3. Pôle Hépato-digestif, Nouvel Hôpital Civil, Strasbourg, France
    4. Institut Hospitalo-Universitaire, Strasbourg, France
    Search for more papers by this author
  • Mirjam B. Zeisel,

    1. Inserm, U1110, Institut de Virologie, Strasbourg, France
    2. Université de Strasbourg, Strasbourg, France
    Search for more papers by this author
  • Markus H. Heim,

    1. Department of Biomedicine, Hepatology Laboratory, University of Basel, Basel, Switzerland
    Search for more papers by this author
    • These authors contributed equally.

  • Thomas F. Baumert

    Corresponding author
    1. Université de Strasbourg, Strasbourg, France
    2. Pôle Hépato-digestif, Nouvel Hôpital Civil, Strasbourg, France
    3. Institut Hospitalo-Universitaire, Strasbourg, France
    • Inserm, U1110, Institut de Virologie, Strasbourg, France
    Search for more papers by this author
    • These authors contributed equally.


  • See Editorial on Page 1200

  • Potential conflict of interest: Nothing to report.

  • This work was supported by the European Union (ERC-2008-AdG-233130-HEPCENT and INTERREG-IV-Rhin Supérieur-FEDER-Hepato-Regio-Net 2009 and 2012, EU FP7 HepaMab), Inserm, Laboratoire d'Excellence HEPSYS (ANR-10-LAB-28), the Agence Nationale pour la Recherche contre le SIDA et les Hépatites Virales (2008/354, 2009/183, 2011/132, and 2012/319), Université de Strasbourg, the Conseil Général du Bas Rhin, the Communauté Urbaine de Strasbourg, the Cancéropôle du Grand Est (30/03/09), INCa (2009-143), the Ligue Contre le Cancer (CA 06/12/08), and the Direction Générale de l'Offre de Soins (A12027MS). The authors thank R. Bartenschlager (University of Heidelberg, Heidelberg, Germany), T. Wakita (National Institute of Infectious Diseases, Tokyo, Japan), and F.V. Chisari (The Scripps Research Institute, La Jolla, CA) for providing HCV strain Luc-Jc1 and Huh7.5.1 cells and F. Sinicrope for providing lentiviral shRNA plasmids.

  • Inserm and the University of Strasbourg have filed a patent application on combinations of protein kinase inhibitors and interferons for the treatment and prevention of hepatitis C virus infection.

Address reprint requests to: Thomas F. Baumert, M.D., Inserm Unit 1110, University of Strasbourg, 3 rue Koeberlé, F-67000 Strasbourg, France. E-mail: Thomas.Baumert@unistra.fr; fax: +33 3 68 85 37 24.

Abstract

Interferon-alpha (IFN-α) exhibits its antiviral activity through signal transducer and activator of transcription protein (STAT) signaling and the expression of IFN response genes (IRGs). Viral infection has been shown to result in activation of epidermal growth factor receptor (EGFR)—a host cell entry factor used by several viruses, including hepatitis C virus. However, the effect of EGFR activation for cellular antiviral responses is unknown. Here, we uncover cross-talk between EGFR and IFN-α signaling that has a therapeutic effect on IFN-α-based therapies and functional relevance for viral evasion and IFN resistance. We show that combining IFN-α with the EGFR inhibitor, erlotinib, potentiates the antiviral effect of each compound in a highly synergistic manner. The extent of the synergy correlated with reduced STAT3 phosphorylation in the presence of erlotinib, whereas STAT1 phosphorylation was not affected. Furthermore, reduced STAT3 phosphorylation correlated with enhanced expression of suppressors of cytokine signaling 3 (SOCS3) in the presence of erlotinib and enhanced expression of the IRGs, radical S-adenosyl methionine domain containing 2 and myxovirus resistance protein 1. Moreover, EGFR stimulation reduced STAT1 dimerization, but not phosphorylation, indicating that EGFR cross-talk with IFN signaling acts on the STATs at the level of binding DNA. Conclusions: Our results support a model where inhibition of EGFR signaling impairs STAT3 phosphorylation, leading to enhanced IRG expression and antiviral activity. These data uncover a novel role of EGFR signaling in the antiviral activity of IFN-α and open new avenues of improving the efficacy of IFN-α-based antiviral therapies. (Hepatology 2013;58:1225–1235)

Abbreviations
3D

three-dimensional

Abs

antibodies

CI

combination index

DMSO

dimethyl sulfoxide

EGFR

epidermal growth factor receptor

EMSA

electrophoretic mobility shift assay

HBV

hepatitis B virus

HCV

hepatitis C virus

HCVcc

cell-culture–derived HCV

HDV

hepatitis D virus

IC50

half-maximal inhibitory concentration

IFN-α

interferon-alpha

IL

interleukin

IRG

IFN response gene

IFNAR

IFN-α receptor

Jak

Janus kinase

mRNA

messenger RNA

Mx1

myxovirus resistance protein 1

PHHs

primary human hepatocytes

P-STAT

phospho-STAT

qPCR

quantitative polymerase chain reaction

RSAD2

radical S-adenosyl methionine domain containing 2

RTK

receptor tyrosine kinase

shRNA

short hairpin RNA

siRNA

small interfering RNA

SOC

standard of care

SOCS

suppressors of cytokine signaling

STAT

signal transducer and activator of transcription protein

TCID50

half-maximal tissue culture infectious dose

Tyk2

tyrosine kinase 2.

Interferons (IFNs) are pleiothrophic cytokines produced by the immune system that have a key role in combating viral infections. The IFNs that show the most antiviral properties are the type I and III IFNs alpha (IFN-α) and lambda (IFN-λ).[1, 2] It has been suggested that the inhibitory effect on hepatitis C virus (HCV) replication is mediated through distinct patterns of signal transduction resulting in IFN response gene (IRG) expression, and that viral infections themselves trigger IFN production.[1-3] IFN-α leads to recruitment of Janus kinase (Jak) and tyrosine kinase 2 to the IFN-α receptor (IFNAR), causing phosphorylation of the signal transducer and activator of transcription proteins (STATs).[1] Activation of the Jak-STAT pathway by IFN-α involves phosphorylation of STAT1 and STAT2 by Jak, leading to phospho-STAT (P-STAT)1 homodimerization as well as heterotrimerization with P-STAT2 and IFN regulatory factor 9. These complexes migrate to the nucleus, where they bind to cis-acting elements in the promoters of IRGs and act as transcriptional activators.[1, 4] Hundreds of IRGs are transcribed in response to IFN-α, including genes such as radical S-adenosyl methionine domain containing 2 (RSAD2) and myxovirus resistance protein 1 (Mx1). The gene products of these IRGs facilitate antiviral responses.[2, 5-7] IFN-α engagement to the IFNAR also activates STAT3,[8] which is a crucial player in liver regeneration in response to interleukin (IL)−6, promoting cell survival, proliferation, and immune tolerance and counteracting the proinflammatory role of STAT1 in response to IFN-α.[9, 10] Additional regulatory mechanisms of STAT signaling involve suppressors of cytokine signaling (SOCS) proteins. SOCS1 and SOCS3 expression is rapidly and transiently induced upon IFN-α exposure to act as negative regulators of the Jak-STAT pathway in a classic feedback loop.[1]

The antiviral activity of IFN-α can be observed in the treatment of hepatitis virus infections, which are the leading causes of liver cirrhosis and hepatocellular carcinoma.[11] IFN-α and its pharmacologically modified derivate, pegylated IFN-α, are part of the standard of care (SOC) of HCV infection[12] and are also used in the treatment of hepatitis B (HBV) and D virus (HDV) infections, albeit with limited efficacy.[13, 14] Although the outcome of IFN-α-based treatment of chronic hepatitis C has recently markedly improved by the clinical licensing of protease inhibitors, adverse effects and viral resistance remain important challenges.[12] Furthermore, because the therapeutic options for difficult-to-treat patients remain unsatisfactory and IFN-free regimens have not yet been approved, it is likely that IFN-α will remain a relevant component of HCV treatment for some time to come.[12] Resistance to IFN-α therapy has been explored in great detail.[1, 15] In cell-culture models, IFN signaling is impaired by both HCV infection and HCV protein expression.[1, 16-18] In patients, a preactivated and refractory hepatic IFN-α-signaling pathway is a hallmark of IFN-α nonresponse.[19]

Epidermal growth factor receptor (EGFR) has been shown to be a host entry factor for several viruses, including HCV.[20-22] Indeed, inhibition of EGFR function using the clinically licensed kinase inhibitor, erlotinib, has been shown to impair HCV infection in a dose-dependent manner.[21, 23] Based on these observations, host cell kinases have been suggested as targets for anti-HCV therapy.[24-28] Moreover, because host cell kinases have been shown to be critical in the life cycles of a broad range of other viruses,[20, 22, 29-31] kinase inhibitors may have utility as broad antiviral interventions. Most recently, data demonstrate that HCV infection not only requires EGFR as a host cell entry factor, but also results in EGFR activation and signaling.[23] Because signaling of receptor tyrosine kinases (RTKs) and IFN-α share distinct signal transducers,[32, 33] we aimed to investigate whether (i) inhibition of RTK signaling offers an opportunity to improve the therapeutic efficacy of IFN-α and (ii) RTK signaling is relevant for the antiviral activity of IFN-α.

Materials and Methods

Cell Lines and Primary Human Hepatocytes

The sources and culture conditions for Huh7.5.1 cells have been described.[34] HeLa cells were obtained from American Type Culture Collection (Manassas, VA), and PAP HN5 cells were obtained from the Institut de Génétique et de Biologie Moléculaire et Cellulaire (Ilkirch, France). Primary human hepatocytes (PHHs) were isolated and cultured as previously described.[21]

Inhibitors, Ligands, and Antibodies

Erlotinib and sorafenib were obtained from LC Laboratories (Woburn, MA), Cpd188 was obtained from Millipore (Billerica, MA), dimethyl sulfoxide (DMSO) was used at a final concentration of 0.7% for incubation of protein kinase inhibitors and control experiments, IFN-α2b was obtained from Merck (Darmstadt, Germany), IFN-α2a was obtained from Roche (Mannheim, Germany), and human recombinant IL-6 was obtained from PeproTech EC (London, UK). Recombinant EGF and specific antibodies (Abs) targeting β-actin and phospho-EGFR (Tyr1086) were obtained from Sigma-Aldrich (St. Louis, MO). Abs specific for STAT1, STAT3, P-STAT1 (58D6), P-STAT3 (D3A7), and SOCS3 were obtained from Cell Signaling Technology (Danvers, MA). The HCV core Ab (clone C7-50) was obtained from Thermo Fisher Scientific Inc. (Waltham, MA). Horseradish-peroxidase–labeled secondary Abs were obtained from GE Healthcare (Uppsala, Sweden). Secondary Abs IRDye immunoglobulin G (800CW and 680RD) were obtained from LI-COR Biosciences (Lincoln, NE).

Small Interfering RNAs, Short Hairpin RNAs, and Expression Plasmids Used for Functional Studies

siCTRL, siCD81, and siEGFR (Hs-siEGFR_6) have been described.[21] siSTAT3 (Hs-siSTAT3_7) was obtained from Qiagen (Germantown, MD). siSOCS3 (siGenome SMARTpool: 5′-CAG CAU CUC UGU CGG AAG A-3′, 5′-GCG AUG GAA UUA CCU GGA A-3′, 5'-UCA AGA CCU UCA GCU CCA A-3′, and 5′-CCU AUU ACA UCU ACU CCG G-3′) was obtained from Thermo Fisher Scientific. pSIH1-puro-control short hairpin RNA (shRNA) (shCTRL) and shRNA (shSTAT3) were obtained from Frank Sinicrope, M.D. (Mayo Clinic, Rochester, MN) (Addgene plasmids #26597 and #26596).[35] The lentiviral shSOCS3 plasmid was generated by annealing oligonucleotides 3′-GAT CCG GCG GAC CTG GAA TGT GTT TTC AAG AGA AAC ACA TTC CAG GTC CGC CTT TTT G-′5 and 3′-AAT TCA AAA AGG CGG ACC TGG AAT GTG TTT CTC TTG AAA ACA CAT TCC AGG TCC GCC G-5′. The shSOCS3 oligonucleotide was cloned into pSIH1-puro-STAT3 plasmid by replacing the shSTAT3-containing EcoRI/BamHI fragment.

Production and Infection Using Cell-Culture–Derived Recombinant HCV

Recombinant cell-culture–derived HCV (HCVcc) (strain Luc-Jc1) were generated as previously described.[21, 34, 36] Infection of Huh7.5.1 cells with HCVcc (half-maximal tissue culture infectious dose [TCID50]: 104 mL−1) was performed as previously described.[21] Unless otherwise stated, HCVcc infection was assessed by luciferase reporter gene expression.

Analysis of Antiviral Activity of Compounds and Combinations on HCV Infection

The in vitro antiviral activity of each compound was tested individually or in combination with a second compound using the HCVcc-Huh7.5.1 cell culture model, as previously described.[21, 34, 36] Huh7.5.1 cells (cultured in 96-well plates) were persistently infected for 5 days with HCVcc (Luc-Jc1) before incubation with IFN-α2a or IFN-α2b and erlotinib or sorafenib for an additional 5 days at 37°C. Viral infection was analyzed by assessing luciferase activity, as previously described.[21]

Mathematical Analysis of Synergy

Synergy was assessed by two independent methods comprising the combination index (CI)[37] and the method of Prichard and Shipman.[38] The CI was calculated as previously described.[37] A CI of <0.9, 0.9-1.1, and >1.1 indicates synergy, additivity, and antagonism, respectively.[37] The method of Prichard and Shipman was applied as previously described.[38] Surface amplitudes >20% above the zero plane indicate synergistic effects of inhibitor combinations, and surface amplitudes <20% below the zero plane indicate antagonism.[38] The validity of the assay and methods was confirmed by comparative analyses of combinations showing a nonsynergistic or an antagonistic effect.

Silencing, Inhibition, and Ligand Experiments

Silencing was performed using INTERFERin (Polyplus-transfection SA, Illkirch, France) and DharmaFECT4 (Dharmacon, Inc., Lafayette, CO), as previously described.[21] Experiments with EGFR ligand were conducted on serum-starved cells.

Analyses of Messenger RNA Transcription

Quantitative polymerase chain reaction (qPCR) to analyze IRG transcript levels were performed as previously described.[39]

Analyses of Protein Expression

Western blottings of cell lysates using protein-specific Abs (described above) were performed following GE Healthcare protocols using Hybond-P membranes and visualized using ECF substrate and a Typhoon Trio high-performance fluorescence scanner (GE Healthcare) or using the Odyssey Infrared Imaging System (LI-COR Biosciences).

Electrophoretic Mobility Shift Assay

Electrophoretic mobility shift assay (EMSA) using 32P-labeled m67 oligonucleotide probe has been described.[16]

Results

Inhibition of EGFR Activity Using Erlotinib Enhances the Antiviral Efficacy of IFN-α in a Highly Synergistic Manner

We previously identified EGFR as a cofactor for HCV entry.[21] We have demonstrated that the clinical EGFR inhibitor, erlotinib, impairs EGFR signaling in human hepatocytes and inhibits HCV infection and viral spread at a half-maximal inhibitory concentration (IC50) of 0.45-0.53 μM.[21] To investigate the relevance of EGFR signaling for the antiviral activity of IFN-α, we incubated Huh7.5.1 cells that had been persistently infected for 5 days with HCVcc (Luc-Jc1) with IFN-α2a or IFN-α2b and a sub-IC50 concentration of erlotinib, which exerts only minimal effects on infection with HCVcc. Incubation with an unrelated kinase inhibitor, sorafenib, served as an unrelated control. The CI was calculated as a relative measure of synergy, as described previously.[37] We observed an enhanced and synergistic antiviral effect on HCVcc infection of IFN-α2a and IFN-α2b when combined with 0.1 µM of erlotinib (CIs of 0.04 ± 0.01 and 0.04 ± 0.005; Fig. 1A,B). In contrast, an antagonistic effect was observed when IFN-α2a and IFN-α2b were combined with sorafenib (CIs of 1.54 ± 0.16 and 1.45 ± 0.18, respectively; Fig. 1A,B). These data demonstrate that the observed synergy is specific for erlotinib. The addition of a very low dose of erlotinib (0.1 µM) to IFN-α2a or IFN-α2b decreased their IC50 values up to 1,000-fold (from 2.1 ± 0.3 to 0.007 ± 0.002 IU/mL for IFN-α2a and from 2.0 ± 0.4 to 0.002 ± 0.003 IU/mL for IFN-α2b; Fig. 1C,D; Table 1). Furthermore, three-dimensional (3D) modeling, according to the method of Prichard and Shipman, was used to validate synergy.[38] Surface amplitudes >20% above the zero plane confirmed synergistic effects of the indicated combination of IFN-α with erlotinib (Fig. 1E,F).

Figure 1.

IFN-α and a clinically licensed EGFR inhibitor inhibit HCV infection in a synergistic manner. Combination of EGFR inhibitor erlotinib with IFN-α potentiates the antiviral effect of IFN-α. Huh7.5.1 cells were infected for 5 days with HCV Luc-Jc1 before treatment with the indicated compounds. Subsequently, cells were incubated for 5 days with increasing concentrations of (A) IFN-α2a or (B) IFN-α2b and 0.1 µM of erlotinib or sorafenib. HCVcc infection was analyzed by luciferase reporter gene expression as previously described.[21, 36] (A and B) CI at IC50 was calculated as previously described[37] and is indicated in Table 1. Dotted lines at CI values of 0.9 and 1.1 indicate the boundaries of an additive interaction. (C and D) Combination of IFN-α with 0.1 µM of erlotinib resulted in a shift in the IC50 value of IFN-α2a and IFN-α2b up to 1,000-fold. Means ± standard deviation from at least three independent experiments performed in triplicates are shown. (E and F) Synergy was confirmed using 3D analysis of the inhibitor-inhibitor combinations according to Prichard and Shipman.[38] Surface amplitudes >20% above the zero plane highlight inhibitor combinations acting synergistically, as highlighted in green. One representative experiment is shown.

Figure 2.

STAT3 is relevant for HCV infection. Inhibition of STAT3 expression and function decreases HCV infection. (A) Silencing of STAT3 protein expression in Huh7.5.1 cells transfected with STAT3-specific siRNA (siSTAT3) or control siRNA (siCTRL). Immunoblottings of 30 µg of total proteins in cell lysates using a STAT3-specific Ab (upper panel) or a β-actin Ab (lower panel) are shown. (B) Inhibition of HCVcc (Luc-Jc1) infection in Huh7.5.1 cells with silenced STAT3 expression. HCV infection was assessed by luciferase reporter gene expression. Cells with silenced expression of EGFR (siEGFR) or CD81 (siCD81) and cells transfected with a control siRNA (siCTRL) serve as positive and negative controls, respectively. Data are expressed as percentage HCVcc infection relative to siCTRL-transfected cells. Means ± standard deviation from three independent experiments in triplicate are shown. (C) STAT3-specific inhibitor Cpd18842 impairs HCVcc infection in Huh7.5.1 cells. Cells were preincubated for 1 hour with Cdp188 (1 and 10 µM) dissolved in 0.7% DMSO or 0.7% DMSO alone (control). Subsequently, cells were incubated with HCVcc Luc-Jc1 in the presence of the inhibitor and infection was assessed by luciferase reporter gene expression. Data are expressed as percentage HCVcc infection relative to cells treated with drug solvent 0.7% DMSO. Means ± standard deviation of three experiments in triplicate are shown. (D) Viability of cells treated with Cpd188 shown in panel (D), as measured by methyl thiazol tetrazolium assay.[21] Means ± standard deviation of three experiments in triplicate are shown.

Table 1. Synergy of Erlotinib and IFN-α2a or IFN-α2b on Inhibition of HCV Infection
Compound 1IC50 (IU/mL)Compound 2IC50 (IU/mL) for CombinationCI
  1. Five days postinfection with HCVcc (Luc-Jc1), Huh7.5.1 cells were incubated with serial concentrations of IFN-α2a or IFN-α2b and 0.1 µM of erlotinib or control reagent for 5 days at 37°C. HCVcc infection was analyzed by luciferase reporter gene expression.[21] CI was calculated as previously described.[37] A CI less than 0.9, between 0.9 and 1.1, and more than 1.1 indicates synergy, additivity, and antagonism, respectively. Means ± standard deviation from at least three independent experiments performed in triplicate are shown. IC50 of erlotinib: 0.5 ± 0.06 µM.

IFN-α2a2.1 ± 0.3Erlotinib0.007 ± 0.0020.04 ± 0.010
IFN-α2b2.0 ± 0.4Erlotinib0.002 ± 0.0030.04 ± 0.005

Taken together, these data suggest that the EGFR inhibitor, erlotinib, enhances the antiviral activity of IFN-α in a highly synergistic manner when added postinfection. The finding that very low doses of erlotinib (0.1 µM) with absent or minimal effects on HCV in preinfection[21] or postinfection experiments (Fig. 1C,D) efficiently potentiated the antiviral effect of IFN in postinfection experiments indicates that EGFR activity has a functional role for the antiviral activity of IFN-α that is independent from its role in HCV entry. Interestingly, we observed that, at high concentrations, erlotinib also inhibited HCV in persistently infected cells (Fig. 1C,D). Because we have previously shown that erlotinib monotherapy does not have a robust effect on HCV replication in replicon models,[21] we conclude that this effect is probably the result of the inhibition of HCV cell-cell transmission, which is required for maintenance of persistent infection.[21]

IFN-α Signal Transducer STAT3 Is a Host Factor for HCV Infection

To investigate the molecular mechanism of the synergistic effects of erlotinib and IFN-α, we investigated the role of IFN signal transducers in HCV infection using functional silencing of gene expression by STAT-specific siRNAs. STAT3 is an important regulator in the STAT1-mediated IFN-α response[40] and has been described as an interaction partner of EGFR in proliferative cells.[41] The effect of STAT3 on HCV infection was studied using RNAi and a pharmacological inhibitor in HCV-permissive Huh7.5.1 cells. Silencing of STAT3 expression in Huh7.5.1 cells (Fig. 2A) decreased infection of HCVcc (Fig. 2B). Pharmacological inhibition of STAT3 function using Cpd188[42] dose dependently impaired HCVcc infection (Fig. 2C). Analysis of cell viability revealed no detectable toxicity, excluding antiviral effects that were the result of unspecific adverse effects (Fig. 2D). Collectively, these data suggest that STAT3 expression and STAT3 function are relevant for HCV infection and that STAT3 is a host factor for HCV infection.

Figure 3.

Erlotinib inhibits IFN-α-induced STAT3 phosphorylation. Inhibition of EGFR function using erlotinib triggers SOCS3 expression that inhibits STAT3 phosphorylation. (A) Twelve-hour serum-starved Huh7.5.1 cells and (B) PHHs were incubated before cell lysis for 15 minutes with 1 µg/mL of EGF, 15 minutes with 1,000 IU/mL of IFN-α, 1 hour with 10 µM of erlotinib, or combinations of compounds as indicated. Cell-culture medium contained 0.7% DMSO. STAT1 phosphorylation at Tyr701 (P-STAT1), STAT3 phosphorylation at Tyr705 (P-STAT3), EGFR phosphorylation at Tyr1086 (P-EGFR), expression of SOCS3, and actin were measured by western blotting. (C) IL-6-induced SOCS3 expression does not influence IFN-α-mediated STAT1 phosphorylation in Huh7.5.1 cells. SOCS3 expression was induced by incubation of Huh7.5.1 cells for 2 hours with 10 ng/mL of IL-6 before induction of STAT signaling for 15 minutes with 1,000 IU/mL of IFN-α. SOCS3 expression and STAT1 phosphorylation at Tyr701 (P-STAT1) were measured by western blotting. (D) Transduction of shSOCS3 inhibited IFN-α-mediated increase of SOCS3 mRNA expression. Huh7.5.1 cells were transduced with lentiviruses coding irrelevant shRNA (shCTRL) or targeting SOCS3 mRNA (shSOCS3) before 1-hour incubation with 1,000 IU/mL of IFN-α or PBS (CTRL). Expression of SOCS3 mRNA is shown relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA. (E) Silencing of SOCS3 expression stimulated HCV replication. Huh7.5.1 cells were transduced with lentiviruses coding shCTRL or shSOCS3 before infection with HCVcc (Luc-Jc1). After 3 days of infection, HCV core protein expression was measured from cell lysates as a measure for HCV replication. (F) EGFR activation and STAT3 phosphorylation in hepatic and nonhepatic cell lines. Human cell lines Huh7.5.1 (hepatoma), HeLa (cervix carcinoma), and HN5 (head neck cancer) were serum starved for 12 hours and then incubated for 15 minutes with 1 µg/mL of EGF before cell lysis. EGFR phosphorylation at Tyr1086 (P-EGFR), STAT3 phosphorylation at Tyr705 (P-STAT3), and actin were measured by western blotting using specific Abs. Western blottings were analyzed using the Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE) (A and C) and chemoluminescence exposure to ECL X-ray films (GE Healthcare Uppsala, Sweden) (B).

Inhibition of EGFR Activity Reduces IFN-α-Induced STAT3 Activity by Induction of SOCS3 Expression

Because the EGFR interaction partner, STAT3, plays a role in HCV infection (Fig. 2), we studied the effect of combinations of IFN-α, EGFR inhibitor erlotinib, and EGFR ligand EGF on STAT1 and STAT3 activity using phosphoimmunoblotting. Figure 3A,B confirmed the biological activity of EGF (P-EGFR, lane 4), erlotinib (P-EGFR, lane 6), and IFN-α (P-STAT1, lane 2) in Huh7.5.1 and PHHs. Incubation of Huh7.5.1 cells or PHHs with IFN-α enhanced the phosphorylation of STAT1 and STAT3 (Fig. 3A,B, lane 2). Although modulation of EGFR activity by EGF or erlotinib had no effect on IFN-α-induced STAT1 phosphorylation, IFN-α-induced STAT3 phosphorylation decreased in the presence of erlotinib (Fig. 3A,B, lanes 7 and 8). Moreover, the expression of SOCS3, a suppressor of STAT3 activity, was induced in the presence of erlotinib (Fig. 3A,B, lanes 6-8). To study whether SOCS3 also modulates STAT1 function in hepatic cells, we analyzed the effect of SOCS3 on IFN-α-mediated STAT1 phosphorylation in Huh7.5.1 cells by inducing SOCS3 expression with IL-6 before stimulation with IFN-α. Because SOCS3 up-regulation did not impair STAT1 phosphorylation in Huh7.5.1 cells (Fig. 3C), we conclude that SOCS3 expression does not have a major role for STAT1 function in hepatic cells.

Figure 4.

SOCS3 silencing impairs the synergistic antiviral effect of IFN-α and erlotinib. Huh7.5.1 cells were transfected with (A) irrelevant siRNA (siCTRL) or (B) siRNA targeting SOCS3 (siSOCS3). After 3 days, cells transfected with siRNAs were preincubated for 1 hour with serial concentrations of IFN-α2a and erlotinib (ERL) before incubation for 72 hours with HCVcc (Luc-Jc1) in the presence of both compounds. HCVcc infection was analyzed by luciferase reporter gene expression, as previously described.[21, 36] Means ± standard deviation of one representative experiment performed in triplicate are shown.

The relevance of SOCS3 in HCV infection was confirmed by lentiviral transduction of SOCS3-specific shRNA in Huh7.5.1 cells. Silencing of SOCS3 messenger RNA (mRNA) expression by shSOCS3 in the absence or presence of IFN-α (1,000 IU/mL) was confirmed by qPCR (Fig. 3D). Silencing of SOCS3 expression increased HCV core protein production in Huh7.5.1 cells infected with HCVcc, which is a measure for HCV replication (Fig. 3E). This observation is in line with our observations that (1) STAT3 function is required for HCV infection (Fig. 2) and that (2) erlotinib-induced enhancement of IFN-α-dependent SOCS3 expression leads to reduced STAT3 phosphorylation (Fig. 3A,B). Interestingly, in contrast to results obtained in nonhepatic human cell lines HeLa and HN5 (Fig. 3F), EGF did not induce STAT3 phosphorylation in Huh7.5.1 cells and PHHs (Fig. 3A,B). These data indicate that, in hepatic cells, EGF does not result in STAT3 activation in the absence of IFN-α.

The role of SOCS3 for the synergistic action of erlotinib on the antiviral activity of IFN was confirmed by functional experiments in cells with silenced SOCS3 expression. Although erlotinib enhanced the antiviral activity of IFN-α (IC50 (IFN-α) = 0.4 IU/mL; IC50 (IFN-α + 0.1 µM of erlotinib) = 0.04 IU/mL) in a highly synergistic manner in Huh7.5.1 cells transfected with irrelevant siRNA not affecting SOCS3 expression (CI = 0.2; Fig. 4A), erlotinib acted only in an additive manner (IC50 (IFN-α) = 0.6 IU/mL; IC50 (IFN-α + 0.1 µM of erlotinib) = 0.6 IU/mL) in cells with silenced SOCS3 expression (CI = 1.1; Fig. 4B). These data clearly confirm a key role of SOCS3 for the observed synergy of erlotinib on the antiviral activity of IFN-α.

Figure 5.

Erlotinib enhances STAT1 homodimerization and expression of IFN-α response genes RSAD2 and Mx1. (A) EMSA measuring IFN-α-induced STAT1:STAT1, STAT1:STAT3, and STAT3:STAT3 dimerization. Before cell lysis, 12-hour serum-starved Huh7.5.1 cells were incubated for 30 minutes with 1 µg/mL of EGF, 30 minutes with 1,000 IU/mL of IFN-α, 1 hour with 10 µM of erlotinib, or combinations of compounds as indicated. Cell-culture medium contained 0.7% DMSO. EMSA was performed using the STAT1/STAT3-specific oligonucleotide probe, m67.[16] Sample 2 (IFN-α treated) was used in lanes 9 and 10 for supershift with specific STAT1 or STAT3 Abs to confirm STAT1 and STAT3 bands on the gel. STAT1:STAT1 and STAT1:STAT3 bands disappeared (supershift) in the presence of STAT1-specific Ab (lane 9). STAT3:STAT3 and STAT3:STAT1 disappeared in the presence of STAT3-specific Ab (lane 10). (B and C) Quantification of RSAD2 and Mx1 mRNAs using qPCR. Serum-starved Huh7.5.1 cells were incubated for 4 hours with 1 µg/mL of EGF, 4 hours with 1,000 IU/mL of IFN-α, 1 hour with 10 µM of erlotinib, or combinations of compounds as indicated. Cell-culture medium contained 0.7% DMSO. mRNAs were quantified from cell lysates relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA using qPCR. **P < 0.005; ***P < 0.0005.

Taken together, these data indicate that IFN-α-dependent SOCS3 expression is enhanced by erlotinib. Enhanced SOCS3 expression results in repression of STAT3 activation with subsequent synergistic enhancement of the antiviral activity of IFN-α.

Inhibition of EGFR Activity Enhances STAT1-Mediated Antiviral Response

Because STAT1 and STAT3 have opposing roles in many aspects of the IFN response,[10] we next studied whether elevated SOCS3 levels resulting from inhibited EGFR activity contribute to an enhanced antiviral response by STAT1 activation. Phosphorylated STATs form dimers or trimers that translocate to the nucleus, where they bind to DNA and direct specific transcriptional initiation.[4] P-STAT1 homodimerization, in the presence of IFN-α, EGF, erlotinib, or combinations of the three compounds, was quantified using EMSA. Therefore, a 32P-labeled oligonucleotide probe (m67) derived from the promoter of the c-fos gene[16] was used, which specifically binds to P-STAT1 and P-STAT3 dimers. The presence of EGF reduced P-STAT1 homodimerization (Fig. 5A, lanes 2 and 5) correlated with reduced transcription levels of the IRGs, Mx1 and RSAD2 (Fig. 5B,C). IFN-α-induced Mx1 and RSAD2 expression play an important role during viral infections.[2, 5-7] This suggests a negative regulatory role of EGFR for STAT1 function (Fig. 5A, lanes 2 and 5). In contrast, inhibition of EGFR activity by erlotinib increased STAT1 homodimerization (Fig. 5A, lanes 5 and 8) and removed the inhibitory effect of EGF on IFN-α-induced STAT1 function, leading to significantly (P < 0.05) increased Mx1 and RSAD2 transcription levels (Fig. 5B,C, bars 5 and 8). These data confirm the regulatory role of EGFR on STAT1 function and suggest that inhibition of EGFR function by erlotinib enhances the IFN-α-induced antiviral function of STAT1. Moreover, these data demonstrate that the regulation of IFN-α-induced gene expression by EGFR occur at the DNA-binding level of STAT1 (Fig. 5), because STAT1 phosphorylation is not affected by EGF treatment (Fig. 3A,B, lanes 2 and 5).

Discussion

Our results uncover a previously undiscovered regulatory role of EGFR for STAT3 activity that counteracts the proinflammatory and antiproliferative function of STAT1. In the cell, STAT1 and STAT3 activity delicately balance IFN-α signaling.[10, 40] Cross-talk influences that balance, which may favor the opposing roles of STAT1 or STAT3. For example, it has been shown that EGF activates STAT3 transcriptional activity.[43] Moreover, Wang et al. have recently demonstrated that STAT3 knockout enhances type I IFN-mediated antiviral activity.[44] Although STAT3 has been described as a binding factor to EGFR in highly proliferating cells,[45] we demonstrate that STAT3 activity is not directly mediated by EGFR engagement in HCV-permissive cells, because, in hepatic cells, EGFR stimulation by EGF in the absence of IFN-α is not sufficient to mediate STAT3 phosphorylation (Fig. 3A,B,F). Rather, our data demonstrate that inhibition of EGFR function in hepatocytes unleashes SOCS3 expression that inhibits IFN-α-induced STAT3 activity. This hypothesis is supported by previous observations showing that EGFR overexpression down-regulates SOCS3 expression in mouse embryonic fibroblasts,[46] consistent with a negative regulatory role of EGFR for SOCS3 expression by acting as a repressor. Interestingly, serum starvation of cells (Fig. 3A,B, lanes 1-3) did not release the repression of IFN-α-induced SOCS3 expression. Because EGF did not further enhance IFN-α-mediated STAT3 phosphorylation, we conclude that, in PHHs and Huh7.5.1 cells, ligand-independent EGFR basal activity (indicated by basal EGFR phosphorylation in serum-starved Huh7.5.1 cells and PHHs, as shown previously[21]; Fig. 3F) is sufficient to maintain SOCS3 repression. This repressive effect by basal EGFR activity is not further enhanced by more-pronounced EGFR activation through EGF (Fig. 3A,B).

Moreover, we observed that EGF inhibited STAT1 homodimerization that correlated with reduced ISG expression, whereas STAT3 homodimerization was unchanged (Fig. 5). Taken together, these observations appear to point toward a regulatory role of EGFR signaling on STAT1/STAT3 in human hepatocytes, which is different from nonhepatic cells. Indeed, in contrast to nonhepatic human cell lines HeLa and HN5 (Fig. 3F), EGF did not enhance STAT3 phosphorylation in Huh7.5.1 cells and PHHs (Fig. 3A,B,F). Although further experiments need to be performed to better understand these findings, it is conceivable that this observation is relevant for the special role of EGFR during liver regeneration[9, 47] and HCV infection.[23] EGFR activation may inhibit the proapoptotic role of STAT1 signaling on a DNA-binding level, resulting in promotion of cell proliferation and viral spread.

Collectively, identification of the cross-talk of EGFR and IFN-α explains the marked synergy of IFN-α and erlotinib through the following findings: (1) Synergy at low doses of erlotinib (0.1 µM) with absent effects on HCV entry and infection in monotherapy (Fig. 1C,D) indicates that EGFR activity has a functional role for the antiviral activity of IFN-α that is independent from its role in HCV entry; (2) SOCS3 expression is required for synergy on antiviral activity (Fig. 4); (3) inhibition of STAT3 phosphorylation by erlotinib through SOCS3 shifts the balance of STAT1/STAT3 activity toward an enhanced antiviral effect of STAT1 that is reflected by enhanced IRG transcription levels (Fig. 5); (4) STAT3 function is relevant for HCV infection (Fig. 2), and inhibition of STAT3 phosphorylation by erlotinib (Fig. 3) may further contribute to the antiviral activity of erlotinib in combination with IFN-α; and (5) inhibition of EGFR function by erlotinib enhances the IFN-α-induced antiviral function of STAT1 (Fig. 5).

There are two implications of EGFR-IFN cross-talk. First, the interference of signaling explains viral evasion during IFN-α treatment. Because HCV infection has been shown to result in EGFR activation,[23] virus-induced EGFR signaling and cross-talk may contribute to viral evasion and resistance to IFN-α. Furthermore, EGFR expression is up-regulated in response to treatment, presumably as a homeostatic cellular response to antiproliferative stimuli by IFN-α.[48] Because EGFR is a cofactor for HCV entry, and EGFR expression correlates directly with enhanced viral entry and propagation,[21] elevated EGFR levels in IFN-α-treated patients might be further exploited by the virus to counteract the antiviral effect of IFN-α. Collectively, these findings suggest an important role for EGFR signaling in viral evasion and IFN-α resistance. Second, our results open a novel possibility of improving the efficacy of IFN-α-based therapies. Although the clinical licensing of protease inhibitors has improved the outcome of IFN-α-based therapy for HCV, adverse effects, drug-drug interactions, and a low genetic barrier to resistance are important limitations.[49] These limitations result in unsatisfactory treatment options in patients with advanced liver disease, liver transplantation, comorbidity, and human immunodeficiency virus coinfection.[50] Although promising IFN-α-sparing regimens are in clinical development, early clinical trials demonstrate that nonresponders to previous IFN-based therapies appear only partially to clear viral infection.[51] Thus, it is likely that a clinically relevant subset of patients will require IFN-α in the future, despite the anticipated approval of IFN-free regimens.

Our mechanistic data suggest that a clinical combination therapy of IFN-α and erlotinib may result in unique benefits to address the limitations of IFN-α. First, as shown here, erlotinib synergistically enhances the antiviral activity of IFN-α by interfering with IFN-α signaling. Enhanced efficacy may contribute to break the tolerance to IFN-α in patients with slow virological response or primary nonresponders.[19] Second, erlotinib may decrease IFN resistance by overcoming a potential proviral effect of enhanced EGFR expression induced by IFN.[48] Third, erlotinib may increase the genetic barrier to viral resistance to therapy by inhibiting a host cell HCV entry factor.[21, 23] In contrast to the very high genetic variability of the virus, the genetic variability of host factors, such as EGFR, is low.[26, 28]

It is important to note that the combination of high doses of erlotinib and IFN-α did not result in detectable toxicity in PHHs (data not shown). In clinical use, the safety of erlotinib in cancer patients is comparable to HCV protease inhibitors.[52-54] An investigator-initiated phase I clinical trial is in preparation at Strasbourg University Hospital (Strasbourg, France) to specifically evaluate the safety, antiviral activity, and resistance profile of erlotinib in HCV-infected patients.

Because the IFN-α-sensitizing effect of SOCS3/STAT3 and STAT1 is independent of the specific host-targeting activity of erlotinib on HCV entry, it is possible that the synergistic action of erlotinib also potentiates the antiviral activity of IFN-α against other viruses, such as HBV or HDV. Collectively, our data reveal a novel concept and strategy to enhance the antiviral efficacy of IFN-α and decrease resistance of IFN-α-based therapies.

Acknowledgment

The authors thank L. Heydmann (Inserm U1110, Strasbourg, France) and A. Weiss (IGBMC, Illkirch, France) for their excellent technical work, J. Neyts and P. Leyssen (Rega Institute for Medical Research, KU Leuven, Leuven, Belgium), C. Schuster, S. Fafi-Kremer, and H. Barth (Inserm U1110) for their helpful discussions, and D.J. Felmlee (Inserm U1110) for critical reading and editing of the manuscript.

Ancillary