Apolipoprotein E on Hepatitis C Virion Facilitates Infection through Interaction with Low Density Lipoprotein Receptor

Abstract

Hepatitis C virus (HCV) infection is a major cause of liver disease. HCV associates with host apolipoproteins and enters hepatocytes through complex processes involving some combination of CD81, claudin-I, occludin, and scavenger receptor BI. Here we show that infectious HCV resembles very low density lipoprotein (VLDL) and that entry involves co-receptor function of the low density lipoprotein receptor (LDL-R). Blocking experiments demonstrate that β-VLDL itself or anti-apolipoprotein E (apoE) antibody can block HCV entry. Knockdown of the LDL-R by treatment with 25-hydroxycholesterol or siRNA ablated ligand uptake and reduced HCV infection of cells, whereas infection was rescued upon cell ectopic LDL-R expression. Analyses of gradient-fractionated HCV demonstrate that apoE is associated with HCV virions exhibiting peak infectivity and dependence upon the LDL-R for cell entry. Our results define the LDL-R as a cooperative HCV co-receptor that supports viral entry and infectivity through interaction with apoE ligand present in an infectious HCV/lipoprotein complex comprising the virion. Disruption of HCV/LDL-R interactions by altering lipoprotein metabolism may therefore represent a focus for future therapy.

Introduction

Hepatitis C virus (HCV) infects nearly 200 million people around the world. Exposure to HCV most often leads to a chronic infection in the liver, eventually causing cirrhosis, cancer, and liver failure requiring transplant (Lauer and Walker, 2001). Current treatment with ribavirin and pegylated interferon-α is suboptimal, achieving sustained virologic response in only 50% of cases (Hoofnagle and Seeff, 2006), such that new and targeted therapies based on the molecular biology of the HCV lifecycle are needed. HCV is an enveloped virus containing a 9600 nt single strand positive-sense RNA genome encoding a polyprotein that is processed by host and viral proteases into a set of structural and nonstructural proteins. The structural proteins include the core nucleocapsid protein and the envelope glycoproteins E1 and E2. E1 and E2 are components of the virus lipid envelope and function to mediate binding to co-receptors on hepatocytes, the target cell of HCV infection (Pileri et al., 1998; Flint et al., 1999; Scarselli et al., 2002). The nonstructural proteins NS2, NS3/4A, NS4B, NS5A, and NS5B form the virus replicase and also function in virus-host interactions that modulate host defenses and cellular permissiveness for infection (Gale, Jr. et al., 1998;Foy et al., 2003; Loo et al., 2006). HCV replicates and assembles in association with lipid droplets thought to be associated with or derived from the endoplasmic reticulum membrane (Miyanari et al., 2007). Once released from infected cells the mature virus particles are found in low density fractions in the serum of infected individuals (Miyamoto et al., 1992;Andre et al., 2002).

The process of cell interaction with the HCV virus particle, and the mechanisms of cell entry by HCV are incompletely understood. However, several HCV co-receptors have been identified using soluble envelope protein binding assays and HCV pseudoparticles expressing E1 or E2 (Bartosch et al., 2003a). These receptors include CD81, SR-BI, claudin-I, and occludin (Bartosch et al., 2003b;Zhang et al., 2004;Evans et al., 2007). CD81 is a member of the tetraspanin family of membrane proteins. The high density lipoprotein (HDL) receptor, SR-BI, also binds E2 and is required for cell entry by HCV pseudoparticles, while blocking antibodies against SR-BI reduce HCV entry into cultured hepatoma cells (Kapadia et al., 2007;Catanese et al., 2007). Claudin-I is a tight junction protein that is required for HCV entry into cells expressing CD81 and SR-BI, and may facilitate direct cell to cell virus spread during infection (Timpe et al., 2008). Recently another tight junction protein, human occludin, has been identified as an HCV entry factor and may play a role in restricting species specificity of HCV infection (Ploss et al., 2009).

The low density lipoprotein receptor (LDL-R) has also been proposed to function as a co-receptor for HCV entry wherein LDL-R-HCV interaction would be facilitated through virus association with host lipoprotein components (Thomssen et al., 1992;Monazahian et al., 1999;Chang et al., 2007;Huang et al., 2007;Gastaminza et al., 2008) Indeed, in previous studies to assess in vitro binding of patient-derived HCV particles to cultured hepatocytes, virus binding was strictly associated with the levels of LDL-R expression despite high expression of CD81 and SRBI on the target cells (Wunschmann et al., 2000;Martin et al., 2008). Cell entry of HCV RNA of patient derived virus isolates was also dependent on LDL-R expression by the primary hepatocyte target cells (Agnello et al., 1999;Molina et al., 2007). However primary hepatocytes and virtually all HCV clinical isolates studied to date do not support efficient productive infection in vitro, thereby limiting our understanding of LDL-R/virus/lipoprotein interactions in HCV infection.

The development of a tissue culture system for productive HCV infection based on genotype 2a strain JFH1 (Wakita et al., 2005) has afforded the possibility of testing the role of these interactions in productive infection using virus particles resembling those produced in vivo. Recent studies using this HCV cell culture infectious system indicate a role for HCV/lipoprotein interactions in HCV infection. The HCV replicase has been shown to localize to sites of very low density lipoprotein (VLDL) assembly within infected cells, from which release of infectious virus is dependent on the microsomal transfer protein (MTP) and secretion of apolioproteins (apo)B and apoE (Huang et al., 2007;Nahmias et al., 2008;Gastaminza et al., 2008;Chang et al., 2007). Importantly, hepatocytes secrete VLDL particles which are composed of a hydrophobic core of triglycerides and cholesteryl esters surrounded by a surface coat containing phospholipids, free cholesterol and two predominant lipoproteins, apoB (present in a single copy) and apoE (multiple copies). Nascent VLDL particles released into plasma are not ligands for LDL-R. However, upon processing by lipoprotein lipase which hydrolyzes the triglycerides in the core of the lipoprotein particles, a large proportion (~70%) of the resulting intermediate density lipoproteins (IDL), is efficiently removed from plasma by LDL-R on hepatocytes. This process depends on the interaction between LDL-R and apoE located on IDL. The remaining IDL in the circulation is converted to LDL by a reaction catalyzed by hepatic lipase, which further reduces the amount of triglycerides in the lipoprotein particles (Chappell and Medh, 1998). β-VLDL is a class of lipoprotein particles enriched in cholesteryl ester that is similar to IDL in that cellular uptake of β-VLDL is dependent on the interaction between LDL-R and apoE associated with the lipoprotein particles. β-VLDL is frequently used as a substitute for IDL to study LDL-R-mediated endocytosis as it can be easily isolated from animals fed with diet enriched in cholesterol, while IDL is difficult to isolate. Once formed from either IDL or -VLDL, LDL is taken up by LDL-R on hepatic as well as nonhepatic cells in a process that relies on interaction between the receptor and apoB associated with LDL. Thus, HCV interaction with host lipoproteins, and formation of HCV/lipoprotein particles, could explain the low density of HCV particles present in patient sera and may define a role for HCV/lipoprotein particles in infection.

In the present study we examined HCV/lipoprotein interaction and function in HCV infection in vitro. Our observations present a role for the LDL-R in mediating HCV entry of cultured cells through an apoE-dependent process of HCV/lipoprotein complexes.

Results

β-VLDL blocks HCV infection via apoE

To assess the role of lipoproteins in supporting HCV infection from putative HCV/lipoprotein particles we infected Huh7.5 cells with the JFH1 stain of HCV at multiplicity of infection (MOI) = 0.5 in the presence of increasing concentrations of HDL, LDL, or β-VLDL. The percentage of infected cells were then determined via flow cytometry assay. When added to cultures prior to HCV infection, β-VLDL reduced the number of infected cells by approximately 60%. In contrast, LDL treatment of cells only moderately reduced the percentage of HCV infected cells, while HDL treatment had no effect on reducing infection (Fig. 1A and B). HDL is a complex containing apolipoproteins AI-AIII that serve to bind SR-BI. β-VLDL and LDL particles both contain apoB and bind to the LDL-R, but β-VLDL particles also contain the higher affinity LDL-R ligand apoE ((Chappell and Medh, 1998); see Fig. 1C). That β-VLDL blocked HCV infection indicates that apoE could preferentially compete for cell binding with HCV/lipoprotein particles. To test this idea, we infected cells in the presence of increasing concentrations of goat polyclonal apoE or sheep polyclonal apoB antibody or nonimmune control antibodies. The presence of apoE but not apoB antibodies rendered a dose-dependent block of HCV infection (Fig. 1D). These results suggest that apoE plays an important role in HCV infection.

Figure 1. β-VLDL competes with apoE for HCV binding and entry.

A) HCV supernatants from infected Huh7.5 cells were mixed with increasing concentrations of HDL, LDL, or β-VLDL and were incubated with Huh7.5 cells at MOI=0.5 for 90 minutes at 37°. After washing with PBS, cells were cultured for an additional 48 hours and intracellular HCV protein was detected by flow cytometry. Data are presented from a single representative experiment (out of three total experiments). B) The percentage of cells infected with HCV in the presence of increasing concentrations of indicated lipoprotein relative to mock-treated cells as determined by flow cytometry. The mean and standard deviation from 3 separate experiments are shown. C) Selected lipoproteins, associated apoproteins and cellular endocytic receptors. D) Cells were infected in the presence of increasing concentrations of α-apoE, α-apoB, or isotype control antibodies and the relative number of infected cells was quantified by focus forming unit assay 48 hours later.

Infectious HCV particles contain apoE

To determine if apoE was physically associated with infectious HCV particles we examined protein, HCV RNA, and overall infectivity of apoE or apoB-associated virus. Immunocomplexes were recovered from immunoprecipitation reactions of cell culture supernatant from HCV-infected Huh7.5 cells using polyclonal antibodies specific for apoE, apoB, or Sendai virus (unrelated virus negative control). The components of the unbound, supernatant fractions and the immunocomplex (pellet fractions) of each immunoprecipitation reaction were then examined in parallel. Anti-apoB or anti-apoE but not anti-Sendai virus antibodies removed the apoB or apoE containing particles from the supernatant fraction and were recovered in the pellet fraction (Fig. 2A). Analysis of the relative amount of HCV RNA associated with each of these samples demonstrated an approximate 65% recovery of the total input HCV RNA within the anti-apoE immunocomplex. 12% of HCV RNA was recovered from anti-apoB immunocomplexes, and only background levels of HCV RNA were present in anti-Sendai virus antibody immunocomplexes (Fig. 2B). The number of infectious HCV particles remaining in the supernatant after each immunoprecipitation reaction were quantified using a sensitive focus-forming unit assay of infected Huh7.5 cells. Supernatants from anti-apoE but neither anti-Sendai virus nor anti-apoB immunoprecipitation reactions exhibited an approximate 65% reduction in infectivity (Fig. 2C, bottom panel). This reduction paralleled the relative amount of both apoE-associated HCV RNA (Fig. 2B) and the reduction of HCV infectivity achieved when infection was carried out in the presence of β-VLDL or apoE antibody (see Fig 1). These results indicate that apoE is a component of the infectious HCV particle and forms a stable virion complex that serves to support or enhance virus particle infectivity. These properties of apoE are relatively specific for HCV, as infection with Sendai virus, another enveloped RNA virus, was reduced by immunoprecipitation of infectious supernatant with anti-Sendai virus antibodies but not by anti-apoE antibodies (Fig 2C, top panel). It was somewhat unexpected that immunoprecipitation using anti-apoB antibodies did not result in recovery of significant amounts of HCV RNA since β-VLDL particles contain both apoB and apoE. However this observation is consistent with results shown in Fig 2Ain which anti-apoB did not deplete apoE from virus supernatants, suggesting that within β-VLDL particles the single apoB molecule may be relatively inaccessible to antibody.

Figure 2. ApoE associates with HCV RNA and infectious HCV particles.

A) Immunoblot analysis of immunocomplexes or residual supernatants recovered from the indicated immunoprecipitation reactions. B) Supernatant from JFH1 infected cells was immunoprecipitated with antibody to apoB, apoE, or Sendai virus. RNA was extracted from the resulting pellet fractions or residual supernatant, and HCV genome copies were quantified by RT-qPCR. C) The number of infectious virions remaining in the supernatant after immunoprecipitation was determined by focus forming assay after infection of Huh7.5 cells (bottom panel). Similar immunoprecipitation reactions in media containing Sendai virus were also assessed by focus forming unit assay (top panel). P values were determined by Student’s t-test comparison against negative control (beads, no antibody).

LDL-R is a co-receptor for HCV

The presence of apoE within β-VLDL and IDL confers greater affinity for LDL-R binding than LDL(Chappell and Medh, 1998), and the increased ability of β-VLDL or apoE antibody to block HCV infection compared to LDL or apoB antibody implicates the LDL-R as a possible common binding site on hepatocytes for both β-VLDL particles and infectious HCV virions. Since HCV infection was dependent on access to apoE binding site(s) on target cells, we sought to determine if LDL-R participates in the HCV-host cell interaction. We therefore prepared Huh7.5 cell lines stably expressing either an empty cytomegalovirus immediate-early promoter plasmid (control cells) or a similar plasmid expressing the LDL-R (LDL-R1 cells), and we assessed the requirement for the LDL-R in HCV infection. 25-hydroxycholesterol (25-HC), an antagonist of the sterol response element binding protein pathway, blocks the expression of the LDL-R by suppressing Ldl-r promoter activity within treated cells (Adams et al., 2004). We note that 25-HC treatment of cells also blocks the synthesis of geranylgeraniol, a prenyl lipid that is essential for HCV RNA replication (Ye et al., 2003;Wang et al., 2005). Thus, in order to retain HCV replication competence of cells all treatments with 25-HC were carried out in culture media supplemented with 10uM geranygeraniol, which supports HCV replication in the presence of high levels of 25-HC (Ye et al., 2003;Wang et al., 2005). 25-HC treatment resulted in a dose-dependent decrease in the expression of the LDL-R within control cells, with an 85% reduction in expression observed at 1μg/mL treatment while LDL-R1 cells maintained LDL-R expression(Fig. 3A). 25-HC did not affect the expression levels of claudin-1, SR-BI or CD81 (Fig. 3A and B). To assess the functional impact of 25-HC treatment on ligand uptake by the LDL-R we measured the uptake of LDL labeled with a fluorescent lipid, 3-pyrenemethyl-23, 24-dinor-5-cholen-22-oate-3 beta-yl (PMCA) oleate. Increasing concentrations of 25-HC had no significant effect on PMCA oleate uptake by LDL-R1 cells, but uptake was reduced by approximately 60% in control cells (Fig. 3C and D, Supplemental Fig. S1). Importantly, when 25-HC-treated cells were challenged with HCV (at MOI=0.5–1.0) we observed an approximate 60% reduction in the frequency of HCV-infected cells (Figs. 3E and Supplemental Fig. S1) that mirrored the reduction in ligand binding and uptake by the LDL-R (see Fig 3C). The reduction in HCV infection paralleled that mediated by the β-VLDL competition and anti-apoE immunoprecipitation experiments (see Fig 1A and Fig 2C, respectively). The effect of 25-HC on HCV infection was specific for the HCV entry process, as treatment with up to 1μg/mL 25-HC had no effect on intracellular HCV replication and viral protein expression in cells harboring an HCV subgenomic replicon (Supplemental Fig. S2).

Figure 3. Inhibition of HCV infection through suppression of LDL-R expression and function.

Control cells and an LDL-R overexpressing stable cell line (LDL-R1) were treated for 16 hours with increasing amounts of 25-HC (shown above each lane) and analysed for protein expression, PMCA oleate uptake, and HCV infection. NM, normal media. A) Immunoblot analysis of LDL-R and HCV co-receptor abundance. B) CD81 expression was measured by flow cytometry and is presented as mean fluorescence relative to untreated cells. C) Uptake of PMCA oleate, a fluorescent LDL analogue, was measured by flow cytometry. Graphs show the fluorescence peaks of treated (black line) versus untreated (0μg, gray line) cells from a representative experiment. D) Mean PMCA oleate uptake by control and LDL-R1 cells treated with increasing 25-HC. Graph shows relative mean fluorescence from five separate experiments. E) Cells treated with 25-HC were infected with HCV at MOI=1. Graphs show the percent of HCV positive cells as measured by flow cytometry staining for intracellular HCV proteins. Data are from a representative experiment. F) The mean relative percentage of infected cells from five combined experiments as described in E.

In order to define the potential role of the LDL-R in cell binding and entry by HCV, and to compare LDL-R functions to the various HCV co-receptors, we conducted expression knockdown experiments using siRNA targeting the LDL-R, CD81, claudin-I or SR-BI. Knockdown of each receptor target was verified by immunoblot analysis (Fig. 4A) or flow cytometry assay of treated cells (Fig. 4B). We achieved a level of knockdown of CD81 or claudin-1 expression (Fig. 4C) that significantly reduced HCV infection, consistent with their known function as HCV co-receptors. Importantly, siRNA knockdown of LDL-R expression also reduced the frequency of infected cells and suppressed infection by approximately 30–40% overall in independent experiments (Figs. 4D and 4E). This effect was less that the 60% reduction of HCV infection that occurred in cells treated with β-VLDL or 25-HC (see Fig 1 and Fig 2), likely reflecting the background level of HCV infection resulting from less than 100% transfection of siRNA among all cells in the culture and/or variable knockdown within individual transfected cells. Knockdown of SR-BI expression by more than 90% (Fig. 4C) did not significantly affect HCV infection in our studies. Taken together, our results indicate that LDL-R can function as a co-receptor for HCV infection to facilitate cell binding and/or entry processes through interactions with an HCV/lipoprotein virion complex.

Figure 4. Knockdown of LDL-R expression decreases HCV infection.

Huh7.5 cells were treated with 0.4nM of siRNA targeting the indicated mRNAs or a negative control siRNA. In each case we achieved 60–80% transfection efficiency of target cells as measured by parallel transfection of fluorescently labeled control siRNA and immunofluorescence microcopy analysis of the transfected cells. A) Immunoblot analysis of cells harvested three days after siRNA transfection. B) CD81 expression in siRNA-treated cells was measured by flow cytometry. Data are presented as mean fluorescence relative to non-transfected control cells. C) Reduction of targeted protein expression relative to negative control siRNA treated cells as measured by quanitative densitometry from immunoblot (LDL-R, Claudin-1, SR-BI) or mean fluorescence signal (CD81) D) siRNA-treated Huh7.5 cells were infected with HCV at MOI =0.5. Two days later the percentage of HCV-infected cells was determined by flow cytometry assay. Data are shown from a representative experiment. E) Combined HCV infection results from three separate experiments as described in C. P values were determined by Student’s t-test comparison against siRNA control-treated cells.

Characteristics of HCV infectious particles

In order to determine if the HCV virions that associate with apoE are the same as those which depend on LDL-R for virus binding and/or entry of target cells, we carried out density fractionation and infectivity studies of virus supernatant from HCV-infected cells. Infectious cell supernatants were subject to ultracentrifugation through iodixanol density gradients from which fractions were collected and subjected to immunoprecipitation with anti-apoE or nonimmune control antibodies. We measured the relative level of HCV RNA present in the immunocomplex recovered from each fraction or that remained in the non-bound supernatant from each reaction. In parallel we measured the infectivity of the immunoprecipitation supernatant and input material from each gradient fraction. RNA analysis of the supernatant and pellet from each fraction demonstrated a peak of apoE-associated HCV RNA within the 1.06 g/mL fraction (Fig. 5A, upper panel). The HCV RNA association with apoE was specific since only nonsignificant levels of HCV RNA were recovered within the respective control immunocomplexes. (Fig. 5A, lower panel). We found that the peak of total HCV RNA present within the gradient had a density of approximately 1.09–1.10 g/mL (Fig 5B, upper panel), whereas the peak of infectious HCV particles occurred at a density of 1.06 g/mL, corresponding to the density of the peak apoE-associated HCV RNA (compare Fig. 5B, lower panel with Fig 5A). To determine if the 1.06 g/mL apoE-associated, infectious HCV was dependent on the LDL-R for cell entry, we divided infectious HCV supernatant between two sets of gradients and subjected each to separation by ultracentrifugation. The resulting fractions were then recovered and used to infect Huh7.5 cells alone or cells that had been pretreated for 16 hours with 1μg/mL 25-HC to reduce LDL-R expression. In untreated cells the peak of infectivity occurred at a similar density of 1.06 g/mL (Fig. 5C, left panel). The corresponding 1.06 g/mL peak recovered from the parallel gradient was significantly reduced in its infectivity when used to infect 25-HC treated cells, indicating that HCV virions of this density are dependent on LDL-R for infection (Fig. 5C, right panel). Thus the HCV virions exhibiting a density of approximately 1.06 g/mL are associated with apoE and mediate high infectivity that is dependent on the LDL-R.

Figure 5. ApoE is a component of infectious HCV particles.

Supernatant from HCV-infected Huh7-5 cells was ultracentrifuged through a 5–50% iodixanol gradient. 1mL fractions were collected from the gradient and analysed for apoE association, HCV RNA content, and overall infectivity of Huh7.5 cells. A) 0.5mL of each fraction was immunoprecipitated with apoE or isotype control antibody. HCV RNA recovered within the pellet or residual supernatant of each fraction was determined by RT-qPCR. B) Total HCV RNA within each gradient fraction was quantified by RT-qPCR. C) Infectivity of gradient fractions. D) 1 ml each of a common HCV supernatant stock were separated by ultracentrifugation through 10–40% iodixanol density gradients The resulting fractions were used to infect Huh7.5 cells that were cultured alone or pretreated for 16 hours with 25-HC to suppress LDL-R expression. HCV infection of cells was quantified 48 hrs later by focus forming unit assay.

Discussion

Our studies reveal that cell culture produced HCV JFH1 infectious virions contain apoE which allows them to productively infect hepatocytes through interactions with the LDL-R. These observations confirm and extend the observations of Chang, et al., who demonstrated that apoE antibody was able to block the entry HCV RNA from cell culture derived virus into Huh7.5 cells (Chang et al., 2007). Our results reveal that anti-apoE antibody blocks an HCV entry pathway of productive infection. We also showed that apoE antibody was directed against infectious particles, as infectivity was reduced following immunoprecipitation of particles containing apoE and HCV RNA. Although primary hepatocytes do not efficiently support productive infection with patient HCV isolates, it has been demonstrated that entry of patient derived HCV RNA depends on expression of LDL-R in cultured hepatocytes (Molina et al., 2007). Similarly, we observed that modulating LDL-R levels with 25-HC, siRNA or by LDL-R overexpression controlled productive infection with HCV. Collectively, these results support the idea that HCV entry contributed through the LDL-R pathway leads to productive virus infection as opposed to nonspecific uptake of lipoprotein associated virions.

Early studies of the physical properties of HCV virions derived from patient sera demonstrated a lower than expected density (1.08g/mL) compared to other flaviviruses (Miyamoto et al., 1992). It was subsequently determined that this low density was due to HCV association with β-lipoproteins which include LDL and VLDL (Thomssen et al., 1992). Prince et al. found that most of the patient derived HCV RNA was in the VLDL fraction (Prince et al., 1996). Other studies have demonstrated that HCV RNA contained within the VLDL and LDL serum fractions of HCV patients was able to be endocytosed by target cells and that this apparent virus entry activity could be blocked by anti-LDL-R and anti-ApoE antibodies (Agnello et al., 1999). Consistent with these observations, we observed that infectious HCV fractionated at a similar density (1.06g/mL) in vitro and importantly was associated with apoE. Furthermore, pretreatment of cells with the LDL-R ligands, β-VLDL (50ug/mL treatment) and to a lesser extent LDL (100ug/mL treatment), reduced subsequent HCV infection. These observations agree with other reports that pretreatment of cultured cells with either VLDL (50–62.5μg/mL treatment) or with LDL (125–200μg/mL treatment), but not HDL, blocked binding of patient-derived HCV to cultured human fibroblasts (Monazahian et al., 1999;Germi et al., 2002;Andre et al., 2002). Thus patient-derived and cell culture derived HCV particles can compete with LDL and VLDL/β-VLDL for binding to human cells. It should be noted that HDL has been implicated as a factor that enhances the infectivity of HCV and HCV pseudoparticles in vitro (Catanese et al., 2007). While we did not directly assess the role of HDL for enhancement f HCV infectivity, our studies show that in the presence of exogenous excess HDL, HCV infectivity is not reduced. Thus, HDL may serve to enhance HCV infectivity through processes independent of cell-binding and entry. Our studies indicate that cell binding by HCV may more closely resemble the VLDL/IDL interaction with target cells than the LDL interaction inasmuch as highly infectious virus contains apoE and is most sensitive to competition for target cells by exogenous VLDL.

VLDL is produced in hepatocytes wherein lipids stored in droplets within the endoplasmic reticulum (ER) are transferred to a growing apoB core through the actions of microsomal transfer protein (MTP) (Hussain et al., 2008). The growing lipoprotein particle acquires additional triglycerides, cholesterol, and apoE, and is secreted from the cell through the golgi apparatus. HCV core protein localizes to the surface of lipid droplets and is able to interact with viral structural proteins assembled on the ER (Miyanari et al., 2007). Furthermore, intracellular membranes containing the HCV replicase are enriched in MTP, apoB and apoE (Huang et al., 2007), and inhibition of the expression or activity of either of these factors blocks the release of infectious HCV (Chang et al., 2007; Gastaminza et al., 2008). Thus, the release of infectious HCV is dependent on virions being packaged as a VLDL-like particle, which ultimately facilitates infection efficiency through LDL-R interactions of target cells expressing the full complement of HCV co-receptors.

We propose a model in which HCV is secreted from hepatocytes as a VLDL-like lipoviral particle (LVP) containing at least HCV core, RNA, E1, E2 apoB and apoE. In common with natural VLDL derived particles, the HCV LVP could bind initially to cell surface glycosaminoglycans, such as heparin sulfate, in a relatively non specific interaction mediated by either E2 or apoE as both have been shown to bind heparin sulfate (Morikawa et al., 2007;Saito et al., 2003). In terms of the LDL-R, these initial interactions with the target cells could permit the LVP to a specific interaction between apoE and/or apoB and the LDL-R. Such an interaction could depend the processing/lipid removal of the LVP by lipoprotein lipase, which has also been reported to be involved in HCV entry (Andreo et al., 2007). By this model the infectious virus particle could mediate a stable interaction with the target cell that promotes the E2-CD81 interaction, subsequent endocytosis, pH-dependent fusion, and the final entry steps directed by claudin-1 and occludin (Evans et al., 2007;Ploss et al., 2009). Whereas apoE- LDL-R interaction is probably not essential for entry in vitro since infection was not completely blocked by excess β-VLDL, apoE antibody, or LDL-R knockdown, and because viral E2 protein can interact directly with surface expressed receptors, we propose that efficient and perhaps natural HCV infection are supported by the LVP/LDL-R interaction on target cells during the intial processes of cell binding by the virus. Indeed our siRNA knockdown experiments confirm that both CD81 and claudin-1 function in HCV entry, and recent studies define human occludin as the species-specific factor that is essential for HCV infection (Ploss et al., 2009). While we did not observe any defects in HCV entry upon knockdown of the majority of SR-BI expression, it is likely that only a small amount of SR-B1 is required to facilitate infection. SR-BI may additionally represent an alternative pathway for entry possibly involving E2 or apoE binding since antibodies against SR-BI have been reported to reduce HCV cell entry (Kapadia et al., 2007;Catanese et al., 2007;Maillard et al., 2006). Interestingly, a mutation in E2 has been reported which shows decreased dependence on SR-BI and increased dependence on CD81 for entry, indicating that SR-BI usage by HCV is conditional and might be modulated by viral adaptive mutations (Grove et al., 2008). Variations in E2 have also been reported to influence HCV genotype differences in SR-BI dependent entry of HCV pseudoparticles (Lavillette et al., 2005).

The production of HCV as an LVP may serve as a mechanism to both enhance infection and escape immune detection by co-opting the host lipid delivery system. It may also help to explain the hepatotropism of HCV, as apolipoproteins regulate the recycling of lipid particles to the liver. Moreover, the nature of the HCV virion as a VLDL derived LVP implies that modulating lipoprotein metabolism could play a role in the treatment or management of HCV infection. Indeed, microsomal transfer protein and long chain acylCoA synthetase-3 have already been identified as enzymes required for efficient release of HCV from infected cells (Huang et al., 2007;Yao and Ye, 2008). The grapefruit flavonoid naringenin has also been shown to reduce HCV release from cells by interfering with lipid metabolism, which could possibly alter the composition of the LVP (Nahmias et al., 2008). Omega-3 fatty acid supplementation alters lipid metabolism and reduces VLDL production in humans (Bays et al., 2008). Together these observations suggest that therapeutic or dietary modulation of specific lipid could be considered as an avenue to regulate LVP production and impart control of HCV infection.

Methods

Cell culture

Huh7.5 cells were maintained in DMEM containing 10% FBS. The LDL-R1 stable cell line was constructed by transfecting plasmid encoding LDL-R under the control of a CMV promoter into Huh7.5 and selecting the G418 resistant clone with highest LDL-R expression by western blot. LDL-R1 cells were maintained in DMEM containing 10% FBS and 400ug/mL G418.

Virus production and infections

HCV was collected from the supernatant of an Huh7 derived cell line that constitutively produces genotype 2a strain JFH1 RNA from a ribozyme-modified HCV genome integrated into the host genome (kind gift from George Luo). For large scale virus preps, these cells were incubated with DMEM containing 2% FBS for 48 hours and then the HCV containing supernatant was filtered through a .22uM filter and concentrated 100x using Centricon 100,000MW cut-off filters (Millipore). Virus stocks were titered using the focus forming unit assay and stored at −80°C. For infections at a defined MOI, concentrated stocks were diluted with an appropriate amount of DMEM and incubated with Huh7.5 cells for 1.5 hours. Cells were washed 1 time with PBS and then maintained in DMEM containing 10% FBS and analyzed 48 hours post infection, unless otherwise indicated. Sendai virus (Cantrell strain) was purchased from Charles River Labs.

Immunoprecipitation

For recovery of immunocomplexes, up to 500uL of HCV-containing supernatant was incubated with the indicated antibody for 1 hour and then 50uL of Protein G Plus Agarose Suspension (Calbiochem) was added. Tubes were rocked at 4°C overnight and then centrifuged at 1000g for 2min. Supernatant was removed and used for infection, RNA extraction, or mixed with SDS sample buffer for western blotting. Pellets were washed 3 times with PBS and then used for RNA extraction or western blotting.

Focus forming unit assay

100uL of virus sample were used to infect 2×10⁴ Huh7.5 cells in triplicate in a 48 well dish. After 48 hours, plates were washed with PBS and fixed for 30 minutes in 3% paraformaldehyde. Cells were permeablilized with 0.2% TritonX-100 in PBS for 15 minutes. Cells were then blocked in 10% FBS for 10 minutes and stained using 1:1500 human α-HCV polyclonal serum for 1 hour, followed by 1 hour of 1:500 peroxidase-conjugated donkey α-human antibody (Jackson ImmunoResearch). Infected foci were visualized using the Vector VIP peroxidase substrate staining kit according to the manufacturer’s protocol (Vector Labs) and counted by light microscopy on a Nikon eclipse TE2000-E microscope. For Sendai virus quantitation a similar assay was performed, but plates were fixed 24 hours post infection, polyclonal goat α-Sendai serum was used at 1:1000, and peroxidase-conjugated donkey α-goat was used as the secondary antibody.

Flow cytometry

To determine the percentage of HCV-infected cells within in vitro cultures, 2×10⁵- 1×10⁶ infected cells were trypsinized, washed with PBS, and fixed with 3% PFA. Cells were stained for intracellular HCV protein following the Current Protocols in Immunology intracellular staining protocol using 1:2000 human polyclonal α-HCV serum and 1:2000 Alexa Fluor 488 Goat α-human secondary (Molecular Probes). CD81 expression was measured using PE-conjugated α-human CD81 antibody according to the manufacturer’s protocol (Pharmingen). Cells were analyzed on a FACSscan2 flow cytometer and data was collected using Cell Quest Pro (Becton Dickson). Data was analyzed using FlowJo (Tree Star, Inc.). For PMCA oleate LDL-R function assays, cells were incubated in media containing 10ug/mL PMCA oleate for 16 hours. Cells were trypsinized, washed with PBS, and analysed for PMCA oleate uptake by measuring fluorescence on the pacific blue channel of an LSRII flow cytometer (Becton Dickson).

25-hydroxycholesterol treatment

To modulate LDL-R expression, cells were treated for 16 hours in DMEM containing 10% NCL-PPS (newborn calf lipoprotein deficient serum), 50uM compactin, 50uM mevalonate, 10uM geranylgeraniol, 10ug/mL cholesterol and the indicated amount of 25-hydroxycholesterol. We note that 25-HC treatment of cells suppresses the synthesis of geranylgeraniol, a prenyl lipid that is essential for HCV replication (Ye et al., 2003; Wang et al., 2005). Thus, in order to retain HCV replication competence of cells all treatments with 25-HC were carried out in the presence of 10uM geranylgeraniol.

Immunoblot analysis

Immunoblot analysis was carried out according to standard protocols using the following antibodies and conditions: Goat α-apoE (Calbiochem) 1:2000, Sheep α-apoB 1:1000, Mouse α-LDL-R HL1 1:2000 (gifts from Dr. Jin Ye), Rabbit α-SR-BI ab396 (Abcam) 1:2000, Mouse α-Claudin-1 2H10D10 (Zymed) 1:500. Peroxidase conjugated secondary antibodies were used at 1:10,000 (Jackson ImmunoResearch) and bands were visualized using ECL Plus reagent (GE Healthcare).

RNA analysis

HCV RNA was extracted from 140uL of supernatant or agarose bead slurry using the QIAamp viral RNA mini kit (Qiagen). Quantitative RT-PCR (RT-qPCR) reactions were set up using the GeneAmp EZ rTth RNA PCR kit (ABI), JFH1 primer, and Taqman probe (ABI). Reactions were run in triplicate using an Applied Biosystems 7300 Real Time PCR System.

siRNA treatment

Previously described siRNA duplexes targeting CD81 (UGAUGUUCGUUGGCUUCCU), SR-BI (GCAGCAGGUCCUUAAGAAC), and Claudin-1 (UAACAUUAGGACCUUAGAA) were purchased from Dharmacon. The member of the siGENOME ON-TARGETplus SMART pool (Dharmacon) targeting the LDL receptor, which resulted in the greatest knockdown of Ldl-r mRNA expression, was identified (GGACAGAUAUCAUCAACGA). 0.4nmols of these siRNA duplexes were transfected into 1×10⁵ Huh7.5 cells using RNAiMAX transfection reagent (Invitrogen). Cells were analyzed for knockdown or used for infection at 3 days post transfection.