Article Text

Download PDFPDF

Original article
Ribavirin restores IFNα responsiveness in HCV-infected livers by epigenetic remodelling at interferon stimulated genes
  1. Barbara Testoni1,2,3,
  2. David Durantel1,2,
  3. Fanny Lebossé1,2,3,
  4. Judith Fresquet1,2,
  5. François Helle4,
  6. Francesco Negro5,
  7. Maria Francesca Donato6,
  8. Massimo Levrero7,8,9,
  9. Fabien Zoulim1,2,3,10
  1. 1INSERM U1052, Cancer Research Center of Lyon (CRCL), Lyon, France
  2. 2University of Lyon, UCBL, UMR_S1052, Lyon, France
  3. 3Hospices Civils de Lyon (HCL), Lyon, France
  4. 4EA4294, Laboratoire de Virologie, Centre Hospitalier Universitaire et Université de Picardie Jules Verne, Amiens, France
  5. 5Division of Gastroenterology and Hepatology and of Clinical Pathology, University Hospitals, Geneva 4, Switzerland
  6. 6IRCSS Foundation Ca’ Granda, Maggiore Hospital Policlinico and University of Milan, Milan, Italy
  7. 7Department of Internal Medicine (DMISM) and the IIT-CNLS, Sapienza University of Rome, Rome, Italy
  8. 8EAL INSERM U785, Villejuif, France
  9. 9EAL INSERM U785, Rome, Italy
  10. 10Institut Universitaire de France (IUF), Paris, France
  1. Correspondence to Professor Fabien Zoulim, Centre de Recherche en Cancérologie de Lyon (CRCL), UMR_INSERM 1052-CNRS 5286, 151, cours Albert Thomas, Lyon 69424, CEDEX 03, France; fabien.zoulim{at}inserm.fr

Abstract

Objectives Caveats in the understanding of ribavirin (RBV) mechanisms of action has somehow prevented the development of better analogues able to further improve its therapeutic contribution in interferon (IFN)-based and direct antiviral agent-based regimens for chronic HCV or other indications. Here, we describe a new mechanism by which RBV modulates IFN-stimulated genes (ISGs) and contributes to restore hepatic immune responsiveness.

Design RBV effect on ISG expression was monitored in vitro and in vivo, that is, in non-transformed hepatocytes and in the liver of RBV mono-treated patients, respectively. Modulation of histone modifications and recruitment of histone-modifying enzymes at target promoters was analysed by chromatin immunoprecipitation in RBV-treated primary human hepatocytes and in patients’ liver biopsies.

Results RBV decreases the mRNA levels of several abnormally preactivated ISGs in patients with HCV, who are non-responders to IFN therapy. RBV increases G9a histone methyltransferase recruitment and histone-H3 lysine-9 dimethylation/trimethylation at selected ISG promoters in vitro and in vivo. G9a pharmacological blockade abolishes RBV-induced ISG downregulation and severely impairs RBV ability to potentiate IFN antiviral action and induction of ISGs following HCV infection of primary human hepatocytes.

Conclusions RBV-induced epigenetic changes, leading to decreased ISG expression, restore an IFN-responsive hepatic environment in patients with HCV, which may also prove useful in IFN-free regimens.

  • HEPATITIS C
  • INTERFERON
  • CHRONIC VIRAL HEPATITIS
  • GENE REGULATION
View Full Text

Statistics from Altmetric.com

Significance of this study

What is already known on this subject?

  • The mechanism underlying the ability of ribavirin (RBV) to potentiate the antiviral effect of pegylated interferon-α (PEG-IFNα) and new generation direct antiviral agents (DAAs) in patients infected with HCV remains largely unknown.

  • RBV potentiates IFN-stimulated genes (ISGs) induction by IFN but the mechanism involved is still unclear.

  • Specific ISGs are preactivated in the liver of patients who will not respond to PEG-IFNα/RBV therapy and cannot be further activated by exogenous IFN treatment.

What are the new findings?

  • RBV downregulates abnormally preactivated ISGs following HCV infection of primary human hepatocytes and in the liver of RBV mono-treated patients infected with HCV.

  • This downregulation is achieved by chromatin remodelling and modulation of histone methylation following the recruitment of the G9a histone methyltransferase on target promoter regions.

How might it impact on clinical practice in the foreseeable future?

  • The new concept that RBV resets the abnormal preactivation and IFN-unresponsiveness of specific ISGs in the liver of patients infected with HCV may explain its contribution to higher sustained virological response rates.

  • In the context of the new DAA-based regimen, our results may explain the role of RBV in restoring responsiveness to endogenous IFN to enhance antiviral activity and increase sustained virological response rates especially in difficult-to-treat patients, that is, patients with cirrhosis with genotype 3 infection, or in regimen combining paritaprevir, ombitasvir and dasabuvir.

Introduction

Ribavirin (RBV), a synthetic triazole analogue of guanosine, has played a crucial role in anti-HCV therapy since its approval by Food and Drug Administration (FDA) in 1998. Despite its long-term and successful clinical use, how RBV potentiates the antiviral effect of pegylated interferon-α (PEG-IFNα) in patients infected with HCV is still a matter of debate. Several, not mutually exclusive, modes of action have been proposed, including: (A) a direct inhibition of the viral RNA-dependent RNA-polymerase; (B) depletion of the cellular guanosine triphosphate (GTP) pool by inhibiting the enzymatic activity of the inosine monophosphate dehydrogenase (IMPDH); (C) direct mutagenesis of viral nucleic acids leading to error catastrophe; (D) immunomodulation by favouring a TH2 to TH1 T lymphocyte shift.1 ,2 Studies of RBV monotherapy in patients infected with HCV not eligible for IFN-based regimens have shown that RBV alone has little or no effect on HCV viral load.3–5 Although an increased rate of G-to-A and C-to-U transitions, specifically induced by RBV, could be detected by next generation sequencing in RBV monotreated patients, there is no evidence for continuous accumulation of mutations and generation of defective viruses.5–9 Interestingly, approximately half of the patients undergoing long-term RBV monotherapy show a significant reduction in serum alanine aminotransferase (ALT) levels and a liver histology improvement.10–13 This suggests that RBV may be beneficial in controlling the progression of chronic hepatitis C through indirect mechanisms of action not directly related to the control of viral replication.

RBV was shown to be required with first generation direct antiviral agents (DAAs) in combination with PEG-IFNα by reducing the development of resistance and virological breakthroughs.14 Furthermore, newly available data indicate that RBV administration might still be required to optimise the response to IFN-free DAA combinations15 and, in particular, to improve the clinical response to some of the newly developed IFN-free combinations16 ,17 and in difficult-to-treat patients, that is, HCV genotype 3 patients with cirrhosis and genotype 1 infected patients with decompensated cirrhosis.18 ,19

Increasing evidence indicate that RBV might also potentiate IFNα signalling. A 3-day pretreatment with RBV followed by PEG-IFN administration has been shown to greatly improve interferon-stimulated gene (ISG) induction in the livers of patients chronically infected with HCV.20 In vitro experiments performed with uninfected hepatoma cells or cell lines harbouring HCV subgenomic-replicons confirmed the strong synergism with IFNα, but the direct effect of RBV on ISGs transcriptional activation remained controversial and the underlying mechanisms elusive.21–23 A more recent study on hepatic gene expression in patients undergoing a 4-week lead-in RBV monotherapy confirmed the potentiation of intrahepatic ISG response to subsequent exposure to PEG-IFN and the lack of generalised RBV-triggered ISG activation.24 The same study has shown, however, that RBV alone rather inhibits the expression of some ISGs, thus raising the possibility that RBV might favour liver susceptibility to IFNα by lowering the baseline expression of specific ISGs rather than directly activating IFNα signalling. This interpretation is also consistent with the notion that specific ISGs, including RSAD2, CXCL10, IFI27, among others, are preactivated in the liver of patients who will not respond to PEG-IFNα/RBV therapy and will not be further activated by exogenous IFN treatment.25–27

Here, we provide novel evidences that RBV induces epigenetic changes contributing to the restoration of a hepatic environment favourable to IFN stimulation in patients chronically infected with HCV in vivo. This is achieved through the specific downregulation of abnormally preactivated ISGs by chromatin remodelling and modulation of histone marks following the recruitment of the G9a histone methyltransferase on their promoter regions.

Materials and methods

Cell cultures and drug treatments

HepaRG cell line was maintained in William's medium (Life Technologies) supplemented with 10% FetalClone II (Thermo Scientific), 1% penicillin/ streptomycin and 1% glutamine (Life Technologies), as well as 5 μg/mL insulin and 5×10−7 M hydrocortisone hemisuccinate (Sigma Aldrich). To induce HepaRG differentiation towards the hepatocytic lineage, complete medium was supplemented with 1.8% of dimethyl sulfoxide (DMSO) (Hybrimax, Sigma Aldrich) for 2 weeks. Huh7 hepatoma cells were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% FetalClone II, 1% penicillin/streptomycin and 1% glutamine. PEG-IFNα (Roferon A, Roche), RBV (Sigma Aldrich), mycophenolic acid (Sigma Aldrich), guanosine (Sigma Aldrich) and BIX01294 (Sigma Aldrich) were used at a final concentration of 1000 IU/mL, 100 μg/mL, 10 ng/mL, 100 µg/mL and 5 μM, respectively, and added directly to the culture medium alone or in combination. Neutral red assay was used for the assessment of cell viability after drug treatments.28

Primary culture of human hepatocytes and HCV infection

Primary human hepatocytes (PHHs) were prepared from adult patients undergoing lobectomy or segmental liver resection for medically required purposes unrelated to this research programme. Detailed protocol is found in online supplementary materials and methods.

siRNA transfection

Dharmafect Reagent (GE Healthcare) was used to transfect siRNA into dHepaRG cells according to manufacturer's instructions. The SMARTpool ON-TARGETplus EHMT2 siRNA (GE Healthcare #L-006937) was used to specifically target G9a, while ON-TARGETplus Non-Targeting Pool (GE Healthcare #D-001810-10) was used as a control.

RNA extraction, RT-qPCR and Taqman low density arrays

After drug treatments of PHHs, differentiated HepaRG and Huh7 cells, total mRNAs were extracted at different time points with Nucleospin RNA II (Macherey & Nagel) according to the manufacturer's instructions. RNA extraction from liver biopsies was performed with High Pure Paraffin Kit (Roche) to maximise RNA yield and quality. Primers and detailed instructions for qPCR are described in online supplementary materials and methods.

Western blot and ELISA analysis

Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer (10 mM Tris HCl pH 7.5, 140 mM NaCl, 1 mM EDTA, 0.5 mM ethylene glycol tetraacetic acid (EGTA), 1% Triton, 0.1% sodium dodecyl sulphate (SDS), 0.1% Na-Deoxycholate, 1 mM phenylmethanesulfonylfluoride (PMSF), protease inhibitors), incubated on ice for 10 min, then centrifuged to pellet the cellular debris and supernatant was quantified using Bradford reagent (Sigma). Antibodies used for western blotting were the same used in chromatin immunoprecipitation (ChIP) except for anti-β-ACTIN (Cell signalling #8457), OAS1 (Sigma Aldrich #SAB1105258), USP18 (Cell signalling #4813), and Hsp60 (Abcam #ab46798). CXCL10 ELISA tests on HepaRG supernatants were performed according to the manufacturer's instructions (R&D Systems #DY266).

ChIP and real-time quantification

ChIP and µChIP analyses were based on the protocol described in Testoni et al29 and Dahl and Collas,30 with minor modifications. See online supplementary materials and methods for detailed protocol.

IL28B/IFNL3 rs12979860 genotyping

Genomic DNA was extracted from sera of patients using the MasterPure Complete DNA and RNA Purification Kit (Epicentre, Illumina) according to the manufacturer's instructions and then analysed for IL28B rs12979860 polymorphisms using the LightMixKit IL28B (40-0588-32, Roche) following the manufacturers’ recommendations.

Intrahepatic HCV RNA quantification

RNA from patients’ biopsies was extracted and retrotranscribed as detailed above. For qPCR analysis, specific HCV primers were used: RC1 5′-GTCTAGCCATGGCGTTAGTA-3′ and RC21 5′-CTCCCGGGGCACTCGCAAGC-3′. 2ΔΔCt method was used to calculate fold changes with respect to the housekeeping gene β-ACTIN and RPLP0 and expressed as arbitrary units.

Patients and statistical analysis

A detailed description of patients’ characteristics and statistical methods used in the study can be found in online supplementary materials and methods.

Results

RBV represses a subset of ISGs in non-transformed hepatic cells

The recent observations made by Rotman et al,24 who failed to find a generalised RBV-triggered ISG activation in vivo, conflict with the RBV-mediated activation of ISGs reported by others in hepatic and non-hepatic transformed cell lines.21 ,31 ,32 The mechanisms underlying ISG activation remain controversial and the ability of RBV to enhance signal transducer and activator of transcriptions (STATs) phosphorylation and interferon sensitive response element (ISRE)-dependent transcription has been strongly challenged.22 To further investigate the molecular basis of RBV-triggered repression observed in vivo, we took advantage of more physiological hepatic in vitro cellular models, that is, PHHs and differentiated HepaRG cells (dHepaRG),33 together with the hepatoma Huh7 cell line used as a control. Dose-response and time-response experiments with RBV, PEG-IFNα and their combination were performed to define the optimal doses used to analyse the expression profiles of 95 ISGs by customised Taqman low density arrays (see online supplementary table S1 and figure S1). No evident cell toxicity was detected at the RBV concentrations selected for subsequent experiments in PHH and HepaRG cells (ie, 100 µg/mL; see online supplementary figure S2A). The main effect of RBV in PHHs and dHepaRG cells was to repress ISGs, with only few activated genes (15% of total), (figure 1A, lane 1 of upper and lower panels, and figure 1B). A complete list of activated and repressed genes is provided in online supplementary table S2. Regarding the spectrum of modulated genes, the effect of PEG-IFNα and RBV combination was not significantly different from that of IFNα alone (figure 1A, lane 3 vs lane 2 in upper and lower panels). In fact, RBV potentiated IFN-driven gene activation rather than enlarging the spectrum of activated genes (see online supplementary table S2). Among RBV-repressed genes in PHHs and dHepaRG cells (figure 1A, lane 1 of upper and lower panels), many were ISGs that are overexpressed in patients non-responders to PEG-IFNα/RBV treatment.25–27 Quantitative RT-PCR performed on three of these genes (CXCL10, IFI27 and RSAD2) confirmed their repressed expression by RBV alone in non-transformed liver cells (figure 1C and see online supplementary figure S1) and activated expression, as previously shown,22 in Huh7 hepatoma cells (figure 1C). The ability of RBV to reduce CXCL10 steady-state mRNA levels translated into a significant reduction in the corresponding IP-10 protein levels (figure 1D). Similarly, RBV-triggered transcriptional repression of OAS1 and USP18 (see online supplementary table S2) was accompanied by reduced protein levels (figure 1D). RBV also increased to some extent ISG activation in response to IFNα, and this effect was maximal when RBV was administered before IFN stimulation and greater in either PHHs or dHepaRG cells as compared with Huh7 hepatoma cells (figure 1C). Meanwhile, IL-8 and other genes that were activated by RBV alone were not further activated when cells were treated with the combination of PEG-IFNα and RBV (figure 1C and online supplementary table S2).

Figure 1

Ribavirin (RBV) modulates interferon-stimulated gene (ISG) expression in non-transformed liver cells. (A) Heat map showing ISG expression in primary human hepatocytes (PHHs) (upper panel) and dHepaRG cells (lower panel) after (1) 24 h of RBV treatment (100 μg/mL), (2) 8 h interferon-α (IFN-α) treatment (1000 IU/mL), or (3) 24 h of RBV followed by 8 h of IFNα. (B) Dot plot showing the mRNA levels of RBV repressed genes in PHHs (36 genes) and HepaRG cells (23 genes) after RBV, IFNα or RBV+IFNα treatments. Two-tailed p value was calculated for a risk threshold=0.05 using Mann-Whitney test. Individual absolute gene expression values are listed in online supplementary table S2. (C) RT-qPCR on selected ISGs in PHHs, dHepaRG and Huh7 cells after different combinations of RBV (100 μg/mL) and IFNα (1000 IU/mL) treatments. mRNA expression levels have been normalised over RPLP0 and β-ACTIN housekeeping gene expressions. Experiments have been repeated thrice with dHepaRG and Huh7 cells and with four different batches of PHHs. (D) ELISA test for CXCL10/IP-10 protein secretion (left panel) and western blot analysis of OAS1, USP18 and Hsp60 protein expression (right panel) in HepaRG cells after RBV treatment (100 µg/mL) for 24 h or 48 h. OAS1 and USP18 proteins two major isoforms are presented in the blot: a full length and N-truncated form for USP18, a p42 and a p46 isoform for OAS1. Graphs represent mean values±SD. Two-tailed p value was calculated for a risk threshold=0.05 using the two/K sample permutation test with Monte Carlo resampling approximation. p Value <0.05*, <0.01**, <0.001***.

Inhibition of IMPDH, and the consequent depletion of intracellular guanosine pools, have been proposed as one of the possible RBV modes of action and have been very recently invoked as a main determinant of a potential direct anti-HCV activity of RBV.34 Interestingly, the capacity of RBV to repress ISGs and to potentiate IFN activation in differentiated hepatocytes does not rely on its activity as IMPDH inhibitor. Indeed, mycophenolic acid, a well-characterised and specific IMPDH inhibitor, neither repressed ISG expression nor enhanced ISG activation by IFNα (see online supplementary figure S2B). Guanosine was reported to abrogate RBV-mediated and IMPDH-dependent activation of ISGs in transformed cells.22 Interestingly, in our model, the administration of guanosine did not revert CXCL10, IFI27 and RSAD2 ISG inhibition by RBV in PHHs and dHepaRG cells (see online supplementary figure S2C).

RBV-triggered ISG repression is mediated by G9a histone methyltransferase activity

The repertoire of RBV-repressed ISGs largely overlaps with the one upregulated in patients infected with HCV who do not respond to PEG-IFNα/RBV therapy. We speculated that RBV-triggered downregulation could be functional to counteract the abnormal activated status of these ISGs. Since these ISGs share a common chromatin organisation,29 we investigated whether RBV could change their chromatin status in order to restore their sensitivity to IFN induction. Histone modifications regulate promoter accessibility to the transcriptional machinery and the balance between histone H3 methylation at lysine 4 (H3K4me) and lysine 9 (H3K9me) seems to play a pivotal role in regulating the recruitment of transcription factor and cofactors.35 ,36 Interestingly, H3K9 dimethylation and trimethylation (H3K9me2/3) and G9a, the histone methyltransferase responsible for these histone modifications,37 have been recently shown to play a key role in the regulation of innate immunity genes.38 As a consequence, we investigated whether H3K9me2/3 could be involved in ISG repression by RBV. To this aim, PHHs and dHepaRG cells were treated with the G9a inhibitor BIX01294,37 which specifically diminishes the degree of H3K9 dimethylation/trimethylation without affecting G9a mRNA and protein levels (see online supplementary figure S3A,B). G9a functional inhibition by BIX01294 completely reverted RBV-induced ISG repression, whereas it did not affect RBV-induced ISG activation, as shown for IL8 (figure 2A) and had no effect in Huh7 cells (see online supplementary figure S3C). In order to link unequivocally G9a and G9a-dependent histone modifications to RBV-induced ISG repression, we showed that efficient G9a depletion by targeted siRNA in HepaRG cells (figure 2B) fully restored CXCL10, IFI27 and RSAD2 expression (figure 2C). Thus, chromatin remodelling involving H3K9me2/3 may be critical for ISG repression by RBV, whereas it is not involved in ISG activation. Importantly, BIX01294 also completely abolished RBV potentiation of IFN-dependent ISG activation, strongly suggesting that G9a-dependent RBV-triggered repression is a prerequisite for RBV potentiation of IFNα stimulation (figure 2D).

Figure 2

H3K9 methylation by the G9a HMT mediate ribavirin (RBV)-induced transcriptional repression and potentiation of interferon (IFN) antiviral activity (A) RT-qPCR on selected IFN-stimulated genes (ISGs) in primary human hepatocyte (PHH) and dHepaRG cells exposed for indicated time to RBV (100 μg/mL), BIX01294 (5 μM) or their combination. (B) Left panel, RT-qPCR on G9a mRNA expression in dHepaRG cells after transfection of a scrambled siRNA (siCTL) or a siRNA targeting G9a for 24 h or 48 h. mRNA expression levels have been normalised over RPLP0 and β-ACTIN housekeeping gene expressions. Experiments have been repeated thrice. Graphs represent mean values ±SD. Right panel, western blot analysis of G9a and β-ACTIN protein expression in dHepaRG cells following 48 h of siCTL or siG9a transfection (25 nM). (C) RT-qPCR on selected ISGs mRNA in dHepaRG cells after transfection with control (siCTL) or G9a-specific (siG9a) siRNAs and RBV treatment. Cells were first transfected with 25 nM of siCTL or siG9a for 48 h and then treated with 100 µg/mL RBV for 24 h. RNA expression levels have been normalised over RPLP0 and β-ACTIN housekeeping gene expressions. Graphs represent mean values ±SD. Experiments have been repeated thrice. (D) RT-qPCR on selected ISGs in PHH and dHepaRG cells exposed for indicated time to IFN (1000 IU/mL), RBV (100 μg/mL), BIX01294 (5 μM) or their combination. mRNA levels were normalised over RPLP0 and β-ACTIN housekeeping gene expressions. Graphs represent mean values±SD. Experiments have been repeated thrice with dHepaRG cells and with four different batches of PHHs. (E) RT-qPCR of selected ISGs following treatments with IFN (1000 IU/mL), RBV (100 μg/mL), BIX01294 (5μM) or their combination in HCV infected PHH; (F) HCV RNA quantification in PHHs infected with HCV (strain JFH1) at a multiplicity of infection (MOI) of 1–5 and treated at day 3 post infection with IFN (1000 IU/mL), RBV (100 μg/mL), BIX01294 (5 μM) or their combination for 24 h. Experiments shown in (E and F) have been repeated thrice. Two-tailed p value was calculated for a risk threshold=0.05 using the two/K sample permutation test with Monte Carlo resampling approximation. p Value <0.05*, <0.01**, <0.001***.

In HCV-infected PHH, CXCL10, IFI27 and RSAD2 expression was strongly induced (see online supplementary figure S3D) and IFNα was not able to further increase their expression levels (figure 2E). As shown in figure 2E, RBV treatment inhibited CXCL10, IFI27 and RSAD2 activation and restored IFNα capacity to enhance ISG activation. These effects of RBV were reverted by BIX01294-mediated inhibition of G9a activity (figure 2E), thus linking RBV-dependent epigenetic modulation with its anti-HCV activity.

The relevance of RBV-induced G9a-dependent repression of selected ISGs for the control of HCV replication is also supported by the ability of BIX01294 treatment to counteract RBV-induced potentiation of IFN antiviral effect in PHHs infected with HCV (ie, strain JFH1) 39 ,40 (figure 2F and see online supplementary figure S3E,F). BIX01294+IFNα+RBV combined treatment shows no antiviral effect (figure 2F). We do not have a mechanistic explanation for this unexpected result, which is, however, not due to experimental variability among different infections in PHHs, as demonstrated by the dot plot representation (figure 2F and see online supplementary figure S3F).

Recruitment of G9a HMT and methylation at H3K9 mediate RBV-induced transcriptional repression

To explore in more detail chromatin organisation at ISG promoters in response to RBV treatment, we performed ChIP experiments in PHHs treated either with RBV, or IFN and their combination. ChIPs were performed with antibodies against G9a and the dimethylated or trimethylated forms of lysine 9 of histone 3 to assess G9a recruitment and G9a-directed histone methylation at target promoters, as well as antibodies against STAT2 and its active phosphorylated form (P-STAT2) to assess if RBV had an impact on the recruitment of transcriptionally active complexes, the final step of type I IFN signalling. We confirmed our previous observations29 that selected ISGs possess a basal level of chromatin bound STAT2/P-STAT2, which is significantly increased upon IFNα simulation (figure 3A and see online supplementary figure S4A). We also showed that RBV does not significantly affect STAT2 protein levels (see online supplementary figure S4C), has a modest effect on STAT2 and P-STAT2 occupancy on target promoters (figure 3A), and that IFNα/RBV combined treatment does not increase their recruitment as compared with IFNα alone (figure 3A). These results indicate that RBV exerts its repressive activity without impacting on type-I IFN signalling. Similarly, BIX01294 and BIX01294/RBV have no or a very limited effect on STAT2 and P-STAT2 occupancy on target promoters (data not shown and figure 3A). The lack of modulation of STAT2 and P-STAT2 occupancy on the IL8 promoter, transcriptionally activated by RBV (figure 3A), is in agreement with the recent observation that RBV-dependent activation of IL8 is mediated by AP1 rather than STAT transcription factors.23 ISG promoters also displayed a basal binding level of G9a and H3K9me2/3, which was strongly increased in response to RBV administration, and accordingly lost after IFN, RBV+IFN and BIX01294/RBV treatments (figure 3B, see online supplementary figure S4B). No increase in H3K9me2/3 levels was observed on IL8 promoter, which is activated by RBV (figure 3B). Western blot analysis showed that RBV does not change the global levels of H3K9 dimethylation/trimethylation which are reduced by IFNα and BIX, thus suggesting that RBV treatment results in a redistribution of H3K9 repressive marks on specific target genes (see online supplementary figure S4D).

Figure 3

Epigenetic modulation of interferon-stimulated gene (ISG) promoters in primary human hepatocytes (PHHs) by ribavirin (RBV). Chromatin immunoprecipitation (ChIP) was performed with specific antibodies against Stat2 and phospho-Stat2 (A) or G9a, H3K9me2 and H3K9me3 (B). Signal enrichment is expressed as % of input chromatin. Control of antibodies background binding levels on GAPDH promoter is shown in online supplementary figure S4A, B. Results were derived from three independent ChIPs on distinct PHH preparations. Graphs represent mean values±SD. Two-tailed p value was calculated for a risk threshold=0.05 using the two/K sample permutation test with Monte Carlo resampling approximation. p Value <0.05*, <0.01**, <0.001***.

ISGs epigenetic signature in patients infected with HCV treated with RBV

PHHs and dHepaRG cells are not susceptible to persistent HCV infection and this prevents testing the epigenetic effects of RBV in the context of a chronic viral infection in vitro.41 To overcome this limitation, we performed micro-ChIPs on matched pretreatment (preTT) and on-treatment (onTT) liver biopsy samples from six patients infected with HCV. Three of them belong to a cohort of 24 patients enrolled in an RBV monotherapy protocol due either to ineligibility to IFN therapy (8/24) or IFN treatment interruption for severe side effects/treatment failure (16/24)13 (see online supplementary table S3A, B; hereafter referred to as the chronic hepatitis C (CHC) group). The other three belong to a second group of 10 patients who underwent RBV monotherapy following HCV recurrence after orthotopic liver transplantation (OLT) (see online supplementary table S3A, C; hereafter referred to as the OLT group). In the CHC group, the preTT biopsies taken on IFN-treated patients were performed at least 6 months after the last IFN administration and all patients underwent a second liver biopsy 6 months to 1 year after the beginning of the RBV monotherapy (onTT biopsies). In the OLT group, biopsies were available at the time of HCV recurrence (preTT biopsies) and the onTT biopsies were taken 3–6 months after beginning of RBV monotherapy. Response to RBV treatment was evaluated as amelioration of liver damage: patients were classified according to ALT decrease13 into ALT-normalisers (ALT-N; normalisation of ALT levels) and ALT non-normalisers (ALT-NN). The six liver specimens analysed by micro-ChIP were derived from three ALT-N and three ALT-NN patients. In the three ALT-N patients, RBV treatment diminished H3K4me3 and augmented H3K9me2/3 and G9a recruitment at ISG promoters, therefore switching the ‘epigenetic balance’ towards a transcriptional repressive context (figure 4, and see online supplementary figure S5 for control gene promoters). Conversely, in the ALT-NN patients, we failed to find a significant decrease in H3K4me3, and the H3K9me2/3 mark was overall far less increased as compared with the ALT-N patients (figure 4 and see online supplementary figure S5), thus suggesting that there is a good correlation between the lack of biochemical response to RBV in ALT-NN patients and the failure to activate this epigenetic shift.

Figure 4

Ribavirin (RBV)-induced epigenetic changes at interferon-stimulated gene (ISG) promoters in patients infected with HCV. Micro-chromatin immunoprecipitation (μChIP) was performed with specific antibodies against G9a, H3K4me3, H3K9me2 and H3K9me3 on liver biopsies from RBV-monotherapy patients. Signal enrichment is expressed as % of input chromatin. Control of antibodies background binding levels on GAPDH promoter (for G9a and H3K9me2/3) and on DHFR 3′ untranslated region (UTR) (for H3K4me3) is shown in online supplementary figure S5. ALT-N, normalisation of ALT levels (ie, complete biochemical responders); ALT-NN, ALT non-normalisers (ie, partial biochemical responders or non-responders to RBV treatment); onTT, on treatment; preTT, pretreatment. Patients ALT-N6, ALT-N8 and ALT-NN1 belong to the CHC group, while patients ALT-N13, ALT-NN6 and ALT-NN7 belong to the orthotopic liver transplantation group. ChIP experiments from each biopsy have been repeated twice. Graphs represent mean values±SD.

RBV-triggered ISG downregulation correlates with liver biochemical amelioration in patients infected with HCV responders to RBV monotherapy

We also took advantage of the 34 aforementioned RBV mono-treated patients (see online supplementary table S3) to study the effects of RBV on hepatic ISG expression in vivo. The expression profiling of the 95 ISGs contained in our customised Taqman low density arrays (see online supplementary table S1) showed that the modulation of gene expression (onTT vs preTT) by RBV differed according to patients’ responses. In ALT-N patients, many ISGs were strongly downregulated, while in ALT-NN patients, the same ISGs were either not affected or induced by RBV treatment (figure 5A, B; see online supplementary table S4). Similar results were obtained with CHC (see online supplementary figure S6A, B) and OLT (see online supplementary figure S6D, E) patient groups when analysed separately. Importantly, many ISGs whose overexpression is associated with non-response to PEG-IFNα/RBV treatment and found to be repressed by RBV in non-transformed liver cells (figure 1A, lane 1 in upper and lower panels and figure 1B), were also strongly repressed in RBV-treated patients and in particular in ALT-N patients. Figure 5C and online supplementary figure S6C, S6F show the expression levels of RSAD2, IFI27 and CXCL10 ISGs, which are among the strongest predictors of response to PEG-IFN and RBV combination therapy,42 in all patients, as well as in CHC and OLT patients, respectively. Their basal expression level was higher in patients infected with HCV compared with non-infected controls. RBV decreased their expression level in RBV ALT-N patients, whereas in ALT-NN patients their negative modulation was absent or much weaker. RBV effect on ISGs was independent of HCV viral load, intrahepatic HCV-RNA levels and was not influenced by IL28B/IFNL3 polymorphisms, an important factor impacting on the outcome of HCV therapy,42 or the response to previous IFN treatments (see online supplementary table S3). Importantly, we confirmed that not all ISGs are downregulated by RBV in vivo and IL8 exemplifies ISGs that are activated by RBV in patients treated with RBV monotherapy (figure 5C and see online supplementary table S4).

Figure 5

Ribavirin (RBV) modulates interferon-stimulated gene (ISG) expression in vivo, in patients infected with HCV. (A) Heat map displaying gene expression ratios during/before RBV in 34 patients with HCV (24 patients with CHC and 10 patients with orthotopic liver transplantation, see online supplementary table S3) treated with RBV monotherapy. Green squares indicate a ratio <1 (downregulation), red squares indicate a ratio >1 (activation). (B) Dot plot showing the gene expression ratio (onTT/preTT) in patients grouped according to response to RBV treatment (C) Box plots of selected ISG mRNA expression in matched liver biopsies of patients undergoing RBV monotherapy (preTT, before RBV therapy; onTT, during therapy; 6–12 months of treatment). CTL, biopsies derived from non-infected patients without liver disease; ALT-N, normalisation of ALT levels (ie, complete biochemical responders); ALT-NN, ALT non-normalisers (ie, partial biochemical responders or non-responders to RBV treatment); onTT, on treatment; preTT, pretreatment. Two-tailed p values were calculated for a risk threshold=0.05 using the Wilcoxon's test for matched samples. Relative RNA levels were normalised over rRNA-18s, RPLP0 and β-ACTIN housekeeping gene expression.

Discussion

RBV is a guanosine analogue exhibiting a broad spectrum of antiviral activities against a variety of viruses, including HCV, hepatitis E virus (HEV),43 respiratory syncytial virus (RSV)44 and haemorrhagic fever viruses.45–48 The clinical importance of RBV stems from its long-term and successful use in combination with PEG-IFNα to achieve sustained virological response in patients infected with HCV.49 ,50 Recently, a number of phase 3 clinical studies have shown that the efficacy of some IFN-free combinations of NS3/4A protease, NS5A and NS5B polymerase inhibitors can be significantly improved by RBV, in particular in difficult-to-treat patient populations: (A) regimens containing non-nucleoside inhibitors of the NS5B polymerase which have a somewhat lower barrier to resistance;16 ,17 (B) patients with cirrhosis infected with HCV genotype 3 receiving the combination of sofosbuvir (anti-NS5B nucleotide analogue) and ledipasvir (second generation NS5A inhibitor);51 (C) and HCV genotype 1b patients with decompensated cirrhosis.18 ,19 The recent european association for the study of the liver (EASL) clinical practice guidelines stated that, in IFN-free regimens, RBV should be considered in patients with predictor of poor response to anti-HCV therapy, especially prior non-responders and/or patients with cirrhosis, either as a first-choice option or as a facultative addition allowing significant reduction of treatment duration (http://www.easl.eu/research/our-contributions/clinical-practice-guidelines/detail/recommendations-on-treatment-of-hepatitis-c-2015). Thus, a full understanding of RBV mechanism of action, which has not yet been completely elucidated, is instrumental for a full exploitation of RBV properties even in the era of IFN-free DAA regimens.

Our results demonstrate that the main effect of RBV monotherapy in infected and non-infected PHHs and in non-transformed hepatic cell lines (HepaRG cells) in vitro is to repress ISGs (85% of analysed genes) rather than activate them. This ‘repressive’ effect of RBV has likely been overlooked until now, because in the more widely used transformed hepatoma cells lines, including Huh7 cells, RBV does not inhibit any of the ISGs we analysed. The spectrum of ISGs included in our analyses was enriched for ISGs whose overexpression has been associated with non-response to PEG-IFNα/RBV treatment.25–27 These genes are abnormally upregulated in patients non-responding to therapy and are refractory to further activation by IFN. Our results suggest that RBV ability to repress those ISGs might be crucial to re-establish responsiveness to IFN and immunomodulation within the intrahepatic environment. Interestingly, we previously demonstrated that the ISGs belonging to this category share a common chromatin organisation pattern, which suggests a shared mechanism of IFN regulation.29 The hypothesis that epigenetic modulation could play a role in RBV-triggered repression of these targets has been confirmed by the observation that functional inactivation of histone methyl transferase (HMT) G9a completely impairs RBV-induced downregulation. Moreover, the loss of its repressive activity prevents RBV to potentiate the levels of IFN activation on ISGs following the triple treatment with a specific G9a inhibitor, suggesting that RBV-triggered repression requires G9a activity, which is in turn instrumental to IFN and RBV combined action. Notably, G9a, which is the HMT mainly responsible for H3K9me2/3, has been recently shown to play a key role in the regulation of innate immunity genes, as the dominant mechanism specifically involved in the inhibition of ISG expression.38 Moreover, H3K9me2/3 has been proposed to be essential in reforming positioned nucleosomes over the transcribed regions after the transit of RNA polymerase to finely tune internal transcriptional initiation.35 Accordingly, RBV might exploit H3K9me2/3 regulation to reset ISG promoters to a basal chromatin status which is ‘ready to be activated’ by IFN stimulation (figure 6). In agreement with this hypothesis, ChIP analysis of selected ISGs in RBV-treated PHH indicates that RBV induces G9a recruitment and H3K9 methylation at their promoters, independently from the activation of STATs signalling.

Figure 6

Schematic representation of ribavirin (RBV) mode of action at the chromatin level. RBV-triggered repression of preactivated interferon-stimulated genes (ISGs) involves chromatin remodelling and the modulation of histone marks. In a setting of actively transcribed ISG promoters following HCV infection and type I/III interferon (IFN) induction, RBV favours the repositioning of nucleosomes over the transcribed regions by increasing G9a-mediated H3K9me2/3 over H3K4me3 levels in order to reset ISG promoters to a basal chromatin status which is again ‘ready to be activated’ by IFN stimulation.

The comparison of the epigenetic profiles of ISG promoters in paired biopsies derived from patients infected with HCV, collected before and during RBV monotherapy, further demonstrates that RBV is able to modify histone methylation in vivo. In particular, the RBV-triggered switch of balance between ‘activating’ H3K4me3 and ‘repressive’ H3K9me2/3 marks seems to be associated to a better response of patients to RBV regimen. Gene expression profiling of the same liver samples and a wider cohort of patients infected with HCV receiving RBV monotherapy indicates that ISG repression following RBV therapy is strongly associated to response to therapy, evaluated as a significant reduction of ALT levels and liver histological scores, mirroring reduction of hepatic inflammation and damage. Notably, a biochemical response during RBV monotherapy lead-in has been correlated by Rotman et al with a better response to IFN and RBV combination therapy.24 Since our patients were not eligible for IFN therapy, we could not investigate the relation between RBV-induced epigenetic changes in the liver, the intrahepatic ISG repression and the clinical response to IFN/RBV combination. The decline in serum IP-10/CXCL10 levels observed by Rotman et al in patients with biochemical response to RBV and in virological responders to IFN/RBV together with the RBV-associated decrease of CXCL10 intrahepatic RNA levels in our ALT-N patients support the notion that RBV activity is mediated by a resetting of IFN responsiveness in the liver. More recently, Meissner et al52 have shown that HCV clearance was accompanied, in patients treated with the potent anti-NS5B agent sofosbuvir (SOF) and RBV, by a decreased expression of selected ISGs and antigen presentation pathways. The authors hypothesise that the strong decrease in viral load may explain the reversal of HCV-triggered ISG activation. Along the same line of reasoning, the ISG downregulation by RBV monotherapy might simply reflect the ability of RBV to reduce HCV replication and, therefore, blunt HCV-induced activation of the same ISGs. However, in their study it is not possible to discriminate between the effect of SOF and that of RBV on ISG expression. Moreover, RBV has a relatively low-level antiviral activity (half-log reduction of HCV viraemia) in patients infected with HCV,3–5 much lower than what was expected with SOF or SOF+RBV,15 and we confirmed that RBV monotherapy had a very limited effect, if any, on intrahepatic HCV RNA levels in our patients (see online supplementary table S3). Therefore, the two mechanisms (ie, the reversal of HCV induced ISG activation secondary to a rapid reduction of HCV replication and the direct epigenetic effect of RBV on ISG promoters) are not mutually exclusive and they may contribute to the resetting of intrahepatic IFN responsiveness and HCV clearance.

We do not know whether the novel mechanism of action of RBV described in this manuscript might be relevant to explain the well known antiviral activity of RBV against HEV or RSV.43 ,44 Indeed, the knowledge on the involvement of IFN signalling as an antiviral response and/or as part of viral pathogenicity is not comparable in the three viral infections. In patients with HEV RBV monotherapy is associated with a rapid HEV clearance in solid organ transplant recipients43 and there are several significant differences between the HCV and HEV pathophysiology. RSV does not cause a chronic infection and, due to the type of patients affected, there are no studies addressing the interplay between the RSV and ISG modulation in vivo. Thus, RBV-mediated decrease of ISGs might help to re-establish a ‘chromatin’ environment favourable for further IFN activation in HEV or RSV infections, in addition to other known or yet uncovered RBV-driven mechanisms associated to viral clearance.53 ,54

In conclusion, our results unravel a unique mechanism to explain how RBV, a nucleoside analogue, enhances IFNα effect in patients infected with HCV by an unexpected epigenetic remodelling at ISG promoter regions. Converging evidence from the study of non-transformed cultured hepatocytes and liver samples from chronically infected individuals indicates that RBV resets the abnormal preactivation and IFN-unresponsiveness of specific ISGs in the liver of patients infected with HCV by directing epigenetic changes that restore a chromatin environment favourable to be newly activated by exogenous IFNα (figure 6). In the perspective of IFN-free regimens as preferred strategy to cure most chronically infected patients,55 it will be of great interest to evaluate whether the epigenetic reprogramming properties of RBV described in this paper will enhance the capability of the new anti-HCV DAAs to restore innate immunity translating in higher sustained virological response rates especially in difficult-to-treat patients.

Acknowledgments

The authors thank Prof M Rivoire and his staff at the Centre Léon Bérard (CLB) for access to liver resections. The authors also thank P Berthillon, F Berby and I Bordes for their help in patients’ specimens and clinical data retrieval and the Taqman low density array platform from UMS3444 for technical assistance.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors BT planned the study, performed experiments, analysed data, wrote the manuscript; DD performed experiments, wrote the manuscript; FL collected patients’ data; JF and FH performed experiments; FN and MFD collected patients’ samples and data; ML supervised the study, critically revised and approved the manuscript; FZ: planned and supervised the study, critically revised and approved the manuscript.

  • Funding This work was supported by grants from ANRS (French national agency for research on AIDS and viral hepatitis to DD, ML and FZ), FINOVI (Foundation for innovation in infectiology; project call n°#4 to FZ and DD, FRM (Foundation for medical research; DEQ20110421327 to FZ and DD), and by INSERM core grants. FZ and DD were also supported by the DevWeCan LABEX (ANR-10-LABX-0061) of Université de Lyon, within the programme ‘Investissements d’Avenir’ (ANR-11-IDEX-0007) operated by the French National Research Agency (ANR) and by the Institut Hospitalo-Universitarire (IHU) OPeRa.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Competent institutional ethical committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.