A Cell Culture System for Distinguishing Hepatitis C Viruses with and without Liver Cancer-related Mutations in the Viral Core Gene

Abstract

Background and Aims

Although patients infected by genotype-1b hepatitis C virus (HCV) with Q⁷⁰ and/or M⁹¹ core gene mutations have an almost five-fold increased risk of developing hepatocellular carcinoma (HCC) and increased insulin resistance, the absence of a suitable experimental system has precluded direct experimentation on the effects of these mutations on cellular gene expression.

Methods

Huh-7 cells were treated long-term with human serum to induce differentiation and to produce a model system for testing high-risk and control HCV. For clinical validation, profiles of infected cells were compared to each other and to those of liver biopsies of patients with early-stage HCV-related cirrhosis followed prospectively for up to 23 years (n=216).

Results

Long-term culture in human serum produced growth-arrested, hepatocyte-like cells whose gene profile overlapped significantly with that of primary human hepatocytes. High-risk (Q⁷⁰/M⁹¹) and control (R⁷⁰/L⁹¹) viruses had dramatically different effects on gene expression of these cells. The high-risk virus enhanced expression of pathways associated with cancer and type II diabetes, while the control virus enhanced pathways associated with oxidative phosphorylation. Of special clinical relevance, the transcriptome of cells replicating the high-risk virus correlated significantly with an HCC high-risk profile in patients (Bonferroni-corrected P=0.03), whereas no such association was observed for non-HCC-related clinical outcomes.

Conclusions

The cell-based system allowed direct head-to-head comparison of HCV variants and provided experimental support for previous clinical data indicating an oncogenic effect of core gene mutations. This simple experimental system distinguished HCV variants and will enable future mechanistic analysis and exploration of interventional approaches.

Introduction

Hepatocellular carcinoma (HCC) is the second leading cause of site-specific cancer-related death worldwide [1] and the most rapidly increasing cause of cancer-related death in the United States (US) [2]. Chronic hepatitis C virus (HCV) infection is a leading risk factor for HCC in the US, Japan, and other parts of the world [3]. HCV will remain a major cause of HCC for the foreseeable future despite the introduction of highly effective direct-acting antiviral drugs that increase the percentage of patients who achieve a sustained virological response (SVR) [4]. Many barriers to HCV treatment exist, including the high percentage of patients who are unaware of their infection and the high cost of treatment. Moreover, for poorly understood reasons, successful HCV treatment does not eliminate the HCC risk [5]. Patients with advanced fibrosis/cirrhosis who achieve an SVR are advised to continue life-long screening for HCC [6]. Currently, many patients with HCV-related HCC are diagnosed at a late stage. Only 12% of patients with HCV-related HCC are diagnosed through surveillance in the US [7]. Methods for stratifying HCC risk would allow HCV treatment and HCC surveillance to be focused on the individuals most likely to benefit.

Data about factors modulating the risk of other types of cancer might provide clues about risk stratification for HCC. Approximately 12% of human cancers are associated with oncogenic viruses [8]. Experimental studies suggest that the HCV core gene has oncogenic potential. Expression of the core gene in transgenic mice leads to insulin resistance, hepatic steatosis, and liver cancer [9, 10]. In cell culture systems, the core gene causes cellular transformation and disruption of several growth control pathways [11].

In the case of human papillomavirus, the cancer risk varies by viral strain [12], demonstrating the importance of sequence- and strain-specific differences in viral oncogenesis. In the case of HCV, genotype-1b is reported to increase the HCC risk [13]. Among patients infected by genotype-1b HCV, those with virus harboring mutations in codons 70 and/or 91 in the core gene have the highest risk [14]. Patients infected with viruses carrying a mutation in codon 70 (Q⁷⁰) and/or codon 91 (M⁹¹) had a 4.65-fold increased occurrence of HCC (P=0.017) [14]. The close association between codon 70 mutations and HCC risk has been observed in several independent patient cohorts [15-22], primarily in Asian (Japanese) populations. The impact in non-Asian population is unclear. Significantly, a recent study showed that the elevated HCC risk among patients with Q⁷⁰ persisted even after they achieved an SVR [25]. The Q⁷⁰/M⁹¹ mutations are also associated with insulin resistance [23] and failure of interferon-based therapies [24].

Despite clinical data showing an association between Q⁷⁰/M⁹¹ mutations and HCC, the absence of supporting laboratory data has raised concern that it is an artifact of uncontrolled confounding in the clinical data. To address this concern, we developed a cell culture system for testing the impact of high-risk HCV on the expression of cancer-related molecular pathways. The model utilizes Huh-7 cells cultured long-term in media containing human serum (HS), as was done previously for Huh-7.5 cells [26]. Under these conditions, Huh-7 cells acquire the capacity for contact inhibition and they express many gene pathways that are typical of primary human hepatocytes (PHH). These growth-arrested, hepatocyte-like Huh-7 cells support replication of full-length HCV. Matched viruses with and without the high-risk mutations were indistinguishable in their virological properties; however, cells replicating the high-risk variant had enhancement of gene pathways associated with oncogenesis and insulin resistance compared to cells replicating the control HCV, which had enhancement of pathways associated with normal metabolism, such as oxidative phosphorylation. In addition, cells replicating the high-risk virus had a global transcriptome that correlated significantly with a high-HCC risk profile among 216 HCV-infected patients with early-stage cirrhosis who were followed prospectively for up to 23 years to determine clinical outcome. The transcriptome of cells replicating the control virus did not correlate with HCC or any other clinical outcome. Thus, we introduce a simple experimental system that distinguished high-risk and control HCV. When combined with each other, the clinical and experimental data emphasize the importance of HCV mutations as prognostic indicators of liver cancer risk. The simple experimental system will enable detailed biochemical investigation of cellular pathways disrupted by high-risk HCV strains and provide a platform for the discovery of drugs that reverse these disruptions.

Materials and Methods

Huh-7 cultures

Huh-7 cells cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (FBS) are referred to as FBS/Huh-7 cells. To induce differentiation, Huh-7 cells were maintained in media containing 2% AB human serum (HS) (Corning-Mediatech), without sub-culturing and with regular media changes as previously described [26] and are referred to as HS/Huh-7 cells. Additionally, a range of dimethyl sulfoxide (DMSO) concentrations (1% - 2.5%) (Sigma-Aldrich) was also tested [27]. Primary human hepatocytes (PHH) were cultured as described in Supplementary methods.

Quantification of hepatocyte differentiation markers and innate antiviral gene expression

RNeasy Plus Mini kits (Qiagen) were used to purify RNA. Reverse transcription (RT) was performed using SuperScript III First-Strand Synthesis (Invitrogen) and random hexamers. RT products were then used for quantitative-PCR (q-PCR) using the LightCycler 480 SYBR Green I Master kit. Expression of hepatocyte-specific genes, albumin and A1AT, and innate immune antiviral genes was determined and calculated using the ΔΔC_t method. RPS11 was used for normalization. Primer sets were previously reported [27, 28]. Statistical analysis of RNA quantitation was performed using Student's t-Test. A two-tailed P value < 0.05 was regarded as statistically significant.

Viral constructs

APP144, a Con1/JFH-1 chimeric construct (Apath_LLC), has five enhancing mutations (L831F, I857T, I1312V, K2073R and I2266L) and core, E1, E2, P7 and part of NS2 from Con1 (genotype 1b; accession number: AJ238799) and other regions from JFH-1 (genotype 2a; accession number: AB047639). APP144 has Q⁷⁰ (glutamine) and L⁹¹ (leucine). Mutagenesis was performed as previously described [29], creating constructs with R⁷⁰/L⁹¹ and Q⁷⁰/M⁹¹, which were verified by sequencing.

Infectivity assays

Infectious titers were determined by a tissue culture infectious dose 50 assay (TCID₅₀) previously described [30]. Standard methods were used for preparation of viral stocks, intracellular HCV RNA quantitation, Western blotting and flow cytometry to assess percentage of infected cells; detailed in Supplementary methods.

Gene-expression profiling and bioinformatics analysis

Genome-wide gene expression profiling was performed using the Human-HT-12 Expression Beadchip (Illumina); 200 ng/sample. Hybridized chips were scanned using HiScan Array Scanner, raw data were extracted using Genome Studio software ver.3 (Illumina), and normalized by cubic spline algorithm implemented in Illumina Normalizer module of GenePattern genomic analysis toolkit (www.broadinstitute.org/genepattern). The dataset is available at NCBI Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo, accession number GSE64605). The transcriptome profile of PHH was previously described [31] (www.ncbi.nlm.nih.gov/geo, accession number GSE29907, samples GSM740485 and GSM740486).

The global transcriptome profile of HCV-infected cells was compared to that of liver biopsy specimens from 216 patients with early stage HCV-associated cirrhosis and no prior history of HCC or complications of cirrhosis. These patients were followed longitudinally for a median of 10 years to identify patients who developed HCC, died from any cause, progressed in Child-Pugh class, and/or experienced hepatic decompensation. The cohort and the transcriptome profile of the liver specimens have been described previously (NCBI Gene Expression Omnibus accession number: GSE15654) [32]. For each gene in the dataset, the association with each clinical outcome was calculated using the Cox score, as previously described [33]. Vectors of Cox scores were merged to create a matrix, and the association between HCV core gene variants and clinical outcomes were assessed by using a Subclass Mapping algorithm, a bi-directional gene signature enrichment-based subclass association determination method, as previously described [34]. A mutation-outcome association with a Bonferroni-corrected p-value <0.05 was regarded as statistically significant.

Differential modulation of molecular pathways in the Kyoto Encyclopedia of Genes and Genomes database (KEGG) (http://www.genome.jp/kegg/) was determined by using Gene Set Enrichment analysis (GSEA) [35]. MicroRNA and reactome databases were also investigated (http://www.broadinstitute.org/gsea/msigdb/collections.jsp). Gene sets with a false discovery rate (FDR) <0.05 were regarded as statistically significant.

Results

Differentiation of Huh-7 cells by long-term culture in human serum

We used human serum (HS) to induce differentiation of Huh-7 cells, as described previously for Huh-7.5 cells [26]. Cells were plated in media containing 10% FBS and cultured until they reached 60% to 70% confluence. Cells were then cultured in media containing 2% HS for three weeks or more to produce monolayers of growth-arrested HS/Huh-7 cells (Fig. 1A). The HS/Huh-7 cells exhibited contact inhibition and grew in a cobble-stone pattern, unlike FBS/Huh-7 cells (Fig. 1B).

Fig. 1. Differentiation of Huh-7 cells by long-term culture in human serum.

(A) Experimental conditions for Huh-7 cells cultured in media containing 2% HS (HS/Huh-7). (B) Phase contrast microscopy of FBS/Huh-7 (left), HS/Huh-7 (middle) and primary human hepatocytes (PHHs) (right). (C) Expression of hepatocyte-specific genes and antiviral genes in HS/Huh-7 cells and PHHs relative to that in FBS/Huh-7 cells. Shared metabolic pathways significantly (D) enhanced or (E) down-regulated in HS/Huh-7 cells and PHHs compared to FBS/Huh-7 cells. (F) Model of the impact of long-term culture of Huh-7 cells in HS. * P<0.05, † P< 0.01 and ‡ P< 0.001.

The HS/Huh-7 cells had increased levels of hepatocyte-specific mRNAs compared to FBS/Huh-7 cells, including an approximately 20-fold increase in albumin mRNA (P = 0.03) and an approximately 6-fold increase in alpha-1 antitrypsin (A1AT) mRNA (P = 0.01) (Fig. 1C). The level of albumin mRNA in HS/Huh-7 cells was comparable to that of cultured PHH (shown in Fig. 1C); the levels were 19- and 20.8-fold higher than in FBS/Huh-7 cells, respectively. Compared to FBS/Huh-7 cells, HS/Huh-7 cells expressed up to 120-fold higher levels of antiviral genes, such as DDX58 (encoding RIG-I) (P = 0.001), MDA5 (P = 0.001), IFNAR1 (P = 0.0001), STAT1 (P = 0.04), IRF9 (P = 0.02), EIF2AK2 (encoding PKR) (P = 0.006), OAS1 (P = 0.005), ISG15 (P = 0.02), MX1 mRNA (P = 0.002), IFNB (P = 0.02) and IFNL1 (P = 0.01) (Fig. 1C). FBS and HS cells had similar levels of TLR3 mRNA, as expected due to a known lesion in this gene [36]. They also had similar levels of IFITM3 mRNA. Cultured PHH expressed higher levels of several antiviral genes, but lower levels of PKR and STAT1. The robust antiviral response accords with data implicating IFN signaling in HCV control [37].

To investigate global similarities between HS/Huh-7 cells and primary human hepatocytes (PHH), we analyzed the KEGG pathways differentially regulated in both HS/Huh-7 cells and PHH compared to FBS/Huh-7 cells. Eight of the ten pathways enhanced in HS/Huh-7 cells versus FBS/Huh-7 cells were also enhanced in PHH versus FBS/Huh-7 cells (Fig. 1D and table 1). Many of the shared pathways carry out hepatocyte functions, such as steroid biosynthesis and drug metabolism by cytochrome P450. Conversely, four of the six pathways down-regulated in HS/Huh-7 cells versus FBS/Huh-7 cells were also down-regulated in PHH versus FBS/Huh-7 cells (Fig. 1E and table 2). These down-modulated pathways included cell cycling, aminoacyl tRNA biosynthesis, TGF-β signaling, and RNA degradation. The cobble-stone morphology of the HS/Huh-7 cells combined with their capacity for contact inhibition and their gene expression profile indicate that long-term culture in HS produced cells with many features typical of PHH (Fig. 1F). Two different lots of HS induced similar cell growth properties and similar enhancement of the expression of hepatocyte-specific genes in HS/Huh-7 cells, establishing the consistency of the approach (Supplementary Fig. 1).

Table 1.

Gene Set Enrichment Analysis (GSEA) of significantly enriched gene sets of the Kyoto Encyclopedia of Genes and Genomes database (KEGG) in PHHs and HS/Huh-7 cells when compared to FBS/Huh-7 cells.

Enriched in PHH compared to FBS/Huh-7		Enriched in HS/Huh-7 compared to FBS/Huh-7
Gene set	FDR*	Gene set	FDR*
AMINO SUGAR AND NUCLEOTIDE SUGAR METABOLISM	0.019	AMINO SUGAR AND NUCLEOTIDE SUGAR METABOLISM	0.01
ANTIGEN PROCESSING AND PRESENTATION	0.012	ANTIGEN PROCESSING AND PRESENTATION	0.003
ASCORBATE AND ALDARATE METABOLISM	<0.001	ASCORBATE AND ALDARATE METABOLISM	0.03
DRUG METABOLISM CYTOCHROME P450	<0.001	DRUG METABOLISM CYTOCHROME P450	0.002
LYSOSOME	0.017	LYSOSOME	0.003
METABOLISM OF XENOBIOTICS BY CYTOCHROME P450	<0.001	METABOLISM OF XENOBIOTICS BY CYTOCHROME P450	0.006
PENTOSE AND GLUCURONATE INTERCONVERSIONS	<0.001	PENTOSE AND GLUCURONATE INTERCONVERSIONS	0.008
STEROID HORMONE BIOSYNTHESIS	<0.001	STEROID HORMONE BIOSYNTHESIS	0.01
ALLOGRAFT REJECTION	0.034	STEROID BIOSYNTHESIS	0.002
APOPTOSIS	0.019	TERPENOID BACKBONE BIOSYNTHESIS	0.007
ARACHIDONIC ACID METABOLISM	<0.001
AUTOIMMUNE THYROID DISEASE	0.02
BETA ALANINE METABOLISM	0.042
BUTANOATE METABOLISM	0.047
COMPLEMENT AND COAGULATION CASCADES	<0.001
CYTOSOLIC DNA SENSING PATHWAY	0.018
DRUG METABOLISM OTHER ENZYMES	<0.001
FRUCTOSE AND MANNOSE METABOLISM	0.029
GLYCEROLIPID METABOLISM	0.018
GLYCOLYSIS GLUCONEOGENESIS	0.016
GLYCOSAMINOGLYCAN DEGRADATION	0.008
GRAFT VERSUS HOST DISEASE	0.013
INTESTINAL IMMUNE NETWORK FOR IGA PRODUCTION	0.018
LINOLEIC ACID METABOLISM	<0.001
PENTOSE PHOSPHATE PATHWAY	0.019
PEROXISOME	0.017
PORPHYRIN AND CHLOROPHYLL METABOLISM	0.006
PPAR SIGNALING PATHWAY	0.034
PRIMARY BILE ACID BIOSYNTHESIS	0.018
PRION DISEASES	0.012
RETINOL METABOLISM	<0.001
TOLL LIKE RECEPTOR SIGNALING PATHWAY	0.007
TRYPTOPHAN METABOLISM	0.017

^*False discovery rate (FDR) <0.05 were regarded as statistically significant. Shared gene sets are in boldface type. Click gene set for details.

Table 2.

GSEA of significantly down-regulated gene sets of the KEGG in PHHs and HS/Huh-7 cells when compared to FBS/Huh-7 cells.

Enriched in FBS/Huh-7 compared to PHH		Enriched in FBS/Huh-7 compared to HS/Huh-7
Gene set	FDR*	Gene set	FDR*
AMINOACYL TRNA BIOSYNTHESIS	0.026	AMINOACYL TRNA BIOSYNTHESIS	<0.001
CELL CYCLE	0.001	CELL CYCLE	0.01
RNA DEGRADATION	0.028	RNA DEGRADATION	0.013
TGF BETA SIGNALING PATHWAY	0.005	TGF BETA SIGNALING PATHWAY	<0.001
OOCYTE MEIOSIS	0.017	NOD LIKE RECEPTOR SIGNALING PATHWAY	0.043
DNA REPLICATION	0.006	WNT SIGNALING PATHWAY	0.029
RIBOSOME	0.005
SPLICEOSOME	0.032
MISMATCH REPAIR	0.034

^*False discovery rate (FDR) <0.05 were regarded as statistically significant. Common gene sets are in boldface type. Click gene set for details.

DMSO was tested as an alternative method for inducing differentiation of Huh-7 cells [27]. Similar to the process used to create HS/Huh-7 cells, cultures were initiated in media containing 10% FBS. When the cells reached 60% -70% confluence, they were switched to media containing 1% - 2.5% DMSO (Fig. 2A). The only concentration that yielded growth-arrested cells with hepatocyte-like morphology was 2% (Fig. 2B). Similar to HS/Huh-7 cells, the 2% DMSO/Huh-7 cells had significantly increased mRNA levels of hepatocyte-specific markers and antiviral genes (Fig. 2C) compared to FBS/Huh-7 cells. The induction of albumin was only 3.2-fold, which was significantly lower than for HS/Huh-7 cells (P=0.048) (Figs. 2C, 1C). The global transcriptome of the DMSO/Huh-7 cells was not investigated.

Fig. 2. Differentiation of Huh-7 cells by long-term culture in DMSO.

(A) Experimental conditions for Huh-7 cells cultured in media containing 2% DMSO (DMSO/Huh-7). (B) Phase contrast microscopy of differentiated DMSO/Huh-7 cells. (C) Expression of hepatocyte-specific genes and antiviral genes in DMSO/Huh-7 cells compared to that in FBS/Huh-7 cells. * P<0.05, † P< 0.01 and ‡ P< 0.001.

HCV replication and infectious particle production

Huh-7 cells grown under a variety of conditions were tested for their susceptibility to APP144, a chimeric virus with a genotype 1b core gene (Fig. 3A-F). Cells were either infected after they had been cultured in HS or DMSO (Fig. 3C) or they were infected prior to culture in HS or DMSO, with subsequent differentiation (Fig. 3D). The percentage of infected cells was determined by flow cytometry using an antibody (9E10) directed against the HCV NS5A protein. FBS/Huh-7 cells served as controls (Fig. 3B). The percentage of positive cells ranged from 93.6% to 1.6% (Fig. 3E). Only about 7% of the cells differentiated in DMSO expressed NS5A, versus 45% of the cells differentiated in HS (Fig. 3E). Although the HS/Huh-7 cells contained less HCV RNA and produced fewer infectious particles than FBS/Huh-7 cells (Fig. 3F), their level of production was high enough to support investigation of high-risk and control HCV variants.

Fig. 3. Susceptibility of differentiated Huh-7 cells to HCV APP144 infection.

(A) Diagram of the APP144 virus. Black arrows indicate adaptive mutations. (B) Infection of undifferentiated control FBS/Huh-7 cells. (C) Infection of Huh-7 cells cultured in media containing either 2% human serum (HS/Huh-7) or 2% DMSO (DMSO/Huh-7) post-differentiation. (D) Infection of HS/Huh-7 or DMSO/Huh-7 cells prior to differentiation with subsequent exposure to differentiating conditions. (E) Flow cytometry analysis of the proportion of NS5A positive cells among control FBS/Huh-7 cells, HS/Huh-7 or DMSO/Huh-7 cells that were infected either post-differentiation or prior to differentiation with subsequent exposure to differentiating conditions. (F) Levels of intracellular HCV RNA replication and the amount of infectious particles released into the media within a 24 hour window were determined 4 days post infection in FBS/Huh-7 cells and 24 days post infection in HS/Huh-7 cells.

High-risk and control viruses have similar fitness in HS/Huh-7 cells

A pair of matched viruses was derived from APP144. One harbored high-risk core gene mutations, Q⁷⁰/M⁹¹; the other had R⁷⁰/L⁹¹ (Fig. 4A). FBS/Huh-7 cells were infected, cultured for four days to allow the infection to spread, and then they were switched to HS-containing media and cultured for an additional three weeks to produce differentiated HCV-infected cells (Fig. 4B). In assays of virological fitness, the Q⁷⁰/M⁹¹ and R⁷⁰/L⁹¹ viruses were indistinguishable. They persisted in a similar percentage of cells (44.7% versus 44.2%) (Fig. 4C), produced similar levels of intracellular HCV RNA (6.1 log₁₀ versus 6.0 log₁₀ copies/100 ng) (Fig. 4D), released similar quantities of infectious particles (4.0 log₁₀ versus 4.1 log₁₀ TCID₅₀/ml) (Fig. 4E), expressed similar levels of intracellular HCV p21 core protein (Fig. 4F) and, as shown by immunofluorescence/confocal microscopy, had similar subcellular localization of the core protein (data not shown).

Fig. 4. Viral fitness of the high-risk and control viruses in HS/Huh-7 cells.

(A) Diagrams of the high-risk (Q⁷⁰/M⁹¹) and the control (R⁷⁰/L⁹¹) viruses. (B) Infection of HS/Huh-7 cells. (C) Flow cytometry analysis of the proportion of NS5A positive cells among HS/Huh-7 cells infected either with R⁷⁰/L⁹¹ or Q⁷⁰/M⁹¹ viruses. (D) Levels of intracellular HCV RNA replication and (E) amount of released infectious particles of R⁷⁰/L⁹¹ and Q⁷⁰/M⁹¹ viruses were determined in HS/Huh-7 cells 24 days after infection. (F) Expression levels of p21 core protein of R⁷⁰/L⁹¹ and Q⁷⁰/M⁹¹ viruses were determined in HS/Huh-7 cells 24 days after infection. NS; not significant.

Differential impact of core gene mutations on the global gene profile of HS/Huh-7 cells

The two viruses had very different effects on cellular gene expression, despite their similar virological fitness. Of great interest, the global transcriptomic profiles of cultures replicating the high-risk virus (Q⁷⁰/M⁹¹) were significantly correlated with those of liver biopsies of patients with early-stage HCV-related cirrhosis (n=216) who had an increased incidence of HCC when followed prospectively for up to 23 years [32] (Bonferroni-corrected P=0.03) (Fig. 5A). The transcriptome of the cells replicating the control virus (R⁷⁰/L⁹¹) did not correlate with increased HCC incidence (Bonferroni-corrected P= 0.60) or any of the other clinical outcomes that were monitored (overall death, progression of Child-Pugh class from A to B or C, and development of symptoms of hepatic decompensation).

Fig. 5. Differential gene signature in HS/Huh-7 cells replicating either the high-risk virus or the control virus.

(A) Heatmap of association between global transcriptome of HS/Huh-7 cells that were infected either with the high-risk (Q⁷⁰/M⁹¹) or the control (R⁷⁰/L⁹¹) viruses with long-term clinical outcomes of HCV-related early-stage cirrhosis patients. Bonferroni-corrected p-values are indicated inside each box. (B) GSEA of site-specific cancer pathways in the KEGG set in HS/Huh-7 cells infected with the high-risk virus (Q⁷⁰/M⁹¹) compared to cells infected with the control virus (R⁷⁰/L⁹¹). All 14 gene sets were significantly enhanced in HS/Huh-7 cells replicating high-risk virus while none in cells replicating control virus. GSEA for differential expression of gene sets related to (C) type II diabetes mellitus and (D) Oxidative phosphorylation between HS/Huh-7 cells infected either with the high-risk or control viruses. Genes contributed to pathway enrichment (leading edge genes) are enclosed in red rectangle with the heat map of these genes displayed beneath the enrichment plot. NES; normalized enrichment score. False discovery rate (FDR) was regarded as statistically significant. For detailed description of enrichment blot, see Supplementary information.

To obtain additional information about the impact of the high-risk virus on cellular gene expression, the global transcriptome of HS/Huh-7 cells replicating this variant (Q⁷⁰/M⁹¹) was compared to that of HS/Huh-7 cells replicating the control virus (R⁷⁰/L⁹¹). In GSEA, all 14 site-specific cancer-related KEGG pathways were enhanced in the cells replicating the high-risk virus, while none were enhanced in cells replicating the control virus (Fig. 5B and Supplementary table 1).

Additionally, GSEA revealed that HS/Huh-7 cells replicating the high-risk virus (Q⁷⁰/M⁹¹) had greater induction of pathways related to altered metabolism, such as type II diabetes and insulin signaling, than cultures replicating the control virus (R⁷⁰/L⁹¹) (Fig. 5C and Supplementary table 1). Cells replicating the high-risk virus had altered expression of pathways regulated by miR26, miR181, miR202, and miR519 (FDR <0.001; data not shown). By contrast, cells infected by the control virus showed greater expression of the oxidative phosphorylation pathway, which is a feature of normal aerobic metabolism (Fig. 5D and Supplementary table 1). These cells also tended to have greater expression of pathways associated with physiological functions of primary hepatocytes, such as metabolism of xenobiotics by cytochrome P450 (FDR = 0.06), DNA repair (KEGG homologous recombination) (FDR=0.08) and reactome-P53 independent G1-S DNA damage checkpoint (FDR<0.001); data not shown.

Differential impact of core gene mutations on activation of ISGs

Mutations in codons 70 and 91 (Q⁷⁰/M⁹¹) of the genotype 1b core gene predict poor response to IFN-based therapy [24]. To seek evidence of a differential effect of high-risk and control viruses on IFN responses in HS/Huh-7 cells, mRNA levels of five ISGs (MX1, OAS1, IFITM3, STAT1 and ISG15) were determined in the presence or absence of exogenous IFN treatment. In the absence of IFN treatment, HS/Huh-7 cells infected with either variant had higher ISGs expression than uninfected cells (Fig. 6), demonstrating the capacity of the differentiated cells to respond to viral infection by increasing IFN signaling. Cells infected with the high-risk virus had significantly lower expression of MX1 and OAS1 than cells infected with the control virus (Fig. 6A and B). IFN treatment increased ISG expression in cells infected with either variant compared to infected cells without IFN treatment. The high-risk virus blunted the IFN-mediated ISG induction, with the exception of ISG15 (Fig. 6E), consistent with the association between higher ISG15 expression in liver tissue of patients who do not respond to IFN-based treatments [38].

Fig. 6. Differential activation of ISGs in HS/Huh-7 cells infected with either the high-risk or the control virus.

Expression of (A) MX1, (B) OAS1, (C) IFITM3, (D) STAT1 and (E) ISG15 in HS/Huh-7 cells infected with either the high-risk virus (Q⁷⁰/M⁹¹) or the control virus (R⁷⁰/L⁹¹) in the absence or presence of IFN treatment relative to that in uninfected HS/Huh-7 cells. For IFN treatment, infected-differentiated HS/Huh-7 cells were treated with 100 IU/ml of IFN-α for 24 hours before extracting total RNA. The P values of different comparisons are displayed beneath each graph.

Discussion

Our results shed new light on the clinical data that first established an association between point mutations in the HCV core gene and adverse outcomes, such as failure of IFN-based therapy [24], insulin resistance and type II diabetes [23], and HCC [14]. Our study found a remarkable difference between the impact of the high-risk (Q⁷⁰/M⁹¹) and control (R⁷⁰/L⁹¹) viruses on cellular gene expression, although the two viruses are identical in all but two amino acids out of 3,200. These two viruses and the novel cell culture system we developed to study them provide unique tools for dissecting the mechanistic links between altered metabolism and abnormal cellular growth.

Of particular clinical relevance, we demonstrated that the global transcriptome profile of cells replicating the high-risk virus correlated significantly with the profile of liver biopsy specimens from patients with early-stage HCV-related cirrhosis who had an elevated incidence of HCC during longitudinal follow up. The HS/Huh-7 cells infected with the high-risk virus had an induction of many KEGG pathways associated with cancer and altered metabolism. In particular, cultures infected by the high-risk virus had induction of the type II diabetes gene set; whereas cultures infected by the control virus had greater expression of the oxidative phosphorylation pathway. These results might indicate that the high-risk virus promotes the use of glycolytic pathways, recapitulating an emerging hallmark of cancer [39]. Notably, insulin resistance is one of the major risk factors associated with HCC development in HCV patients [40].

The HS/Huh-7 cell system provides a novel and advantageous model for scoring the pathological potential of HCV mutations present in clinical isolates and for investigating the cellular targets of oncogenic HCV strains. In non-1b subtypes, codons 70 and 91 are generally conserved and thus do not vary according to clinical outcomes [41]; however, it is possible that other codons modulate the oncogenic potential of non-1b subtypes. Well-designed clinical and molecular virology studies are needed to explore this possibility. Genotype 2 HCV is much more sensitive to interferon-based therapy than genotype 1 [42] and genotype 3 HCV is associated with steatosis [43]. The HS/Huh-7 cells provide a new experimental system for investigating the molecular basis of these genotypic differences. Simplicity is an important feature of the experimental system.

Increasing evidence indicates that many cancer cells have the plasticity required for reprogramming [44], focusing attention on the tumor microenvironment as a source of regulatory factors. We aimed to attenuate the neoplastic phenotype of Huh-7 cells by culturing them in adult human serum (HS) rather than fetal bovine serum (FBS). Using this approach, the phenotype of Huh-7 cells changed, as demonstrated by an alteration in morphology from spindle-shaped to cobble-stone, by a switch from unrestrained growth to contact inhibition, and by the induction of hepatocyte-specific genes (such as albumin and A1AT) and genes mediating antiviral and antiproliferative responses. The HS/Huh-7 cells had a global gene profile including enhanced expression of pathways typical of PHH and greatly attenuated expression of pathways associated with cell cycling. A unique feature of this experimental system is its use of a single cell type. Notably, this mono-cell system, when replicating high-risk HCV, acquired a gene signature that was significantly associated with that of liver biopsies of patients at high risk for developing of HCC, even though the biopsies contain multiple cell types. This suggests that among different types of liver cells, hepatocytes are likely to be a key source of the oncogenic signature.

We selected HS/Huh-7 cells for our studies after considering several alternatives. Humanized mice, primary human hepatocytes and hepatocyte-like cells induced from pluripotent stem cells were potential options [45]; however, donor-to-donor variability, high-cost and technical complexities make these challenging to use. We investigated DMSO as a differentiating agent. Previously, Bruno et al. [27] demonstrated that 1% DMSO produced differentiated Huh-7 cells that were highly permissive to HCV, however, we were unable to reproduce this result, most likely because of differences between sublines of Huh-7 cells [46]. Thus, we investigated adult human serum as a differentiating agent. Steenbergen et al. used human serum, rather than DMSO, to differentiate Huh-7.5 cells [26]. We followed their important lead, but elected to use the parental, Huh-7, cell line. Huh-7.5 cells are highly permissive to HCV, but they have a mutation in the gene encoding RIG-I [47] and we wanted to avoid this lesion.

Since a suitable experimental system is now in hand, future detailed biochemical experiments can be carried out to determine the molecular basis of the differential oncogenic and diabetogenic effects on cellular gene expression of the high-risk virus compared to the control. In the meantime, several features of the core gene merit discussion. The high-risk mutations could act at either the protein- or RNA-level and they might, or might not, require viral replication. At the protein-level, it is noteworthy that codon 70 lies within the RNA binding domain of p21 [48]. The substitution of arginine (R), which has a positive charge, with glutamine (Q), which is neutral, might reduce the affinity for RNA; however, the mutation did not affect production of infectious particles, suggesting that it had minimal effect on interactions with genomic RNA. It is possible that the newly discovered HCV minicores play a role since minicores 70 and 91 are predicted to have N-termini at amino acid 70 and 91, respectively [29]. Recent data indicate that minicores circulate in blood [49], suggesting a possible role in infectivity and/or in modulation of host responses to infection. RNA-level effects warrant investigation, perhaps with a focus on differential binding to cellular microRNAs.

In summary, we present the first direct evidence for the sequence-specific effects of HCV variants that were first reported in clinical and epidemiological studies. Our findings were made possible by the development of a novel experimental system that combined HS-differentiated cells and full-length infectious HCV harboring disease-associated mutations. The major advantages of this system are its simplicity, low cost, and ability to recapitulate important clinical findings. This system will enable further biochemical investigation of the biology underlying the cellular pathways disrupted by high-risk HCV strains. This experimental model would be a useful platform for high throughput screening of novel anti-HCC therapies.

List of abbreviations

HCC

hepatocellular carcinoma

HCV

hepatitis C virus

FBS

fetal bovine serum

HS

human serum

A1AT

alpha-1 antitrypsin

RIG-I

retinoic acid-inducible gene 1

DMSO

dimethyl sulfoxide

RPS11

ribosomal protein subunit 11

TLR3

toll-like receptor 3

MDA5

melanoma differentiation-associated protein 5

IFNAR1

interferon alpha receptor 1

STAT1

signal transducer and activator of transcription 1

IRF9

interferon regulatory factor 9

PKR

protein kinase RNA-activated

OAS1

2′,5′-oligoadenylate synthetase

ISG15

interferon stimulated gene 15

MX1

myxovirus resistance 1

IFNB

interferon beta

IFNL1

interferon lambda 1

IFITM3

interferon-induced transmembrane protein 3

TCID₅₀

tissue culture infectious dose 50

TBP

TATA-binding protein

KEGG

Kyoto Encyclopedia of Genes and Genomes

GSEA

Gene Set Enrichment Analysis

NES

normalized enrichment score

FDR

false discovery rate