Systems Virology Identifies a Mitochondrial Fatty Acid Oxidation Enzyme, Dodecenoyl Coenzyme A Delta Isomerase, Required for Hepatitis C Virus Replication and Likely Pathogenesis

JOURNAL OF VIROLOGY, Nov. 2011, p. 11646–11654 0022-538X/11/$12.00 doi:10.1128/JVI.05605-11 Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Vol. 85, No. 22

Systems Virology Identifies a Mitochondrial Fatty Acid Oxidation Enzyme, Dodecenoyl Coenzyme A Delta Isomerase, Required for Hepatitis C Virus Replication and Likely Pathogenesis䌤† Angela L. Rasmussen,1 Deborah L. Diamond,1 Jason E. McDermott,2 Xiaoli Gao,3 Thomas O. Metz,3 Melissa M. Matzke,2 Victoria S. Carter,1 Sarah E. Belisle,1 Marcus J. Korth,1 Katrina M. Waters,2 Richard D. Smith,3 and Michael G. Katze1* Department of Microbiology, University of Washington School of Medicine, Seattle, Washington 981951; Computational Biology & Bioinformatics, Pacific Northwest National Laboratory, Richland, Washington 993522; and Biological Sciences Division, Pacific Northwest National Laboratory, Richland, Washington 993523 Received 4 July 2011/Accepted 25 August 2011

We previously employed systems biology approaches to identify the mitochondrial fatty acid oxidation enzyme dodecenoyl coenzyme A delta isomerase (DCI) as a bottleneck protein controlling host metabolic reprogramming during hepatitis C virus (HCV) infection. Here we present the results of studies confirming the importance of DCI to HCV pathogenesis. Computational models incorporating proteomic data from HCV patient liver biopsy specimens recapitulated our original predictions regarding DCI and link HCV-associated alterations in cellular metabolism and liver disease progression. HCV growth and RNA replication in hepatoma cell lines stably expressing DCI-targeting short hairpin RNA (shRNA) were abrogated, indicating that DCI is required for productive infection. Pharmacologic inhibition of fatty acid oxidation also blocked HCV replication. Production of infectious HCV was restored by overexpression of an shRNA-resistant DCI allele. These findings demonstrate the utility of systems biology approaches to gain novel insight into the biology of HCV infection and identify novel, translationally relevant therapeutic targets. Lipids play a role in numerous steps of the hepatitis C virus (HCV) replication cycle, including RNA replication associated with the lipid droplet (6, 35, 48), virus uptake, assembly, and secretion in association with cellular apolipoproteins (2, 9, 16, 23, 33, 37, 44), fusion with host membranes during virus entry (13), endocytic trafficking (3, 8), and reorganization of cellular membranes associated with replication and assembly (1, 4, 49, 51). Furthermore, patients with HCV often exhibit hepatic steatosis and the upregulation of a number of genes involved in hepatic lipid metabolism (11, 52). Recently, using integrated modeling efforts combining proteomic and lipidomic data, we identified two enzymes, dodecenoyl coenzyme A (CoA) delta isomerase (DCI) and hydroxyacyl-CoA dehydrogenase beta subunit (HADHB), that were upregulated during in vitro infection and in patients with histological evidence of fibrosis. These two enzymes were predicted to be bottleneck proteins, key regulators of the HCVinduced temporal alterations in cellular metabolic homeostasis during viral replication (11). DCI and HADHB are localized to the inner mitochondrion and catalyze the degradation of longchain fatty acids during fatty acid ␤-oxidation (7, 38, 47). Previously, ␤-oxidation was shown to be required for measles virus (46) and dengue virus replication (14), although DCI and

* Corresponding author. Mailing address: Department of Microbiology, University of Washington School of Medicine, Box 358070, Seattle, WA 98195. Phone: (206) 732-6136. Fax: (206) 732-6056. E-mail: [email protected]. † Supplemental material for this article may be found at http://jvi .asm.org/. 䌤 Published ahead of print on 14 September 2011.

HADHB were not directly implicated in this process. Several proteins involved in lipid catabolism, including lipases, esterases, acyl-CoA dehydrogenases, and palmitoyltransferases have also been identified as important host factors in HCV replication by RNA interference-based screening (5, 8, 31, 40, 45, 49). DCI and HADHB were previously identified as host factors producing a subtle decrease in RNA production in a replicon system, although their function in HCV replication remains unknown (45). In this study, we evaluated the role of DCI by using short hairpin RNA (shRNA)-expressing lentiviruses to stably silence DCI gene expression, rather than the transient-knockdown approaches used in small interfering RNA (siRNA)-based screens. Although lipid-related metabolic dysfunction has been observed in the context of HCV infection, no studies have directly linked fatty acid oxidation with HCV replication or pathogenesis. Here, we describe an extension of sophisticated modeling techniques to clinical proteomic data confirming that DCI is computationally predicted to be a key cellular protein required for HCV pathogenesis in vivo. Moreover, we identified additional proteins predicted to further regulate cellular metabolic alterations augmenting HCV infection. We also present biological validation data proving that DCI is essential to HCV replication, confirming our computational predictions postulating a critical role for mitochondrial fatty acid oxidation in the HCV life cycle. Together, these findings illustrate the power of systems biology approaches to identify novel host factors that provide both biological insight into the processes of HCV replication and unique targets for the development of antiviral therapeutics.

11646

VOL. 85, 2011

DCI IS REQUIRED FOR HCV REPLICATION AND PATHOGENESIS MATERIALS AND METHODS

Generation of a correlation network. Our clinical modeling efforts leveraged quantitative proteomic data generated from a previous study describing global proteomic alterations accompanying liver disease progression in patients with chronic hepatitis C virus (HCV) infections (10). Briefly, quantitative data obtained on a total of 1,641 proteins identified in human liver biopsy tissue specimens were filtered for those proteins showing a minimal 1.5-fold change (P ⬍ 0.05) in at least 6 of 15 patients represented in the prior study. We created a correlation network on these data by repeating the analysis we had previously described in the in vitro cell culture model (11). Briefly, we removed abundance values below 1.5 (corresponding to a similar threshold used previously), calculated correlation between abundance profiles of all pairs of proteins, and then calculated correlation between all pairs of proteins. The correlation values are based on comparing the protein abundance profiles, and we filtered these to keep only comparisons in which the two proteins being compared were observed in the same 6 or more patients. We then filtered the correlation matrix to include correlation values above 0.9 (highly correlated profiles). Protein-protein interactions from the Human Interactome were integrated into the resulting network, considering interactions between observed proteins. It is important to note that the correlation-based edges in this network are being used only for topological analysis and may not represent true interactions between the proteins. Topology was calculated in the resulting network using the igraph library in R, and proteins were ranked on the basis of their betweenness (a topological measure that identifies highly central proteins in the networks that restrict flow through the network) to identify bottlenecks. Cell culture. Huh7 cells were routinely cultured in Dulbecco modified Eagle medium (DMEM) (Invitrogen) containing 10% fetal bovine serum, penicillin (100 IU/ml) (Invitrogen), and streptomycin (100 ␮g/ml) (Invitrogen). To generate stable short hairpin RNA (shRNA)-expressing cell lines, SMARTvector lentiviruses expressing shRNA targeting dodecenoyl coenzyme A delta isomerase (DCI) or a scrambled target sequence were obtained from Thermo Scientific. The following sequences in DCI were targeted: DCI-1 (5⬘-CATTCCAGACCA TGCTCGA-3⬘), DCI-2 (5⬘-CCAGGGAGGTCTTAAACAA-3⬘), and DCI-3 (5⬘AGGTACTGCATAGGACTCA-3⬘). Huh7 cells were transduced at a multiplicity of infection (MOI) of 10 with shRNA-expressing lentiviruses and Polybrene (10 ␮g/ml). After 48-h incubation with lentiviruses, fresh medium containing puromycin (2 ␮g/ml) (Sigma) was applied. After 3 weeks of drug selection, target protein knockdown was evaluated by Western blotting. Stable knockdown cell lines were routinely maintained in medium containing puromycin (2 ␮g/ml). HeLa S3 cells and Huh7.5 cells were also routinely cultured in DMEM containing 10% fetal bovine serum, penicillin (100 IU/ml), and streptomycin (100 ␮g/ml). BHK-21 cells were routinely cultured in minimal essential medium (MEM) (Invitrogen) containing 10% fetal bovine serum, 1% nonessential amino acids (Invitrogen), penicillin (100 IU/ml), and streptomycin (100 ␮g/ml). Western blotting. Western blotting was performed using anti-DCI antibody at 3 ␮g/ml (Abcam) with horseradish peroxidase-conjugated secondary donkey anti-mouse IgG diluted 1:10,000 (Jackson ImmunoLabs). Loading control blots were performed using horseradish peroxidase-conjugated anti-GAPDH (antibody against glyceraldehyde-3-phosphate dehydrogenase [GAPDH]) (0.5 ␮g/ml) (Abcam). Immunoreactivity on the Western blots were detected using ECL Plus Western blotting detection system (GE Biosciences). Quantitation of protein abundance was performed by densitometry using ImageJ software (National Institutes of Health) to determine DCI expression relative to wild-type Huh7 cell expression. Cloning and expression of shRNA-resistant DCI construct. A DCI clone (shRNA-resistant DCI-1 [SRDCI-1]) containing the following silent mutations in the DCI-1 target site was synthesized by Blue Heron: 5⬘-TATCCCGGATCAC GCGCGC-3⬘. This clone was amplified using the following primers: forward primer, 5⬘-GC GCGAAGCTTATGGCGCTGGTGGCTTCTGTGCCA-3⬘; reverse primer, 5⬘GCGGGATCCTTTAAGTTGAAAAATACC-3⬘. The resulting PCR product was digested with HindIII and BamHI (New England BioLabs), ligated into HindIII/BamHI-digested pcDNA3.1 (⫹)/Hygro (Hygro stands for hygromycin) using T4 DNA ligase (New England BioLabs), and transformed into electrocompetent Escherichia coli DH5␣ (Invitrogen) using a GenePulser (Bio-Rad). Intact, properly oriented clones were selected and confirmed by DNA sequencing. Huh7/DCI-1 stable knockdown cells were transfected with pSRDCI-1 or empty pcDNA3.1 vector using FuGENE 6 transfection reagent (Roche) per the manufacturer’s instructions. After 48 h, hygromycin B (50 ␮g/ml) was applied to

11647

the cells, and stably transfected cells were selected by 3 weeks of drug treatment. The cells were routinely maintained with drug selection. Virus propagation. HCV genotype 2a/SJ virus stocks were generously provided by Michael Gale at the University of Washington. Working stocks were generated by infecting Huh7.5 cells, harvesting supernatant after 5 or 6 days, and concentrating supernatant in Centricon-70 concentrators (Millipore). Concentrated virus stocks were titrated by a focus-forming assay. Briefly, 100-␮l aliquots of 10-fold serial dilutions of virus stock were used to infect Huh7.5 cells previously seeded in triplicate in 24-well plates at a density of 1 ⫻ 105 cells/well. Following 1-h adsorption time, 0.5 ml of fresh medium was added to each well, and cultures were incubated at 37°C and 5% CO2. After 48 to 72 h of incubation, cells were fixed for 30 min in 4% paraformaldehyde. The cells were washed in phosphate-buffered saline (PBS) with 1 mM glycine, permeabilized for 15 min in PBS containing 0.2% Triton X-100, and blocked for 15 min with PBS containing 10% donor equine serum. The cells were incubated overnight with pooled HCV patient sera diluted 1:1,000 at 4°C with gentle rocking. The cells were washed and incubated with horseradish peroxidase-conjugated donkey anti-human IgG secondary antibody (Jackson ImmunoLabs) for 1.5 h at room temperature with gentle rocking. The cells were washed, and focus-forming assays were developed with VIP ImmPACT peroxidase substrate (Vector Laboratories). The virus titer was determined by calculating the number of focus-forming units per milliliter of virus stock. Plasmid pT7M bearing a full-length infectious molecular clone of poliovirus type I/Mahoney was generously provided by Vincent Racaniello at Columbia University. pT7M was linearized with EcoRI (New England BioLabs), and viral RNA was transcribed in vitro using the Ambion MEGAscript T7 kit (Applied Biosystems). Viral RNA was transfected into HeLa S3 cells using DEAE-dextran (Sigma). Supernatant was harvested after total cytopathic effect was apparent on the plate, and virus was frozen and thawed three times and then centrifuged to pellet debris. Clarified supernatants were titrated on HeLa S3 cells by plaque assay. The virus titer was determined by calculating the number of PFU per ml of virus stock. Dengue virus serotype 2/NGC was also generously provided by Michael Gale at the University of Washington. Stocks were grown at 33°C in C6/36 cells. Supernatant was harvested after the appearance of cytopathic effect, clarified by centrifugation to pellet cell debris, and concentrated using Centricon-70 concentrators (Millipore). Concentrated stocks were titrated on BHK-21 cells by plaque assay. Growth curves were set up in general by plating 2 ⫻ 104 cells per well in 12-well tissue culture plates prior to inoculation. The cells were then infected with 100 ␮l of inoculum at a multiplicity of infection of 1. Following infection, the inoculum was washed off in 1 ml of infection medium, and the cells were fed with 1 ml of fresh medium per well. Growth curves were titrated by harvesting and freezing cell culture supernatants, followed by low-speed centrifugation to remove particulate matter. Quantitation of viral RNA. RNA was harvested from infected cells using TRIzol reagent (Invitrogen) per the manufacturer’s instructions. RNA was reverse transcribed using the Quantitect kit (Qiagen), and viral genomes were quantitated relative to 18S RNA by TaqMan quantitative real-time reverse transcription-PCR (RT-PCR). For quantifying positive-sense single-stranded viral genomes, we used the following primers: forward primer, 5⬘-TCCCGGCAA TTCCGGTGTAC-3⬘; reverse primer, 5⬘-TCCCGGAGAGCCATAGTG-3⬘. We used the following probe: 6FAM-5⬘-TCT GCG GAA CCG GTG-3⬘-MGBNFQ (6FAM stands for 6-carboxyfluorescein, and MGBNFQ stands for molecular groove-binding nonfluorescence quencher). We normalized the values using primer-probe sets for the endogenous control human 18S rRNA per the manufacturer’s specifications (Applied Biosystems). All reactions were performed using standard TaqMan protocols and reagents supplied by the manufacturer and run on an ABI 7500 real-time PCR instrument (Applied Biosystems). Pharmacologic inhibition of fatty acid oxidation. Etomoxir (Sigma) was dissolved in dimethyl sulfoxide (DMSO) (Sigma) to generate a 10 mM stock solution. We performed a titration to determine the dose at which etomoxir was the most inhibitory to HCV and least cytotoxic. Twenty-four hours prior to infection, etomoxir was added to the culture medium at a final concentration of 100 ␮M. Mock-treated cells were treated with an equivalent volume of DMSO. The cells were treated again at the time of infection. Statistical analysis. All data are expressed as mean ⫾ standard error of the mean. Differences between control and experimental groups were assessed by one-way analysis of variance (ANOVA), and means were compared by Student’s t test with Bonferroni’s multiple test correction using JMP 9 (SAS, Inc.). P values less than 0.05 were considered significant. Metabolomics data collection and statistical and functional analyses. Huh7, Huh7/NT, Huh7/DCI-1, and Huh7/DCI-3 cells were washed in ammonium bi-

11648

RASMUSSEN ET AL.

carbonate (Sigma), pelleted by centrifugation, and snap-frozen. Metabolites and lipids were extracted from six replicate cell pellets using chloroform-methanol (2:1 [vol/vol)], and the water- and lipid-soluble layers were isolated by centrifugation and then dried in vacuo. Metabolites in the dried water-soluble layers were derivatized using methyoxyamine and N-methyl-N-(trimethylsilyl)trifluoroacetamide prior to gas chromatography-mass spectrometry (GC-MS) analysis in duplicate. GC-MS data were processed using the software program MetaboliteDetector to identify GC-MS features (15), align these chromatographically across multiple data sets, and then match these to the Fiehn Metabolomics RTL Library included with the Agilent GC-MS instrument (28). This software program also generates abundance for each GC-MS peak. Lipids in the dried lipid-soluble layers were reconstituted in 200 ␮l of isopropanol and analyzed by liquid chromatography (LC)-MS. A small trapping column (180 ␮m by 2 cm) packed with reversed-phase particles (Symmetry C18 column; 5 ␮m; Waters, Milford, MA) was used prior to the analytical column for fast loading of lipid samples (within 1.5 min at a flow rate of 10 ␮l/min), followed by washing of the column-bound lipids to remove chemical impurities and nonlipid sample components. Reconstituted lipids (4 ␮l) were loaded onto the trapping column under the following isocratic conditions: 93% acetonitrile– water (40:60) containing 10 mM ammonium acetate (solvent A) and 7% acetonitrile–isopropanol (10:90) containing 10 mM ammonium acetate (solvent B). The lipids retained on the trapping column were then back-flushed to the analytical column using 10% solvent B at a flow rate of 1 ␮l/min. The analytical column (150 ␮m by 20 cm) was slurry packed (12) with 1.8-␮m particles (HSS T3; Waters) and maintained at 40°C in a column oven. Gradient elution was performed as follows: initial conditions, 10% solvent B; 0 to 2 min, ramp to 30% solvent B; 2 to 10 min, ramp to 40% solvent B; 10 to 20 min, ramp to 55% solvent B; 20 to 40 min, ramp to 60% solvent B; 40 to 70 min, ramp to 99.5% solvent B; and 70 to 90 min, hold at 99.5% solvent B. The LC system was interfaced to an LTQ-Orbitrap mass spectrometer (Thermo Scientific, San Jose, CA) using a chemically etched electrospray ionization (ESI) emitter (25), and the ESI emitter and MS inlet capillary potentials were ⫺2.2 kV and ⫺12 V, respectively. Datadependent tandem MS (MS-MS) (collision-induced dissociation [CID]) scan events (top five ions) were performed in the ion trap using a normalized collision energy of 35% and were set with a maximum charge state of 2 and an isolation width of 2 m/z units. An activation Q value of 0.18 was used, and dynamic exclusion in the ion trap was enabled as follows: repeat count of 2, repeat duration of 30 s, exclusion list size of 200, and exclusion duration of 60 s. The full scan mass range for negative ESI mode was 200 to 2,000 m/z, respectively. One biological replicate from each cell type was analyzed separately in order to determine the appropriate loading of the LC column, and then the remaining 5 biological replicates were analyzed in random order. The PRISM Data Analysis system (26), a series of software tools freely available at http://ncrr.pnl.gov/software/ and developed in-house, was used to process and analyze the LC-MS lipid data. The first step involved deisotoping of the raw MS data to give the monoisotopic mass, charge state, and intensity of the major peaks in each mass spectrum using Decon2LS (18). The data were next examined in a two-dimensional (2-D) fashion using MultiAlign to identify groups of mass spectral peaks that were observed in sequential spectra using an algorithm (36) that computes a Euclidean distance in n-dimensional space for combinations of peaks. Each group, generally ascribed to one detected species and referred to as a “feature,” has a median monoisotopic mass, central normalized elution time (NET), and abundance estimate computed by summing the intensities of the MS peaks that comprise the entire LC-MS feature. LC-MS features were then chromatographically aligned across all replicates for each sample using the LCMSWARP algorithm (19) in MultiAlign, and the identities of detected lipids were determined by searching entries in the Lipid Maps database within a search tolerance of ⫾10 ppm mass tolerance. Tentative fatty acid identifications were confirmed by comparison with the retention times or ranges of authentic fatty acid standards. The metabolite and fatty acid data were loaded into the software package DAnTE (39), the feature abundances were transformed to log2 scale, and the data were normalized using mean centering. ANOVA was then performed to identify metabolites and fatty acids that differed quantitatively with q values (adjusted P values found using an optimized false discovery rate approach) of ⬍0.05. To assist in visualization and interpretation of the heat map of significantly different metabolites and fatty acids, we performed a z-score transformation of the abundances for each metabolite or fatty acid. For GC-MS-based data, a G-test was also performed to identify statistically significant differences in the data based on metabolite occurrence.

J. VIROL.

RESULTS Clinical data support a critical role for DCI in HCV pathogenesis. To explore the role of dodecenoyl coenzyme A (CoA) delta isomerase (DCI) as a key cellular protein involved in hepatitis C virus (HCV) pathogenesis in vivo and thus define a role for DCI in clinical disease pathology, we employed a computational strategy to model proteomic data generated from core needle liver biopsy specimens taken from 15 patients chronically infected with HCV (10). Network analysis of patterns of protein expression was used to identify points of restriction, bottlenecks, representing points of control for important processes related to disease progression. The abundance “profiles” of proteins, their abundance in each of the 15 patient samples, were used to determine patterns of similar expression using correlation with all other proteins identified. Briefly, we removed abundance values below 1.5 and calculated correlation between abundance profiles of all pairs of proteins excluding correlation values based on fewer than 6 comparisons where a single comparison is valid if both proteins were observed in the same patient sample. We then filtered to retain only highly correlated pairs of proteins (⬎0.9 correlation). These parameters are derived from a more-extensive characterization of network inference from proteomic data (J. E. McDermott et al., submitted for publication), which is not focused on DCI and describes similar network analysis in more-general terms. Protein-protein interactions from the Human Interactome were integrated into the resulting network to account for interactions between observed proteins. Topology was calculated in the resulting network, and proteins were ranked on the basis of their betweenness to identify bottlenecks that are predicted to be key regulators of metabolic reprogramming in chronically infected patients (Fig. 1A). Betweenness is a topological measure that identifies highly central proteins in the networks that restrict flow through the network. Bottlenecks are predicted to be more important to the functional processes underlying the network because of their position in the network (34). Consistent with our in vitro modeling results, DCI was the second highest-ranked bottleneck after histidine triad nucleotide-binding protein 2 (HINT2) in the clinical proteomic network (Fig. 1B), indicating a physiologically relevant role for DCI during HCV infection in vivo. We chose to focus on DCI because of its role in lipid metabolism, which was highlighted in our previous work (11), and the fact that it is one of a small number of bottlenecks shared between clinical and cell culture networks (McDermott et al., submitted). This demonstrates the utility of a systems biology approach for identifying translationally relevant host factors based both on in vitro experimental models and patient specimens. Our hypothesis resulting from this analysis is that DCI represents an important point of control for metabolic processes critical for HCV replication and pathogenesis. Knockdown of DCI inhibits HCV RNA replication in cultured hepatoma cells. To better understand DCI’s role in the HCV replication cycle, we generated Huh7 human hepatoma cells stably expressing shRNAs targeting DCI. Lentiviruses expressing three distinct shRNA constructs targeting different regions of the DCI transcript were used to transduce Huh7 cells. After puromycin selection for transduced Huh7 cells

VOL. 85, 2011

DCI IS REQUIRED FOR HCV REPLICATION AND PATHOGENESIS

FIG. 1. Correlation network showing metabolic bottlenecks identified in proteomic data from liver biopsy specimens of patients with chronic hepatitis C virus (HCV) infections by computational modeling efforts, integrating observed proteins with known protein-protein interactions. (A) The full network of 103 connected proteins. Proteins are colored by their rank according to betweenness centrality; red is highest, and blue is lowest. Edges represent correlation associations greater than 0.9 or known protein-protein interactions. Colored ovals represent regions of significantly (P value ⬍ 0.001) enriched biological process functions in the network. The rectangle represents the subnetwork shown in panel B. (B) Subnetwork showing the two proteins with highest betweenness centrality, dodecenoyl coenzyme A delta isomerase (DCI) and histidine triad nucleotide-binding protein 2 (HINT2).

expressing the shRNA, DCI knockdown was evaluated by Western blotting and relative quantitation was performed using ImageJ software, which demonstrated that Huh7/DCI-1 cells express 18% of the amount of DCI expressed by wild-type Huh7 cells (Fig. 2A). The DCI-1 construct demonstrated the greatest reduction in DCI protein levels, so Huh7/DCI-1 cells were used for all subsequent experiments. The DCI-3 construct did not result in observable knockdown and was used as an additional control. Infection of the Huh7/DCI-1 cell line with the cell culture HCV genotype 2a strain SJ resulted in a substantial reduction in virus production compared to infection of wild-type cells, cells expressing a scrambled nontargeting control shRNA, and cells expressing the DCI-3 shRNA (Fig. 2B). Huh7/DCI-1 cells also supported lower levels of HCV RNA replication compared to the controls. Quantitative PCR for viral genomes demonstrated a substantial reduction in the DCI knockdown cells compared to the controls (Fig. 2C). Huh7/DCI-1 cells did not support increases in positive-sense single-stranded HCV genomes, indicating that viral replication is blocked at or prior to viral RNA production in DCI-1 knockdown cells. Pharmacologic inhibition of fatty acid oxidation blocks HCV replication. All fatty acid ␤-oxidation in cells can be abrogated

11649

FIG. 2. DCI knockdown blocks replication in human hepatoma cells. (A) Western blot showing DCI protein expression in Huh7 cells bearing stable short hairpin RNA (shRNA)-expressing lentiviral constructs. Western blotting for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was performed as a loading control. (B) Huh7 cells expressing shRNAs were infected with HCV genotype 2a (HCV2a)/SJ at a multiplicity of infection (MOI) of 1. Virus was titrated at 1, 3, and 5 days postinfection by a focus-forming assay. Values that are significantly different are indicated by brackets and asterisks as follows: *, P ⫽ 0.00315; **, P ⫽ 0.0025; ***, P ⫽ 0.000067. (C) DCI knockdown inhibits replication of viral RNA. Viral RNA levels were quantified by TaqMan real-time reverse transcription-PCR (RT-PCR) relative to cellular 18S rRNA. Values that are significantly different are indicated by brackets and asterisks as follows: *, P ⬍ 0.005; **, P ⬍ 0.0002; ***, P ⬍ 0.00007.

by treatment with etomoxir, an irreversible inhibitor of carnitine palmitoyltransferase 1 (CPT-1). CPT-1 is an integral membrane protein localized in the outer mitochondrial membrane and transports fatty acids to the mitochondrion. Blocking CPT-1 activity with etomoxir effectively blocks the transport of any fatty acid substrates to the mitochondrial matrix and thus inhibits activity of the ␤-oxidation enzymes, including DCI. Huh7 cells treated with etomoxir 24 h prior to and over the course of infection did not support production of infectious HCV compared to mock-treated cells (Fig. 3). These data demonstrate that the mitochondrial fatty acid oxidation pathway is required for HCV replication.

11650

RASMUSSEN ET AL.

J. VIROL.

FIG. 3. Pharmacologic inhibition of fatty acid oxidation blocks HCV replication. Huh7 cells mock treated or treated with 100 ␮M etomoxir were infected with HCV2a/SJ at an MOI of 1. Virus was titrated by a focus-forming assay on Huh7.5 cells. FFU, focus-forming unit.

DCI is not required for the replication of other positivesense, single-stranded RNA viruses. To determine whether DCI supports HCV replication by a mechanism that generally augments the assembly of viral RNA replication complex assembly, we tested the effect of DCI knockdown on other viruses with similar biological properties. Poliovirus is an enterovirus in the family Picornaviridae, which like flaviviruses have positive-sense, single-stranded RNA genomes. Like HCV, poliovirus genomes replicate on specialized vesicular structures derived from the endoplasmic reticulum (41). Furthermore, both enteroviruses and flaviviruses require a specialized lipid microenvironment in the membranes on which RNA replication complexes assemble (17). Therefore, we hypothesized that poliovirus might similarly be inhibited in DCI knockdown cells. However, virus production in Huh7/DCI-1 cells infected with poliovirus type 1/Mahoney (P1/M) was similar to that seen in P1/M-infected Huh7, Huh7/NT, and Huh7/DCI-3 control cells (Fig. 4A). We then investigated whether or not DCI is required for replication of dengue virus serotype 2/New Guinea C (DENV2/NGC), which also cannot replicate in cells treated with etomoxir (14) and like HCV is a member of the family Flaviviridae. DENV2/NGC production was similar in Huh7/ DCI-1 knockdown cells to infected controls (Fig. 4B), indicating that DCI is not required to support replication of all flaviviruses, including ones that require ␤-oxidation of fatty acids to replicate. While these data do not demonstrate a measurable effect of DCI on poliovirus or dengue virus replication, they indicate that the requirement for DCI is highly specific to HCV infection. It also indicates that Huh7/DCI-1 cells are competent to support replication of viruses that are not dependent on DCI activity, demonstrating the specificity of DCI’s mechanism of action for HCV infection. Restoration of DCI in shRNA-expressing knockdown cells rescues HCV propagation. To determine the specificity of DCI in regulating HCV replication, we generated Huh7/DCI-1 cells stably transfected with a clone expressing a shRNA-resistant allele of DCI (SRDCI-1). SRDCI-1 contains silent mutations in the region of the DCI transcript targeted by the DCI-1 shRNA. This expresses a DCI protein with no changes in the amino acid sequence but contains multiple nucleotide substitutions in the mRNA that prevent recognition and consequent gene silencing by the mature small interfering RNA (siRNA) (Fig. 5A). Infection of Huh7/DCI-1 cells expressing SRDCI-1

FIG. 4. DCI deficiency does not block growth of other RNA viruses. (A) Cells were infected with poliovirus type 1/Mahoney strain (P1/M) at an MOI of 1. P1/M growth was titrated by plaque assay on HeLa S3 cells. (B) Cells were infected with dengue virus serotype 2/New Guinea C strain (DENV2/NGC) at an MOI of 1. DENV2/NGC growth was titrated by a plaque assay on BHK-21 cells.

restored HCV growth compared to Huh7/DCI-1 cells or Huh7/ DCI-1 cells bearing an empty vector pcDNA3.1 control (Fig. 5B). This clearly demonstrates that inhibition of HCV replication is specific to DCI activity and not the result of off-target gene silencing or alterations induced by constitutive shRNA expression. DCI deficiency causes alterations in the cellular metabolome. Our previous computational models predicted that DCI is an essential bottleneck protein regulating overall reprogramming of cellular metabolism to create a favorable environment for HCV replication. To gain a preliminary understanding of the effect of DCI on global cellular metabolism, we performed gas chromatography-mass spectrometry (GC-MS) analysis of aqueous metabolites in uninfected Huh7/DCI-1 cells and the associated controls. Using ANOVA (q ⬍ 0.05), we identified 37 metabolites with significantly altered abundance in Huh7/ DCI-1 cells (Fig. 6A; see Table S1 in the supplemental material). These data indicate that DCI has pleiotropic effects on a variety of metabolic processes, including biogenesis and homeostasis of amino acids, nucleotides, vitamins, and sugars. We further identified 64 metabolites with differences based on the presence or absence in the Huh7/DCI-1 cells compared to the controls by the G-test (P ⬍ 0.05) (Table S1), confirming that cells lacking DCI have a distinct metabolic profile compared to other cell lines expressing DCI that are permissive to HCV infection. We also analyzed the cellular lipidome to determine the

VOL. 85, 2011

DCI IS REQUIRED FOR HCV REPLICATION AND PATHOGENESIS

FIG. 5. HCV replication is restored in DCI-deficient cells when complemented in trans by shRNA-resistant DCI-1 (SRDCI-1). (A) Western blot demonstrating DCI expression in Huh7 cells bearing stable shRNA-expressing lentiviruses with shRNA-resistant DCI allele SRDCI-1 added in trans. (B) Growth of HCV in cells expressing SRDCI-1 infected at an MOI of 1was determined by a focus-forming assay on Huh7.5 cells.

effect that DCI knockdown would have on fatty acid composition, since its substrates are long-chain mono- and polyunsaturated fatty acids. Lipid features detected in the mass range of free fatty acids were matched to entries in the Lipid Maps database and subsequently identified on the basis of comparison of retention times to those of free fatty acid standards. This approach identified 26 differentially abundant fatty acid species from Huh7/DCI-1 cells compared to the controls (ANOVA, q ⬍ 0.05). Notably, Huh7/DCI-1 cells showed dramatic increases in medium- to long-chain monoand polyunsaturated fatty acids (Fig. 6B). This is consistent with our hypothesis that without DCI present to degrade polyunsaturated fatty acids, these molecules accumulate in the Huh7/DCI-1 knockdown cells. DISCUSSION We used a combination of gene silencing and pharmacologic approaches to validate our previous computational modeling predictions that dodecenoyl coenzyme A (CoA) delta isomerase (DCI)-mediated mitochondrial fatty acid oxidation plays a critical role in hepatitis C virus (HCV) replication. Both approaches confirmed the importance of DCI to the HCV replication cycle, demonstrating the value of systems biology as a means of gaining insights into HCV pathogenesis that have been overlooked by conventional approaches. DCI knockdown completely blocked viral RNA production, indicating that its mechanism of action in augmenting HCV replication occurs at or before assembly of RNA replication complexes and initiation of viral transcription by the HCV RNA-dependent RNA polymerase (RdRp). Because of DCI’s essential role in the catabolism of long-chain fatty acids and its initial identification

11651

by a model built using proteomic and lipidomic data, we propose that DCI exerts its effects on HCV replication by modulating lipid content in the cell. Alterations in lipid metabolism have long been observed in both experimental and clinical HCV infection. Specific lipid species undergo changes in composition and abundance that facilitate molecular interactions required for HCV infection. Patients with chronic HCV develop steatosis and alterations in serum lipid levels, particularly as disease progresses. Though DCI has previously not been implicated in mediating these changes, mitochondrial ␤-oxidation could be involved in a number of cellular processes required for HCV replication. Numerous studies have observed changes in the lipidome of infected cells corresponding with known interactions between replicating HCV and multiple host lipids or lipid-related machinery. Lipid rafts rich in sphingomyelin are required for the assembly of RNA replication complexes and activity of the HCV RdRp (1, 51). Lipid droplets are specialized organelles designed for the storage of neutral lipids, which attach to HCV core protein and facilitate assembly of viral replication complexes and viral particles (35). These lipid droplets are attached to membranes where RNA replication complexes assemble. Both the induction of lipid droplet formation and the substantial organelle and membrane remodeling required for RNA replication and virion assembly at these sites necessitate substantial reprogramming of cellular lipid biosynthetic and metabolic pathways, all of which may be mediated by DCI. However, despite the wealth of information concerning the role of lipids in HCV replication and the major role of fatty acid oxidation in both meeting the cellular energetic requirements and generating building blocks for new lipid species in lipogenesis, conventional approaches have never implicated this pathway in playing an essential role in HCV replication. Without employing a systems-level approach, fatty acid oxidation would not have been identified as a key pathway regulating metabolic reprogramming required for HCV replication and pathogenesis. Activating fatty acid oxidation pathways seems contrary to numerous reports that lipogenesis, including fatty acid synthesis, is required at virtually all stages of HCV replication. DCI may be required to break down more-complex lipid species to favor synthesis of other lipid species facilitating HCV infection. Some of these lipid species might be membrane phospholipids, such as phosphatidylinositol-4-phosphate, which nucleates assembly of RNA replication complexes on membrane surfaces (17). Additionally, lipidation of host and viral proteins is known to play an essential role in HCV infection. For example, geranylgeranylation of host proteins is required for HCV RNA replication (24, 50, 53), and the viral core and NS4B proteins are palmitoylated to facilitate interactions with membranes and other proteins required to assemble functional replication complexes (32, 54). By mediating the degradation of lipids not required for HCV replication, DCI generates energy and materials for the biosynthesis of other lipid molecules required during the viral life cycle. Another possibility is that DCI specifically reduces the quantity of polyunsaturated fatty acids (PUFA) in the cell, creating a favorable environment for HCV replication. DCI specifically isomerizes mono- and polyunsaturated fatty acids with cis double bonds at odd-numbered carbon atoms into their 2-trans-

11652

RASMUSSEN ET AL.

J. VIROL.

FIG. 6. DCI deficiency causes pronounced alterations in the cellular metabolome. (A) Aqueous metabolites demonstrating significant differences in abundance by ANOVA (q ⬍ 0.05). (B) Free fatty acids (FA) demonstrating significant differences in abundance by ANOVA (q ⬍ 0.05).

VOL. 85, 2011

DCI IS REQUIRED FOR HCV REPLICATION AND PATHOGENESIS

enoyl-CoA forms (20, 21). HCV RNA replication is inhibited specifically by accumulation of polyunsaturated fatty acids normally degraded by DCI (30). A characteristic feature of fasting DCI knockout mice is hepatic accumulation of unsaturated long-chain fatty acids (21). DCI deficiency results in the accumulation of PUFA in Huh7/DCI-1 cells, subsequently inhibiting HCV replication, possibly by altering lipid rafts or membrane structure sufficiently as to prevent the assembly and function of RNA replication complexes. Our metabolomic data provide insight into the molecular basis for HCV inhibition in DCI-deficient Huh7 cells. Consistent with DCI’s function in channeling unsaturated fatty acids into the mitochondrial ␤-oxidation pathway, DCI knockdown results in an increase in abundance of the monounsaturated DCI substrate oleic acid. Exogenous oleic acid enhances RNA synthesis by a full-length HCV genotype 1b (HCV1b) replicon (24), although the molecular mechanism for this enhancement and the impact of DCI-mediated catabolism during oleic acid supplementation remain undetermined. Fatty acid oxidation may exert pleiotropic effects on the HCV life cycle, including potentially significant roles in enhanced energy production and modulating the cellular composition of lipid species with proor antiviral effects. Either one of these potentially significant roles could influence the global metabolic reprogramming known to occur during HCV infection (11). During conditions of impaired lipid catabolism, these effects would presumably be abrogated, forcing the cell to rely on alternative energy sources. Indeed, the observed aqueous metabolic profile further revealed that Huh7/DCI-1 cells exhibited increases in numerous amino acids and exemplary intermediates of urea synthesis. These metabolic alterations are consistent with those observed in DCI knockout mice (21) and suggest an increased dependence on amino acids for energy production. Alternatively, increasing evidence indicates that fatty acid remodeling substantially impacts the composition, integrity, and function of biological membranes (22, 27, 29, 42, 43). For example, the incorporation of PUFA, and to a lesser extent oleic acid, into phosphatidylethanolamine (PE) was recently shown to modify the biophysical organization and fluidity of lipid rafts (43). This may be due to unfavorable interactions between the disordered acyl chain of PE with sphingomyelin, resulting in PE-rich nonraft microdomains that can impact protein conformation (22). Such membrane reorganizations adversely impact the localization of molecules that need to be clustered in close physical proximity, such as immunological synapse proteins required for antigen presentation and T cell activation (27). The relative contribution of DCI deficiency on the composition of lipid species comprising the various structural entities supporting HCV replication remains an important unanswered question. We are actively addressing this question with additional metabolomic studies to better understand both the full breadth of the cellular lipidome in Huh7/ DCI-1 cells as well as the influence of DCI deficiency on the metabolome in the context of HCV infection. As the phenotypic impact of impaired lipid catabolism appears to be most prominent under conditions of metabolic challenge, such as dietary fasting, it will be crucial to perform additional studies aimed at characterizing global host, gene, protein, and metabolite changes occurring in DCI-deficient cells during the cellular stress induced during HCV infection.

11653

Comparative analyses of the molecular changes occurring in purified subcellular structures, including lipid rafts, lipid droplets, viral RNA replication complexes, mitochondria, and mitochondrion-associated membranes (MAM) during HCV infection of wild-type versus DCI-deficient Huh7 cells will help to refine our understanding of how metabolic reprogramming influences current models of HCV infection and pathogenesis. Further investigation of the importance of fatty acid oxidation in vivo may lead to the development of novel HCV therapeutics targeting host factors rather than viral proteins. The HCV therapies now in development primarily target viral enzymes such as proteases and polymerases and must be used in combination with PEGylated alpha interferon and ribavirin therapy. These regimens are not tolerated well by patients and can select for drug-resistant viral mutants. Targeting host factors such as DCI is an approach that could potentially eliminate the need for multidrug regimens with severe side effects in favor of a treatment that does not select for drug-resistant viruses, may be used as monotherapy, and completely blocks HCV replication. Together, these studies demonstrate our use of systems biology to identify and evaluate a novel host factor that both advances our understanding of the biology of HCV infection and provides an attractive potential new target for treating this highly prevalent and deadly disease. ACKNOWLEDGMENTS We thank V. Racaniello (Columbia University, New York, NY) for providing the infectious clone of poliovirus type 1/Mahoney strain and M. Gale (University of Washington, Seattle, WA) for providing stocks of HCV genotype 2a/SJ and dengue virus serotype 2/New Guinea C strain. This study was supported by National Institute on Drug Abuse grant 1P30DA01562501 to M.G.K. Portions of this work were performed at the Environmental Molecular Sciences Laboratory, a national scientific user facility located at the Pacific Northwest National Laboratory (PNNL) and sponsored by the U.S. Department of Energy (DOE) Office of Biological and Environmental Research. PNNL is operated by Battelle for the DOE under contract DE-AC06-76RLO-1830. REFERENCES 1. Amemiya, F., et al. 2008. Targeting lipid metabolism in the treatment of hepatitis C virus infection. J. Infect. Dis. 197:361–370. 2. Benga, W. J., et al. 2010. Apolipoprotein E interacts with hepatitis C virus nonstructural protein 5A and determines assembly of infectious particles. Hepatology 51:43–53. 3. Berger, K. L., et al. 2009. Roles for endocytic trafficking and phosphatidylinositol 4-kinase III alpha in hepatitis C virus replication. Proc. Natl. Acad. Sci. U. S. A. 106:7577–7582. 4. Berger, K. L., and G. Randall. 2009. Potential roles for cellular cofactors in hepatitis C virus replication complex formation. Commun. Integr. Biol. 2:471–473. 5. Borawski, J., et al. 2009. Class III phosphatidylinositol 4-kinase alpha and beta are novel host factor regulators of hepatitis C virus replication. J. Virol. 83:10058–10074. 6. Boulant, S., et al. 2008. Hepatitis C virus core protein induces lipid droplet redistribution in a microtubule- and dynein-dependent manner. Traffic 9:1268–1282. 7. Carpenter, K., R. J. Pollitt, and B. Middleton. 1992. Human liver long-chain 3-hydroxyacyl-coenzyme A dehydrogenase is a multifunctional membranebound beta-oxidation enzyme of mitochondria. Biochem. Biophys. Res. Commun. 183:443–448. 8. Coller, K. E., et al. 2009. RNA interference and single particle tracking analysis of hepatitis C virus endocytosis. PLoS Pathog. 5:e1000702. 9. Cun, W., J. Jiang, and G. Luo. 2010. The C-terminal alpha-helix domain of apolipoprotein E is required for interaction with nonstructural protein 5A and assembly of hepatitis C virus. J. Virol. 84:11532–11541. 10. Diamond, D. L., et al. 2007. Proteomic profiling of human liver biopsies: hepatitis C virus-induced fibrosis and mitochondrial dysfunction. Hepatology 46:649–657. 11. Diamond, D. L., et al. 2010. Temporal proteome and lipidome profiles reveal

11654

12.

13.

14. 15.

16. 17. 18.

19.

20.

21.

22.

23. 24.

25.

26. 27.

28.

29.

30. 31.

32. 33.

RASMUSSEN ET AL.

hepatitis C virus-associated reprogramming of hepatocellular metabolism and bioenergetics. PLoS Pathog. 6:e1000719. Ding, J., et al. 2008. Application of the accurate mass and time tag approach in studies of the human blood lipidome. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 871:243–252. Haid, S., T. Pietschmann, and E. I. Pecheur. 2009. Low pH-dependent hepatitis C virus membrane fusion depends on E2 integrity, target lipid composition, and density of virus particles. J. Biol. Chem. 284:17657–17667. Heaton, N. S., and G. Randall. 2010. Dengue virus-induced autophagy regulates lipid metabolism. Cell Host Microbe 8:422–432. Hiller, K., et al. 2009. MetaboliteDetector: comprehensive analysis tool for targeted and nontargeted GC/MS based metabolome analysis. Anal. Chem. 81:3429–3439. Hishiki, T., et al. 2010. Infectivity of hepatitis C virus is influenced by association with apolipoprotein E isoforms. J. Virol. 84:12048–12057. Hsu, N. Y., et al. 2010. Viral reorganization of the secretory pathway generates distinct organelles for RNA replication. Cell 141:799–811. Jaitly, N., et al. 2009. Decon2LS: an open-source software package for automated processing and visualization of high resolution mass spectrometry data. BMC Bioinformatics 10:87. Jaitly, N., et al. 2006. Robust algorithm for alignment of liquid chromatography-mass spectrometry analyses in an accurate mass and time tag data analysis pipeline. Anal. Chem. 78:7397–7409. Janssen, U., T. Fink, P. Lichter, and W. Stoffel. 1994. Human mitochondrial 3,2-trans-enoyl-CoA isomerase (DCI): gene structure and localization to chromosome 16p13.3. Genomics 23:223–228. Janssen, U., and W. Stoffel. 2002. Disruption of mitochondrial beta-oxidation of unsaturated fatty acids in the 3,2-trans-enoyl-CoA isomerase-deficient mouse. J. Biol. Chem. 277:19579–19584. Jenski, L. J., P. K. Nanda, P. Jiricko, and W. Stillwell. 2000. Docosahexaenoic acid-containing phosphatidylcholine affects the binding of monoclonal antibodies to purified Kb reconstituted into liposomes. Biochim. Biophys. Acta 1467:293–306. Jiang, J., and G. Luo. 2009. Apolipoprotein E but not B is required for the formation of infectious hepatitis C virus particles. J. Virol. 83:12680–12691. Kapadia, S. B., and F. V. Chisari. 2005. Hepatitis C virus RNA replication is regulated by host geranylgeranylation and fatty acids. Proc. Natl. Acad. Sci. U. S. A. 102:2561–2566. Kelly, R. T., et al. 2006. Chemically etched open tubular and monolithic emitters for nanoelectrospray ionization mass spectrometry. Anal. Chem. 78:7796–7801. Kiebel, G. R., et al. 2006. PRISM: a data management system for highthroughput proteomics. Proteomics 6:1783–1790. Kim, W., et al. 2008. n-3 polyunsaturated fatty acids suppress the localization and activation of signaling proteins at the immunological synapse in murine CD4⫹ T cells by affecting lipid raft formation. J. Immunol. 181:6236–6243. Kind, T., G. Wohlgemuth, D. Y. Lee, Y. Lu, M. Palazoglu, S. Shahbaz, and O. Fiehn. 2009. FiehnLib: mass spectral and retention index libraries for metabolomics based on quadrupole and time-of-flight gas chromatography/ mass spectrometry. Anal. Chem. 81:10038–10048. Langelier, B., A. Linard, C. Bordat, M. Lavialle, and C. Heberden. 2010. Long chain-polyunsaturated fatty acids modulate membrane phospholipid composition and protein localization in lipid rafts of neural stem cell cultures. J. Cell. Biochem. 110:1356–1364. Leu, G. Z., T. Y. Lin, and J. T. Hsu. 2004. Anti-HCV activities of selective polyunsaturated fatty acids. Biochem. Biophys. Res. Commun. 318:275–280. Li, Q., et al. 2009. A genome-wide genetic screen for host factors required for hepatitis C virus propagation. Proc. Natl. Acad. Sci. U. S. A. 106:16410– 16415. Majeau, N., et al. 2009. Palmitoylation of hepatitis C virus core protein is important for virion production. J. Biol. Chem. 284:33915–33925. Mancone, C., et al. 2011. Hepatitis C virus production requires apolipopro-

J. VIROL.

34.

35. 36.

37.

38.

39. 40. 41.

42.

43.

44.

45. 46.

47.

48.

49. 50.

51. 52.

53.

54.

tein A-I and affects its association with nascent low-density lipoproteins. Gut 60:378–386. McDermott, J. E., M. Costa, D. Janszen, M. Singhal, and S. C. Tilton. 2010. Separating the drivers from the driven: integrative network and pathway approaches aid identification of disease biomarkers from high-throughput data. Dis. Markers 28:253–266. Miyanari, Y., et al. 2007. The lipid droplet is an important organelle for hepatitis C virus production. Nat. Cell Biol. 9:1089–1097. Monroe, M. E., et al. 2007. VIPER: an advanced software package to support high-throughput LC-MS peptide identification. Bioinformatics 23:2021– 2023. Owen, D. M., H. Huang, J. Ye, and M. Gale, Jr. 2009. Apolipoprotein E on hepatitis C virion facilitates infection through interaction with low-density lipoprotein receptor. Virology 394:99–108. Partanen, S. T., et al. 2004. The 1.3 A crystal structure of human mitochondrial Delta3-Delta2-enoyl-CoA isomerase shows a novel mode of binding for the fatty acyl group. J. Mol. Biol. 342:1197–1208. Polpitiya, A. D., et al. 2008. DAnTE: a statistical tool for quantitative analysis of -omics data. Bioinformatics 24:1556–1558. Randall, G., et al. 2007. Cellular cofactors affecting hepatitis C virus infection and replication. Proc. Natl. Acad. Sci. U. S. A. 104:12884–12889. Schlegel, A., T. H. Giddings, Jr., M. S. Ladinsky, and K. Kirkegaard. 1996. Cellular origin and ultrastructure of membranes induced during poliovirus infection. J. Virol. 70:6576–6588. Schley, P. D., D. N. Brindley, and C. J. Field. 2007. (n-3) PUFA alter raft lipid composition and decrease epidermal growth factor receptor levels in lipid rafts of human breast cancer cells. J. Nutr. 137:548–553. Shaikh, S. R., D. S. Locascio, S. P. Soni, S. R. Wassall, and W. Stillwell. 2009. Oleic- and docosahexaenoic acid-containing phosphatidylethanolamines differentially phase separate from sphingomyelin. Biochim. Biophys. Acta 1788:2421–2426. Shimizu, Y., et al. 2010. Lipoprotein lipase and hepatic triglyceride lipase reduce the infectivity of hepatitis C virus (HCV) through their catalytic activities on HCV-associated lipoproteins. Virology 407:152–159. Tai, A. W., et al. 2009. A functional genomic screen identifies cellular cofactors of hepatitis C virus replication. Cell Host Microbe 5:298–307. Takahashi, M., E. Watari, E. Shinya, T. Shimizu, and H. Takahashi. 2007. Suppression of virus replication via down-modulation of mitochondrial short chain enoyl-CoA hydratase in human glioblastoma cells. Antiviral Res. 75: 152–158. Takahashi, Y., Y. Hirata, Y. Burstein, and I. Listowsky. 1994. Delta 3, delta 2-enoyl-CoA isomerase is the protein that copurifies with human glutathione S-transferases from S-hexylglutathione affinity matrices. Biochem. J. 304: 849–852. Targett-Adams, P., S. Boulant, and J. McLauchlan. 2008. Visualization of double-stranded RNA in cells supporting hepatitis C virus RNA replication. J. Virol. 82:2182–2195. Vaillancourt, F. H., et al. 2009. Identification of a lipid kinase as a host factor involved in hepatitis C virus RNA replication. Virology 387:5–10. Wang, C., et al. 2005. Identification of FBL2 as a geranylgeranylated cellular protein required for hepatitis C virus RNA replication. Mol. Cell 18:425– 434. Weng, L., et al. 2010. Sphingomyelin activates hepatitis C virus RNA polymerase in a genotype-specific manner. J. Virol. 84:11761–11770. Wu, J. M., N. J. Skill, and M. A. Maluccio. 2010. Evidence of aberrant lipid metabolism in hepatitis C and hepatocellular carcinoma. HPB (Oxford) 12:625–636. Ye, J., et al. 2003. Disruption of hepatitis C virus RNA replication through inhibition of host protein geranylgeranylation. Proc. Natl. Acad. Sci. U. S. A. 100:15865–15870. Yu, G. Y., K. J. Lee, L. Gao, and M. M. Lai. 2006. Palmitoylation and polymerization of hepatitis C virus NS4B protein. J. Virol. 80:6013–6023.