|Home | About | Journals | Submit | Contact Us | Français|
Cell culture systems reproducing virus replication can serve as unique models for the discovery of novel bioactive molecules. Here, using a hepatitis C virus (HCV) cell culture system, we identified neoechinulin B (NeoB), a fungus-derived compound, as an inhibitor of the liver X receptor (LXR). NeoB was initially identified by chemical screening as a compound that impeded the production of infectious HCV. Genome-wide transcriptome analysis and reporter assays revealed that NeoB specifically inhibits LXR-mediated transcription. NeoB was also shown to interact directly with LXRs. Analysis of structural analogs suggested that the molecular interaction of NeoB with LXR correlated with the capacity to inactivate LXR-mediated transcription and to modulate lipid metabolism in hepatocytes. Our data strongly suggested that NeoB is a novel LXR antagonist. Analysis using NeoB as a bioprobe revealed that LXRs support HCV replication: LXR inactivation resulted in dispersion of double-membrane vesicles, putative viral replication sites. Indeed, cells treated with NeoB showed decreased replicative permissiveness for poliovirus, which also replicates in double-membrane vesicles, but not for dengue virus, which replicates via a distinct membrane compartment. Together, our data suggest that LXR-mediated transcription regulates the formation of virus-associated membrane compartments. Significantly, inhibition of LXRs by NeoB enhanced the activity of all known classes of anti-HCV agents, and NeoB showed especially strong synergy when combined with interferon or an HCV NS5A inhibitor. Thus, our chemical genetics analysis demonstrates the utility of the HCV cell culture system for identifying novel bioactive molecules and characterizing the virus-host interaction machinery.
IMPORTANCE Hepatitis C virus (HCV) is highly dependent on host factors for efficient replication. In the present study, we used an HCV cell culture system to screen an uncharacterized chemical library. Our results identified neoechinulin B (NeoB) as a novel inhibitor of the liver X receptor (LXR). NeoB inhibited the induction of LXR-regulated genes and altered lipid metabolism. Intriguingly, our results indicated that LXRs are critical to the process of HCV replication: LXR inactivation by NeoB disrupted double-membrane vesicles, putative sites of viral replication. Moreover, NeoB augmented the antiviral activity of all known classes of currently approved anti-HCV agents without increasing cytotoxicity. Thus, our strategy directly links the identification of novel bioactive compounds to basic virology and the development of new antiviral agents.
Natural products possess a wide range of structural and functional diversity, with many of them exhibiting drug-like properties (1,–4). Thus, natural products have been a rich source of new drugs for treating many diseases, while also serving as probes for characterizing molecules and pathways critical for biological processes. Among compounds approved by the U.S. FDA from 1981 to 2010, approximately 34% of the total, and 47% of the anti-infective small molecules, are compounds derived from natural products or their analogs (3). Isolation and identification of bioactive compounds are among the most fundamental steps of drug development, necessitating the screening of compounds via cell-based, in vitro, and/or in silico assays. Models that permit the identification of both bioactivity and modes of action are limited in number and therefore especially need to be developed. In the present study, we employed a viral replication cell culture system to screen a natural product library for novel bioactivities. This cell culture-based screen provided several advantageous features, as we note here. First, virus replication, which depends on a wide variety of cellular processes, is an especially sensitive indicator of bioactivity (5). Second, the use of different virus cell culture systems permits the determination of the step(s) in the viral life cycle that is targeted by novel bioactivities (6). Third, the targets of bioactive compounds can be readily identified using the information of a panel of cellular factors known to be involved in viral replication (5, 7). In the present study, we used the hepatitis C virus cell culture (HCVcc) system to identify the bioactivity and target molecule of a fungus-derived natural product known as neoechinulin B (NeoB).
Chronic HCV infection affects approximately 170 million people worldwide. HCV infection is a major cause of liver cirrhosis and hepatocellular carcinoma and constitutes a significant public health problem. In addition to the anti-HCV treatment using pegylated alpha interferon (IFN-α) combination with ribavirin, newly approved direct-acting antivirals (DAAs) that directly target HCV-derived proteins, including NS3 protease, NS5A, and NS5B polymerase, significantly improve clinical outcomes of HCV-infected patients (8, 9). However, the problems of these DAAs include the huge cost and thus the low availability of drugs, especially in disadvantaged countries. Another approach to antiviral drug development is to target cellular factors that are essential for HCV propagation. This line of trials has yielded promising developments of cyclophilin inhibitors and microRNA-122 inhibitors, which are classified as so-called host-targeting antivirals (HTAs) (8, 9). So far, we have characterized the anti-HCV activity of cyclophilin inhibitors and the role of cyclophilin in HCV replication (10, 11). Despite this progress in the development of novel antivirals, new medications, notably the DAAs, are very expensive, making these drugs difficult to use in disadvantaged countries with a significant HCV-infected population. To eradicate HCV infection worldwide, new low-cost anti-HCV drugs are greatly needed. So far, searches have been made for drug development using natural products as lead compounds (2, 12,–16). However, most of these studies have not been successful in clarifying the mode of action of the studied compounds, especially with regard to target molecules.
In this study, we prepared an in-house natural product library consisting of compounds isolated from fungal strains and screened this library in cell-based functional assays using the HCV cell culture system (2, 17). We identified NeoB as a fungus-derived small molecule with the strongest potency to inhibit the production of infectious HCV. Further analysis demonstrated that NeoB is a novel inhibitor of liver X receptors (LXRs). Using NeoB as a probe, LXRs were revealed to be critical for HCV replication in host cells: LXR inactivation reduced the formation of double-membrane vesicles (DMVs), where the HCV replication complex is proposed to form. Thus, the linking of the HCV cell culture system to a chemical genetic analysis permits the identification of a new bioactive molecule which is a valuable tool for cell biological research and may serve as a lead for the development of novel antiviral drugs.
Huh-7, Huh7.5.1, Huh7-25, LucNeo#2, HuSE2, and 293T cells, as well as primary human hepatocytes, were cultured as described previously (2, 18,–20). Huh7.5.1 and LucNeo#2 cells were kindly provided by Francis Chisari at The Scripps Research Institute and Kunitada Shimotohno at National Center for Global Health and Medicine. Huh7-25 cells, an Huh-7 cell clone deficient for CD81 expression, as well as a human hepatoma Huh-7 and a human embryonic kidney 293T cell line, were described previously (2). HuSE2 cells are immortalized human primary hepatocytes that can support HCV infection (18). Primary human hepatocytes were isolated from PXB mice (PhoenixBio). For detecting SREBP1 protein, Huh-7 cells were precultured for 24 h in medium supplemented with 10% lipoprotein-deficient serum (Sigma-Aldrich) and 10 μM compactin (Wako) and then incubated for 48 h with compounds and compactin (21).
Natural products were extracted from culture broths of fungal strains isolated from seaweeds, mosses, and other plants as described previously (2, 17). We prepared an in-house natural product library consisting of approximately 200 isolated compounds.
NeoB is a secondary metabolite of Aspergillus amstelodami (22). We used a preparation of NeoB isolated from the fungus in all the experiments described in this study.
Infectious HCV was recovered from the culture supernatant of Huh-7 cells transfected with HCV RNA, as described previously (2). Huh7.5.1 cells infected with HCV at a multiplicity of infection (MOI) of 0.15 for 4 h were left untreated or treated with compounds for 72 h. The infectivity of HCV and the amount of HCV core protein in the culture supernatant were quantified by infectious focus formation assay and chemiluminescent enzyme immunoassay (Lumipulse II HCV core assay), respectively (2). HuSE2 cells were infected with HCV at an MOI of 1 for 4 h.
Dimethyl sulfoxide (DMSO), cyclosporine, acriflavine, lamivudine, troglitazone, TO-901317, all-trans-retinoic acid (ATRA), 4-hydroxytamoxifen, GW6471, and GSK0660 were purchased from Sigma. Wy14643 was purchased from Cayman Chemical. 17β-Estradiol and FH535 were purchased from Merck. Ro41-5253 and GW501516 were from Enzo Life Sciences. Bafilomycin A1 and 5CPPSS-50 were from Wako. IFN-α was obtained from Schering-Plough. Recombinant full-length LXRβ protein was purchased from OriGene. Telaprevir, sofosbuvir, and daclatasvir were purchased from Selleckchem.
Huh7.5.1 cells were infected with HCV at a multiplicity of infection (MOI) of 0.15 for 4 h. After free virus was washed out, cells were treated with compounds at 10 μM in growth medium for 72 h. Infectivity of HCV in the resultant culture supernatant and cell viability were quantified by infectious focus formation assay and MTT [3-(4,5-dimethylthiazol-2-yl)2 2,5-diphenyl tetrazolium bromide] assay, respectively (2). Normalized infectivity was calculated as HCV infectivity divided by cell viability. Compounds that decreased the normalized infectivity to less than 40% were selected as primary hits and were further evaluated for data reproducibility and dose dependency (2). Compounds that decreased cell viability to less than 50% of that observed for nontreated cells were considered cytotoxic and eliminated from evaluation.
Immunoblot analysis was performed as described previously (10). Mouse anti-HCV core protein (2H9), rabbit anti-NS5A, rabbit anti-ISG56 (Abcam), rabbit anti-MxA (Santa Cruz Biotechnology), anti-LXR (Abcam), rabbit anti-SCD-1 (Cell Signaling Technology), mouse anti-SREBP1 (BD Pharmingen), and mouse anti-β-actin (Sigma) antibodies were used as primary antibodies. The intensities of the bands were quantified with ImageQuant TL (GE Healthcare).
Indirect immunofluorescence analysis was performed essentially as described previously using anti-HCV core protein antibody (2H9) as the primary antibody (2).
The cell viability was quantified by using a Cell Proliferation Kit II XTT (Roche Diagnostics) as described previously (23).
An HCV pseudoparticle (HCVpp) assay, which evaluates HCV envelope-dependent entry activity, was performed as described previously (2, 24). HCVpps were prepared with expression plasmids for HCV JFH-1 E1E2, murine leukemia virus Gag-Pol, and luciferase protein, which were kindly provided by Francois-Loic Cosset at L'Université de Lyon. Compounds were evaluated by treating cells for 2 h prior to infection and for 4 h during HCV infection.
An HCV replicon assay, which reproduces HCV translation and replication, was performed as described previously using an HCV JFH-1 subgenomic replicon (SGR) RNA (SGR-JFH1/Luc) or cells carrying an HCV NN (genotype-1b) subgenomic replicon RNA (LucNeo#2 cells) (2, 19). An RNA polymerase I (Pol I)-based HCV expression system (pHH-JFH1) was used as a plasmid-driven replicon (25). For evaluation of HCV translation, an HCV JFH-1 subgenomic replicon encoding a GND substitution at its polymerase active GDD motif in the NS5B region (SGR-JFH1/Luc GND) (2, 26) was used.
Huh-7.5.1 cells were treated with compounds (0.2% DMSO and 20 μM NeoB) for 24 h, and total RNA was recovered. Total RNA (500 ng) was amplified using a total RNA amplification kit (Ambion, Carlsbad, CA). cRNA was hybridized to an Illumina Human HT-12, version 4, array (Illumina, San Diego, CA). Scanning of the chip was performed using Illumina BeadScan and BeadStudio software packages, and data were generated using BeadStudio (version 1.6). Three biological replicates were analyzed for the microarray experiments. Bioconductor packages limma and lumi combined with Illumina's Human HT-12, version 4.0, annotation set were used for differential expression analysis (27, 28). Expression levels were background corrected and quantile normalized using the neqc function of the limma package. Probes annotated as being low quality by HT-12, version 4.0, were removed. Nonresponsive probes with interquantile ranges less than or equal to 0.15 across all samples were removed as well, leaving 21,737 probes in the end. Thresholds for differential expression analysis were set at a log2 fold change of 1 and a minimum false discovery rate of 0.05. Gene ontology (GO) analysis was performed using the mean log P value (29).
Huh7-25 cells were transfected with a reporter plasmid carrying an interferon-sensitive response element (ISRE) and the binding elements for peroxisome proliferator-activated receptor (PPAR), liver X receptor (LXR), retinoic acid receptor (RAR), or estrogen receptor (ER) upstream of the firefly luciferase and a reporter carrying the herpes simplex virus thymidine kinase promoter upstream of the Renilla luciferase, with or without an expression plasmid for a nuclear hormone receptor (either PPARα, PPARβ/δ, PPARγ, RARα, LXRα, or LXRβ) along with the heterodimeric partner retinoid X receptor alpha (RXRα) or ERα (30,–32). The cells were then either left untreated or treated with compounds in the presence or absence of their agonist for 48 h. The cells were lysed, and the luciferase activity was measured (23).
For quantification of Gaussia luciferase (Gluc) activity, Huh7-25 cells were transfected with the reporter vector carrying a stearoyl coenzyme A (CoA) desaturase-1 (SCD-1) gene promoter region upstream of the Gluc gene (GeneCopoeia), with or without the LXR expression plasmid, and then either left untreated or treated with NeoB for 72 h. Gluc activity was measured using a Secrete-Pair dual luminescence assay kit (GeneCopoeia) according to the manufacturer's protocol.
Reverse transcription-PCR (RT-PCR) analysis was performed as described previously, using RNA prepared from Huh-7 cells treated for 48 h with NeoB or 5CPPSS-50 in the presence or absence of an LXR ligand, TO-901317 (10). The primers used in this study were 5′-GATCAAAGAGGAGCCAGTGC-3′ and 5′-TAGATGGTGGCTGCTGAGTG-3′ for SREBP-1c transcripts, 5′-CCTGCTGTACTTGGGGATCGGGAACG-3′ and 5′-CCAGCGCGGCAAACAGCACAAAG-3′ for ABCG1 transcripts, 5′-GCGGAGCCATGGATTGCAC-3′ and 5′-CTCTTCCTTGATACCAGGCCC-3′ for SCD-1 transcripts, and 5′-CCATGGAGAAGGCTGGGG-3′ and 5′-CAAAGTTGTCATGGATGACC-3′ for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcripts. The intensities of the bands were quantified with ImageJ software.
Small interfering RNAs (siRNAs) used in this study were the following: 5′-GGAUGCUAAUGAAACUGGU-3′ and 5′-GAACAGAUCCGGAAGAAGA-3′ (si-LXRα); 5′-AGCUAACAGCGGCUCAAGA-3′ and 5′-AGAUCGUGGACUUCGCUAA-3′ (si-LXRβ) (Applied Biosystems). For siRNA screening of putative LXR downstream genes, we used a mixture of four different siRNAs against each gene (siGENOME SMARTpool; Dharmacon). siRNAs used were those against SCD-1 (si-SCD-1; siGENOME SMARTpool [Dharmacon]) and randomized control siRNAs (siGENOME Nontargeting siRNA Pool 1; Dharmacon). Cells were transfected with siRNAs using Lipofectamine RNAiMAX (Life Technologies) or DharmaFECT 4 transfection reagent (Dharmacon) according to the manufacturers' protocols.
The kinetics of NeoA and NeoB binding to LXRα and LXRβ were analyzed using a Biacore 3000 instrument (GE Healthcare). LXRα and LXRβ were immobilized onto the surface of a CM5 sensor chip using an amine coupling kit (GE Healthcare). Different concentrations of NeoA and NeoB diluted with HBS-P (10 mM HEPES, 0.15 M NaCl, 0.005% surfactant P20) containing 5% DMSO were injected for 120 s at a flow rate of 20 μl/min at 25°C. The bulk effect of DMSO was subtracted using reference flow cells. Kinetic parameters were determined by analyzing the data using BIAevaluation, version 4.1, software (GE Healthcare).
Huh7-25 and Huh7.5.1 cells, either uninfected or infected with HCV, and primary human hepatocytes were either left untreated or treated with 20 μM NeoB for 72 h in medium with or without 25 μM chloroquine to facilitate lipid accumulation. Cellular lipids were then evaluated with a steatosis assay kit (Cayman Chemical) according to the manufacturer's protocol.
Electron microscopic analysis was performed essentially as described previously (33). In brief, the cells were prefixed with 2.5% glutaraldehyde, 2% paraformaldehyde, and 0.1 M phosphate buffer (pH 7.4) for 2 h at room temperature, postfixed with 1% osmium tetroxide, and embedded in Epon. Ultrathin sections were stained with uranyl acetate and lead citrate and then examined with a transmission electron microscope (HT7700; Hitachi, Ltd., Japan). For the quantification data, we counted the cells that carried clear double- or multimembrane vesicles (DMVs or MMVs) over 200 observed cells and calculated the percentage of DMV- or MMV-positive cells.
Huh-7 cells infected with dengue virus type 1 (DENV-1) at an MOI of 0.125 for 1 h were treated with compounds for 72 h. DENV RNA in the culture supernatant was quantified by real-time RT-PCR using a TaqMan Fast Virus 1-Step Master Mix (Applied Biosystems) with 5′-GAACATGGRACAAYTGCAACYAT-3′ and 5′-CCGTAGTCDGTCAGCTGTATTTCA-3′ as primers and 5′-ACACCTCAAGCTCC-3′ as a probe (34).
Poliovirus replication was assessed as described previously using a poliovirus replicon RNA carrying the luciferase gene (35).
HepG2.2.15 cells (23) were treated with compounds for 6 days in medium that was changed every 3 days. Compounds were supplemented in the fresh medium when the medium was changed. Intracellular DNA was recovered with a QIAamp DNA minikit (Qiagen), and HBV DNA was quantified by real-time PCR as described previously (23).
To examine whether the effects of the drug combinations were synergistic, additive, or antagonistic, MacSynergy (kindly provided by Mark Prichard), a mathematical model based on Bliss independence theory, was conducted as described previously (2). The theoretical additive effects were compared to the actual experimental effects at various concentrations of the two compounds and were plotted as a three-dimensional differential surface. On the plot, additive effects show the zero plane at the z axis. Synergistic and antagonistic effects show peaks above and below this plane, respectively. The 95% confidence interval of the experimental results is shown.
Statistical significance was determined using Student's t tests. P values of less than 0.05 were considered statistically significant.
Microarray data were deposited in the Gene Expression Omnibus (GEO) database under accession number GSE63026.
We prepared an in-house natural product library consisting of approximately 200 isolated compounds derived from fungal secondary metabolites, as described in Materials and Methods (2, 17). We employed the infectious HCV cell culture (HCVcc) system, which reproduces the whole HCV life cycle (36,–38), to identify bioactive natural organic compounds. Huh7.5.1 cells persistently infected with HCV were treated with compounds for 72 h, and the infectivity of HCV produced from these cells was quantified (2). In the chemical screening, compounds that reduced the infectivity of the HCV produced to less than 40% were regarded as primary hits, which were then validated by evaluating their reproducibility, dose dependency, and cytotoxicity. Neoechinulin B [(3Z)-3-[[2-(2-methylbut-3-en-2-yl)-1H-indol-3-yl]methylidene]-6-methylidenepiperazine-2,5-dione] (NeoB) (22) (Fig. 1A) was among the most potent compounds in reducing viral infectivity and was the focus of the following analyses.
To characterize the anti-HCV activity of compounds, Huh7.5.1 cells were infected with HCV at a multiplicity of infection (MOI) of 0.15 for 4 h and treated with the compounds for 72 h. Infectivity of HCV in the resulting culture supernatant was evaluated by reinoculating the supernatant into naive Huh7.5.1 cells and detecting HCV core and NS5A proteins by immunoblotting (Fig. 1B) and immunofluorescence (Fig. 1C). The infectivity of HCV produced from NeoB-treated Huh7.5.1 cells was significantly reduced, similar to results in cells treated with cyclosporine, a known HCV replication inhibitor (Fig. 1B and andC).C). In quantitative analysis, HCV infectivity and core protein level were reduced to approximately 11% and 29% of the control levels, respectively, by NeoB treatment (Fig. 1D). This effect was dose dependent without significant cytotoxicity (Fig. 1E). The anti-HCV activity of NeoB was also observed in immortalized primary human hepatocytes, HuSE2 cells (Fig. 1F). Thus, NeoB decreased the production of infectious HCV particles.
We next investigated which step in the HCV life cycle was inhibited by NeoB. The life cycle of HCV can be divided into the following steps: (i) early steps, attachment and entry; (ii) middle steps, translation and RNA replication; and (iii) late steps, assembly and release (5) (Fig. 2A). We first evaluated the effects of NeoB on the early steps using an HCV pseudoparticle (HCVpp) system that models the HCV envelope-mediated attachment and entry (see Materials and Methods) (24). As shown in Fig. 2B, NeoB had no significant effect on the luciferase activity driven by HCVpp infection, in contrast to bafilomycin A1, a known HCV entry inhibitor (39) (Fig. 2B, left). The middle steps of the HCV life cycle, including translation and RNA replication, were evaluated using an HCV subgenomic replicon (see Materials and Methods) (40). The replicon carrying a mutation in the NS5B polymerase active motif (replicon GND) that allows translation but not RNA replication (40) was used to evaluate HCV translation, without replication (Fig. 2A). NeoB was shown to reduce the activity of the replication-competent replicon (Fig. 2B, middle) but not that of the replication-deficient replicon GND (Fig. 2B, right), suggesting that NeoB inhibits the RNA replication step of the HCV life cycle. IFN-α is known to suppress HCV replication through activation of IFN-sensitive response element (ISRE)-mediated transcription and the resultant induction of downstream IFN-stimulated genes (ISGs) (41,–44). As shown in Fig. 2C, NeoB did not enhance the level of ISRE-mediated transcription in either the presence or absence of IFN-α (Fig. 2C, left) and had no significant effect on protein expression levels of ISGs, ISG56 and MxA (Fig. 2C, right). Thus, NeoB inhibits RNA replication in a manner independent of the IFN pathway.
HCV RNA replication is regulated by numerous cellular factors and environments (6, 7, 45). To identify the target molecules of NeoB, we performed a transcriptome analysis that compared cellular gene expression between NeoB-treated and untreated Huh-7.5.1 cells. In NeoB-treated cells, 36 and 25 genes out of 17,509 genes showed more than 2-fold upregulation and downregulation, respectively (Table 1). Gene ontology (GO) enrichment analysis based on a mean log P value indicates that the GO terms such as drug metabolism, lipid metabolism, and hormone metabolism are significantly overrepresented among genes downregulated by NeoB (Table 2). Many of the genes involved in these metabolic processes are regulated by transcription factors, including nuclear hormone receptors (46, 47). Moreover, the HCV life cycle is known to be modulated by multiple nuclear hormone receptors such as peroxisome proliferator-activated receptor (PPAR), estrogen receptor (ER), and farnesoid X receptor (32, 48,–50), leading us to question whether NeoB affected the function of nuclear hormone receptors that, in turn, modulate HCV replication.
We performed reporter assays to examine the effect of NeoB on the transcriptional activity of specific nuclear hormone receptors, including PPARs, retinoic acid receptor (RAR), ER, and LXR, in uninfected Huh7-25 cells (Fig. 3A). NeoB did not have a significant effect on the transcriptional activity of PPARα, PPARβ/δ, PPARγ, RAR, or ER in either the presence or absence of receptor-specific agonists, in contrast to the transcriptional repression by known antagonists as positive controls (GW6471, GSK0660, FH535, Ro41-5253, and 4-hydroxytamoxifen) (Fig. 3A). Among these, NeoB decreased the reporter activity mediated by LXRα and LXRβ (Fig. 3A, panels f and g) without affecting the level of the LXR protein itself (Fig. 3A, panel h). Consistent with this observation, transcripts for LXR downstream genes, including ATP-binding cassette subfamily G member 1 (ABCG1) (Fig. 3B), sterol regulatory element-binding protein 1c (SREBP1c) (Fig. 3B; see also Fig. 5B), and stearoyl-CoA desaturase-1 (SCD-1) (see Fig. 7D), clearly decreased with NeoB treatment. The inhibition of LXRα and LXRβ upon NeoB treatment was also observed in HCV-infected Huh-7 cells and HuSE2 cells (Fig. 3C and andD).D). The half-maximal inhibitory concentrations (IC50s) for inhibition of LXRα and LXRβ were 5.5 and 7.6 μM, respectively, in HuSE2 cells (Fig. 3D) although these values varied among cell types. These data suggest that NeoB specifically inhibits the transcriptional activity of LXRs.
A derivative analysis of NeoB revealed that a series of analogs, including neoechinulin A (NeoA), fumitremorgin C (FTMC), 3-indoleethanol (3-IET), and N-acetyltryptamine (N-ATTP), did not inhibit LXR activity (Fig. 4A and andB).B). NeoA is the closest derivative among these, with a double bond in the diketopiperazine structure replaced by a single bond (Fig. 4A). Intriguingly, we found by surface plasmon resonance analysis that NeoB interacted with recombinant LXRα and LXRβ, with KD (equilibrium dissociation constant) values of 2.9 ± 1.4 and 12.5 ± 1.1 μM, respectively (Fig. 4C). In contrast, NeoA did not show any in vitro affinity to either LXRα or LXRβ (Fig. 4C), in agreement with the lack of an NeoA effect on the transcriptional activity of these receptors (Fig. 4B).
LXRs are implicated in a wide variety of physiological events, including lipid metabolism, cholesterol metabolism, glucose homeostasis, inflammation, and neurological homeostasis (51,–53). LXRs induce the expression of SREBP1c, a master regulator of lipid and cholesterol metabolism (51, 54), and mediate lipid accumulation in hepatocytes, as can be observed by oil red O staining (55, 56). Employing this methodology, we treated uninfected Huh7.5.1 and Huh7-25 cells with compounds and then stained them with oil red O to observe intracellular lipid accumulation. As shown in Fig. 5A, lipid staining was markedly reduced in both Huh7.5.1 and Huh7-25 hepatocyte-derived cell lines that were treated with 5CPPSS-50, a known pharmacological inhibitor of LXR (Fig. 5A, panels d and h). Treatment with NeoB similarly reduced lipid accumulation in these cells, but NeoA did not (Fig. 5A, panels b, c, f, and g). These phenotypes were accompanied by, and correlated with, the decreased accumulation of SREBP1c mRNA (Fig. 5B) and of the cleaved form of SREBP1 protein (Fig. 5C, arrow) (57). Similar observations were obtained in HCV-infected Huh-7 cells and primary human hepatocytes (Fig. 5D and andE).E). We therefore concluded that NeoB is a novel LXR inhibitor that impairs LXR-mediated transcription, thereby modulating LXR-dependent cellular physiology.
We then examined whether the repression of LXR was related to the observed reduction in HCV replication by NeoB treatment. To date, the significance of LXRs in the regulation of the HCV life cycle has not been fully understood: one paper showed that LXR depletion resulted in a slight reduction in HCV RNA levels in the replicon system although no further analysis was presented (48); other studies have reported that TO-901317 and GW3965, known as LXR antagonists, modulate HCV entry (58, 59). We showed that NeoB decreased HCV replicon activity but that NeoB derivatives lacking LXR inhibitory activity did not affect HCV replication (Fig. 4B and and6A).6A). LXR was inactivated by another antagonist, 5CPPSS-50, which also supported reduced levels of HCV replication without cytotoxicity (Fig. 6B). Similarly, knockdown of endogenous LXR in HCV-infected cells yielded decreased levels of HCV proteins such as HCV core and NS5A, consistent with decreased replication activity (Fig. 6C). Moreover, in cells overexpressing LXR and RXR, HCV replication levels were increased in a dose-dependent manner (Fig. 6D). The above results clearly suggest that LXR transcriptional activity supports efficient HCV replication and that dysregulation of LXR activity by antagonists results in decreased HCV replication in host cells.
We further examined the mechanisms whereby LXR mediates efficient HCV replication. As previously reported, electron microscopic analysis of HCV-infected cells showed remarkable accumulations of membrane compartments, known as double-membrane vesicles (DMVs) or multimembrane vesicles (MMVs), which are putative sites of the viral replication complex (60,–64), while these membrane compartments were never seen in uninfected cells (Fig. 7A, compare panels a and b to panels c to e). However, upon inactivation of LXRs of HCV-infected cells by NeoB, DMVs and MMVs were severely dispersed, and small vesicles (SVs) were remarkably increased (Fig. 7A, compare panels c to e to panels f to h). The frequency of DMV/MMV-positive cells relative to the total number of cells was quantified to be significantly reduced by NeoB treatment (Fig. 7B). A similar observation was obtained in cells transfected with a mutant HCV RNA [JFH1(GND)] that produces HCV proteins but is deficient for viral replication (data not shown), suggesting that the NeoB effect was not the result of a decrease in HCV RNA replication but, rather, a direct effect on the machinery for producing and/or maintaining these membrane structures. In support of this, we observed that LXR inhibition also attenuated the efficient replication of DMV-associated viruses but not DMV-unrelated viruses. Specifically, poliovirus is known to replicate in the DMVs, while another plus-stranded RNA virus, dengue virus, replicates via a different type of membrane structure, the invaginated vesicles (64). Notably, NeoB reduced the replication level of poliovirus but not that of dengue virus (Fig. 7C, left and center panels). Replication of hepatitis B virus, a hepatotropic DNA virus that replicates independently of intracellular membrane compartments, was not affected by NeoB treatment (Fig. 7C, right). These data are in agreement with the observation that the antiviral effect of NeoB is specific for viruses that replicate in association with DMVs.
The observed effects of NeoB treatment were accompanied by the downregulation of stearoyl CoA desaturase-1 (SCD-1), a key enzyme in the synthesis of monounsaturated fatty acids from saturated fatty acids, thereby altering membrane fluidity (63, 65). Among 22 genes that were reported as downstream genes of LXR (51,–53), SCD-1 was suggested to contribute to efficient HCV replication: knocking down endogenous SCD-1 levels reduced the replication of HCV in both replicon (Fig. 8A) and HCV-infected cells (Fig. 8B) without apparent cytotoxic effects (Fig. 8A and andB,B, lower graphs). Actually, SCD-1 was transcriptionally regulated by LXRs (Fig. 8C and andD).D). Overproduction of SCD-1 significantly rescued the reduction in HCV replication in LXR-depleted cells although it did not fully restore the replication level (Fig. 8E), suggesting that SCD-1 was at least one of the targets through which LXR supports HCV replication. Indeed, SCD-1 has been proposed to be important for the formation of DMVs (63, 65), but it has not been known which stimulus and signal regulate DMV formation. Considered together, our results support the possibility that LXR regulates the formation of DMVs at least in part through induction of SCD-1, thereby promoting the efficient replication of HCV.
To explore the significance of NeoB-mediated LXR inhibition as an antiviral strategy, we investigated the effect of NeoB on the replication of different HCV genotypes, 1b (NN) and 2a (JFH-1) (19, 37). As shown in Fig. 9A, NeoB decreased the replication activity of both of these genotypes (Fig. 9A).
We then evaluated the anti-HCV activity of all the major classes of clinically available anti-HCV agents when combined with NeoB in HCV-infected cells: interferon (Fig. 9B), a protease inhibitor (telaprevir) (Fig. 9C), a polymerase inhibitor (sofosbuvir) (Fig. 9D), and an NS5A inhibitor (daclatasvir) (Fig. 9E). Supplementation with NeoB led to a greater reduction in HCV infectivity without a significant increase in cytotoxicity at any given concentration, compared with results achieved by these anti-HCV agents alone (Fig. 9B to toE).E). Notably, synergy/antagonism analysis with the Bliss independence model showed that IFN-α and an NS5A inhibitor especially exerted strong synergistic anti-HCV effects in combination with NeoB at most of the tested doses (in this analysis, a peak above the zero plane in the z axis indicates a synergistic antiviral effect by cotreatment of two drugs, while a valley under the zero and zero plane itself designate antagonistic and additive antiviral effects, respectively) (Fig. 9B and andE).E). Thus, NeoB is highly effective in treating multiple HCV genotypes and in multidrug treatment combinations with currently approved anti-HCV agents.
Identification of relevant bioactivities of natural products is a fundamental and critical step in linking natural source compounds to biological applications that can lead to development of new drugs. HCV is highly dependent on cellular factors for its efficient propagation, and many compounds that target cellular factors therefore inhibit the HCV life cycle (5, 7, 8). We here took advantage of this fact by using an HCV cell culture system to screen a natural product library for novel antiviral activities. This screen identified NeoB as an anti-HCV compound and demonstrated that this effect is mediated by inhibition of LXR. This study additionally revealed that LXRs regulate the formation of membrane structures that are associated with HCV replication. LXRs are known to be involved in a wide variety of physiological processes, including lipid metabolism, cholesterol metabolism, glucose homeostasis, inflammation, and neurological homeostasis, making it a potential drug target for diseases, including atherosclerosis, diabetes, hepatic steatosis, and Alzheimer's disease (51,–53). The compound identified here, NeoB, was originally isolated from Aspergillus amstelodami (22). Although NeoA, a NeoB derivative (Fig. 4A), has been reported to show multiple physiological activities, including antioxidant and cytoprotective activities in neuronal cells and anti-inflammatory effects in macrophages (66, 67), there are no reports on the bioactivity of NeoB so far. In the present work, we demonstrated that NeoB, but not NeoA, interacted with LXRs and inhibited their transcriptional activities. While the use of LXR ligands for drug development has been an area of interest for over a decade (51,–53), it will be interesting to observe in the future whether NeoB or related compounds affect LXR-mediated cellular functions other than those implicated in HCV replication. For instance, the present work demonstrates that NeoB modulates lipid accumulation in hepatocytes, independently of HCV infection (Fig. 5).
Our analysis using NeoB as a probe has clarified the significance of LXR in HCV replication (Fig. 10). Inhibition of LXR correlated with anti-HCV activity, suggesting that an LXR downstream gene(s) is involved in HCV replication. One such target gene is SCD-1 (Fig. 8) although other genes may also contribute to the modulation of HCV replication. Recent reports proposed that unsaturated fatty acids produced by SCD-1 are important for the formation and/or maintenance of the membrane structures used by the HCV replication complex (63, 65). However, the factor(s) that triggers or regulates the formation of the HCV replication site has remained poorly understood. Our results suggest that LXR inactivation and the resulting attenuation of LXR downstream gene induction disrupted the formation of the membrane compartments that are associated with HCV replication.
The primary strategy for developing anti-HCV drugs has been the targeting of virus-derived factors such as NS3 protease, NS5A, and NS5B polymerase (8, 9). However, since the huge cost of the resulting anti-HCV agents constitutes a significant social problem, the identification of new molecular targets for the development of additional antiviral agents is an appealing approach. To date, natural products reported to inhibit HCV replication/production include silymarin/silibinin, epigallocatechin gallate, ladanein, naringenin, and quercetin (12,–16). However, while these compounds likely act upon cellular factors, their actual modes of action, structure-activity relationships, and especially their target molecules remain poorly defined. The identification of LXRs as a cellular target of NeoB raises the possibility that LXRs can serve as potential targets for the development of antiviral agents, not only against HCV but also against other DMV-associated viruses. Moreover, the fact that the LXR inhibitor identified here displays enhanced antiviral effects, including synergy, in combination with other antiviral drugs is a significant finding for drug development. In conclusion, viral cell culture systems are powerful tools for identifying unknown bioactivities of natural products and for linking such compounds to the development of new classes of antiviral agents.
Huh7.5.1 and LucNeo#2 cells were kindly provided by Francis Chisari at The Scripps Research Institute and Kunitada Shimotohno at National Center for Global Health and Medicine, respectively. The expression plasmids for preparing HCVpp were generous gifts from Francois-Loic Cosset at L'Université de Lyon.
This study was supported by grants-in-aid from the following: the Ministry of Health, Labor, and Welfare, Japan; the Ministry of Education, Culture, Sports, Science, and Technology, Japan; Japan Society for the Promotion of Science (KAKENHI 26460565 and 26102747); Core Research for Evolutional Science and Technology, Japan Science and Technology Agency; a research grant from the ONO Medical Research Foundation; and Research Program on Hepatitis from the Japan Agency for Medical Research and Development.
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.