Combined Analysis of Metabolome, Proteomes, and Transcriptomes of HCV-infected Cells and Liver to Identify Pathways Associated With Disease Development

Abstract

Background & Aims:

The mechanisms of hepatitis C virus (HCV) infection, liver disease progression, and hepatocarcinogenesis are only partially understood. We performed genomic, proteomic, and metabolomic analyses of HCV-infected cells and chimeric mice to learn more about these processes.

Methods:

Huh7.5.1^dif (hepatocyte-like cells) were infected with culture-derived HCV and used in RNA-Seq, proteomic, metabolomic, and integrative genomic analyses. uPA/SCID mice were injected with serum from HCV-infected patients; 8 weeks later, liver tissues were collected and analyzed by RNA-seq and proteomics. Using differential expression, gene set enrichment analyses, and protein interaction mapping, we identified pathways that changed in response to HCV infection. We validated our findings in studies of liver tissues from 216 patients with HCV infection and early-stage cirrhosis and paired biopsies from 99 patients with hepatocellular carcinoma, including 17 patients with histologic features of steatohepatitis. Cirrhotic liver tissues from patients with HCV infection were classified into 2 groups based on relative peroxisome function; outcomes assessed included Child-Pugh class, development of hepatocellular carcinoma, survival and steatohepatitis. Hepatocellular carcinomas were classified according to steatohepatitis; the outcome was relative peroxisomal function.

Results:

We quantified 21,950 mRNAs and 8297 proteins in HCV-infected cells. Upon HCV infection of hepatocyte-like cells and chimeric mice, we observed significant changes in levels of mRNAs and proteins involved in metabolism and hepatocarcinogenesis. HCV infection of hepatocyte-like cells significantly increased levels of mRNAs, but not proteins, that regulate the innate immune response—we believe this was due to the inhibition of translation in these cells. HCV infection of hepatocyte-like cells increased glucose consumption and metabolism and the STAT3 signaling pathway and reduced peroxisome function. Peroxisomes mediate beta-oxidation of very long-chain fatty acids (VLCFAs); we found intracellular accumulation of VLCFAs in HCV-infected cells, which is also observed in patients with fatty liver disease. Cells in livers from HCV-infected mice had significant reductions in levels of mRNAs and proteins associated with peroxisome function, indication perturbation of peroxisomes. We associated defects in peroxisome function with outcomes and features of HCV-associated cirrhosis, fatty liver disease, and hepatocellular carcinoma in patients.

Conclusions:

We performed combined transcriptome, proteome, and metabolome analyses of liver tissues from HCV-infected hepatocyte-like cells and HCV-infected mice. We found that HCV infection increases glucose metabolism and the STAT3 signaling pathway and thereby reduces peroxisome function; alterations in expression of peroxisome genes were associated with outcomes of patients with liver diseases. These findings provide insights into liver disease pathogenesis and might be used to identify new therapeutic targets.

INTRODUCTION

Viruses have developed sophisticated strategies to promote their life cycle, evade the antiviral defense systems and cause disease. As a result, some viruses can persist beyond the stage of acute infection and develop a state of co-existence with the host through either chronic or latent infection. Viral and cellular gene expression are adjusted over time to meet the requirements of persistence. Considering chronic infection as merely an enduring acute phase is thus fundamentally inaccurate. In fact, the modulations of host biology in the long run are profound and continually damaging the host cell and its microenvironment¹. This contributes to pathogenic phenotypes including chronic inflammation, tissue injury and cancer. Viral reprogramming of host cells can be investigated by systematic genome-wide profiling of gene-products. Such analysis does not only increase our understanding of the rearrangements of cellular architecture and functions, but also provides mechanistic insight into disease development.

Chronic hepatitis C virus (HCV) infection is an intriguing prototype to study general mechanisms of immune evasion and disease pathogenesis because of the refined strategies to evade antiviral responses, and alteration of metabolic pathways as well as regulatory cell circuits including signaling pathways, translation machinery and RNAi^2–5. This may have profound impact on the host cell’s proteome and transcriptome resembling patterns induced by other pathologies such as alcohol or obesity. HCV establishes acute and chronic infection of the liver, which is a leading cause of liver disease progressing from chronic inflammation and metabolic disease to fibrosis, cirrhosis and ultimately hepatocellular carcinoma (HCC). It is assumed that HCV contributes to liver disease directly by viral factors and indirectly through signaling. Indeed, tumor expression profiling of cirrhotic livers revealed similar genetic profiles between HCV-associated HCCs and other etiologies⁶. Novel direct antivirals achieve very high cure rates, but despite HCV elimination patients with advanced liver disease remain at high risk for HCC development⁷. Treatment is costly and is currently only available for a fraction of all HCV-infected patients. Furthermore, resolved infection does not provide protection against reinfection, emphasizing the need for increased understanding of immune evasion for effective vaccine development⁸.

During recent years thematically focused studies have revealed that virus-host interactions and microbial immune evasion involve the manipulation of both host gene transcription and RNA translation^{9, 10}. However, a global and integrated view on the multiple biological layers is required to understand the host antiviral response and mechanisms leading to disease. Using HCV infection as a model we assessed perturbations of the cellular homeostasis contributing to chronic inflammatory disease and virus-induced cancer. We combined a state-of-the art HCV infection model with cutting edge screening technologies including proteomics, RNA sequencing (RNA-Seq), metabolomics and mathematical modeling to gain a multilayered insight into virus-host interaction and its impact for liver disease biology at the systems level. This approach sheds new light on how viruses evade innate immune responses, reprogram host cell metabolism and trigger chronic inflammatory and metabolic disease as well as cancer.

MATERIALS AND METHODS

HCV infection of human hepatocyte chimeric mice.

Mouse uPA+/+/SCID+/+ (uPA/SCID) liver samples used including ethical approval and informed consent have been described¹¹. Mice with human albumin levels >10mg/mL (~80–90 % human hepatocytes repopulation) were used for this study. Briefly, mice were intravenously inoculated with HCV+ patient serum (10⁵ HCV particles, genotype 1b). Five viremic mice (>10⁷ HCV copies/mL) and five control mice were sacrificed at week eight. RNA-seq data on a subset of these mice (three HCV-infected and three non-infected mice) have been described¹¹. All liver samples were snap frozen and stored at −80 °C prior to analysis.

Immunofluorescence microscopy.

Cells were fixed with 4 % paraformaldehyde for 20 min prior permeabilization (15 min) with 0.1 % Triton X-I00, 0.5 % BSA. Antibody incubation was performed as recommended. Slides were mounted with Fluoroshield including DAPI (Sigma). Fluorescence dots (puncta) of 25 uninfected and 25 HCV-infected cells were quantified as described¹².

Differentiation and infection of liver cells.

Huh7.5.1¹³ and HepG2-NTCP¹⁴ cells have been described. For proliferation arrest and differentiation (Huh7.5.1^dif cells), Huh7.5.1 cells were cultured in DMEM containing 1 % DMSO¹⁵ for 10 days before infection. 2.5 × 10⁴ Huh7.5.1^dif cells (RNA-Seq and metabolomics) or 1.5 × 10⁶ Huh7.5.1^dif cells (proteomics) were infected with a MOI of 8 using affinity purified HCVcc (TCID₅₀ 6.7 × 10⁵ / mL) or mock-inoculated with FLAG-peptide elution buffer. After 6 h incubation, the inoculum was replaced by fresh medium supplemented with 1 % DMSO. This MOI is >10× higher than the minimal MOI (0.12) required to infect 100 % of Huh7.5.1^dif cells after 7 days (Supplementary Fig. S1C).

Proteome analysis.

Protein lysates were harvested (duplicates) at day 0, 3, 7 and 10 post-infection (pi) for both HCV-infected and mock-infected Huh7.5.1^dif cells. Cells were lysed for 15 min at room temperature in 8.0 M urea buffer containing 75 mM NaCl, 50 mM Tris pH 8, 1 mM EDTA, aprotinin 2 μg/mL (Sigma-Aldrich), leupeptin 10 μg/mL (Roche), PMSF 1 mM (Sigma-Aldrich), phosphatase inhibitor cocktail 2 and phosphatase inhibitor cocktail 3 (Sigma-Aldrich). Snap-frozen mouse liver tissues were crushed in liquid nitrogen and lysed, accordingly. Lysates were cleared at full speed in a bench top centrifuge for 5 min, and stored at −80 °C. Proteomic analyses of Huh7.5.1^dif lysates were performed at the Broad Institute of MIT and Harvard (Cambridge, MA, USA) and the tissue lysates were analyzed at the Max Delbruck Center for Molecular Medicine/Berlin Institute of Health (Berlin, Germany) using 10-plex tandem mass tag (TMT) labeling as described¹⁶.

Transcriptome profiling.

RNA samples were collected daily between days 0 and 10 pi (triplicates). Cells were lysed in TCL buffer (Qiagen) supplemented with 1 % 2-mercaptoethanol. Cellular mRNA was isolated and analyzed as described¹⁷. RNA-Seq was performed at the Broad Institute of MIT and Harvard (Cambridge, MA, USA). Therefore, HCV RNA was co-amplified with cellular mRNA using a SMART-compatible primer (sequence: 5’-Biotin-AAG CAG TGG TAT CAA CGC AGA GTA CTC TGC GGA ACC GGT GAG TA-3’)¹⁸. Cellular mRNA was isolated and analyzed according to the SmartSeq2 protocol as described^{17, 19}. RNA sequencing paired-end reads were aligned to the human hg19 UCSC reference using TopHat software (v2.0.14). HCV RNA sequencing paired-end reads were aligned to the HCV Jc1E2^FLAG genome using Bowtie2 (v2.2.5) software. Gene expression levels for 21,950 human genes were estimated using Cufflinks (v2.2.1) with the fragments per kilobase of transcript per million mapped reads method.

Metabolomics and lipid analysis.

Analysis of polar metabolites was performed in Huh7.5.1^dif cells infected with HCV Jc1E2^FLAG (MOI=8). Intra/extra-cellular metabolites were analyzed by mass spectrometry as described²⁰ by the Barteesy lab at the Massachusetts General Hospital (Boston, MA, USA). Methyl derivatives of very long-chain fatty acids (VLCFA) were extracted in the presence of internal standards by the Ricci and Dali-Youcef labs and analyzed by gas chromatography-mass spectrometry (GC-MS) at the Hôpitaux Universitaires de Strasbourg (France) as described²¹.

Bioinformatical analyses

Proteins and transcripts were mapped despite possible many-to-many relationships (i.e. isoforms) by constructing “analysis groups”²² Host responses to HCV infection were assessed over a time course on a temporal population level. Pre-ranked gene set enrichment analysis (GSEA)²³ was performed on a temporal population level, taking time points as continuous class label, and taking the control time course as day 0 samples. To compare the two different sources, an indirect comparison was applied. Each time point for both sources were mapped independently to a functional metaspace by performing GSEA, implemented in GenePattern genomic analysis toolkits²⁴ for each gene set in the collection. For each time point, we compared the HCV-infected samples with the mock samples using the Signal-To-Noise ratio and the default setting for GSEA. Correlation to time point for the HCV-infected samples was used to obtain a general enrichment over the entire time course. Next, we compared the significant enrichment scores (ES) (p-value <0.005) for the transcriptional and proteomic data sets in this metaspace. The following gene sets from the Molecular Signature Database²⁵ (MSigDB, ver.4.0) were included for the analysis: BIOCARTA, BIOPLEX, KEGG, NABA, PID, REACTOME, SA, SIG, ST, CORUM, HCVpro, HALLMARK, E1, E2, E3, DUB. “Leading-edge” genes were extracted from all previous GSEA analyses, followed by overlap analyses using the hypergeometric test. The data set was compared to HCVpro database signatures identifying significant overlaps with the current knowledge in HCV-host interactions. Raw reads of patient’s samples with low ISG²⁶ were trimmed using cutadapt²⁷ and mapped to the human genome hg19 using HISAT2. Reads mapping to GENCODE v19 genes were counted with htseq-count. Differentially expressed genes were analyzed using DESeq2. Common transcription factors as potential regulators of HCV-impaired peroxisome expression were identified using Enrichr²⁸. The top 30 transcription factors were retrieved from a combined score ranking (calculated by multiplying p-value (Fisher extract test) and z-score) of the protein-protein interaction (PPI) database and those profiled by ChIP-Seq in mammalian cells (ChEA). These were then successively compared to transcription factors retrieved manually using GeneCards (www.genecards.org), that potentially regulate more than 2 peroxisomal genes. Inference of the transcriptional regulatory networks underlying HCV infection using AMARETTO is described in supplementary information.

RESULTS

An integrated proteogenomic approach reveals the spatiotemporal map of HCV-hepatocyte interactions during infection

We used an hepatocyte-like cell culture model consisting of DMSO-differentiated Huh7-derived liver cells (Huh7.5.1^dif) because it is suitable for robust, long-term culture and has been shown to have a similar phenotype than primary human hepatocytes in cell culture^{15, 29}. Indeed, Huh7.5.1^dif are quiescent and display enhanced hepatocyte-specific marker expression compared to undifferentiated Huh7.5.1 cells (Supplementary Fig. S1A–B). Huh7.5.1^dif were infected with HCVcc (Jc1E2^FLAG) prior sampling during a 10 days culture period (Fig. 1A). HCV protein expression was visualized by a HCV peptide time course analysis (Fig. 1B) and quantified 59 peptides from 8 viral proteins (Fig. 1C). During infection all HCV peptides increased in abundance until day 7 pi (Fig. 1C) simultaneously (p<0.0001, Chi-Square Test), with indifferent expression levels or kinetics (p>0.05, U-Test). These results suggest comparable half-lives of the viral proteins indicating that the HCV polyprotein abundance defines the amount of individual cleavage products. We therefore focused our investigations on these first 7 days. In total we quantified 21,950 mRNAs and 8,297 proteins providing a multidimensional atlas of persistent HCV infection. The atlas reveals a time-resolved proteogenomic state of HCV-infected cells. 7,416 proteins (90.8 %) were uniquely mapped to overlapping mRNAs using integrative functional genomic analyses (Fig. 2A, Supplementary table S1). The replicates (n=2) demonstrated excellent sensitivity and technical quality of the approach superior to recent studies for other viral infections²² (Supplementary Fig. S2).

Fig. 1. Mapping of HCV protein expression by time-resolved proteogenomics.

(A) Hepatocyte-like Huh7.5.1^dif cells were infected with HCVcc (Jc1E2^FLAG) over 10 days. Sampling intervals for transcriptomics, metabolomics and proteomics are indicated. (B) Quantification of HCV-specific peptides over the infection time course relative to non-infected and mock-infected (Mock) cells. The data represent TMT ratios with the 131 channels in the denominator (internal reference = mix of all samples). All ratios were normalized for the median of all 120,000 distinct peptides for each time point. The FLAG peptide is indicated by asterisk. (C) Increase of HCV protein abundance between 3–7 days pi. Data are displayed as relative expression (mean) of all peptides corresponding to a given HCV protein over the time course. HCV replication is depicted in red as median of total HCV peptide abundance ± standard error of the median.

Fig. 2. Persistent HCV infection manipulates host pathways and triggers an attenuated innate immune response.

HCV-infected Huh7.5.1^dif cells relative to mock-infected cells until 7 days pi. (A) Proteogenomic mapping of HCV infection. (B) Gene sets obtained from the MSigDB that are modulated by HCV infection on the RNA and protein level. (C) Stimulation (red) and suppression (blue) of host pathways (gene sets) by HCV infection.

Disparate dynamics between host cell mRNA and proteins upon HCV infection

We then analyzed 2,006 predefined gene sets representing specific host pathways, cellular functions (MSigDB database) and protein complexes (Bioplex) with our proteogenomic data set using GSEA²³. This approach classifies differentially expressed genes or proteins according to their representation within a pre-defined gene set associated to a phenotype. 47.4 % (956 gene sets, transcriptomic data set) and 31 % (621 gene sets, proteomic data set) of these gene sets were significantly altered by HCV over 7 days pi (Fig. 2B). Most of these gene sets were generally downregulated (negatively enriched). In gene set collections with >50 gene sets (BIOCARTA, BIOPLEX, KEGG, PID, REACTOME, CORUM, HALLMARK) a median of 62 % and 79 % of RNA and protein, were negatively enriched, respectively. Moreover, 83 % of these downregulated gene sets on the RNA level were also significantly impaired on the protein level. In contrast, only 17 % of the upregulated pathways on the RNA level matched to the corresponding protein trends (Fig. 2C). This suggests that HCV infection shuts down most of the non-vital processes on the transcriptional and/or at the posttranscriptional level to divert the resources toward viral replication and persistence. Upregulated pathways include the antiviral response and inflammation as well as pro-viral signaling pathways as discussed in the supplementary information (Supplementary Fig. S3, Supplementary table S2).

Persistent HCV infection impairs peroxisome function, lipid metabolism, fatty acid and bile acid metabolism

A striking observation from our proteogenomic atlas of HCV infection is a strongly impaired peroxisomal function as suggested by the GSEA at the RNA and protein levels (Fig. 3A). Peroxisomes are involved in lipid synthesis, signaling, ß-oxidation of VLCFAs and the detoxification of hydrogen peroxide³⁰. Accordingly, we observed an impaired expression of genes involved in peroxisomal biogenesis, bile acid metabolism, fatty acid metabolism and cholesterol biosynthesis (Fig. 3A). We confirmed these findings in a transcriptomic database comprising liver biopsies of 25 patients with chronic HCV infection and 6 non-infected individuals²⁶ and in livers of HCV-infected chimeric mice (n=3) (Fig. 3A). Comparing the GSEA results of HCV-infected patients with Huh7.5.1^dif we identified 46 leading-edges of the HALLMARK_PEROXISOME gene signature that are impaired in infected Huh7.5.1^dif cells and in liver tissue of HCV-infected patients. The expression of 11 of these leading-edge genes changed significantly (p<0.05, Wald test) in HCV-infected Huh7.5.1^dif cells and in the livers of HCV-infected patients (Fig. 3B, Supplementary Fig. S4). We further validated this finding by demonstrating impaired catalase expression in infected cells (Fig. 3C, Supplementary Fig. S5), which is a peroxisome-specific enzyme. Consistently, quantification of catalase-stained peroxisomes revealed a significantly (p<0.005, T-Test) lower number of catalase puncta formed in HCV-infected cells compared to uninfected hepatocytes (Fig. 3C) suggesting an impaired metabolic function of these organelles and an accumulation of fatty acids in infected cells. Whether less peroxisomes are formed during HCV infection cannot be concluded for sure, however a positive correlation between catalase expression and peroxisome abundance has been recently suggested³¹. Moreover, we demonstrate that HCV infection of Huh7.5.1^dif increases intracellular concentrations of very long-chain fatty acids with a chain length of 20–26 carbons, while shorter fatty acids (C16–C18) are less accumulated (Fig. 4A). This is consistent with the formation of intrahepatic lipid droplets during infection, which are important during viral assembly³² and the accumulation of hepatic lipids during steatosis. At the same time, impaired peroxisomes will increase oxidative stress imposed by HCV infection increasing the oncogenic pressure on infected cells. In contrast to HCV, persistent HBV infection increased peroxisomal function and its associated metabolic processes (Fig. 3D), which was confirmed by GSEA of gene expression in HBV-infected primary human hepatocytes (data set GSE69590)³³. This is consistent with the clinical pathology of chronic HBV infection where steatosis is much less common³⁴.

Fig. 3. HCV but not HBV infection impairs expression of peroxisomal genes in Huh7.5.1^dif cells and liver tissue of patients.

(A) HCV infection impairs metabolic pathways associated to peroxisomal function and lipid homeostasis. GSEA of RNA-Seq data from liver tissue of 25 chronic HCV-infected patients vs. 6 non-infected individuals²⁶, and transcriptomics and proteomics of HCV-infection time-course of Huh7.5.1^dif relative to mock-infected cells (n=2). (B) Expression of 11 peroxisomal genes is significantly (p<0.05, Wald test) suppressed by HCV in Huh7.5.1^dif and in liver tissue of patients with chronic HCV infection²⁶. Log fold change of leading-edge gene expression of the HALLMARK_PEROXISOME gene set (A). (C) Peroxisome marker expression is significantly (p=0.0048, T-Test) perturbed in HCV-infected hepatocyte-like cells. Immunofluorescence microscopy of Huh7.5.1^dif cells infected for 3 days with HCVcc. The peroxisomal marker catalase (CAT) is stained in red, nuclear DNA (DAPI) in blue and HCV (NS5A) in green (see also supplementary Fig. S5). Quantification of catalase-stained peroxisomes in 25 random HCV-infected cells and 25 mock cells is shown in box/whiskers; bar represents 200 μm. (D) HBV infection promotes peroxisome function. GSEA of transcriptomics of HBV-infection of HepG2-NTCP for 10 days (n=3), and primary human hepatocytes (n=3) infected for 40 days with HBV (genotype D) (GSE69590). NES are displayed in red (increased), blue (decreased), and gray (no significant change). Temporal analysis of infected Huh7.5.1^dif are presented as global trend (global) and individual timepoints. Statistical cut-off for GSEA of liver tissues was FDR q<0.05 and for Huh7.5.1^dif p<0.005.

Fig. 4. Persistent HCV infection enhances metabolism of intracellular very long-chain fatty acids and glucose and creates a Warburg-like shift of the lactate flux.

(A) HCV infection induces intracellular accumulation of very long-chain fatty acids (VLCFA). Data expressed as mean fold change of intracellular C16 (palmitic acid), C18 (oleic acid), C20 (arachidic acid), C22 (behenic acid), C24 (lignoceric acid), C25 (pentacosanoic acid), C26 (cerotic acid) fatty acids relative to mock-infected cells ±SEM (2–3 independent experiments). (B) HCV-infection increases the concentration of glucose metabolites in infected Huh7.5.1^dif except for malate. Mean fold change of intracellular polar metabolites per 100,000 cells ±SD (4 independent experiments in triplicates). (C) HCV infection of Huh7.5.1^dif increases glucose consumption ~6 fold. Glucose consumption of HCV- and mock-infected (CTRL) cells was measured at day 7 pi in supernatants. Mean glucose consumption per 100,000 cells ±SD (2 independent experiments in triplicates). (D) HCV creates a Warburg-like shift in infected cells. Lactate uptake (negative value) from culture supernatant inverses to lactate secretion (positive value) in infected Huh7.5.1^dif cells. Mean lactate flux ±SEM (mole/h/100,000 cells; 4 independent experiments in triplicates). * p<0.05, T-Test.

Impaired PPAR signaling is a regulator of peroxisomal function during HCV infection

We hypothesized that impaired peroxisomal function was the result of direct, virus-related and indirect effects linked to host cell metabolism. The nuclear receptor peroxisomal proliferator-activated receptor (PPAR) is associated with peroxisome function and highly expressed in the liver. VLCFAs activate PPAR-alpha³⁵ and high glucose levels impair its activity³⁶. Indeed, we observed an impaired PPAR signaling expression signature in HCV-infected Huh7.5.1^dif cells and in the liver of patients on both RNA and the protein level (Fig. 3A), suggesting a perturbed switch between fatty acid and glucose metabolism in HCV-infected hepatocytes. This is consistent with suppressed PPAR-alpha expression by HCV infection³⁷. Therefore, we studied the metabolic status of HCV-infected Huh7.5.1^dif cells using mass spectrometry-based metabolomic profiling²⁰. Analysis of polar metabolites revealed elevated concentrations of the glucose metabolites succinate, pyruvate, 3-phosphoglycerate and citrate in HCV-infected Huh7.5.1^dif cells (Fig. 4B, Supplementary table S4), consistent with the high glucose-dependence of the HCV replication³⁸. Indeed, our analyses revealed an increased glucose consumption (Fig. 4C) and an accumulation of lactate and decrease of malate arguing for impaired gluconeogenesis in infected cells (Fig. 4B). In healthy individuals, lactate is produced and secreted by glycolysis-dependent tissues including skeletal muscle, bone marrow and hypoxic tissue. Lactate is then rapidly metabolized in the liver to glucose and energy. As shown in Supplementary Fig. S6 HCV infection induces hypoxia presumably contributing to lactate production. When measuring the hepatocellular lactate flux, a significant accumulation inside the cells was observed (Fig. 4D). Collectively, these data demonstrate that persistent HCV infection causes elevated glucose levels in infected cells contributing to impaired peroxisomal functions. Moreover, infection creates a Warburg-like metabolic shift of the host metabolism which is a hallmark of cancer or cells undergoing carcinogensis³⁹. Next, we aimed to identify common transcription factors of peroxisomal genes that are potentially interacting with HCV proteins. Thereto, we identified common transcription factor binding sites among the combined 85 leading edge genes of the HALLMARK_PEROXISOME gene set in livers of HCV-infected patients and Huh7.5.1^dif cells (Supplementary table S5) by using Enrichr28. Furthermore, we added to the 11 leading edge genes described in Fig. 3B common transcription factors potentially binding to more than 2 leading edge genes. As a result, we predicted 15 transcription factors potentially involved in the transcription of peroxisomal genes and suppressed by HCV infection (Supplementary table S5). Among these HCV-sensitive regulators were PPAR-alpha (PPARA) and its functional partner Retinoid X Receptor Beta (RXRB). Moreover, a comparison of the identified transcription factors with the HCVpro protein interaction database of HCV-host interactions⁴⁰, revealed that HCV core protein associates to PPAR-alpha and RXR-beta. In line with these computational analyses, we observed that HCV infection strongly inhibits PPAR-alpha expression in Huh7.5.1^dif (Fig. 5A). Finally, we analyzed the RNA-Seq profiles of the HCV infection and non-infected control time courses using the AMARETTO algorithm⁴¹. This regulatory network inference tool learns regulatory modules by connecting known regulatory driver genes with co-expressed target genes that they control. Interestingly, AMARETTO highlighted interleukin 6 receptor (IL-6R) as a driver candidate of a module comprising a significant (FDR<0.001) functional negative enrichment of pathways associated to peroxisome function (7 genes), fatty acid (12 genes) and lipid metabolism (20 genes) (Fig. S7, Supplementary Table S6–7). This suggests a functional role for IL-6/STAT3 signaling in the regulation of peroxisome function in HCV infection. Indeed, infection induces a STAT3 transcriptional signature⁵ (Supplementary Fig. S6) and STAT3 activation cause a rapid inhibition of IL-6R transcripts (Fig. 5B) as part of a negative feedback regulation. Moreover, inhibition of STAT3 activity by niclosamide (Fig. 5C) rescued the virus-induced inhibition of peroxisomal genes (Fig. 5D) suggesting a regulatory link between STAT3 signaling and peroxisomes. Indeed, inflammatory pathways regulate PPAR family members⁴².

Fig. 5. HCV inhibits peroxisomal gene expression by suppressing PPAR-alpha function via STAT3 signaling.

(A) HCV infection of Huh7.5.1^dif significantly (p<0.05, U-Test) inhibits PPAR-alpha expression. Western blotting for PPAR-alpha and actin after 7 days of infection with HCVcc (Jc1). Quantification of band intensities of PPAR-alpha and actin. Mean relative PPAR-alpha intensity ±SD (2 independent experiments in triplicates). (B) Activation of the IL-6/STAT3 pathway rapidly downregulates IL-6 receptor (IL-6R). Huh7.5.1 cells were incubated 6 h with 10 ng/mL IL-6 prior RNA extraction and qPCR analysis. Mean fold change (FC) ±SEM triplicates. * p<0.05 (T-Test). (C) Niclosamide inhibits IL-6-induced STAT3 phosphorylation. Incubation of Huh7.5.1 cells with 2 μM niclosamide for 24 h and/or 10 ng IL-6 (30 min). Western blotting for pSTAT3 (Y705) and total STAT3. (D) HCV-induced inhibition of peroxisomal genes is rescued by the STAT3-inhibitor niclosamide. Huh7.5.1^dif cells were infected with HCVcc (Jc1) for 7 days. At day 6 pi cells were treated with solvent control or 2 μM niclosamide. Niclosamide reversed the HCV-induced inhibition of 5 top leading-edge genes of the HALLMARK_PEROXISOME gene set (Fig. 3B, Supplementary Fig. S4), while it had no effect on HCV replication. Mean fold change of copy number normalized to GAPDH from triplicates ±SD. *p<0.05 T-Test.

Taken together, our data suggest that HCV infection suppresses peroxisomal function at different levels by interfering with PPAR-alpha expression and function. This impacts the metabolic switch between fatty acid and glucose metabolism and thus contributes to glucose-dependence and lipid accumulations in infected hepatocytes that is a hallmark of chronic liver disease and HCC development.

Impaired peroxisomal function is associated with HCV clinical liver disease including cirrhosis and HCC

As perturbed peroxisome function might be relevant for outcome of HCV-associated liver disease, we analyzed the gene expression in liver biopsies from 216 patients with HCV-related early-stage liver cirrhosis⁴³ as well as in paired liver biopsies of patients with HCV-associated HCC⁴⁴ (Supplementary table S8–9). Comparing the enrichment of the HALLMARK_PEROXISOME gene set in each individual sample to the clinical outcome of the patients, we revealed a significant association of peroxisomal gene expression with liver cirrhosis (p=0.001, log-rank test), HCC development (p=0.03, log-rank test) and patient survival from HCC (p=0.006, log-rank test) (Fig. 6A). This suggests that HCV patients with low peroxisomal function have high risk for poor clinical outcomes. Moreover, peroxisomal gene expression significantly correlates with collagen, type I, alpha 1 expression (COL1A1) (R=−0.16, p=0.02, Spearman correlation test) and with α-smooth muscle actin (ACTA2) (R= −0.19, p=0.004, Spearman correlation test) (Fig. 6B), suggesting elevated extracellular matrix deposition and liver injury in patient livers with low peroxisomal activity⁶. This also reflects the observed negative association of the peroxisomal gene expression with the Child-Pugh liver disease score in Fig. 6A. Interestingly, peroxisomal gene expression was reduced in tumor (T) and adjacent tissue (A) of HCC patients with steatohepatitis (SH-HCC) (Fig. 6C) suggesting that an impaired peroxisome is not only associated to fatty liver development but also a hallmark of tumors reflecting their glucose dependence and the shutting down of beta-oxidation^{45, 46}. We validated key findings in Fig. 6 using CAT expression (Supplementary Fig. S8) thus, confirming the reliability of CAT as functional peroxisomal marker in patients. Reactivation of peroxisomal function during liver disease may therefore reduce oxidative stress and the risk of disease progression toward liver cirrhosis and HCC.

Fig. 6. Significant association of hepatic peroxisome expression with clinical outcomes and phenotypes in viral and metabolic liver disease.

(A) Patients with impaired peroxisomal function showed worse outcome compared to those with intact function (log-rank test). Early-stage HCV cirrhosis patients (n=216)⁴³ are classified into 2 groups based on relative peroxisomal function. Therefore, modulation of the HALLMARK_PEROXISOME gene set²⁵ in each individual sample was used to infer peroxisomal function in the liver tissues. Impaired peroxisomal function was defined by coordinated suppression of the gene set determined by modified gene set enrichment analysis with statistical significance (false discovery rate <0.10)⁶. (B) Expression of fibrosis-related genes, COL1A1 and ACTA2, tend to be higher in the livers with a suppressed peroxisome pathway (Spearman correlation test). Induction or suppression of the HALLMARK_PEROXISOME gene set were measured by gene set enrichment index (GSEI)⁶. GSEI was calculated from gene set enrichment p value based on iterative random gene permutations (1,000 times). GSEI of +3 indicates induction at enrichment p= 0.001, GSEI of −3 indicates suppression at enrichment p=0.001, and GSEI of 0 indicates no modulation at enrichment p=1.0. (C) The peroxisome is significantly (p<0.05) suppressed in histological steatohepatitis HCC (SH-HCC) (n=17) compared to other histological types (n=82) in tumor samples (T) (NES=−1.38) and in tissue adjacent to tumors (A) (NES=−1.43) of paired liver biopsies⁴⁴. GSEA using the HALLMARK_PEROXISOME gene set.

DISCUSSION

Chronic liver disease generally progresses from steatosis, inflammation to fibrosis, cirrhosis and ultimately to HCC. The striking similarities in liver disease progression independent of the underlying etiology, suggests the presence of common drivers and deregulated pathways that promote liver pathogenesis. In this study, we have unraveled the temporal proteogenomic atlas of persistent HCV infection that sheds light to open questions in HCV-host interactions and uncovers virus-induced perturbations of host cell circuits driving viral pathogenesis and disease progression. Using a highly efficient and reproducible infection model, we mapped and quantified for the first time the entire viral proteome in the early and steady-state phase of infection. Surprisingly, infection of the quiescent hepatocyte-like cells did mount a robust interferon response on the RNA level including expression signatures from dsRNA sensors RIG-I, MDA5, TLR3 and their downstream effector IRF3. This is consistent with reports demonstrating MDA5 being more critical for HCV sensing than RIG-I^{47, 48}. HCV replication intermediates are also partially recognized by the TLR3 protein, however not sufficient enough to mount a fully effective host response against HCV⁴⁹. However, all these antiviral gene expression patterns do not translate into protein in Huh7.5.1^dif. These findings support a model where HCV infection overcomes the antiviral defense in hepatocytes mainly by inhibiting the translation of ISG mRNAs via the induction of PKR activity as previously suggested⁵⁰. Indeed, HCV-infection of our model strongly induced PKR (EIF2AK2) expression at the RNA level which led to a self-limiting expression of PKR protein that spiked at day 1 pi (Supplementary table S1). Why in HCV-infected chimeric mice this evasion strategy seems to be distinctly different from HCV-infected patients and Huh7.5.1^dif is an interesting observation and may be useful to refine our understanding of viral immune evasion. This shows that the Huh7.5.1^dif is a useful model to study and understand the innate immune responses to HCV infection and serves as an example of how this data set can serve as a resource file to understand temporal changes and putative discrepancies in the proteogenomic landscape of HCV-host interactions. Using this atlas, we validated cancer relevant pro-oncogenic pathways and gene signatures in a thus far unparalleled depth and comprehension (Supplementary Fig. S6) and correlated it with metabolic features resembling cancer, i.e. a Warburg-like shift of the lactate flux in infected cells. Of note, the transcriptomic patterns identified in HCV-infected Huh7.5.1^dif were similar to those in liver tissue of patients and HCV-infected human liver chimeric mice (Figs. 3A, Supplementary Figs. S3 and S6) further emphasizing the suitability of this atlas to identify previously unrecognized disease-relevant processes in vivo.

Peroxisomes are key organelles for VLCF metabolism and detoxification of membrane-permeable peroxides and oxidative stress. We observed a marked decrease of the peroxisomal function in HCV-infected hepatocytes, in the livers of HCV patients and in infected chimeric mice, which is distinctly different from patients with chronic hepatitis B where peroxisomal function is increased (Fig. 3). Consistently, HCV but rarely HBV infection is associated with fat accumulations in the liver leading to NAFLD and NASH⁵¹. Both HBV and HCV induce NF-kB signaling⁵², contributing to chronic liver inflammation, injury and disease progression. Inflammation is a regulator of host metabolism⁵³. However, only HCV but not HBV infection impairs peroxisomal function and predominantly induces fat inclusions in hepatocytes arguing for an additional HCV-specific molecular mechanism suppressing fatty acid metabolism. Complementary to previous lipidomic analysis that suggested an HCV-induced accumulation of phospholipids and sphingomyelins³⁸ our data revealed that HCV cause a phenotype of VLCFA accumulation, which corresponds to impaired peroxisomal beta-oxidation. Consequently, HCV infection of differentiated cells as used here mimics a phenotype resembling fatty liver disease and its complication NASH, which is characterized by fatty acid accumulations and chronic inflammation⁵⁴. HCV infection requires lipid droplets for particle assembly⁵⁵. The suppressed peroxisomal activity in HCV-infected cells is thus of potential advantage to the virus since it contributes to accumulation of long-chain fatty acids in infected hepatocytes with potential impact on liver pathogenesis. Indeed, HCV-infected cells display a higher abundance of phosphatidylcholines and triglycerides with longer chain fatty acids⁵⁶, which are preferentially metabolized in the peroxisome. This resembles the situation in NASH patients, where increased systemic phosphatidylcholine levels are observed⁵⁷ and suggests that HCV infection is a model to study the molecular mechanisms of fatty liver disease independent of the underlying etiology. Moreover, high levels of VLCFA in serum provoked hepatic steatosis, NASH, and HCC in an animal model⁵⁸. Our data suggest that re-activation of HCV-impaired nuclear receptors like PPAR-alpha has a potential clinical relevance for HCC prevention. Furthermore, targeting of PPAR-alpha is in development for NASH, where treatment options are currently limited. Indeed, PPAR stimulation improves steatosis, inflammation and fibrosis in pre-clinical models of NAFLD^{59, 60}, albeit with limited clinical efficacy. Alternatively, our data highlight a potential new strategy for a restoration of peroxisomal function using clinical STAT3 inhibitors. Collectively, these data indicate that HCV and NASH share similar pathways driving the pathogenesis of liver disease, which are distinctly different from HBV-associated liver disease. The highly similar results obtained in livers of HCV-infected chimeric mice and livers and HCV patients suggest the hepatocyte-like HCV-Huh7.5.1^dif as suitable model for the discovery of targets and compounds of metabolic liver disease. Our atlas validated previous transcriptomic studies demonstrating an association of HCV with genes involved in lipid metabolism and reactive oxygen species^{61, 62} in an unprecedented temporal resolution, which allowed the prediction of regulatory networks (Supplementary data S7). The multi-omics approach in this study integrated transcriptome, proteome, polar metabolites and lipid analysis revealed novel mechanistic insights in the regulation of peroxisomal function by an interplay of glucose levels and HCV-induced cytokine signaling.

Finally, the temporal proteogenomic atlas of HCV infection is a useful and unique resource data set for researchers to validate individual hypotheses in virus-host interactions and liver disease biology. Moreover, the convenient upscaling, the high reproducibility, and the high similarity with gene expression profiles in livers of HCV patients emphasizes the potential of this model for screening approaches targeting drivers of liver disease pathobiology and cancer-risk to identify therapeutics for liver disease in general.

Abbreviations:

ATCC

American Type Culture Collection

ChEA

DNA sequencing after chromatin immunoprecipitation in mammalian cells

ChIP-Seq

DNA sequencing after chromatin immunoprecipitation

CTRL

control

DAPI

4’,6-diamidino-2-phenylindole

DMSO

dimethyl sulfoxide

dsRNA

double-stranded RNA

EDTA

ethylenediaminetetra-acetic acid

EMT

epithelial-mesenchymal transition

ES

enrichment score

GC-MS

gas chromatography-mass spectrometry

GEO

Gene expression Omnibus

GSEA

gene set enrichment analysis

HBsAg

hepatitis B virus surface antigen

HBV

hepatitis B virus

HCC

hepatocellular carcinoma

HCV

hepatitis C virus

HCVcc

cell culture-derived hepatitis C virus

IL-6

Interleukin 6

ISGs

interferon-stimulated genes

LC-MS/MS

liquid chromatography-tandem mass spectrometry

MAPK

mitogen-activated protein kinase

MOI

Multiplicity of Infection

MSigDB

The Molecular Signatures Database

NAFLD

non-alcoholic fatty liver disease

NASH

non-alcoholic steatohepatitis

NES

normalized enrichment score

NF-kB

Nuclear factor kappa B

pi

post-infection

PMSF

phenylmethylsulfonyl fluoride

PPARA

Peroxisome proliferator-activated receptor alpha

PPI

protein-protein interaction

RNAi

RNA interference

RNA-Seq

RNA sequencing

SD

standard deviation

SH

steatohepatitis

SH-HCC

steatohepatitis-associated hepatocellular carcinoma

SRA

Sequence Read Archive

STAT3

Signal transducer and activator of transcription 3

TCID₅₀

50 % tissue culture infective dose

TMT

tandem mass tag

uPA/SCID

urokinase-type plasminogen activator/severe combined immunodeficiency

VLCFA

very long-chain fatty acid.