Potential conflict of interest: Dr. Manns is on the speakers' bureau of, consults for, and received grants from Roche, Gilead, Bristol-Myers Squibb, Merck, and Janssen. He is also on the speakers' bureau of GlaxoSmithKline. He consults for, and received grants from, Novartis and Boehringer-Ingelheim. He consults for Tibotec and Vertex. Dr. Pietschmann consults for Biotest and Janssen. Dr. Ciesek is on the speakers' bureau of MSD and received grants from Novartis.
This study was funded by the DFG Emmy Noether Program (grant no.: HA 4393/2-1; to T.v.H.) as well as SFB900 (project A6) and a grant from the Initiative and Networking Fund of the Helmholtz Association (grant no.: SO-024; to T.P.) and a DFG grant (no. Ci 171/2-1; to S.C.). The authors thank Jens Bukh (University of Copenhagen, Denmark), Alfredo Nicosia (Okairos, Basel, Switzerland), and Charlie Rice (The Rockefeller University, New York, NY) for providing reagents and Heiner Wedemeyer and the HepNet serum Bank for providing sera.
Oxidized low-density lipoprotein (oxLDL) has been reported as an inhibitor of hepatitis C virus (HCV) cell entry, making it the only known component of human lipid metabolism with an antiviral effect on HCV. However, several questions remain open, including its effect on full-length cell-culture–grown HCV (HCVcc) of different genotypes or on other steps of the viral replication cycle, its mechanism of action, and whether endogenous oxLDL shares the anti-HCV properties of in vitro–generated oxLDL. We combined molecular virology tools with oxLDL serum measurements in different patient cohorts to address these questions. We found that oxLDL inhibits HCVcc at least as potently as HCV pseudoparticles. There was moderate variation between genotypes, with genotype 4 appearing the most oxLDL sensitive. Intracellular RNA replication and assembly and release of new particles were unaffected. HCV particles entering target cells lost oxLDL sensitivity with time kinetics parallel to anti-SR-BI (scavenger receptor class B type I), but significantly earlier than anti-CD81, suggesting that oxLDL acts by perturbing interaction between HCV and SR-BI. Finally, in chronically HCV-infected individuals, endogenous serum oxLDL levels did not correlate with viral load, but in HCV-negative sera, high endogenous oxLDL had a negative effect on HCV infectivity in vitro. Conclusion: oxLDL is a potent pangenotype HCV entry inhibitor that maintains its activity in the context of human serum and targets an early step of HCV entry. (HEPATOLOGY 2013)
Chronic hepatitis C virus (HCV) infection affects an estimated 160 million people worldwide.1 Chronic HCV is a frequent cause of end-stageliver diseases (ESLDs) and hepatocellular carcinoma as well as a common indication for liver transplantation (LT). Treatment remains problematic because of side effects and high cost. A particular problem is graft reinfection that occurs immediately after LT for HCV-associated ESLD and often leads to transplant hepatitis and graft loss.2
The HCV replication cycle is intricately linked to host lipid metabolism.3 The production of infectious viral particles is dependent on the cellular very-low-density lipoprotein (VLDL) assembly machinery, and infectious particles in the blood are reminiscent of VLDL in that they are of low density, triglyceride (TG) and cholesterol rich, and tightly associated with apolipoproteins (Apos) ApoB and ApoE.3 Three lipid transport molecules on the hepatocyte surface have been implicated in viral entry: Low-density lipoprotein (LDL) receptor mediates cellular uptake of HCV RNA, but may be nonessential for productive infection.4 Scavenger receptor class B type I (SR-BI) is essential for HCV infection in vitro5 and is discussed in more detail below. Finally, Niemann-Pick C1–like 1 (NPC1L1) has been shown to support HCV entry, but its exact role is, as yet, unclear.6 Cell-surface factors essential for HCV entry, but not directly involved in lipid trafficking, include the tetraspanin, CD81,7 and the tight-junction components, claudin-18 and occludin.9
SR-BI is a 509-amino-acid protein with two transmembrane domains and a single large extracellular loop that is highly expressed in the liver and in tissues producing steroid hormones.10 It is the key high-density lipoprotein (HDL) receptor and mediates a flux of free cholesterol from the lipoprotein into the cellular membrane.10, 11 Besides HDL, SR-BI interacts with diverse other ligands, including VLDL, native LDL, oxidized LDL (oxLDL), lipopolysaccharides, TG-rich lipoprotein (TRL) remnants, and serum amyloid alpha (SAA) through multiple binding sites.10, 12-15
First proposed as an HCV receptor because of its interaction with the viral glycoprotein, E2,16 SR-BI was subsequently shown to be essential for HCV entry in numerous studies.5, 9, 17, 18 Animal data indicate that SR-BI is also important for HCV infection in vivo,19 and an inhibitor of HCV/SR-BI interaction is the only HCV-entry inhibitor currently in clinical trials.20 Recent work has suggested that SR-BI fulfils at least two distinct functions during viral entry: one related to initial attachment and another during post-binding.5, 21, 22
Among SR-BI ligands, HDL has a modest enhancing effect on HCV entry,23 whereas oxLDL has been found to be a strong inhibitor.24 oxLDL is a modified form of LDL that is generated early on in the pathogenesis of atherosclerosis, when native LDL deposited in early atherosclerotic lesions (“fatty streaks”) undergoes oxidative modifications.25 The exact nature and driving forces behind these modifications are incompletely understood,26 but they allow LDL to be taken up in great quantities by macrophages through an LDL-receptor–independent route creating foam cells, a hallmark of atherosclerotic lesions.27
Previous work on oxLDL has almost exclusively relied on HCV pseudoparticles (HCVpp), and there may be important differences in the cell-entry properties of HCVpp and full-length HCV particles grown in cell culture (HCVcc).6 Moreover, the mechanism of HCV-entry inhibition by oxLDL is unclear. An SR-BI-dependent mechanism of action has been suspected, but there is no experimental evidence for this. We undertook the present study with three main aims: (1) characterize the effect of oxLDL on the full HCV replication cycle; (2) test whether oxLDL inhibition is indeed linked to SR-BI; and probe the anti-HCV properties of endogenous oxLDL present in patient sera.
Isolation of the LDL fraction of human plasma was performed by sequential gradient ultracentrifugation.28 Briefly, the LDL fraction was dialyzed at 4°C against 1× phosphate-buffered saline, and protein concentration was determined by Bradford's method. Native LDL (100 μg/mL) was incubated with 10 μmol/mL of copper (II)-sulfate for 24 hours at 4°C. LDL peroxidation was stopped by adding 20 μmol/L of butylhydroxytoluol and 5 mmol/L of ethylenediaminetetraacetic acid. Oxidation state was determined by detection of conjugated diene formation by measuring UV absorbance at 234 nm.29
Virus Production and Infection Assay.
Firefly reporter (CSFlucW2) lentiviral pseudoparticles were generated as previously described.30 Production of HCVcc was performed by electroporation of Huh-7.5 cells with 5 μg of RNA of a full-length HCV genome, as previously described.31 Supernatant was harvested 48 hours post-transfection. Infections were performed in the presence or absence of oxLDL, mostly at a concentration corresponding to the 90% inhibitory concentration of that specific batch of oxLDL (range, 10-50 μg/mL). Activity varied from batch to batch, most likely because of the oxidation state of the particular batch. Infection was quantified after 72 hours. For the tissue culture 50% infectious dose (TCID50) assay, infected cells were stained with an NS5A antibody (9E10; a kind gift from C. Rice). Luciferase assay was performed as previously described.32 Intra- and extracellular HCV core was measured by a commercially available enzyme-linked immunosorbent assay (ELISA) (Architect HCV Ag Assay; Abbott Laboratories, Abbott Park, IL).
Serum oxLDL levels were quantified using two commercial ELISA systems offered by Mercodia (Uppsala, Sweden) and Immundiagnostik (Bensheim, Germany), as indicated in the manufacturers' manuals.
oxLDL Inhibits Infection of Hepatitis C Virus Across Genotypes.
oxLDL, but not native LDL, inhibits HCVcc of genotype 2a (Fig. 1A). Using a single batch of oxLDL, we found similar half-maximal inhibitory concentration (IC50) of 0.74 ± 0.35 and 0.40 ± 0.22 μg/mL for HCVcc and HCVpp, respectively (Fig. 1B). As expected, there was some batch-to-batch variation in the IC50 value, most likely reflecting the degree of oxidation of a given batch. Chimeras of the structural region of genotypes 1a-7a and the Japanese fulminant hepatitis type 1 nonstructural region33 showed a clear inhibitory effect of oxLDL treatment on all genotypes (Fig. 1C). There was limited intergenotype variation, with genotype 6 appearing least and genotype 4 most sensitive when using the same multiplicity of infection (MOI) (Fig. 1D). Thus, oxLDL is a potent inhibitor of HCV infection in vitro with pangenotype activity.
oxLDL Inhibits High- and Low-Density Fractions of HCV.
Next, we assessed the effect of oxLDL on different density fractions of HCVcc isolated by ultracentrifugation over a density gradient. In all fractions containing detectable infectivity (more than 2 TCID50/mL, i.e., fractions 1-7 corresponding to 1.03-1.18 g/mL), we observed inhibition by oxLDL (Fig. 2 and Supporting Fig. 1). In the lowest density fraction (fraction 1, corresponding to 1.03-1.05 g/mL), the degree of oxLDL inhibition appeared slightly lower; however, this fraction also contained low baseline infectivity, compared to the following fractions, which may have impaired our ability to quantify the degree of inhibition (Supporting Fig. 1).
oxLDL Has No Effect on HCV RNA Replication or On Assembly and Release of New Viral Particles.
In Huh-7.5 cells transfected with Fluc-Jc1, virus RNA application of oxLDL or the highly potent antibody, C167, directed against SR-BI had no effect on RNA replication (Fig. 3A). Moreover, intracellular core and core released from transfected cells was not different (Fig. 3B). In addition, we performed a TCID50 assay using the supernatant from oxLDL treated or untreated producer cells and found that similar infectious titers were released (Fig. 3C). Thus, the inhibitory effect of oxLDL on HCVcc appears solely to be the result of an inhibition of viral entry.
oxLDL Treatment Does Not Affect Protein Expression Level of HCV-Entry Receptors.
Previously, it was reported that oxLDL inhibition of HCV entry is not mediated through competition for the E2-binding site on SR-BI.24 Another conceivable mechanism for the observed entry inhibition would be down-regulation of an essential cellular HCV entry factor. We performed western blotting and fluorescent-activating cell sorting (FACS) analyses to determine the expression levels of five known HCV-entry factors (SR-BI, CD81, CLDN1, OCLN, and NPC1L1) in untreated and oxLDL treated Huh-7.5 cells and found similar expression levels in all cases (Fig. 4A,B). These findings argue against altered expression of a cellular entry factor being responsible for oxLDL inhibition of HCV entry.
oxLDL Inhibition Parallels Anti-SR-BI Inhibition.
In a time-course experiment with synchronized infection of target cells by multiple virions and timed addition of inhibitors, oxLDL lost its effect in parallel to anti-SR-BI34 and significantly earlier than anti-CD81 (Fig. 5). This suggests that oxLDL mediates entry inhibition in a SR-BI-dependent manner.
Next, we tested the effect of oxLDL on two HCV variants that have been reported to be less dependent on SR-BI, compared to wild-type (WT) HCV. The IC50 value for oxLDL inhibition of the ΔHVR1 variant, where the entire hypervariable region 1 (HVR1) has been deleted from the E2 glycoprotein,35 was approximately 18-fold higher, compared to unmodified HCV (Fig. 6A), but this observation did not reach statistical significance. The IC50 value for oxLDL against the G451R point mutant36 was similar to WT (Fig. 6B).
oxLDL Inhibition Is Maintained in the Presence of Altered SR-BI Levels.
Next, we were interested whether modulation of cellular SR-BI levels would affect the sensitivity of HCV to oxLDL. When cellular SR-BI expression levels were increased by overexpression or decreased by RNA interference (RNAi), no major change in oxLDL sensitivity was noted (Supporting Fig. 2).
oxLDL Inhibition Is Maintained in the Presence of SR-BI Variants.
Because SR-BI is a multiligand receptor, we hypothesized that it might be possible by deleting the oxLDL-binding site from SR-BI to generate a variant that functions as an HCV receptor, but is insensitive to oxLDL inhibition. The location of the oxLDL-binding site in SR-BI is unknown, but for the closely related CD36, it has been reported on.37 Alanine scanning of the homologous region in SR-BI (amino acids 154-183) was performed by replacing blocks of five amino acids with alanine residues and expressing WT SR-BI or variants in Huh-7.5 cells. Overexpression of WT SR-BI resulted in a 2- to 3-fold enhancement of HCV infection, as previously reported on,18, 31 whereas SR-BI variants containing alanine blocks in the putative oxLDL-binding site region did not—suggesting that they are nonfunctional as HCV receptors (Supporting Fig. 3). Moreover, oxLDL inhibition was maintained in all cases. Thus, we were not able to generate an SR-BI variant that is functional as an HCV receptor yet oxLDL insensitive.
Next, we were interested to see whether the sensitivity of HCV to oxLDL would be changed in the presence of SR-BI variants with altered E2-binding properties. Thus, we overexpressed murine SR-BI (mSR-BI), which has been reported to not bind sE2 yet support HCV infection, and two point mutants where individual amino acids from mSR-BI were introduced into human SR-BI. From the set of point mutants generated by Catanese et al.,34 we chose the ones with the reportedly highest (S101A) and lowest (E210A) sE2 binding. When overexpressed in Huh-7.5 cells, mSR-BI and the point mutants, but not CD36, showed enhanced HCV entry with maintained oxLDL inhibition (Fig. 7A,B). Similar results were observed when the molecules were expressed in a Huh-7.5 subclone where endogenous SR-BI expression was down-regulated by RNAi (Fig. 7C,D). When overexpressed in Chinese hamster ovary cells, human SR-BI, the point mutants, and mSR-BI appeared to bind HCVcc (Supporting Fig. 4A). However, as previously reported on by others, neither mSR-BI nor CD36 bound sE2 (Supporting Fig. 4B).
Endogenous oxLDL Has No Effect on Virus Load, but Correlates With In Vitro Infectivity.
Low levels of oxLDL are present in human serum, although its exact chemical composition is incompletely defined.38 Using the Mercodia ELISA assay, we measured serum oxLDL levels in three different patient cohorts: healthy volunteers (n = 29); chronically hepatitis B virus (HBV)-infected individuals (n = 36); and chronically HCV infected individuals (n = 67; Supporting Table 1). In HCV-infected individuals with a high viral load (>800,000 IU/mL), oxLDL levels were slightly lower, but this was not statistically significant (Fig. 8A). However, when sera from the healthy cohort were heat inactivated and added at 1:10 final dilution to an in vitro infection of Huh-7.5 cells with HCVcc (Fluc-Jc1), infectivity was significantly reduced in the one third of individuals with the highest serum oxLDL, compared with the lowest and middle one third (Fig. 8B). oxLDL measurements, using a second commercial oxLDL ELISA offered by Immundiagnostik, did not correlate with the measurements of the Mercodia assay (Supporting Fig. 5) and hence failed to show the same correlation. However, using the Mercodia system, the observation that high endogenous oxLDL levels contained in human serum reduce HCV infectivity in vitro was reproduced in the HBV-infected cohort (Fig. 8C). In a reverse experiment, we found that in vitro–generated oxLDL retains its inhibitory effect in the presence of human serum (data not shown). Thus, both in vitro–generated and endogenous oxLDL seem to have an anti-HCV effect and are active in the context of human serum.
We found that oxLDL is an effective inhibitor of HCV-cell entry across a broad range of genotypes. Moreover, we provide evidence suggesting that it acts through an SR-BI-dependent mechanism, and that both in vitro generated and endogenous oxLDL have anti-HCV activity.
Our data show that oxLDL is an entry inhibitor that is equally or more active against HCVcc, compared with HCVpp, and does not affect RNA replication or viral particle assembly and release. Moreover, oxLDL inhibits all HCV genoytpes, although the effect may be more pronounced for some genoytpes, such as genotype 4. This would be of some interest, because genotype 4 is difficult to treat with currently available regimens.
When the inhibitory effect of oxLDL on HCV was discovered, it was thought likely that oxLDL perturbs the interaction between HCV and SR-BI on the grounds that oxLDL is a known ligand of SR-BI.24 However, there was no experimental evidence supporting this assumption, and, in fact, binding of sE2 to SR-BI was found to occur normally in the presence of oxLDL. We can now show that the timing when infecting virions become inhibitor insensitive is parallel for oxLDL and anti-SR-BI and occurs significantly earlier, compared to anti-CD81.
The IC50 value of oxLDL against two HCV mutants with reduced—but not absent—SR-BI dependency, ΔHVR135 and G451R,36 did not differ significantly from WT. However, in the case of ΔHVR1, numerically, the IC50 value was markedly (18-fold) higher, compared to WT. We believe that there may, in fact, be a true difference in oxLDL sensitivity between ΔHVR1 and WT, but given the variability of the assay, our experiments were not sufficiently powered to prove or refute this.
Although deletion of the entire HVR1 is a drastic alteration to the viral glycoprotein, ΔHVR1 is, in our hands, a variant with only slightly (approximately 5-fold) decreased titers.35 Interestingly, ΔHVR1 and G451R share a decreased sensitivity to anti-SR-BI sera and increased sensitivity to soluble CD81. Of note, their major difference—G451R, but not ΔHVR1, directly bind SR-BI—has been shown only with sE2, but not complete virions.35, 36
Given that SR-BI is thought to act at more than one step of the HCV-entry process,21, 22 further work will clearly be needed to define the exact effects of the different viral variants and inhibitors of the virus-SR-BI interactions. In addition, it is interesting to speculate how oxidative modification of apolipoproteins, such as ApoE and ApoC1, currently known to be associated with HCV, would affect HCV infectivity.
In samples from chronically HCV-infected individuals, we did not observe a significant correlation between serum oxLDL and viral load. Given that in vivo HCV viral load varies between and within individuals over a wide range and is likely influenced by a multitude of variables, this is not very surprising. Differently from our data, a recent study reported on a positive correlation between oxLDL and HCV viral load39; however, it appears that this hinged strongly on a small number of outliers with very high oxLDL levels and viral load, whereas the majority of the population was similar to our cohort in showing a weak negative correlation that likely falls short of statistical significance.
Human sera with a high oxLDL content, as measured by the Mercodia ELISA system, was associated with a mild inhibitory effect on HCV infectivity in vitro. This observation was made in two independent cohorts: one healthy and one chronically infected with HBV. Although correlation does prove causality, we see this as suggestive that endogenous oxLDL has an inhibitory effect on HCV entry similar to in vitro–generated oxLDL. Most studies investigating the biological effects of oxLDL, including ours, use copper-oxidized LDL because it is thought to share many properties of endogenous oxLDL.25 However, oxidation likely introduces a multitude of chemical alterations to all parts of the LDL particle, and to what extent these are truly alike in copper oxidized and endogenous oxLDL is unknown. Our observation that the available oxLDL ELISA systems offered by Mercodia and Immundiagnostik do not correlate is in line with what has been reported on by others40 and underscores the complex nature of oxLDL.
The introduction of NS3/4A protease inhibitors into HCV therapy has improved the response rate for genotype 1 infection, but side effects, contraindications, drug-drug interactions, selection of resistant variants, and high cost remain significant shortcomings. Moreover, HCV-entry inhibitors would be useful to prevent the presently universal graft reinfection after LT. ITX 5061, an SR-BI inhibitor with an unclear mechanism of action, is in phase I and is currently the only HCV entry inhibitor in clinical trials (www.clinicaltrials.gov). The published in vitro data indicate up to 90% inhibition, which is likely less than what can be achieved with oxLDL.20 Given its chemical complexity and potential involvement in the pathogenesis of atherosclerosis, oxLDL is an unlikely therapeutic lead compound. However, a smaller chemically defined compound that mimics its anti-HCV mechanism of action would be an attractive lead. Encouragingly, we have previously shown that oxidized phospholipids are similarly effective, compared to whole oxLDL particles.24
In conclusion, oxLDL is a potent SR-BI-dependent pangenotype HCV-entry inhibitor that maintains its activity in the context of human serum. Further studies should address whether its mechanism of action can be mimicked by smaller chemically defined compounds and whether these show promise as leads for anti-HCV therapy.
The authors thank Janina Kirschner and Mirja Sirisko for logistical and technical support.