Supported in part by Merit Review grants from the Veterans Administration (K.E.B.), the National Institutes of Health R21 DK068453-01A1 and VA Merit Review (W.N.S.), RO1 DK058597-4 (B.A.L.), and the University of Iowa Carver Trust Foundation (W.N.S. and A.P.M.), and the American Cancer Society Seed Award (Z.Z.).
Presented in part at the American Association for the Study of Liver Diseases, 50th Annual Meeting, San Francisco, CA, 2008 and 2009, abstracts #1905 and #791, respectively.
Potential conflict of interest: Dr. Schmidt is on the speakers' bureau of Schering and is a consultant for Gilead.
Induction of heme oxygenase-1 (HO-1) inhibits hepatitis C virus (HCV) replication. Of the products of the reaction catalyzed by HO-1, iron has been shown to inhibit HCV ribonucleic acid (RNA) polymerase, but little is known about the antiviral activity of biliverdin (BV). Herein, we report that BV inhibits viral replication and viral protein expression in a dose-dependent manner in replicons and cells harboring the infectious J6/JFH construct. Using the SensoLyte 620 HCV Protease Assay with a wide wavelength excitation/emission (591 nm/622 nm) fluorescence energy transfer peptide, we found that both recombinant and endogenous nonstructural 3/4A (NS3/4A) protease from replicon microsomes are potently inhibited by BV. Of the tetrapyrroles tested, BV was the strongest inhibitor of NS3/4A activity, with a median inhibitory concentration (IC50) of 9 μM, similar to that of the commercial inhibitor, AnaSpec (Fremont, CA) #25346 (IC50 5 μM). Lineweaver-Burk plots indicated mixed competitive and noncompetitive inhibition of the protease by BV. In contrast, the effects of bilirubin (BR) on HCV replication and NS3/4A were much less potent. Because BV is rapidly converted to BR by biliverdin reductase (BVR) intracellularly, the effect of BVR knockdown on BV antiviral activity was assessed. After greater than 80% silencing of BVR, inhibition of viral replication by BV was enhanced. BV also increased the antiviral activity of α-interferon in replicons. Conclusion: BV is a potent inhibitor of HCV NS3/4A protease, which likely contributes to the antiviral activity of HO-1. These findings suggest that BV or its derivatives may be useful in future drug therapies targeting the NS3/4A protease. (HEPATOLOGY 2010;52:1897–1905)
Chronic hepatitis C virus (HCV) infection is an important cause of liver disease worldwide. A significant number of infected patients develop persistent viremia that leads to cirrhosis, end-stage liver disease, and hepatocellular carcinoma.1 Current standard treatment for chronic HCV infection, pegylated α-interferon and ribavirin, achieves viral eradication in only approximately half of patients treated.2 Structurally, the virus has a plus-stranded ribonucleic acid (RNA) genome with a single long open-reading frame containing 5′ and 3′ flanking nontranslated nucleotide regions that are important for translation, replication, and immune recognition.3 The genome contains a serine-activated protease and an RNA-dependent RNA polymerase that are important targets for development of new antiviral drugs. Although anti-protease and anti-polymerase drugs promise to improve treatment outcomes, their efficacy may be limited by the rapid development of viral resistance.4
Hepatocellular damage from HCV has been linked to oxidative stress.5 Consequently, we and others have been interested in the potential role of antioxidant enzymes as cytoprotective agents during HCV infection.6-9 Heme oxygenase-1 (HO-1) is an important cytoprotective enzyme, which is readily induced in response to a variety of stressors and cytotoxins. HO-1 oxidizes heme to equimolar concentrations of biliverdin (BV), carbon monoxide, and iron10 (Fig. 1). After heme oxidation, free BV is rapidly reduced to bilirubin (BR) by the enzyme biliverdin reductase (BVR), which is abundant in the hepatocyte.
We and others have shown that HO-1 induction or overexpression in replicons inhibits HCV replication.9, 11 Although the mechanism of this effect has not been clearly defined, it is reasonable to infer that one or more of the products of the reaction catalyzed by HO-1 may be responsible. Consistent with this prediction, iron has been demonstrated to inhibit the nonstructural 5B (NS5B) RNA-dependent RNA polymerase through inhibition of divalent cation binding.12 There are also accumulating data indicating that BV has antiviral activity,13 which has been linked to the induction of interferon response genes.13 However, these results do not exclude the possibility that BV has additional antiviral effects.
We report that BV potently inhibits viral replication at concentrations remarkably similar to those achieved after induction of HO-1 by heme, suggesting that BV is a primary antiviral agent released during heme oxidation. Furthermore, BV, and to a much lesser extent the BV metabolite, BR, inhibits HCV NS3/4A protease in cell-free assays. These findings provide a plausible mechanism for the antiviral activity of HO-1 in addition to those reported previously and strongly suggest a potential role for BV or structural derivatives in future drug design targeting the HCV NS3/4A protease.
BR, bilirubin; BV, biliverdin; BVR, biliverdin reductase; HCV, hepatitis C virus; HO, heme oxygenase; IC50, median inhibitory concentration; NS, nonstructural; RNA, ribonucleic acid; SEM, standard error of the mean; WB, western blot.
Materials and Methods
Taq DNA polymerase (Perkin-Elmer Cetus, Norwalk, CT), and Moloney murine leukemia virus reverse transcriptase (Gibco/BRL Life Technologies, Gaithersburg, MD) were used in these studies. Bile pigments were purchased from Frontier Scientific, Inc (Logan, UT) and included bilirubin-IX-α (#B584-9), biliverdin-IX-α hydrochloride (#B655-9) and mesobilirubin (B588-9). Bilirubin mixed isomers (>99%) was purchased from Sigma Chemical Co (St. Louis, MO). All preparations of tetrapyrroles were the purest form available (99% purity). The BR mixed isomer preparation contained 93% bilirubin IX-α, 3% bilirubin III-α, 3% bilirubin XIII-α, and traces of β and γ isomers (material safety data sheet information). BV was prepared by oxidation of highly purified α-bilirubin followed by final crystallization in ether (personal communication, Dr. Jerry Bommer, Echo Laboratories, Frontier Scientific, Salt Lake City, UT). All tetrapyrroles were dissolved in 0.2 N NaOH and added in small volumes to achieve the final concentration. Controls received an identical volume of diluted NaOH only. HCV protease assay kits (SensoLyte 620, Cat# 71146) and recombinant NS3/4A protease (Ac-DEDif-EchaC, Cat# 25346) were purchased from AnaSpec.
Antibody to human BVR and all secondary antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) unless otherwise indicated.
Cell Lines and Cell Culture.
The human hepatoma cell line (Huh5-15) with replicating subgenomic HCV RNA14 was a gift of Dr. Volker Lohmann (Institute for Virology, Johannes-Gutenberg University, Mainz, Germany) and was cultivated as described.9 Huh7.5 cells harboring full-length (Huh7.5FL) Con1 replicons15 were a gift of Dr. Charles Rice (Rockefeller University, New York, NY). These cells were passed as recommended by their laboratory of origin.15 An infectious clone of HCV, J6/JFH,16 was inoculated into Huh7.5 cells and the cultures passed as previously described.16 Cells were incubated with BV, BR, or FeCl2 for 24 to 48 hours in Dulbecco's modified essential medium containing 5% fetal bovine serum.
The detailed procedure is described in Supporting Methods, available online.
Cells were fixed in absolute methanol, washed in phosphate-buffered saline, and incubated with positive HCV genotype 2A polyvalent human serum. On western blots, this antiserum specifically recognized core, NS3, and NS5A at their appropriate mobilities. Antibody binding was evaluated after labeling with anti-human secondary antibody–alkaline phosphatase conjugate and results recorded by photomicroscopy.
Western Blot Analysis.
Western blots (WB) were performed as previously described using enhanced chemiluminescence for signal detection (Amersham).17 Signal intensities were quantified by using Image J software (National Institutes of Health, Bethesda, MD).
BVR small interfering RNA and control (scrambled) small interfering RNA were purchased from Santa Cruz Biotechnology (sc-44650 and sc-37007). BVR knockdown was performed as described previously.9 Efficiency of the knockdown was monitored by semiquantitative densitometry of BVR WB.
In Vitro Assay of HCV NS3/4A Recombinant Protease.
Protease activity was determined fluorometrically with the SensoLyte 620 HCV Protease Assay (AnaSpec), using a wide wavelength excitation/emission (591 nm/622 nm, respectively) fluorescence energy transfer peptide according to the manufacturer's instructions. Control incubations with BV or metabolite only were performed to eliminate or correct for autofluorescence or quenching. A competitive inhibitor of the NS3/4A protease, AnaSpec #25346, was used as a positive control.
For assays employing endogenous NS3/4A protease, the detailed procedure is described in Supporting Methods, available online.
Immunoprecipitation of NS5A.
The detailed procedure is described in Supporting Methods, available online.
Proliferation and Cytotoxicity Assays.
These assays were performed as described in detail in Supporting Methods, available online.
Data from individual experiments as well as combined data from separate experiments were expressed as mean ± standard error of the mean. The significance between means was determined by using Student t test and, when applicable, with analysis of variance, using pooled variances. P values less than 0.05 were considered significant. All experimental findings, whether performed singly or in parts, were repeated at least three times.
We have previously shown that induction of HO-1 with hemin results in decreased HCV replication in vitro9; however, it was not known whether physiological concentrations of heme exert antiviral effects. Incubation of replicons with various amounts of hemin demonstrated a concentration-dependent antiviral effect of hemin, apparent at levels as low as 5 μM (Table 1). These concentrations are well within the physiological range of heme in human circulation (10-16 μM) and, in the presence of HO-1, would be expected to yield equimolar quantities of BV, Fe, and carbon monoxide.
Log phase replicon cells were incubated with hemin or control vehicle overnight and the relative amount of HCV RNA then determined by the comparative cycle threshold level (ΔCT) as described in Materials and Methods. Each value is the mean of four determinations.
P < 0.01 all within-group differences.
Antiviral Activity of BV and BR.
We next tested BV and isomers of its metabolite BR for antiviral activity in HCV full-length and nonstructural replicons. In both replicon lines, BV showed significant antiviral activity at concentrations as low as 20 μM. In contrast, concentrations of BR-IX-α or BR mixed isomers required to suppress HCV replication were considerably higher (200 μM) (Fig. 2). For comparison, 20 μM of BV or BR corresponds to a circulating BR level of approximately 1.4 mg/dL. Western blots (Fig. 3A, B) confirmed decreased NS5A in both replicon lines after treatment with BV or BR. Levels of core protein were also reduced by BV or BR in full-length replicons, consistent with reduced replication of HCV. Treatment with BV dose-dependently decreased NS5A when assayed by WB (Fig. 3C) or immunoprecipitation using specific NS5A antibody (Fig. 3D). In accord with prior reports,12, 18 FeCl2 (100 μM) also decreased NS5A and core protein (Fig. 3A, B) as well as diminishing HCV RNA (not shown).
Cellular proliferation and toxicity profoundly affect replicon expression of HCV RNA and protein.19 Consequently, we evaluated whether BV influenced cell growth or was toxic under the current assay conditions. Presentation and description of these experiments are in the Supporting Data, available online. We observed no effect of BV or BR on cellular proliferation or toxicity when cells were incubated with tetrapyrrole in medium containing 5% or 10% fetal bovine serum, the conditions used for incubation of cells throughout the manuscript.
We next tested the effects of BV (20-200 μM) on HCV infection of Huh7.5 cells with J6/JFH infectious HCV construct.16 BV markedly decreased Huh7.5 cell infection with J6/JFH, based on immunoreactivity of HCV polyvalent sera (Fig. 4A-C) and measurement of HCV RNA (Fig. 4D).
Biliverdin Inhibits NS3/4A Protease.
Deconjugated bile pigments are known to inhibit serine-activated pancreatic proteases such as chymotrypsin and trypsin.20 This led us to evaluate the effects of BV and other tetrapyrroles on the HCV NS3/4A protease (Fig. 5A-C). These assays were conducted with wide wavelength excitation/emission (591 nm/622 nm, respectively) transfer peptides. Preliminary experiments established that shorter fluorescence wavelength transfer peptides (340 nm/490 nm or 490 nm/520 nm, excitation/emission, respectively) could not be employed because BV, BR, and other tetrapyrroles showed unacceptable autofluorescence or quenching at the shorter wavelengths.
In an assay using a recombinant protease, BV was a markedly more potent inhibitor than BR (either highly purified BR-IXα or BR mixed isomers) (Fig. 5A). BV also displayed the highest median inhibitory concentration (IC50) (9.3 μM) of any tetrapyrrole tested (Table 2), which was similar to that of the commercial NS3/4A inhibitor, AnaSpec #25346. Notably, the IC50 value for the commercial inhibitor in our hands (4.9 μM) was indistinguishable from the value reported by the manufacturer (5 μM), supporting the accuracy of our assay. Assays conducted in the presence of both BV and #25346 showed an additive effect (Fig. 5B), indicating a mixed inhibitory mechanism of BV on the NS3/4A protease as described later (Fig. 6). A modification of the fluorescence protease assay also was performed in which freshly prepared protease from replicons was used in place of recombinant protease, as described by Yu et al.21 (Fig. 5D). The results of these experiments were similar to those with the recombinant enzyme, although inhibition of the endogenous protease required slightly higher concentrations of BV than the recombinant enzyme, possibly because of conversion of BV to BR by endogenous BVR in the microsomes.
Table 2. Activity of Biliverdin and Derivatives for Inhibition of HCV NS3/4A Protease as Compared With Known Competitive Inhibitor AnaSpec #25346
Protease activity was determined fluorometrically with the SensoLyte 620 HCV Protease Assay, using a wide wavelength excitation/emission (591 nm/622 nm) fluorescence energy transfer peptide. Concentrations of inhibitor required for 50% inhibition (IC50) were determined by regression analysis.
Bilirubin (mixed isomers)
The kinetics of BV inhibition of NS3/4A protease was assessed on Lineweaver-Burk plots (Fig. 6A). These data indicated that BV competitively inhibits NS3/4A protease, based on the characteristic increase in slope with higher concentrations of inhibitor. Slopes (Km/V) and y intercepts (1/Vmax) of the primary reciprocal plots were then used to make secondary plots (Fig. 6B, C) to estimate Ki and Ki′, respectively, as general indices of competitive and noncompetitive inhibition. Note that plots of BV versus either 1/Vap or Km/V showed highly significant linearity, (r1 = 0.975 and r2 = 0.979 respectively, p < 0.005), suggesting that BV has both noncompetitive and competitive inhibitor activity for NS3/4A protease (Ki′ = 1.1 and Ki = 0.6 μM, respectively).
BV is rapidly reduced to BR by the soluble enzyme BVR (Fig. 1). We hypothesized that knockdown of BVR expression would result in increased antiviral activity for BV by diminishing its conversion to the less potent BR. Preliminary WB showed that knockdown of BVR was highly efficient and led to more than 80% reduction of BVR expression in both replicon lines (Fig. 7A). The antiviral activity of BV was significantly enhanced by BVR knockdown compared with control (scramble) RNA knockdown (Fig. 7B, left panel, p < 0.01). In contrast, knockdown of BVR before incubation of replicons with BR had no significant effect on the relatively modest antiviral activity of BR (Fig. 7B, right panel). Taken together, these data support the concept that BVR knockdown augments the antiviral activity of BV by arresting its conversion to BR and thereby maintaining higher intracellular levels of BV.
Because interferon remains a cornerstone of HCV therapy, we examined the effects of BV on the antiviral activity of α-interferon. As shown in Fig. 8, BV had a clear additive effect when exposed to cells in the presence of interferon. These findings indicate that BV does not appear to compromise the action of interferon, but rather to enhance it. They also raise the possibility that the BV or stable derivatives could be used as antiprotease agents in combination with interferon.
Heme oxygenase catalyzes the breakdown of heme to equimolar quantities of BV, iron, and carbon monoxide. Expression of the inducible isoenzyme HO-1, also known as heat shock protein 32, is readily upregulated in response to stressors such as hypoxia, heat shock, heavy metals, and oxidants.22 Along with other investigators, we have shown that HO-1 expression is downregulated in HCV-infected human liver and highly modulated in some in vitro models of HCV.6, 9, 17, 23-25 Furthermore, in cell culture models of HCV, HO-1 modulates both oxidative stress and HCV replication.9, 11
To identify the mechanisms by which exogenous heme or HO-1 overexpression inhibits HCV replication in culture,9, 11, 26 we studied the antiviral activities of heme oxidation products. Two reports have addressed the ability of iron to inhibit HCV replication12, 18; however, little attention has been directed at the other heme degradation products, BV and carbon monoxide. Our data demonstrate that BV has potent antiviral activity against HCV in two separate replicon lines and also inhibits replication in J6/JFH construct-infected Huh7.5 cells. Most importantly, our findings provide evidence that BV is a potent inhibitor of the HCV NS3/4A protease. In addition to our preliminary data,27, 28 Lehman et al.13 recently reported that BV has antiviral activity in replicon cells and noted that antiviral activity was accompanied by a rise in specific interferon-stimulated gene products. These observations are consistent with our data showing that BV inhibits NS3/4A protease. We propose that the rise in interferon-stimulated genes is a direct result of NS3/4A inactivation by BV, which prevents cleavage of adapter molecules and innate immunity recognition sites, thereby restoring signaling for innate interferon production.29, 30 Work is currently underway to further explore this possibility.
Iron also has been shown to inhibit HCV replication through prevention of divalent cation binding to RNA-dependent RNA polymerase.12, 18 Our results showing that FeCl2 inhibits replication (Fig. 3) support these data. Thus, the identification of BV as a strong antiviral agent with activity against the NS3/4A protease demonstrates that heme oxidation by HO-1 liberates at least two antiviral agents, iron and BV. These potent antiviral effects may explain the downregulation of HO-1 by HCV in infected human liver, in contrast to other liver diseases in which HO-1 is frequently upregulated.6 Importantly, the antiviral activity of heme is apparent at physiological serum concentrations, raising the possibility that heme or BV could be used as specifically targeted antiviral compounds. Heme (as hemin) is already commercially available for treatment of the porphyrias. Although antiviral activities of BV have not been formally addressed in vivo, the compound appears to be safe and has been shown to prevent hepatic reperfusion injury and vascular injury–induced intimal hyperplasia in rodent models.31
Since the discovery of HCV, the NS3/4A protease has been an attractive target for antiviral therapies. Structurally, the enzyme is a typical β-barrel serine-activated protease with a canonical Asp-His-Ser catalytic triad.32, 33 Both boceprevir and telaprevir, two promising antiviral agents currently in phase III trials, use an α-ketoamide functional group as a “serine trap” to bind and slowly dissociate from the catalytic serine in the enzyme's active site. However, BV does not contain an α-ketoamide moiety, and work is underway to determine the chemical structure(s) important for its interaction with the protease. Inhibition appears complex, because kinetic studies showed a mixed competitive and noncompetitive mechanism. Consequently, in addition to competitive binding to the substrate active site, BV may exert allosteric effects on enzyme activity, possibly through the known antioxidant or solvent effects of tetrapyrroles.34
The HO reaction releases nearly exclusively BV-IX-α,35 which is then reduced to BR-IX-α,36 the predominant BR isomer produced by adult mammalian liver. The fact that highly purified BR-IX-α and mixed isomers of BR are much weaker inhibitors of NS3/4A protease than BV suggests that BR is unlikely to exert antiviral activity in vivo at normal BR blood levels. Interestingly, BV differs from BR by a lone carbon–carbon double bond at position 10 (Fig. 1, arrow). It is intriguing that this single difference causes such a profound difference in the IC50 values of the two compounds (9 μM vs >300 μM, respectively) (Table 2). We speculate that the fixed planar double bond at position 10 may be crucial for active site binding, and we are pursuing this further with structure–function studies of BV.
The inhibition of NS3/4A protease by BV, and to a lesser extent BR and other tetrapyrroles, is not without precedence. In the bowel, unconjugated BR, but not BV, inhibits chymotrypsin and trypsin in a dose-related fashion at similar concentrations to those reported here for antiviral activity.20 In contrast, BV and BR inhibit human immunodeficiency virus protease with nearly equivalent potency,37 whereas BV has been shown to decrease viral activity of herpesvirus 6 in vitro.38
In summary, we have evaluated the antiviral activity of BV, the primary tetrapyrrole product of heme oxidation. Our findings demonstrate that BV is a potent antiviral agent, likely as a consequence of its ability to inhibit the NS3/4A protease. These findings suggest that heme, BV, or related derivatives may be useful for future drug therapies targeting the NS3/4A protease.