S-adenosylmethionine and betaine correct hepatitis C virus induced inhibition of interferon signaling in vitro

Authors

  • Francois H. T. Duong,

    1. Department of Research and Division of Gastroenterology and Hepatology, University Hospital Basel, Switzerland
    Search for more papers by this author
  • Verena Christen,

    1. Department of Research and Division of Gastroenterology and Hepatology, University Hospital Basel, Switzerland
    Search for more papers by this author
  • Magdalena Filipowicz,

    1. Department of Research and Division of Gastroenterology and Hepatology, University Hospital Basel, Switzerland
    Search for more papers by this author
  • Markus H. Heim

    Corresponding author
    1. Department of Research and Division of Gastroenterology and Hepatology, University Hospital Basel, Switzerland
    • Division of Gastroenterology and Hepatology, University Hospital Basel, Petersgraben 4, 4031 Basel, Switzerland
    Search for more papers by this author
    • fax: (41) 61-265-53-52


  • Potential conflict of interest: Nothing to report.

Abstract

Hepatitis C virus (HCV) infection is an important cause of chronic liver disease. Standard therapy, pegylated interferon α (pegIFNα) combined with ribavirin, results in a sustained response rate in approximately half of patients. The cause of treatment failure in the other half of the patients is unknown, but viral interference with IFNα signal transduction through the Jak-STAT pathway might be an important factor. We have shown previously that the expression of HCV proteins leads to an impairment of Jak-STAT signaling because of an inhibition of STAT1 methylation. Unmethylated STAT1 is less active because it can be bound and inactivated by its inhibitor, protein inhibitor of activated STAT1 (PIAS1). We show that treating cells with S-adenosyl-L-methionine (AdoMet) and betaine could restore STAT1 methylation and improve IFNα signaling. Furthermore, the antiviral effect of IFNα in cell culture could be significantly enhanced by the addition of AdoMet and betaine. In conclusion, we propose that the addition of these drugs to the standard therapy of patients with chronic hepatitis C could overcome treatment resistance. (HEPATOLOGY 2006;43:796–806.)

Hepatitis C virus (HCV) infection is a major cause of chronic liver disease worldwide.1 Chronic hepatitis C (CHC) may lead to cirrhosis and hepatocellular carcinoma. Type I interferons (IFNs) are crucial and potent components of the early host response against virus infection2 and recombinant pegylated IFNα2a and IFNα2b are widely used for the treatment of CHC and chronic hepatitis B. Current standard treatment with pegylated IFNα and ribavirin can cure about 50% of patients with CHC.3, 4 The cause of treatment failures in half of the patients is not fully understood, but viral interference with IFNα signal transduction from the cell surface to the nucleus may be an important factor. The most important signal transduction pathway for IFNα is the Jak-STAT pathway (Fig. 1A).5 Signal transducers and activators of transcription (STAT) proteins are activated by members of the Jak kinase family through the phosphorylation of a single tyrosine residue.6 Activated STATs form dimers, translocate into the nucleus and bind specific DNA elements in the promoters of target genes.7, 8 STATs are deactivated by tyrosine dephosphorylation in the nucleus, followed by the decay of dimers and the nuclear export of STATs.9, 10 Important negative regulators of this signal transduction pathway have been found at two levels. First, the suppressor of cytokine signaling (SOCS) family members, SOCS1 and SOCS3, prevent phosphorylation and activation of IFNα induced STATs by inhibiting the IFNα receptor associated Jak kinases.11 Second, downstream of STAT activation by tyrosine phosphorylation, IFNα induced gene transcription can be inhibited by protein inhibitor of activated STAT1 (PIAS1). PIAS1 inhibits binding of STAT1 dimers to the response elements in the promoters of target genes (Fig. 1B).12, 13 The binding of PIAS1 to STAT1 is regulated by methylation of STAT1 by protein arginine methyl transferase 1 (PRMT1).14 Arginine methylation inhibits binding of PIAS1 to STAT1, whereas demethylation of STAT1 enhances its association with PIAS1.

Figure 1.

IFNα signaling. Binding of IFNα to its receptor on the cell surface induces the activation of the phosphotyrosine kinases Jak1 and Tyk2. STAT1 is recruited to the receptor kinase complex and activated by phosphorylation on tyrosine 701. STAT1 then forms dimers, translocates into the nucleus, and binds to promoter elements of IFNα stimulated genes (ISGs).The protein inhibitor of activated STAT1 (PIAS1) can not bind to STAT1 dimers because the STATs are methylated on an arginine. (A) Unmethylated STAT1 molecules can also be recruited to the receptor kinase complex, and are phosphorylated normally. However, PIAS1 can bind to STAT1 dimers and lower their binding affinity especially for weak STAT1 binding sites in ISG promoters. A number of ISGs are not upregulated when STAT1 is bound by PIAS1.1. (B) Working model of HCV interference with IFNα signaling. HCV infection induces the overexpression of protein phosphatase 2A catalytic subunit (PP2Ac). PP2Ac inhibits protein arginine methyltransferase 1 (PRMT1). As a consquence, STAT1 methylation is reduced. (C) Adenosylmethionine is the methyl group donor for STAT1 methylation catalyzed by PRMT1.

We have shown previously that the expression of HCV proteins in human osteosarcoma cell lines or in liver cells of transgenic mice inhibits IFNα induced signaling through the Jak-STAT pathway.15, 16 HCV induced a hypomethylation of STAT1 that resulted in an increased binding of STAT1 to PIAS1.17 HCV proteins also induced the expression of the catalytic subunit of protein phosphatase 2A (PP2Ac), and PP2Ac overexpression was found to be sufficient to induce STAT1 hypomethylation.17 Figure 1C shows our current model of HCV interference with IFNα signaling. Based on these results, we hypothesized that pharmacological treatment that increases the methylation of STAT1 could improve IFN signaling. Interestingly, the methyl group donor for STAT1 methylation by PRMT1 is S-adenosy-L-methionine (AdoMet or SAMe) (Fig. 1D), a compound that has been used to treat alcoholic liver disease and is available in many countries as a nonprescription drug.

To explore the potential of AdoMet for improving the efficacy of peg IFNα in the treatment of patients with CHC, we wanted to test if (1) PP2Ac is overexpressed in liver biopsies of patients with chronic hepatitis, (2) PP2Ac inhibits PRMT1 directly, (3) STAT1 methylation can be induced by enhancing the intracellular concentration of AdoMet, and (4) such a treatment improves IFNα−induced signaling and increases the antiviral effect of IFNα in cell culture.

Abbreviations

HCV, hepatitis C virus; CHC, chronic hepatitis C; IFN, interferon; STAT, signal transducers and activators of transcription; SOCS, suppressor of cytokine signaling; PIAS1, protein inhibitor of activated STAT1; PRMT1, protein arginine methyltransferase 1; PP2Ac, catalytic subunit of protein phosphatase 2a; AdoMet, S-adenosyl-L-methionine; AdOx, adenosine dialdehyde; CT, threshold cycle; GST, glutathione S-transferase; AdoHcy, S-adenosyl-L-homocysteine; ISG, IFN stimulated genes; VSV, vesicular stomatitis virus; GSH, glutathione.

Materials and Methods

Reagents, Antibodies, Cells, and Plasmids.

Human IFNα (hIFN; Roferon) was obtained from Hoffmann LaRoche (Basel, Switzerland). Purified PP2A and purified histone H4 were purchased from Upstate (LucernaChem, Lucerne, Switzerland). Purified Bcl2 was obtained from SantaCruz (LabForce AG, Nunningen, Switzerland). S-Adenosyl-L-Methionine (AdoMet), betaine and adenosine dialdehyde (AdOx) were from Sigma (Fluka Chemie GmbH, Buchs, Switzerland). 14C-AdoMet was obtained from Amersham Biosciences (Amersham Pharmacia Biotech Europe GmbH, Dübendorf, Switzerland). PIAS1 antibodies were a gift from Dr. K. Shuai. pGEX-HRMT1L2 (human PRMT1) expression vector was a gift from Dr. P. A. Silver. UHCV57.3 cells were from Dr. D. Moradpour. UHCV57.3 cells are tetracycline-regulated cell lines containing the entire HCV open reading frame derived from a HCV H consensus cDNA.18 Huh7 harboring the HCV replicon I377/NS3-3′(clone 9-13) were from Dr. R. Bartenschlager.19 HA-PP2Ac cells were previously described.17

Patients and Biopsies.

From August, 2002, to April, 2005, all patients with CHC referred to the outpatient liver clinic of the University Hospital Basel and who had a liver biopsy were asked for their permission to use part of the biopsy for this study. The protocol was approved by the ethical commission of Basel. Written informed consent was obtained from all patients that agreed to participate in the study. Sample handling and extraction procedures have been previously described.17 Control samples were from patients who underwent ultrasound-guided liver biopsies of focal lesions (mostly metastases of carcinomas) and who were asked for their permission to obtain a biopsy from the normal liver tissue outside the focal lesion. Only samples with histologically confirmed absence of liver disease were used as controls. The relevant clinical and histological data of the HCV patients are shown in Supplementary Table 1. (Supplementary material can be found on the HEPATOLOGY website: http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html.)

Expression and Purification GST-PRMT1.

Bacterial cells were grown overnight in Luria-Bertani medium supplemented with 100 μg/mL ampicillin until OD 0.6-0.7. Glutathione S-transferase (GST)-PRMT1 expression was induced with 100 mmol/L IPTG for 3 hours at 37°C. After cell lysis the protein was purified using GST microSpin column (Amersham) according to manufacturer's instructions.

Preparation of Extracts From Cells and Liver Biopsies.

Whole cell lysates and nuclear extracts were done as described.17 The liver biopsies were homogenized in 100 μL of lysis buffer (100 mmol/L NaCl, 50 mM Tris pH 7.5, 1 mmol/L EDTA, 0.1 % Triton X-100, 10 mmol/L NaF, 1 mmol/L PMSF, and 1 mmol/L sodium orthovanadate), and the lysates were then centrifuged at 14,000 rpm for 5 minutes. Protein concentrations were determined with the BioRad Protein Assay (Bio-Rad Laboratories AG, Reinach, Switzerland).

Immunoprecipitation and Immunoblotting.

Immunoprecipitation and immunoblotting was done as described.17 To measure PP2Ac expression in human liver biopsies, 20, 50, and 100 ng of purified PP2Ac was loaded on each gel. These 3 samples allowed the calculation of a standard curve for each gel (Fig. 1A). The intensity of each band was analyzed by densitometry analysis using NIH Image software. The amount of PP2Ac in each liver biopsy was then determined according to the standard curve.

Electrophoretic Mobility Shift Assays.

1 μg nuclear extracts aliquots were used for EMSAs. SIE-m67 was used as the oligonucleotide probe.15 STAT1 was supershifted with antibody SC346 from Santa Cruz (LabForce AG, Nunningen, Switzerland).

RNA Isolation, Reverse Transcription, and SYBR-PCR.

Total RNA was isolated from the cells using Perfect RNA Eukaryotic Mini Kit (Eppendorf, Vaudaux-Eppendorf, Basel, Switzerland) according to manufacturer's instructions. RNA was reverse transcribed by M-MLV reverse transcriptase (Promega, Promega Biosciences Inc., Wallisellen, Switzerland) in the presence of random hexamers (Promega) and dNTPs. The reaction mixture was incubated for 5 minutes at 70°C and then for 1 hour at 37°C. The reaction was then stopped by heating at 95°C for 5 minutes. SYBR-PCR was performed based on SYBR-Green Fluorescence (SYBR-Green PCR Master Mix, Applied Biosystems, Foster City, CA). To prevent influence from genomic DNA amplification, the primers were designed across exon-intron junctions. The primers for GAPDH were 5′-GCTCCTCCTGTTCGACAGTCA-3′ and 5′-ACCTTCCCCATGGTGTCTGA-3′. The primers for tubulin were 5′-GCCAGTGCGGGAACCA-3′ and 5′-GGTCGATGCCGTGCTCAT-3′. The primers for IP10 were 5′-CGATTCTGATTTGCTGCCTTATC-3′ and 5′-GCAGGTACAGCGTACGGTTCT-3′. The primers for HCV were 5′-CACCCCTGCTCCATAACC-3′ and 5′-CGCTGCTTCTGCTTTCG-3′. The ΔCT value was derived by subtracting the threshold cycle (CT) value for GAPDH or tubulin, which served as an internal control, from the CT values for HCV or IP10, respectively. All reactions were run in duplicate using the ABI 7000 Sequence Detection System (Applied Biosystems). mRNA expression level of IP10 or HCV was expressed as a fold increase or fold decrease according to the formula 2ΔCT(PBS)-δCT(Treatment).

Methylation Assay.

In vitro methylation assay was performed according to a protocol previously described14 with some modifications. Briefly, 4 μg of purified histone H4 was incubated in the presence or in the absence of 5 μg of GST-PRMT1 and 10 μL of 14C-AdoMet in a final reaction volume of 50 μg for 2 hours at 37°C. For the inhibition of GST-PRMT1 activity, 14 μg of purified PP2A was incubated with GST-PRMT1 for 40 minutes at 37°C prior to add to H4 and 14C-AdoMet. Fourteen micrograms of purified Bcl2 was used as a negative control. The reaction was then stopped by adding 20 μL of a sample loading buffer, boiled for 5 minutes, and then separated on a 8 % SDS-polyacrylamide gel. The gel was dried and then exposed to PhosphorImager (Kodak) for 2 days.

Vesicular Stomatitis Virus Infection and Cell Viability.

UHCV57.3 cells (30,000 cells) were cultured in 96 well plates for 24 hours in the absence of tetracycline to induce the expression of HCV proteins. During the last 3 hours, AdoMet (170 nmol/L) and betaine (50 μmol/L) were added. They were then infected with 10 pfu of vesicular stomatitis virus (VSV). From the time point of the infection cells were immediately treated with hIFNα (1000 U/mL) alone or in combination with 170 nmol/L AdoMet and 50 μmol/L betaine for 24 hours. 15 μL of the supernatant were then used to infect Vero cells (30,000 cells, 96 well plates) for 24 hours. Vero cell viability was determined using CellTiter 96 Aqueous non-radioactive cell proliferation assay (Promega) according to manufacturer's instructions. All samples were done in duplicate.

Plaque Assay.

UHCV57.3 cells were cultured in 150 mm plate for 7 hours in the absence of tetracycline to induce the expression of HCV proteins (Supplementary Fig. 2). Cells were trypsinized and seeded in 24 well plates (360,000 cells/well) in the absence of tetracycline for another 17 hours before pretreated with 170 nmol/L AdoMet and 50 μmol/L Betaine for 3 hours. They were then infected with 1 pfu of VSV. From the time point of infection, cells were treated with hIFNα (10, 50, 100, or 1,000U/mL) alone or in combination with 170 nmol/L AdoMet and 50 μmol/L Betaine for 24 hours. Supernatants were collected and used (1:200) to infect Vero cells (monolayer, 6 well plates). At 1 hour post infection, supernatant was removed and 3% methylcellulose was overlayed. At 20 hours post infection, overlay was removed, cells were fixed with 4% formaldehyde for 30 minutes, and stained with 0.2% crystal violet in 20% methanol. Plaques were counted and multiplied by the dilution factor to determine viral titer as pfu/mL.

Figure 2.

Expression of protein phosphatase 2A catalytic subunit (PP2Ac) and STAT1 in human liver biopsies of patients with chronic hepatitis C. (A) Representative example of a Western blot showing the samples with 20ng, 50ng and 100ng of purified PP2Ac used to generate a calibration curve and one sample of a patient with chronic hepatitis C and a control sample from a patient without parenchymal liver disease. (B) Example of a calibration curve. The Western blot signal intensities were measured using NIH Image Software. (C) Box-plot diagram of quantified PP2Ac expression levels in liver biopsies of 96 patients with chronic hepatitis C and from 25 control patients without parenchymal liver disease. The difference in the expression of PP2Ac between HCV and control groups is statistically significant (P= .0033, Mann-Whitney U test). (D) PP2Ac expression is higher in samples from patients infected HCV genotype 1 compared with genotype 3. The difference is statistically significant (PI =.0269, Mann-Whitney U test). (E) Semi-quantitative assessment of STAT1 expression. The STAT1 signal intensity in Western Blots was quantified with NIH Image Software. The results (arbitrary units) are shown in a box plot diagram. The difference between controls and HCV patients is statistically significant (P = .0465, Mann-Whitney U test). (F) Semiquantitative assessment of STAT1 tyrosine phosphorylation. The PY-STAT1 signal intensity (arbitrary units) and STAT1 expression levels (arbitrary units) in Western Blots was quantified with NIH Image Software. For each liver biopsy sample, the ratio of phosphorylated STAT1 to total STAT1 was calculated. The results are shown in a box plot diagram. The difference between controls and HCV patients is not statistically significant (P = .67, Mann-Whitney U test).

Results

Expression of PP2Ac in Human Liver Biopsies From Patients With CHC.

Our working model (Fig. 1) of HCV interference with IFNα signaling was mainly based on studies with cultured cells and transgenic mice. We therefore aimed to investigate the relevance of this model for patients with CHC. We have previously published preliminary data showing that the expression level of PP2Ac is higher in liver biopsy samples from patients with CHC compared with controls.17 To confirm this observation, we developed a semi-quantitative Western blot method (Fig. 2A-B) and measured PP2Ac expression in a large number of new biopsy samples (Fig. 2C). The median concentration of PP2Ac was 6.5 ng/μg total protein in 96 samples from patients with CHC, and 4.5 ng/μg total protein in 25 control samples (Fig. 2C). The difference was statistically significant with a P value of .0033 (Mann-Whitney U test). Of note, the expression levels of PP2Ac in CHC samples varied considerably. We analyzed if age, sex, viral load, histological grade of inflammation, or stage of fibrosis were correlated with the expression levels of PP2Ac, but could not find any statistically significant correlation (Supplementary Table 1 and Supplementary Fig. 1). Only the presence or absence of HCV infection and the HCV genotype were significantly correlated with PP2Ac expression.

An increased expression level of PP2Ac is important in the context of CHC because PP2Ac over-expression in cultured cells inhibits IFNα signaling (Fig. 1).17 Based on these observations, we propose that patients with a low or normal PP2Ac expression level would respond better to IFNα-based therapies than patients with high PP2Ac expression levels. An ongoing study will directly test this hypothesis. Indirect evidence for this hypothesis can be obtained by comparing PP2Ac expression levels in genotype 1 versus genotype 3 infected patients. Clinical studies have consistently shown that the patients infected with genotype 1 have a lower sustained response rate to treatments with (pegylated) IFNα plus ribavirin than patients infected with genotype 2 or 3.3, 4, 20 In our group of patients, we found significantly higher expression levels of PP2Ac in 50 samples from patients infected with HCV genotype 1 (median = 7.8 ng PP2Ac/μg total protein) compared with 22 samples from patients infected with genotype 3 (median = 5.5 ng PP2Ac/μg total protein; P = .027), further supporting an important role of PP2A in HCV induced IFNα resistance (Fig. 2D).

We also determined the expression level of STAT1 in 96 biopsies from HCV patients and the phosphorylation of STAT1 in 83 HCV samples and 25 controls. STAT1 expression was increased in HCV samples (Fig 2E), and the phosphorylation of STAT1 was intact (Fig. 2F).

PP2Ac Directly Binds to and Inhibits PRMT1.

These results are consistent with our previous work where we showed the block in IFNα signaling is not caused by a degradation of STAT1 or by an inhibition of STAT1 phosphorylation, but is further downstream in the signaling pathway, at the level of STAT1-PIAS1 association (Fig. 1). The reversible association of STAT1 with PIAS1 is regulated by STAT1 methylation.14 We have previously shown that HCV protein expression inhibited the methylation of STAT1 and thereby increased the binding of PIAS1 to STAT1.17 Because STAT1 methylation is catalyzed by PRMT1, we wanted to test if PP2Ac could directly inhibit the enzymatic activity of PRMT1. To this end, we expressed and purified PRMT1 and PP2Ac and used the purified proteins in an in vitro methylation assay. As shown in Fig. 3A, the methylation of histone H4 (a standard substrate for this assay) by PRMT1 is completely inhibited by PP2Ac. Based on these in vitro data we conclude that the upregulation of PP2Ac by HCV leads to an inhibition of PRMT1 (Fig. 1C). As a consequence, the methylation of many cellular proteins including STAT1 is reduced.

Figure 3.

The inhibition of protein arginine methyltransferase 1 (PRMT1) by protein phosphatase 2A catalytic subunit (PP2Ac) can be overcome by AdoMet and betaine. (A) PP2Ac directly interacts with PRMT1 and inhibits its methyltransferase activity. Purified histone H4 can be methylated in vitro by purified PRMT1 (GST-PRMT1 fusion protein), lane 2. The addition of purified PP2A completely blocked the enzymatic activity of PRMT1 (lane 3). To control the specificity of the inhibition of PP2A on PRMT1, purified Bcl2 was used and had no effect (lane 4). Lanes 5 and 6 are controls without histone H4. (B) Huh7 cells expressing a constitutive active form of PP2Ac (HA-PP2Ac) have a constant inhibition of PRMT1. In these cells, STAT1 methylation as detected by an IP-Western blot (immunoprecipitation with antibodies to dimethyl-arginine, Western blot with STAT1 antibodies) is diminished (lane 2, upper panel), and binding of PIAS1 to activated STAT1 as detected by a co-IP (immunoprecipitation with STAT1 antibodies, Western blot with PIAS1 antibodies) is enhanced (lane 2, middle panel). A 2-hour pretreatment of HA-PP2Ac cells (lane 3, upper panel) with AdoMet restores STAT1 methylation to normal levels found in Huh7 cells (lane 1, upper panel), and corrects the increased binding of PIAS1 (middle panel). The lower panel shows the loading control. (C) Electrophoretic mobility shift assay (EMSA) using the SIE-m67 oligonucleotide probe. IFNα induced binding of activated STAT1 is impaired in HA-PP2Ac cells compared with the parent cell line Huh7 (lanes 2 and 4). Pretreatment of HA-PP2Ac cells (lanes 5-7) with AdoMet (170 nmol/L, 2 hours) and/or Betaine (50 μmol/L, 2 hours) restores STAT1 shifts to nearly normal levels found in Huh7 cells (lane 2). Lanes 8-10 show controls without IFNα. (D) Improvement of IFNα signaling by AdoMet and betaine in cells expressing HCV proteins. UHCV57.3 cells were cultured in presence of tetracycline (no HCV protein expression, lanes 1 and 2) or without tetracycline for 24 hours (de-repression of HCV protein expression, lanes 3-9). HCV protein expression leads to an inhibition of the STAT1 gel shift (lane 4). A 2-hour pretreatment of de-repressed cells expressing HCV proteins with AdoMet or/and betaine (lanes 5-7) restores STAT1 DNA binding to normal levels found in cells without HCV protein expression (lane 2). Lanes 8-10 show controls. AdoMet, S-adenosyl-L-methionine.

AdoMet and Betaine Increase Methylation of STAT1.

AdoMet is available in many countries as an over-the-counter drug used for the treatment of liver diseases, especially alcoholic and nonalcoholic steatohepatitis. It is used as a precursor for glutathione biosynthesis. AdoMet depletion has been observed in a number of experimental models of liver diseases associated with increased oxidative stress and a depletion of reduced glutathione (GSH).21–24 Oral AdoMet treatment has therefore been used in animal models and in patients to restore GSH content in the liver with the aim to treat liver diseases.25–27 We were interested in AdoMet not because of its role in GSH biosynthesis,28, 29 but because PRMT1 uses AdoMet as the methyl group donor for STAT1 methylation (Fig. 1D). We hypothesized that AdoMet could correct the HCV induced STAT1 hypomethylation by shifting the STAT1 + AdoMet ⇋ Met-STAT1 + S-adenosyl-L-homocysteine reaction equilibrium to the right. To test this hypothesis, we treated HA-PP2Ac cells with AdoMet. HA-PP2Ac cells are stably transfected with a constitutive active catalytic subunit of PP2A, HA-PP2Ac.17 We have shown previously that IFNα induced signaling through the Jak-STAT pathway is inhibited in these cells because of hypomethylation of STAT1 and increased binding of PIAS1 to STAT1.17 Here we show AdoMet could correct the inhibition of IFNα signaling in HA-PP2Ac cells (Fig. 3B). The methylation of STAT1 could be restored to levels found in naïve Huh7 cells, and the increased association of STAT1 with PIAS1 observed in HA-PP2Ac cells was returned to normal levels after treatment with AdoMet (Fig. 3B).

The intracellular AdoMet concentration can also be raised by treating cells with betaine.30–32 Betaine (trimethyl-glycine) is the principle methyl donor for the generation of methionine from homocysteine, a reaction that is central to the recycling of AdoMet (Fig. 4).33 In fact, treatment of HA-PP2Ac cells with betaine could restore STAT1 signaling just as well as AdoMet (Fig. 3C).

Figure 4.

Proposed model for the role of the AdoMet cycle in the chronic HCV infection. Infection by HCV leads to an upregulation of PP2Ac in liver cells. PP2A inhibits PRMT1. PRMT1 catalyzed STAT1 methylation is reduced. Un-methylated STAT1 is a less efficient inducer of IFNα stimulated genes. This deficiency of IFNα signal transduction can be corrected by increasing the intracellular concentration of AdoMet (shifting the equilibrium of PRMT1 catalylized methylation reaction to the right). AdoMet can be increased by adding the substance itself or by adding betaine. Betaine is the methyl group donor for the conversion of L-homocysteine to methionine, a direct precursor of AdoMet., betaine homocysteine methyltransferase. AdoHcy, S-adenosyl-L-homocysteine; AdoMet, S-adenosyl-L-methionine.

We then tested the effect of AdoMet and betaine on IFNα signaling in UHCV57.3 cells. UHCV57.3 cells inducibly express all HCV proteins.34 In these cells, IFNα signaling is impaired by viral protein expression (Fig. 3D, lane 4).15 However, pretreatment of cells with AdoMet and/or betaine restores STAT1 DNA binding even in the presence of viral proteins (Fig. 3D, lanes 5 to 7). To test if the increased DNA binding also improves the transcriptional activation of ISGs, we measured IP-10 mRNA levels after stimulation of cells with IFNα with or without pretreatment with AdoMet and betaine. Both in HA-PP2Ac cells and in UHCV57.3 cells expressing HCV proteins, the induction of IP-10 was impaired. In both cell lines, IP-10 induction could be restored by treating cells with AdoMet and betaine (Fig. 5).

Figure 5.

AdoMet and betaine improve IFNα induced gene expression.Huh7 and HA-PP2Ac cells (A) and UHCV57.3 cells (B) were stimulated with hIFNα (1,000U/mL) for 6 hours with or without a pretreatment with AdoMet (170 nmol/L) and Betaine (50 μmol/L) for 18 hours. The induction of the interferon target gene IP10 was measured with real-time RT-PCR. Samples were done at least in triplicate. Shown are the mean values and the standard errors of the mean. The P values were obtained using the ANOVA test. (A) After treatment with IFNα, the transcriptional induction of IP10 is inhibited in HA-PP2Ac cells (lane 2) compared with the parent cell line Huh7 (lane1). AdoMet and Betaine can restore the response to IFNα even in the presence of PP2Ac over-expression (lane 3). (B) UHCV57.3 cells were cultured either with tetracycline to suppress the expression of HCV proteins (= control, lane 1) or without tetracycline for 24 hours to induce the expression of HCV proteins (lane 2 and 3). Treatment of de-repressed cells with IFNα alone reduced the transcriptional induction of IP10 about 4 fold (lane 2). When cells were pretreated with AdoMet and Betaine, IP10 induction by IFNα was significantly improved (lane 3). AdoMet, S-adenosyl-L-methionine.

AdOx Inhibits STAT1 DNA Binding.

If an increased methylation of STAT1 after treatment with AdoMet and betaine improves IFNα signaling, then one would suppose that the inhibition of STAT1 methylation also inhibits the binding of activated STAT1 in a gel shift assay. We therefore treated Huh7 cells and HA-PP2Ac cells with adenosine dialdehyde (AdOx). AdOx is an inhibitor of S-adenosylhomocysteine hydrolase, and treatment of cells with AdOx leads to a marked intracellular accumulation of S-adenosyl-L-homocysteine (AdoHcy) (Fig. 4).35–38 AdoHcy is a competitive inhibitor of PRMT1 and other methyltransferases. When cells were treated with AdOx, we observed a time-dependent inhibition of STAT1 DNA binding in gel shift assays.(Fig. 6).

Figure 6.

Inhibition of PRMT1 by Adox inhibits IFNα induced Jak-STAT signaling. EMSA with m67 probe showing a time dependent inhibition of the STAT1 gel shift intensity by AdOx, an inhibitor of PRMT1. AdOx, adenosine dialdehyde.

AdoMet and Betaine Increase the Antiviral Effects of IFNα.

We then wanted to test if AdoMet and betaine could improve the antiviral efficiency of IFNα. Huh7 cells harboring a subgenomic HCV replicon are a well-established model for HCV replication.19 In these cells, HCV subgenomic RNA replication can be efficiently inhibited by IFNα in a time- and dose-dependent way.39 Interestingly, we could increase the effect of IFNα about tenfold by adding AdoMet and betaine to IFNα (Fig. 7A). The combination of 100 IU/mL IFNα plus AdoMet and betaine achieved the same reduction in replicon RNA as a monotherapy with a tenfold higher dose (1000 IU/mL) of IFNα. We then confirmed the increase of the antiviral efficacy of IFNα by AdoMet and betaine in a second experimental system (an overview of the experiment is shown in Supplementary Fig. 2). UHCV57.3 cells were first cultured for 24 hours without tetracycline to induce HCV protein expression (and inhibition of IFNα signaling). During the last 3 hours, AdoMet and betaine were added. They were then infected with vesicular stomatitis virus (VSV) and at the same time treated with IFNα alone or in combination with AdoMet and betaine. We hypothesized that the combination treatment would inhibit the replication of VSV in these cells more efficiently than the monotherapy. The supernatant of these cultures was then used to infect Vero cells, and VSV replication was quantified using a cell viability assay. In this assay, cell viability is inversely correlated with the viral titer. Compared with no treatment (Fig. 7B, lane 1), IFNα monotherapy did not significantly improve cell viability (Fig. 7B, lane 2), whereas the combination treatment (Fig. 7B, lane 3) significantly inhibited viral replication in UHCV57.3 cells even in the presence of HCV proteins. Finally, we confirmed these results using a plaque assay (Fig. 7C).

Figure 7.

Impact of AdoMet and betaine on antiviral effects of IFNα. (A) AdoMet and betaine increase the inhibitory effect of IFNα on the replication of a HCV replicon. Huh7 harboring HCV replicon cells were treated for 18 hours with 1000 IU/mL (lanes 1 and 2) or 100 IU/mL IFNα (lanes 3 and 4) without (lanes 1 and 3) or with 170 nmol/L AdoMet and 50 μmol/L betaine (lanes 2 and 4). The amount of replicon RNA was measured by real-time RT-PCR, and calculated as percentage of the baseline value. Mean values from 3 independent experiments are shown. Error bars show standard error of the mean. Of note, the addition of AdoMet and betaine to 100 IU/mL IFNα (lane 2) resulted in the same amount of replicon RNA reduction as a monotherapy with 1000 IU/mL (lane 3). (B) Vero cell viability after infection with supernatant from VSV infected cells. UHCV57.3 cells were first cultured in the absence of tetracycline to induce HCV protein expression. During the last 3 hours, AdoMet (170 nmol/L) and betaine (50 μmol/L) were added. They were then infected with 10pfu VSV and at the same time left untreated (lane 1), treated with 1000 IU/mL IFNα (lane 2), or treated with IFNα and AdoMet (170nmol/L) and betaine (50 μmol/L) (lane 3). After 24 hours incubation, 15 μL of the supernatant were used to infect 30,000 Vero cells. Viability of Vero cells was then tested 24 hours later using a colorimetric method for determining the number of viable cells. Shown is the mean absorbance at 490nm from two duplicate samples (± SEM). Fisher's t test was used for statistical comparison of the groups. P values are indicated. The combination treatment significantly improved cell viability compared with no treatment. (C) Betaine and AdoMet potentiate the antiviral effect of hIFNα. UHCV57.3 cells were cultured in the absence of tetracycline to induce the expression of HCV proteins. They were infected with VSV and treated with different concentrations of hIFNα (10, 50, 100, or 1000U/mL) as indicated, either alone or in combination with 170 nmol/L AdoMet and 50 μmol/LM Betaine for 24 hours. At every concentration of hIFNα tested, the combination treatment inhibited the viral replication more effectively than the monotherapy. The experiments were done in triplicate. Fisher's t test showed statistically significant differences for the samples treated with 10 or with 50 IU/mL hIFNα. AdoMet, S-adenosyl-L-methionine; VSV, vesicular stomatitis virus.

Discussion

Our analysis of 96 biopsy samples from patients with CHC and of 25 control samples provided solid evidence that the expression level of PP2Ac is increased during CHC infection. PP2Ac is an abundant protein that is involved in many signaling pathways, and its expression level and activity are tightly regulated.40, 41 It has been estimated that PP2Ac accounts for 0.3% to 1.0 % of total cellular proteins.42 We found a median increase from 4.5 to 6.5 ng/μg total protein in liver biopsy extracts from patients with CHC compared with controls. Interestingly, the median expression of PP2Ac in a subgroup of patients infected with the difficult-to-treat genotype 1 was even higher (7.8 ng/μg total protein). This further supports our hypothesis that overexpression of PP2Ac is involved in resistance to IFNα treatment. Given the fact that PP2Ac is a very abundant cellular protein, a 50% increase of its expression level is most likely biologically significant, and will have multiple effects on several intracellular signaling pathways.

In the present study, we concentrated our analysis on an aspect of PP2A biology that has not yet been thoroughly studied, i.e., its role in the regulation of PRMT1. We describe here that PP2Ac can directly bind to and inhibit PRMT1 in an in vitro methylation assay with purified proteins. Of note, PRMT1 was involved in many methylation reactions, and an inhibition of its catalytic activity by PP2Ac had profound consequences for many cellular pathways. Most interesting for IFNα signaling is the role of PRMT1 in STAT1 methylation. STAT1 arginine methylation was shown to be important for its reversible association with PIAS1. PIAS1 binds to un-methylated STAT1 and inhibits the DNA binding of STAT1 even after its activation by tyrosine phosphorylation.14 Recently, the role of STAT1 methylation has been disputed in two reports.43, 44 In both reports, no conclusive evidence for STAT1 methylation could be found. In contrast to these reports and in accordance with the report by Mowen et al.,14 we consistently found in our experimental systems that the manipulation of STAT1 methylation had consequences on DNA binding of STAT1 and the induction of IFNα target genes (ISGs).

It has been shown that only a subset of ISGs is regulated by PIAS1-STAT1 interactions.13 One of the better-described PIAS1 dependent ISGs is IP-10, and we therefore analyzed its induction by real-time RT-PCR analysis. We found an inhibition of IP-10 induction in cells overexpressing PP2Ac and in cells that express HCV proteins (Fig. 5). In UHCV57.3 cells that expressed HCV proteins, AdoMet and betaine could restore the normal IP-10 induction by IFNα, and in HA-PP2Ac cells IP-10 induction by IFNα was significantly improved by AdoMet and betaine. Because of the pleiotropic effects of AdoMet and betaine on cells, we can not exclude that methylation of proteins other than STAT1 could be involved in the improvement of IFNα induced target gene induction. However, as shown in Fig. 3, AdoMet and betaine treatment can increase STAT1 methylation, can reduce STAT1-PIAS1 association, and can improve STAT1 DNA binding. We therefore conclude that at least in part, the improved IFNα target gene induction after AdoMet and betaine treatment is mediated by an enhanced DNA binding of STAT1.

We then tested the effect of AdoMet and betaine on the biological activity of IFNα in HCV replicon cells and in UHCV57.3 cells. Strikingly, the combination treatment was about tenfold more potent for inhibiting replication of HCV replicons than the IFNα monotherapy. This effect was consistently found when the replicon RNA levels were analyzed after 18 hours of treatment. Most HCV replicons are very sensitive to IFNα, and longer treatments for 48 hours or 72 hours result in a profound suppression of replicon RNA levels even when low concentrations of IFNα (e.g., 100 IU/mL or 10 IU/mL) are used.39, 45, 46 Because of this high efficacy of IFNα, we found no significant difference when adding AdoMet and betaine in 48-hour or 72-hour treatments (data not shown). We therefore confirmed the effect of AdoMet and betaine in a second system using UHCV57.3 cells infected with VSV (Fig. 7B-C). In these experiments, the addition of AdoMet and betaine also significantly improved the antiviral efficacy of IFNα. The absolute changes in terms of virus titers are not very large (Fig. 7C, logarithmic scale), but often large enough to match the efficacy of the next higher IFNα concentration in this dose response curve. We do not know if the observed improvement of the antiviral efficacy of IFNα (albeit statistically significant) would make a difference in the context of a pegIFNα plus ribavirin treatment of CHC. This will have to be tested in clinical trials.

Over the last years, major progress has been made in the understanding how HCV interferes with the induction of IFNs in infected cells. It has been found that HCV NS3/4A protease inhibits the phosphorylation of IFN regulatory factor 3 (IRF-3), a transcription factor central for the induction of IFNβ.47 Further studies identified RIG-I, a double strand RNA sensor protein, as the sensor of HCV RNA and inducer of IFNβ.47, 48 Recently, the adaptor protein that couples RIG-I to the downstream kinases IKKϵ and TBK1 has been identified and named CARDIF, MAVS, VISA and IPS-1 by four independent groups.49–52 HCV NS3/4 protease can cleave and inactivate Cardif, thereby blocking the induction of IFNβ.52 These findings may explain how HCV can inhibit the activation of the IFN system and establish a persistent infection in a majority of patients. On the other hand, the inhibition of the RIG-I – IRF3 pathway can not readily explain why so many patients do not respond to a therapy with pegIFNα and ribavirin. In this therapeutic setting, the IFN system does not need to be activated through the induction of endogenous IFNβ and IFNαs. The failure of IFN based therapies has to be caused either by a block in the signaling pathway or by blocking of IFN effector systems. We have provided evidence in support of the former, and we show here that at least in cultured cells, the block in IFNα signaling can be alleviated by a co-treatment with AdoMet and betaine.

There are potential clinical implications of these results. Taken together, our results suggest the addition of AdoMet and betaine to the current standard treatment of CHC with pegIFNα and ribavirin could increase the efficacy of the treatment. This hypothesis will have to be tested in clinical trials. In our experiments we used concentrations of AdoMet (170 nmol/L) and betaine (50 μmol/L) that can be obtained in the serum of patients by oral application of these substances in nontoxic doses. For example, after a single oral dose of 400mg AdoMet, peak serum concentrations of 1 to 2 μmol/L are obtained,53 about 6 to 12 times more than the concentrations used in our experiments. In another study, steady state serum concentrations after multiple oral doses (2 × 800 mg/d) over 4 weeks were 790 nmol/L.54 AdoMet and betaine are both nontoxic substances that are available in many countries without prescription, and it might be tempting to add them to the current standard therapy with pegIFNα and ribavirin. However, we would advise to use these substances only in the context of well-designed clinical studies.

Acknowledgements

The authors thank José C. Fernando-Checa for stimulating discussions. We thank Lukas Hunziker for helping us to set up the VSV assay, and Ralf Bartenschlager and Darius Moradpour for helping us with the replicon cell assays.

Ancillary