Potential conflict of interest: D.L.T. owns stock in KMT Hepatech, Inc.
Supported by Canadian Institute for Health Research (CIHR) grants to J.P.P. and D.L.T.
MicroRNAs (miRNAs) are small RNAs that posttranscriptionally regulate gene expression. Their aberrant expression is commonly linked with diseased states, including hepatitis C virus (HCV) infection. Herein, we demonstrate that HCV replication induces the expression of miR-27 in cell culture and in vivo HCV infectious models. Overexpression of the HCV proteins core and NS4B independently activates miR-27 expression. Furthermore, we establish that miR-27 overexpression in hepatocytes results in larger and more abundant lipid droplets, as observed by coherent anti-Stokes Raman scattering (CARS) microscopy. This hepatic lipid droplet accumulation coincides with miR-27b's repression of peroxisome proliferator-activated receptor (PPAR)-α and angiopoietin-like protein 3 (ANGPTL3), known regulators of triglyceride homeostasis. We further demonstrate that treatment with a PPAR-α agonist, bezafibrate, is able to reverse the miR-27b-induced lipid accumulation in Huh7 cells. This miR-27b-mediated repression of PPAR-α signaling represents a novel mechanism of HCV-induced hepatic steatosis. This link was further demonstrated in vivo through the correlation between miR-27b expression levels and hepatic lipid accumulation in HCV-infected SCID-beige/Alb-uPa mice. Conclusion: Collectively, our results highlight HCV's up-regulation of miR-27 expression as a novel mechanism contributing to the development of hepatic steatosis. (Hepatology 2014;58:98–108)
Hepatitis C virus (HCV) is a positive sense RNA virus from the Flaviviridae family that currently infects ∼2.35% of the global population. HCV encodes three structural proteins (core, E1, and E2) and seven nonstructural proteins (p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B), and relies on host pathways to facilitate its lifecycle. HCV-associated host factors include both coding and noncoding genes, such as microRNAs (miRNAs), which are small RNAs, ∼20-25 nucleotides in length, which posttranscriptionally regulate virtually every cellular pathway. Unlike synthetic silencing RNAs that are designed to target individual genes, miRNAs have evolved to regulate many targets, thereby exerting greater regulatory control. Several viruses modulate the host miRNAs for their pathogenesis. HCV displays interactions with components of the RNA silencing pathway,[6-8] and direct interactions with a liver-abundant miRNA, miR-122.[5, 9] Hepatic miRNAs can influence HCV either through direct interactions with the viral genome or regulation of HCV-associated host pathways.
MicroRNA-27 (miR-27) represents a liver-abundant miRNA whose role in HCV pathogenesis is poorly understood. miR-27 regulates lipid metabolism in adipocytes and macrophages and is implicated in atherosclerosis. Furthermore, miR-27 is deregulated in liver metabolic disorders,[12-14] suggesting it plays a role in hepatic lipid metabolism, a critical host pathway hijacked by HCV to facilitate its lifecycle and pathogenesis.[15, 16] HCV-induced modulations of lipid metabolism include increased cellular triglyceride and cholesterol storage to facilitate viral replication.[15-17] Furthermore, both cholesterol and lipoprotein[19, 20] receptors have been implicated as HCV entry factors. Viral particle assembly and secretion also use components of the very-low density lipoprotein (VLDL) pathway. Given this intimate link between HCV and hepatic metabolism, we examined the role of miR-27 in HCV pathogenesis and, herein, establish its role in HCV-induced hepatic steatosis.
Materials and Methods
The pFK-I389luc/NS3-3'/5.1 plasmid containing the HCV subgenomic replicon (genotype 1b isolate Con1, GenBank accession no. AJ242654) and the NS5B active site mutant replicon were kind gifts from Dr. Ralf Bartenschlager (Institute of Hygiene, University of Heidelberg, Heidelberg, Germany). The Huh7.5 cell line stably expressing the full-length HCV genotype 1b replicon with a S2204I adaptive mutation in NS5A (Huh7.5-FGR) was a kind gift from Dr. Charles M. Rice (Rockefeller University, New York, NY) and Apath (St. Louis, MO).
CARS Microscopy Imaging
Imaged cells were washed twice with phosphate-buffered saline (PBS), followed by a 15-minute incubation at room temperature with fixing solution (4% formaldehyde, 4% sucrose, 1 mL). The fixed cells were washed twice with PBS for 3 minutes and then stored at 4°C in PBS prior to imaging. The imaging and subsequent quantitative voxel analysis of TG content was performed as described.[22, 23] Lipid droplet (LD) sizing/counting was performed using ImageJ (NIH, Bethesda, MD).
Immunofluorescence and Oil Red O Staining
Liver frozen sections (at 4 μm thickness) were fixed in 4% freshly made paraformaldehyde for 30 minutes, followed by 5 minutes PBS rinse to remove excess paraformaldehyde. Fixed slides were then permeabilized in PBS containing 0.5% Triton X-100 for 10 minutes and blocked in PBS with 10% normal goat serum for 1 hour. The 1/100 diluted primary rabbit monoclonal antibody specifically recognizing human Cytokeratin 18 (CK-18) (Abcam, Cambridge, MA) was applied to the liver sections and incubated at 4°C overnight. The next day liver sections were incubated in secondary antibody cocktail, including Alexa Fluor 488-conjugated goat antirabbit and DAPI, for 1 hour in the dark. After 3 washes of PBS, slides were immersed in Oil Red O working solution (freshly prepared in 30% triethyl-phosphate), for 30 minutes in the dark, followed by 3 rinses with distilled water. Finally, slides were rinsed in the dark for 10 minutes, air dried, mounted with prolong gold mounting medium (Invitrogen), and coverslipped. Samples were examined with a Leica TCS SP5 confocal microscope. Oil Red O staining of lipids was visualized at far-red wavelength: 633 (ex) and 647 (em). Images were processed using LAS AF Lite software.
HCV Infection Induces miR-27 Expression
Two isoforms of miR-27, miR-27a and 27b, are encoded by separate gene loci and differ by one nucleotide. (Fig. 1A; Supporting Fig. S1). We examined whether HCV modulates the expression of either miR-27 isoform. Huh7.5 cells were transfected with subgenomic replicon (HCV-SGR) from the Con1 isolate (genotype 1b; Fig. 1B). Relative miR-27 expression was analyzed by quantitative reverse-transcription polymerase chain reaction (qRT-PCR). HCV-SGR induced a 2-fold up-regulation of miR-27a expression and 5-fold up-regulation in miR-27b expression (Fig. 1C). Transfection of replication-deficient HCV-SGR ΔNS5B maintained a 2-fold up-regulation of miR-27a (Fig. 1C); however, miR-27b levels did not increase (Fig. 1C). These observations indicate that viral replication is required for miR-27b up-regulation but HCV translation is sufficient to activate miR-27a expression.
Next we examined miR-27 expression during HCV infection. We performed qRT-PCR analysis on Huh7.5 cells infected with JFH-1T, a cell-culture adapted high-titer strain of JFH-1 (genotype 2a). Up-regulation of both miR-27a (2.6-fold; Fig. 1D) and miR-27b levels (1.2-fold; Fig. 1E) was observed. These results confirm that HCV infection induces miR-27 expression, and this induction is conserved across HCV genotypes.
To probe the molecular mechanism by which HCV regulates miR-27, we used an miR-27 sensor plasmid containing a dual-luciferase reporter bearing two fully complementary miR-27b binding sites in the 3′-untranslated region (UTR) of the Renilla luciferase gene. Since miR-27a and miR-27b differ by only one nucleotide, both isoforms regulate luciferase activity. Huh7 cells were cotransfected with HCV proteins and the miR-27 sensor plasmid. HCV core and NS4B expression independently induced a decrease in luciferase signal relative to the controls (Fig. 1F). This down-regulation was reversed upon mutation of the miR-27 binding sites, demonstrating miR-27-specific activity. qRT-PCR confirmed that both core and NS4B overexpression resulted in increased miR-27a/b levels (Supporting Fig. S2).
miR-27b expression can be activated in a PI3K pathway-dependent manner. Since both NS4B and core have previously been shown to activate SREBP by way of the PI3K/Akt pathway,[27, 28] we hypothesized that these proteins may regulate miR-27b expression similarly. Huh7 cells were cotransfected with NS4B and core and miR-27 sensor plasmid and then treated with a PI3K inhibitor, LY294002. The results showed LY294002 impaired HCV proteins' ability to induce miR-27-mediated gene silencing (Supporting Fig. S3), suggesting that HCV activates miR-27 expression in a PI3K-dependent fashion.
miR-27 Regulates Hepatic Lipid Homeostasis
We next examined whether miR-27 plays a regulatory role for lipid metabolism in Huh7 cells by transfecting with control or miR-27 mimics and inhibitors and measuring the effects. The activity of miR-27b mimics and inhibitors was confirmed using the sensor plasmid (Supporting Fig. S4). We used coherent anti-Stokes Raman scattering (CARS) microscopy, a modern multiphoton imaging method, to image miR-27's influence on hepatic lipid content in a highly effective manner. CARS has been used extensively for label-free imaging and quantification of hepatic lipid content in biological systems, thereby avoiding perturbations and artifacts that can be introduced by added dyes and staining protocols.[22, 23] Transfection of miR-27a and miR-27b mimics in Huh7 cells induced an increase in both the size and abundance of LDs (Fig. 2A-C). The average LD diameter increased from 540 ± 10 nm to 600 ± 10 nm (n > 1,900 LDs; P < 0.01) during miR-27b overexpression. Similar results were observed in Huh7.5 cells (Supporting Fig. S5). To exclude the possibility that miR-27 mimics resulted in cytotoxicity, we performed 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide (MTT) assays on transfected Huh7 cells, and no significant changes in cell viability were observed (Supporting Fig. S6A).
Next we sought to identify the relevant endogenous targets of miR-27 that might induce lipid accumulation. We examined messenger RNA (mRNA) levels using qRT-PCR to confirm that they are miR-27 targets. Huh7 cells were transfected with miR-27b or control mimics, and qRT-PCR revealed an inverse correlation between miR-27b activity and the mRNA levels of PPAR-α and angiopoietin-like protein 3 (ANGPTL3) (Supporting Fig. S7A), consistent with previous reports.[14, 29] Both of these genes have conserved miR-27 binding sites (Supporting Fig. S7B), and have known links to triglyceride homeostasis.[14, 29]
PPAR-α is a key nuclear receptor that transcriptionally activates genes associated with fatty acid oxidation. Consistent with previous findings linking PPAR-α inhibition with steatosis, small molecule-based antagonism of PPAR-α signaling in Huh7 cells can induce triglyceride (TG) accumulation (Supporting Fig. S8). If miR-27's induction of hepatic lipid storage relied on inhibition of PPAR-α signaling and the resulting triglyceride accumulation, activating the PPAR-α pathway should reverse the effect. Treatment with a small molecule PPAR-α agonist, bezafibrate, was sufficient to reverse miR-27-induced lipid accumulation to levels observed in control cells, confirming this hypothesis (Fig. 3). Overall, these observations suggest that miR-27 overexpression induces triglyceride accumulation through repression of PPAR-α expression.
miR-27 Regulates the HCV Lifecycle
Our previous work showed that PPAR-α antagonism is effective at inhibiting HCV replication. To examine if miR-27 has a similar effect, we overexpressed miR-27b in Huh7.5 cells stably expressing the HCV full length replicon (Fig. 4A). Interestingly, ectopic miR-27b expression resulted in a 3-fold down-regulation of HCV RNA (Fig. 4B). A similar down-regulation was observed in HCV NS3 and NS5A proteins by western blot (Fig. 4C). No cytotoxicity was observed during miR-27b overexpression (Supporting Fig. S6B). Triglyceride assays revealed that miR-27 overexpression also induced an accumulation of cellular triglyceride in the Huh7.5-FGR cells (Fig. 4D), consistent with our observations in Huh7 cells (Fig. 2). This down-regulation in HCV expression correlated with decreased mRNA levels of PPAR-α and ANGPTL3 (Fig. 4E). Our results confirm that miR-27 overexpression inhibits HCV replication.
Interestingly, we also observed down-regulation of retinoid X receptor alpha (RXR-α), a previously reported target of miR-27 (Supporting Fig. S9A,B). This protein interacts with several nuclear receptors, including PPAR-α, to regulate liver lipid biosynthesis. Therefore, we examined the functional relevance of miR-27-mediated repression of RXR-α expression on HCV replication and lipid metabolism. We performed CARS imaging on Huh7 cells treated with an RXR-α antagonist, UVI-3003, which inhibits the RXR-α's interactions with all other nuclear receptors. Huh7 cells treated with this drug displayed no change in hepatic lipid content (Supporting Fig. S9C). Additionally, RXR-α antagonism in Huh7.5-FGR cells produced no changes in HCV levels.
We were also interested in how miR-27's regulation of PPAR-α signaling would affect viral infectivity. Previous work suggested that increased PPAR-α expression blocks assembly of HCV infectious particles. Huh7.5 cells were cotransfected with JFH-1T RNA and miR-27b mimics and inhibitors, and intracellular HCV RNA levels were measured by qRT-PCR. Neither the miR-27 mimic nor the miR-27 inhibitor had any effect on JFH-1T replication (Supporting Fig. 10), suggesting that miR-27b overexpression has a genotype-specific effect on HCV replication. On the other hand, miR-27b inhibition resulted in a very modest decrease in secretion of infectious HCV, while miR-27b overexpression had no effect on secreted virus' infectivity, consistent with PPAR-α's previously reported antiviral role in HCV secretion. Independent of miR-27's effects on the viral lifecycle, its conserved induction across HCV genotypes manifests globally as a contributor to hepatic steatosis and thus to HCV-associated liver disease.
HCV Infection In Vivo Activates miR-27 Expression
We continued our evaluation of miR-27 expression in a small animal model of acute HCV infection, using the humanized SCID-beige/Alb-uPa mouse model. We infected the chimeric mice with genotype 1a and 2b clinical isolates of HCV (Supporting Fig. S11). qRT-PCR analysis of miR-27b levels revealed a 2.9-fold up-regulation in miR-27b levels 7 weeks postinfection (Fig. 5A). This increase was conserved across both HCV genotypes examined. There was also a 2.0-fold increase in miR-27a levels (Fig. 5B). Oil Red O staining of lipids in the chimeric liver's human hepatocytes revealed a correlation between cellular lipid levels and miR-27 expression in mice (Fig. 5C), and provides further support for our CARS microscopy results in cell culture experiments. Collectively, our in vivo data confirm that HCV infection induces expression of miR-27, and is consistent with miR-27's role as a key molecular determinant of hepatic steatosis.
Endogenous miRNAs posttranscriptionally regulate virtually every cellular process, so it is not surprising that viruses modulate the host miRNA milieu in different ways to facilitate pathogenesis. Herein, we have shown that a liver-abundant miRNA, miR-27, is robustly induced by HCV in both in vitro and in vivo models (Figs. 1, 5), and this modulation is conserved across at least two genotypes (Figs. 1, 2, 5). HCV-induced expression of miR-27b requires replication of the virus while viral translation is sufficient to activate miR-27a expression (Fig. 1C,D), suggesting these isoforms are modulated by HCV through different mechanisms.
In order to understand HCV's induction of miR-27, we studied its effects on hepatocytes. Overexpression of either isoform of miR-27 causes an accumulation of hepatic lipid content in the presence or absence of HCV (Figs. 2, 5). The correlation between miR-27 expression and cellular lipid content was also observed in HCV-infected SCID-beige/Alb-uPa mice (Fig. 5C). This represents, to the best of our knowledge, the first report visualizing HCV-induced hepatic lipid accumulation in SCID-beige/Alb-uPa mice, highlighting the model's utility for studying HCV-associated steatosis. Together, these data demonstrate that the up-regulation of miR-27 by HCV contributes to increased lipid accumulation and larger LDs.
Accumulation of hepatic LDs correlates with increased expression of miR-27 whose predicted target genes are associated with lipid metabolism (PPAR-α and ANGPTL3) (Supporting Fig. S4A). Targetscan predicts that PPAR-α mRNA possesses two miR-27 binding sites in its 3′-UTR, the region generally targeted by microRNAs (Supporting Fig. S7). Previous work suggested that miR-27b regulates PPAR-α largely at the translational level. Our results suggest a direct interaction between miR-27b and PPAR-α mRNA; however, Kida et al. were not able to confirm a functional interaction in their predicted miR-27 binding sites of PPAR-α. Our observation of decreased PPAR-α mRNA during miR-27b overexpression strongly suggests a miR-27-induced effect at the mRNA level as well, and may reflect differences in cells, in transfection efficiency, and in potency of mimics. ANGPTL3 harbors a poorly conserved miR-27 binding site in the 3′-UTR and a highly conserved open reading frame (ORF) site (Supporting Fig. S7) predicted to be functional, as it is preceded by rare codons (Supporting Fig. S7). These rare codons can cause ribosomal pausing and allow stable interactions between miR-27 and the binding site. Our results suggest that miR-27b regulates ANGPTL3 at the RNA level, consistent with previous results.
PPAR-α heterodimerizes with RXR-α to transcriptionally activate genes associated with fatty acid β-oxidation. Our data shows that HCV inhibits the PPAR-α pathway through enhancement of miR-27-mediated repression of PPAR-α expression that also leads to TG accumulation. PPAR-α expression is known to be misregulated during HCV infection. PPAR-α antagonism leads to hepatic lipid accumulation. miR-27's induction of lipid accumulation was also reversed by the PPAR-α agonist bezafibrate (Fig. 3). Therefore, HCV-induced expression of miR-27 represents a novel mechanism by which the virus inhibits PPAR-α signaling and promotes steatosis (Fig. 6).
Overexpression of individual viral proteins revealed that both core and NS4B independently activate miR-27a and miR-27b expression (Fig. 1F; Supporting Fig. S2). Both of these viral proteins have previously been reported to promote lipogenesis. In the case of HCV core, its expression has previously been shown to down-regulate PPAR-α expression. Separate studies demonstrated that HCV core and NS4B promote SREBP activity through the PI3K pathway. Our results suggest that the viral proteins also use the PI3K pathway for activation of miR-27 expression to induce steatosis (Supporting Fig. S3). Furthermore, these results are consistent with a model of steatosis where HCV core modulates PPAR-α expression through up-regulation of miR-27 expression.
The observed repression of ANGPTL3 (Supporting Fig. S4A) may be another mechanism by which HCV-induced miR-27 expression promotes triglyceride accumulation in vivo. A previous study suggested that miR-27b inhibits ANGPTL3 expression in response to dyslipidemia to prevent lipid accumulation in circulation. This is due to its role as an inhibitor of lipoprotein lipase (LPL), a key enzyme in free fatty acid uptake. Decreased ANGPTL3 levels would lead to increased LPL activity and fatty acid uptake into hepatocytes, highlighting an additional mechanism contributing to miR-27's role in HCV-induced steatosis in patients.
Our results also suggest that miR-27 levels can influence the HCV viral lifecycle. At the level of replication, miR-27b appears to play an antiviral role against HCV genotype 1b replication (Fig. 4). As miR-27 is not predicted to have conserved binding sites in the HCV genome, inhibition of HCV replication is most likely dependent on miR-27's regulation of host gene expression. HCV genotype 2a appears less susceptible to miR-27-mediated inhibition (Supporting Fig. 10), consistent with previous observations of sequence-dependent variation in HCV resistance against metabolic inhibitors. Our previous work demonstrated that PPAR-α antagonism is capable of inhibiting genotype 1b HCV replication by inducing hepatic lipid accumulation and blocking the biosynthesis of new lipids required for protein lipidation. This disrupts the HCV-induced cellular lipid environment required for efficient HCV replication. Here we propose an analogous model where miR-27 acts like an endogenous PPAR-α antagonist, resulting in disruption of HCV replication complexes (Fig. 6). An additional antiviral mechanism in vivo for miR-27 may lie in its regulation of ANGPTL3. Due to LPL's proposed inhibitory role against HCV entry, miR-27 may have an additional antiviral effect at the level of entry by decreasing the level of ANGPTL3-mediated inhibition of LPL.
While this article was under review, a study was published reporting activation of miR-27a expression by HCV. Shirasaki et al. focused on miR-27a and showed that it similarly regulates lipid metabolism genes, including PPAR-α, and also observed a correlation between miR-27a expression and severity of steatosis in patients, consistent with our findings. The authors also elegantly demonstrate that ABCA1 is a target of miR-27a, influencing both the viral lifecycle and lipid metabolism. Both studies observed modest influences of miR-27 on viral infectivity (less than one log changes). Moreover, while both studies observed a similar correlation between cellular lipid content and miR-27a expression, Shirasaki et al. suggest miR-27a overexpression results in decreased LD formation, contrary to our observations (Fig. 2D). This apparent discrepancy may be attributed to Shirasaki et al. examining the effect of miR-27a expression in Huh7.5 cells either expressing HCV or supplemented with oleic acid where the cell's metabolic state is shifted. Our data across different cell lines and in HCV infected SCID-beige/Alb-uPa mice using different high-resolution imaging techniques clearly show that miR-27a and miR-27b up-regulate hepatic LD biogenesis and contribute to hepatic steatosis.
It is interesting to consider the multiple mechanisms evolved by the virus to manipulate host lipid homeostasis. These independent mechanisms likely arose out of necessity for the virus to use different cellular components during its lifecycle, such as modified endoplasmic reticulum (ER) membranes, LDs, and the VLDL pathway.[15, 16] In some cases, these effects appear contradictory, but likely arose from competing evolutionary pressures. The overall degree of synergy between these independent mechanisms may be instrumental, at the clinical level, to determining patient susceptibility to HCV-induced steatosis. Future work should examine whether miR-27 is a predictive biomarker of steatosis in vivo, as this would be in line with previous studies reporting a correlation between lower PPAR-α levels and HCV-associated steatosis.
In summary, we have shown that HCV activates miR-27 expression, and this is conserved across genotypes. Expression of both isoforms of miR-27, miR-27a and miR-27b, are activated by HCV infection, and these miRNAs can independently induce lipid droplet biogenesis and accumulation. Our data suggest that HCV-induced miR-27 expression, and the resultant down-regulation of PPAR-α and ANGPTL3, represent a novel mechanism by which the virus induces steatosis.
R.S. thanks the NSERC for funding in the form of a Vanier Scholarship. R.S., N.N., and R.C. thank the NCRTP-HepC for additional training and support. P.S. thanks NSERC for an Undergraduate Student Research Award. R.K.L. thanks OGS for a graduate scholarship. We thank Dr. A. Stolow and Dr. A. Ridsdale for their assistance and useful discussion regarding CARS microscopy.