High plasma level of nucleocapsid-free envelope glycoprotein-positive lipoproteins in hepatitis C patients


  • Potential conflict of interest: Nothing to report.


Hepatitis C virus (HCV) particles associate viral and lipoprotein moieties to form hybrid lipoviral particles (LVPs). Cell culture–produced HCV (HCVcc) and ex vivo–characterized LVPs primarily differ by their apolipoprotein (apo) B content, which is low for HCVcc, but high for LVPs. Recombinant nucleocapsid-free subviral LVPs are assembled and secreted by apoB-producing cell lines. To determine whether such subviral particles circulate in HCV-infected individuals, LVPs complexed with immunoglobulin were precipitated with protein A from low-density plasma fractions of 36 hepatitis C patients, and their lipid content, apolipoprotein profile, and viral composition were determined. HCV RNA in LVPs was quantified and molar ratios of apoB and HCV genome copy number were calculated. LVPs lipidome from four patients was determined via electrospray ionization/tandem mass spectrometry. Protein A–purified LVPs contained at least the envelope glycoprotein E2 and E2-specific antibodies. LVPs were present in every patient and were characterized by high lipid content, presence of apolipoproteins characteristic of triglyceride-rich lipoproteins (TRLs), HCV RNA, and viral glycoprotein. Importantly, save for four patients, LVPs fractions contained large amounts of apoB, with on average more than 1 × 106 apoB molecules per HCV RNA genome. Because there is one apoB molecule per TRL, this ratio suggested that most LVPs are nucleocapsid-free, envelope glycoprotein-containing subviral particles. LVPs and TRLs had similar composition of triacylglycerol and phospholipid classes. Conclusion: LVPs are a mixed population of particles, comprising predominantly subviral particles that represent a distinct class of modified lipoproteins within the TRL family. (HEPATOLOGY 2012;56:39–48)

Hepatitis C virus (HCV) is a member of the Flaviviridae family and a major cause of chronic hepatitis often leading to liver cirrhosis and hepatocellular carcinoma.1 Chronic hepatitis C is a complex disease associated with host metabolic modifications resulting in a unique metabolic syndrome including insulin resistance, liver steatosis, and hypobetalipoproteinemia.2, 3 The most striking link between HCV and lipids resides in the association of HCV particles with lipoproteins.4, 5 Indeed, HCV virions with a density <1.06 g/mL are associated with lipoproteins, thus forming hybrid particles known as lipoviral particles (LVPs). These low-density viral particles are globular, rich in triacylglycerol and total cholesterol (TChol) and contain the viral envelope glycoproteins and nucleocapsid (composed of HCV RNA and core protein). In addition, LVPs contain all the apolipoproteins (apo) that define the triacylglycerol-rich lipoproteins (TRLs). Indeed, apolipoprotein (apo) B, apoE, apoCI, apoCII, and apoCIII, all of which characterize very low-density, intermediate-density, and low-density lipoproteins (VLDL, IDL, and LDL, respectively), also characterize LVPs (for review, see André et al.6 and Bartenschlager et al.7). Interestingly, the proportions of circulating low-density virus vary widely from patient to patient; in some cases, all HCV RNA is recovered in plasma low-density fractions or is coimmunoprecipitated by apoB-specific antibodies.8

The study of LVPs has been hampered by the absence of an in vitro culture system that produces apoB-associated viral particles. Infectious cell culture–produced HCV (HCVcc) that can be propagated efficiently only in the human hepatoma cell line Huh7 has higher density than in vivo circulating viruses.9 HCVcc are associated with apoE and apoC, but only marginally with apoB, in contrast to ex vivo–characterized LVPs.10, 11 Despite these differences, two sets of evidence further ascertain the role of lipoproteins in HCVcc assembly. First, alteration of the lipoprotein pathway by inhibition of the microsomal triglyceride transfer protein (MTP) or of the diacylglycerol acyltransferase-1 (DGAT-1) or silencing of apoB or apoE expression decreases the production of infectious HCVcc virions.12-14 Second, the phospholipid compositions of HCVcc and TRL share similar characteristics, whereas they strikingly differ from those of cellular membranes or envelopes of virus that assemble at cellular membranes.15-17 Furthermore, lipoprotein lipases that specifically hydrolyse lipoprotein triacylglycerol modify HCVcc biochemical and physical features and decrease their infectivity.18, 19 Thus, both in vivo–produced and in vitro–produced HCV particles share many characteristics of lipoprotein association, but with differences in the extent of apoB association.

Recently, we studied the capacity of cell lines to secrete recombinant envelope glycoproteins E1 and E2.20 Only cell lines that produce TRLs such as HepG2, Huh7, and Caco-2 were able to secrete the envelope glycoproteins, in contrast to cells that do not synthesize lipoproteins. The envelope glycoproteins and apoB were present in the same lipoproteins released from HepG2 and Caco-2, but only marginally or not at all with particles released from Huh7. Poor lipidation of apoB in Huh7 compared with other cell lines might explain these differences.21 ApoB-competent cellular models thus produce subviral particles that are recombinant E1E2-containing lipoproteins, associated or not with apoB, depending on the cell. In HCV-infected patients, similar subviral particles might coexist with infectious virions. The aim of this study was to determine whether such subviral particles circulate in the blood of infected hepatitis C patients.


ANCOVA, analysis of covariance; apo, apolipoprotein; DC, dendritic cell; HBV, hepatitis B virus; HCV, hepatitis C virus; HCVcc, cell culture–produced HCV; HDL, high-density lipoprotein; HIV, human immunodeficiency virus; HPLC, high-performance liquid chromatography; IDL, intermediate-density lipoprotein; LDF, low-density fraction; LDL, low-density lipoprotein; LVP, lipoviral particle; PBS, phosphate-buffered saline; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; TChol, total cholesterol; TLC, thin-layer chromatography; TRL, triglyceride-rich lipoprotein; VLDL, very low-density lipoprotein.

Patients and Methods


Unless indicated, all chemicals were from Sigma (Saint Quentin, France). Anti-apoB (clone 1609) monoclonal antibody and peroxidase-conjugated goat anti-human apoB antibody were from BioDesign (Saco, ME). Anti-E2 (H52) monoclonal antibody was obtained from J. Dubuisson (Institut Pasteur de Lille, France). Anti-apoCII polyclonal antibody was purchased from Merck Calbiochem (Darmstadt, Germany). Anti-apoAII and anti-apoCIII polyclonal antibodies and anti-apoE monoclonal antibody were obtained from Millipore-Chemicon (Molsheim, France). Anti-apoCI polyclonal antibody was purchased from LifeSpan Biosciences (Seattle, WA).


Thirty-six chronically infected HCV patients attending the Hepatology Department at the Hospices Civils de Lyon (Lyon, France) were eligible for the study if they were over 18 years old, not coinfected with human immunodeficiency virus (HIV) or hepatitis B virus (HBV), and not given any anti-HCV treatment for the last 6 months. The liver necroinflammatory activity and fibrosis degree had been determined during the last 6 months before inclusion in the study. The study was approved by the Resarch Ethics Committee of the institution. In addition, 200 mL of peripheral venous blood were obtained from four chronically infected HCV volunteers attending the Department of Hepatology, Pr. S. Pol, Cochin Hospital (Paris, France) and who were enrolled in a clinical trial assessing the effect of iron depletion on response to standard treatment. For controls in the purification procedures, noninfected plasmas were obtained from volunteer blood donors (Etablissement Français du Sang, Lyon, France).

Preparation of Lipoprotein Fractions.

Plasma was separated by sequential ultracentrifugation to obtain four low-density fractions whose densities corresponded to those of VLDL, IDL, LDL, and high-density lipoprotein (HDL). Each fraction was obtained by flotation after 4 hours of centrifugation at 4°C and 543,000g with a TLA100.4 rotor and a TL100 ultracentrifuge (Beckman Instruments, Gagny, France). The VLDL top fraction (density <1.0063 g/mL), was obtained after the first centrifugation run. After collection of VLDL, the density of the remaining plasma was successively increased to 1.025, 1.055, and 1.21g/mL with NaBr before each following run, and fractions corresponding respectively to IDL, LDL, and HDL were collected. Each fraction was then dialyzed at 4°C against 150 mM NaCl-0.24 mM ethylene diamine tetra-acetic acid (pH 7.4) buffer and filtered through 0.22-μm pore size filters (Millipore, Saint Quentin, France) before lipid extraction as described below.

Preparation of Low-Density Fractions and LVP Purification.

Plasma was adjusted to 1.055 g/mL with NaBr and centrifuged as described above. After collection, the upper low-density fraction (LDF) was dialyzed as described for lipoprotein fractions and stored at 4°C in the dark, in the presence of 2% protease inhibitor cocktail (P8340; Sigma-Aldrich). LVP purification from normal lipoproteins contained in LDF was performed via protein A immunoprecipitation of the immune complexes that are only found in the LDFs of infected patients as described.4 Briefly, protein A–coated magnetic beads (Miltenyi Biotec, Paris, France) were incubated with LDFs in phosphate-buffered saline (PBS) with gentle rocking for 30 minutes (20 μL of beads per 1 mL LDF). A total of 2 mL sample were then passed through one magnetic column (Miltenyi Biotec), washed with PBS, and collected in 500 μL PBS. For experiments with larger LDF volume, multiple columns were used. Immunocaptured particles (purified LVPs) were subjected to lipid extraction as described below or stored at 4°C in the dark in the presence of 2% protease inhibitor cocktail for biochemical characterization.

Protein, ApoB, and Lipid Quantification.

Protein concentration was determined using Coomassie Plus (Bradford) Protein assay (ThermoScientific, Brebières, France). ApoB concentration in low-density fractions and sera was determined by immunochemical assay (SFRI Diagnostics, Saint-Jean d'Illiac, France). Total cholesterol (TChol), phospholipid, and triacylglycerol concentrations in sera were calculated with Cholesterol RTU, Phospholipid Enzymatic PAP150, and Triacylglycerol Enzymatic PAP150 kits (BioMérieux, Marcy l'étoile, France). ApoB concentration in purified LVPs was determined via enzyme-linked immunosorbent assay (ELISA) as described.20 Mock-prepared LVPs from healthy donors displayed a maximal background of <1% of lipids or apoB detected in LVPs prepared from patients. Lipid extraction of mock LVPs, as described below, has not allowed the detection of any fatty acid by gas chromatography (GC) in the final lipid extract.

HCV RNA Quantification and Genotyping.

RNA was extracted from 150 μL serum, 10 μL lipoproteins, LDFs, or purified LVPs with a NucleoSpin RNA virus kit (Macherey-Nagel, Hoerdt, France) and stored at −80°C. HCV RNA quantification was performed using quantitative real-time polymerase chain reaction of the 5′ HCV noncoding region as described.22 HCV genotyping was performed using the INNO LiPA HCV assay (Innogenetics, Zwijnaarde, Belgium).

Index of HCV RNA Association in LDFs.

Index calculations were determined with apoB included as an internal standard for the lipoprotein compartment as follow: (RNA copy number per milligram of ApoB in LDFs/RNA copy number per milligram of apoB in serum) ×100.4

Lipid Extraction, Phospholipid, and Triacylglyceride Analyses.

Lipid extracts obtained via the Folch procedure23 from 3 mL of LVPs or from 500 μL of each lipoprotein fraction were separated by thin-layer chromatography (TLC) on Silica Gel G60 plates (Merck, Darmstadt, Germany) with hexane/diethyl ether/acetic acid (60/40/1, vol/vol) solvents. Phospholipid and triacylglycerol were scraped off the plate, and the molecular species composition of phospholipids separated by high-performance liquid chromatography (HPLC) on a silica-DIOL column (4 × 250 mm, Agilent 1100) was analyzed via electrospray ionization/tandem mass spectrometry (Q-Trap 2000, Applied Biosystems). Phospholipid classes were eluted subsequently from HPLC as a function of the headgroup polarity using the solvent mixture hexane/isopropanol/aqueous ammonium acetate 5 mM 62.8/34.8/2.4 at the rate of 100 μL/minute. Experimental details have been discussed in recent reviews.24, 25 The method was set to detect the precursors (i.e., the parent phospholipids) of a characteristic fragment ion of each polar headgroup. Mass spectra were processed with Analyst software (v1.4.2, Applied Biosystems). Assignment of the structure to mass peaks and deisotopization correction were performed with LIMSA software26 using a library prepared for the serum circulating lipids.

Analysis of fatty acids in the triacylglycerol fraction separated by TLC (silica G25; solvent mixture hexane/methyl ether/formic acid: 80/20/2 vol/vol) was achieved by GC separation of methyl esters prepared by acid transmethylation according to Christie WW.27 Separation was achieved on a Carbowax 20 M capillary column (0.25 mm, 30 m, Quadrex) fitted on a Thermo-Electron 8000 GC chromatograph.

Anti-LVP Antibody Purification.

A total of 20 μL of protein A–coated magnetic beads (Miltenyi Biotec) were incubated at room temperature with 1 mL of LDF in PBS with gentle rocking for 30 minutes. Samples were then passed through one magnetic column (Miltenyi Biotec) and washed with 2 mL of PBS and then with buffers of decreasing pH (successively, 300 μL of 0.1 M tris-acetate buffer pH 5.0, pH 4.0, and pH 3.0). Each collected fraction was immediately neutralized by addition of 0.1 N NaOH. Fractions eluted at pH 4.0 were used to stain the western blotting membrane.

Western Blotting.

Immediately after preparation, LDFs or LVP fractions were collected in Laemmli buffer and denatured at 95°C for 5 minutes and kept at −20°C until analysis. Positive controls for E1 and E2 were obtained from lysates of cells expressing HCV glycoproteins or from supernatants of E1/E2-Caco2 differentiated cells,20 collected into Laemmli buffer, denatured, and conserved as described above. Samples were fractionated via sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene fluoride membrane. Blots were probed with antibodies against apoAII, apoCI, apoCII, apoCIII, apoE, HCV glycoprotein E2 (H52), or antibodies purified from patients' LDFs, followed by peroxidase-conjugated goat anti-mouse antibody, donkey anti-goat antibody (Santa Cruz Biotechnology), or rabbit anti-human IgG (Jackson ImmunoResearch Laboratories). Blots were developed with enhanced chemiluminescence reagents according to the manufacturer's instructions (SuperSignal West Femto Maximum Sensitivity Substrate, Perbio Science).

Statistical Analysis.

Analysis of covariance (ANCOVA) was used to model the detailed phospholipids composition associated with LVP. The program was run with the software XLSTAT (Addinsoft, France). The test assumes that the associated lipids, specifically phosphatidylcholine molecular species, are unevenly distributed in the lipoprotein fractions (HDL, VLDL, IDL, and LDL) isolated from the patients. Standardized regression coefficients (sometimes referred to as beta coefficients) compare directly the relative influence of the explanatory variables (i.e., the phosphatidylcholine composition of the four lipoprotein fractions) on the dependent variable (phosphatidylcholine comprised in LVPs).


Patient Characteristics.

Patients were mainly infected by genotype 1 HCV with a mean viral load at 5.93 ± 0.74 log10 RNA copies/mL (Table 1). Median fasting serum lipid levels were 208 mg/dL TChol, 71 mg/dL triacylglycerol, and 199 mg/dL phospholipid with 101 mg/dL apoB. The median alanine aminotransferase level was 66 U/L. METAVIR fibrosis and activity mean scores were 2.30 ± 0.95 and 1.19 ± 0.62, respectively.

Table 1. Patient Characteristics
  1. Abbreviations: AST, aspartate aminotransferase; ALT, alanine aminotransferase; BMI, body mass index; GGT, gamma-glutamyl transpeptidase.

No. of patients36
Median age, years (range)51 (36-72)
BMI, kg/m2, mean ± SD25.07 ± 4.66
No. of patients with BMI ≥305
Serum viral load, log10 RNA/mL, mean ± SD (range)5.93 ± 0.74 (4.13-7.52)
HCV genotype 
METAVIR score 
 Activity (A0-A4), mean ± SD (range)1.19 ± 0.62 (0-2)
 Fibrosis (F0-F4), mean ± SD (range)2.30 ± 0.95 (1-4)
ALT, UI/L, mean (range)78.6 (24-250)
AST, UI/L, mean (range)62.9 (19-169)
GGT, UI/L, mean (range)160.3 (15-1,186)
Bilirubin, μmol/L, mean (range)11.4 (5-27)
ApoB, g/L, mean ± SD1.43 ± 1.34
Triacylglycerol, g/L, mean ± SD1.03 ± 0.89
Cholesterol, g/L, mean ± SD2.16 ± 0.54

HCV RNA Association with Lipoprotein Fractions.

The proportion of HCV RNA in LDFs (density <1.055 g/mL) of plasma collected after overnight fasting was estimated by an index of association (see Patients and Methods). The mean index was ≈40% and the median was 12.55% with a large dispersion of values (Fig. 1A).

Figure 1.

HCV RNA distribution in lipoproteins and LVP fractions. For each of the 36 patients, LDFs with density <1.055 g/mL were separated from plasma. (A) HCV RNA and apoB were quantified in plasma and LDFs, and the proportion of HCV RNA recovered in LDFs was expressed by an index of association that included apoB concentration as an internal control as described in Patients and Methods. The distribution of the patient indexes is represented. Values in the upper right corner of the graph refer to the mean and median as percent of the index. (B) LVPs were purified by protein A precipitation from LDFs as described in Patients and Methods before HCV RNA and apoB quantification. The apoB/HCV RNA molar ratio distribution in the patient cohort is presented. Values in the upper left corner of the graph refer to the mean and range of the apoB/HCV RNA ratios.

LVP Purification, Protein Content, and Antibody Specificity Analysis.

LVPs were purified from the plasma LDFs of 36 patients by protein A–mediated precipitation of viral particles naturally coated with patient's antibodies. Quantification of HCV RNA in LVPs indicated that on average, 34% ± 26% of HCV RNA associated with low-density particles circulate as LVP immune complexes as reported.4

ApoB was detected and quantified via ELISA in all but four patient samples (Fig. 1B). The other TRL-associated apolipoproteins, apoE, apoCI, apoCII, and apoCIII were detected in all tested LVP samples by western blotting (representative blot on Fig. 2A). As controls of the purification procedure, apoAII that is distinctive of HDL (density >1.055 g/mL) was not detected in LDF. Furthermore, protein A immunoprecipitation of LDF prepared from uninfected volunteers did not capture any apolipoprotein, and patients' sera did not contain anti-TRL autoantibody (Fig. 2A and Supporting Fig. 1). Despite the inherent difficulties caused by the high variability of HCV glycoproteins, even within a single subtype, the specificity of the LVP-associated antibodies was tested for six patient-purified LVPs and two antibody samples recognizing E2 (representative blot on Fig. 2B). E1 could not be detected in the studied samples (data not shown), even though previous reports have described its presence in LVPs.5, 28 Symmetrically, envelope glycoprotein E2 was detected by a genotype 1a–specific monoclonal antibody in two LVP samples from 12 HCV genotype 1 patients (representative blot on Fig. 2C). No reactivity against E1 could be detected (data not shown). Overall, despite technical limitations due to the lack of autologous glycoproteins and antibodies, these data confirmed that LVP immune complexes contain TRL apolipoproteins and at least the viral envelope glycoprotein E2 against which patient antibodies are directed. Surprisingly for four patients, apoB concentrations in low-density viral particles were below the ELISA detection limit, suggesting the presence of low-density virions not associated with apoB that might resemble those produced by Huh7 cells.

Figure 2.

Characterization of apolipoproteins, viral envelope glycoproteins, and antibodies associated with purified LVPs. (A) Protein A–purified LVPs with a density <1.055 g/mL (LDF) were analyzed via SDS-PAGE under reducing conditions and immunoblotted with apoAII-, apoE-, apoCI-, apoCII-, or apoCIII-specific antibodies. Materials were similarly prepared from the LDFs of uninfected blood donors via protein A immunoprecipitation (mock LDFs and LVPs) and analyzed under the same conditions. The plus symbol (+) indicates control plasma from a blood donor. Representative blots of three experiments are shown. (B) Cell culture supernatants of naïve Caco2 cells and E1E2-expressing Caco2 cells were analyzed via western blotting with a monoclonal antibody against E2 (H52) or antibodies purified from patient LVPs as described in Patients and Methods. (C) LVP protein content was analyzed via western blotting for the presence of E2 glycoproteins by staining with an E2-specific antibody (H52). One positive representative blot is shown that includes mock-prepared LVPs (healthy blood donor), patient 2 (genotype 4), patient 11, and 3 (genotype 1b). Caco2 and E1E2-Caco2 cell lysates were loaded onto the gel to serve as a positive control. LVP E2 appears to have an apparent molecular weight in the upper range of E2 from cell lysate as described.20

HCV RNA/ApoB Molar Ratios in LVPs.

Because only one apoB molecule is present per TRL, molar ratios of apoB to HCV RNA should indicate the proportion of LVPs containing viral genomes. These ratios calculated for the 32 apoB-positive LVP showed a Gaussian distribution that peaked at 6.33 ± 2.64 log10 apoB mol/HCV RNA genome (Fig. 1B) and suggested that the vast majority of circulating LVPs lack viral RNA. Likewise, attempts to detect core protein in these LVPs failed, arguing that they do not contain a nucleocapsid (data not shown). The plasma of 32 out of 36 patients thus contained HCV RNA–negative LVPs that appeared as subviral apoB-positive particles bearing at least E2 at their surface. These particles thus resemble apoB- and E1E2-positive lipoproteins produced in vitro by HepG2 or differentiated Caco-2 cells. LVPs might therefore define a class of modified lipoproteins.

TRLs form a family of lipoproteins that derive from VLDL and chylomicrons assembled respectively in the liver and intestine, which then undergo intravascular modification by lipases with formation of lipoproteins with higher density. TRL lipidome depends thus on the diet of each individual; the lipid compositions of IDL and LDL conserve most features of VLDL.29, 30 HDLs mediate the reverse transport of lipid from tissues to the liver and differ significantly from TRLs. To further evaluate the similarities or differences between LVPs and lipoproteins, we thus compared their lipidomes.

Comparison of LVP and Lipoprotein Lipidomes.

We compared the lipid compositions of LVP and lipoproteins prepared from identical plasma samples. LVPs and lipoproteins were prepared from 200-300 mL of blood collected from patients whose characteristics are listed in Table 2. As above, the proportion of HCV associated with lipoproteins varied between the four patients, but immune-LVP complexes were always captured by protein A. Association of RNA with apoB within the precipitated material was constant (log10 apoB/RNA molar ratio range, 6.14-6.52) (Table 2) and fits to the observation made with the 36 cohort patients (Fig. 1B). Concentrations of triacylglycerol, phospholipid, TChol, and apoB in the lipoprotein fractions and in LVPs are shown in Table 3. As expected, triacylglycerol/apoB ratios in lipoproteins were inversely correlated with the density of lipoproteins, indicating correct separations of lipoproteins. The TChol, phospholipid, and triacylglycerol lipid content was much higher in LVPs than in fractions from which they were purified (LDFs) and even more than in the less dense fraction (VLDL) (i.e., triacylglycerol/apoB = 104.97 ± 25.51 in LVPs versus 1.95 ± 0.22 in LDFs; TChol/apoB = 65.52 ± 19.72 in LVPs versus 1.68 ± 0.17 in LDFs) (Table 3). These ratios indicate that LVPs have greater than 30 times more triacylglycerol and TChol per particle than lipoproteins of the same density, suggesting that LVPs bear a heavier nonlipidic load than their lipoprotein counterpart.

Table 2. Characteristics of the Four Patients Subjected to Detailed Lipidomic Analysis
PatientsHCV GenotypeSerum Viral Load (log10 RNA/mL)LDF Viral Load log10 RNA/mL)ApoB/RNA Molar Ratio LVP (log10)
Table 3. Characteristics of Lipoprotein Fractions and LVPs
CharacteristicMass Ratio*
  • Abbreviations: PL, phospholipids; TAG, triacylglycerol.

  • *

    Ratios are expressed as the mean ± SE of the four patients (A, B, C, and D).

Plasma3.07 ± 0.771.78 ± 0.282.32 ± 0.60
LDL3.04 ± 1.241.34 ± 0.473.62 ± 1.12
VLDL1.39 ± 0.063.28 ± 0.541.4 ± 0.21
Whole fraction (LDF)1.64 ± 0.221.95 ± 0.221.68 ± 0.17
Purified LVP84.23 ± 23.20104.97 ± 25.5165.52 ± 19.72

LVPs and VLDL Have Similar Triacylglycerol Compositions.

We then compared the triacylglycerol fatty acid composition of lipoproteins and LVPs from patients A and D. As expected, the triacylglycerol fatty acid composition of lipoproteins was different between patients, particularly for C18:1 n-9 and C18:0 (Fig. 3 and Supporting Tables 1 and 2), while similar profiles were observed between LVPs and TRLs for each patient (see percentages of fatty acid C18:0, C16:0, C18:1 n-9, and C18:2 n-6). By contrast, differences were observed between the triacylglycerol composition of LVPs and HDL (see C16:0 for both patients or C18:2 n-6 for patient D). Therefore, the triacylglycerol fatty acid compositions of LVPs and TRLs showed close similarities within each individual patient and differences, as TRL, from one patient to another.

Figure 3.

Comparison of triacylglycerol fatty acid compositions in lipoprotein fractions and LVPs. After lipoprotein separation and LVP purification, lipids were extracted from 3 mL of LVP or 500 μL of lipoprotein fractions using the Folch procedure. Triacylglycerol was separated via TLC before transmethylation, and obtained fatty acid methyl esters were analyzed by gas chromatography. Peaks were identified using commercial standard fatty acid methyl esters. Fatty acid composition is expressed as mol% of total fatty acid content in HDL, IDL, LDL, VLDL, or LVP for patient A (A) and patient D (B).

LVPs and TRLs Have Similar Phospholipid Compositions.

We then performed an exhaustive analysis of phospholipids in purified LVPs and lipoproteins via electrospray ionization/tandem mass spectrometry. The relative proportions of phospholipid classes in LVPs and lipoproteins for patients B, C, and D (Fig. 4B) showed similar phospholipid class ratios in lipoproteins and LVPs. Interestingly, no phosphatidylserine was detected in LVPs and lipoproteins, even though the method had a phosphatidylserine detection limit of 1 pmol/L in the injected lipid extract. The minimum phospholipid concentration obtained was 100 μmol/L of phospholipid for the less concentrated LVP preparation. Thus, under these conditions, the phosphatidylserine concentration was less than one molecule per 1 × 108 phospholipid molecules. Therefore, LVPs (as all lipoproteins) were virtually devoid of phosphatidylserine, which is normally present in cellular membranes or in virions such as HIV that acquire their envelope from these membranes (Fig. 4B and Table 4). However, the phosphatidylethanolamine/phosphatidylcholine and sphingomyelin/phosphatidylcholine ratios that distinguish the lipoprotein classes31 differed between LVPs and all other TRLs.

Figure 4.

LVP phospholipid composition is comparable to phospholipid compositions of VLDL, IDL, and LDL, but different from the composition of HDL. Phospholipid classes contained in lipid extracts of VLDL, IDL, LDL, HDL, and LVPs obtained from patient B were separated by liquid chromatography on a silica-diol column and analyzed by tandem-mass spectrometry analysis. (A) Mass spectra of each fraction are presented as relative abundance. Each peak stands for one phospholipid molecular species identified by its carbon:double-bond number. (B) Lipoproteins and LVP phospholipid classes were quantified and expressed as the percentage of total phospholipids. Results are expressed as the mean value for patients B, C, and D. Results are compared with the HIV lipidome.16 (C) Comparison of phosphatidylcholine molecular species composition of LVPs between patients and all lipoprotein classes for each patient. Relative standardized coefficients obtained by ANCOVA reflecting the relative influence of each lipoprotein fraction on phosphatidylcholine molecular species composition of LVPs are plotted. *P < 0.0001. **P = 0.033.

Table 4. Comparison of Phospholipid Class Composition of Lipoproteins and LVPs with HIV and Cellular Membrane Lipidomic Dataset
  1. Data represent the percent of total phospholipids. The means ± SE presented for LDL, IDL, VLDL, and LVP were obtained for three patients. Data for HIV-1 and MT-4 cells are from Brügger et al.16

  2. Abbreviations: ND, not determined; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PS, phosphatidylserine; SM, sphingomyelin.

PE28.8 ± 2.318.8 ± 7.137.5 ± 4.828.7 ± 2.335.232.9
PC48.9 ± 6.268.2 ± 6.650.2 ± 6.961.4 ± 2.91643
PI13.6 ± 6.94.2 ± 2.44.1 ± 2.33.4 ± 3.3NDND
SM8.6 ± 2.88.8 ± 2.18.2 ± 3.96.5 ± 3.433.110.4

Phosphatidylcholine was the most abundant phospholipid class in all four patients' lipoproteins and LVP. ANCOVA statistical analysis was applied to test whether phosphatidylcholine molecular species profiles of LVPs matched the profiles of any lipoprotein class for patients A-D (Fig. 4C). No significant match was found between each patients' LVPs or between LVPs and HDL profiles (null standardized coefficients), while LVPs and IDL, LDL, and VLDL profiles were significantly related to each other (standardized coefficient: 0.339 ± 0.050 [P < 0.0001], 0.453 ± 0.093 [P < 0.0001], and 0.193 ± 0.090 [P = 0.033], respectively). A comprehensive phospholipid molecular species mass spectra of the total phospholipid in LDL, VLDL, and HDL lipoprotein fractions and in purified LVPs of patient B illustrated the similar molecular species profiles of VLDL and LVPs (Fig. 4A). Taken together, these data suggest that LVPs are modified TRLs.


From recent studies, HCV virions are thought to be hybrid particles that result from the combination of lipoprotein and virus moieties.7 Several lipoprotein-producing cell lines secrete the HCV envelope glycoproteins in absence of any other viral components.20 In these models, glycoproteins form low-density subviral nucleocapsid-free HCV particles. The current study reports for the first time that such subviral HCV low-density particles are also present in the blood of infected patients at high concentrations and largely outnumber HCV RNA–positive LVPs. Protein A–purified LVPs are very rich in neutral lipids, TChol, and triacylglycerol, and contain HCV glycoprotein recognized by natural antibodies of the patient and all the apolipoproteins that characterize TRLs, including apoB in large quantity for 90% of the patients. The high ratio of apoB and E1E2-positive, nucleocapsid-free LVPs over HCV RNA–positive LVPs might be overestimated if TRLs could nonspecifically bind to a small number of LVPs. However, this possibility is unlikely. Electron microscopy of LVPs revealed large and single particles.4 Similarly, in vitro–produced apoB and E1E2-positive, nucleocapsid-free particles have two- to three-fold larger diameters than E1E2-negative lipoproteins.32 In addition, the higher molar ratios of neutral lipids on apoB in LVPs compared with TRLs indicates that such particles are not agglomerates of standard lipoproteins with LVPs. The association of apoB with LVPs that resists to detergent treatment further rejects this possibility.33

LVP density and composition in triacylglycerol and phospholipid clearly includes LVPs in the TRL family and distinguishes them from exosomes or circulating microvesicles.34, 35 Nevertheless, differences in phospholipid molecular species composition and higher neutral lipid content distinguish LVPs between specific TRLs defined by their density. Interestingly, most LVPs resemble empty, nucleocapsid-free subviral particles, similar to recombinant subviral envelope particles produced in vitro, whereas nucleocapsid-containing LVPs are only a subset of the whole LVP ensemble. Because all HCV proteins are generated from a unique precursor, it is intriguing that such large excess of two HCV proteins can be secreted and found in the blood without noticeable accumulation of the other peptides in any other sites. Different protein half-life and traffic might explain such disparity.

An excess of subviral particles over infectious virions in plasma is common during viral infections. For instance, HBV surface antigen (HBsAg) circulates in the blood as nucleocapsid-free, envelope-containing subviral particles that also outnumber HBV DNA–positive Dane particles by 1 × 103 to 1 × 105.36 Subviral, nucleocapsid-free particles, bearing the envelope glycoprotein, are also frequently found during dengue virus or tick-borne encephalitis virus Flavivirus infections.37, 38 Subviral particles appear to exert biologically relevant properties. For example, HBsAg inhibits TLR9-mediated activation and interferon-α production in plasmacytoid dendritic cells (DCs).39 Similarly, HCV LVPs interfere with Toll-like receptor 4–triggered maturation of DCs, inducing a shift in DC function that stimulates T helper 2 cells instead of T helper 1 cells.40, 41 Recombinant LVPs also fuse with liposomes in a fusion process leading to the coalescence of internal contents of TRL particles and liposomes.32 The presence of such high proportions of modified lipoproteins during hepatitis C may modify the physiologic functions of lipoprotein, particularly if they have membrane fusion property, and participate to some HCV-induced metabolic dysfunctions.

We also observed the presence of low-density viral particles that did not contain detectable apoB. Because we could not quantify the envelope glycoproteins, and because the number of glycoproteins per particles is not known, the proportion of nucleocapsid-positive and -negative particles could not be estimated. Thus, it remains to be determined whether subviral, nucleocapsid-negative, and apoB-negative low-density particles, either resembling HCVcc or the recombinant glycoprotein subviral particles produced by Huh7 cells, are also produced in vivo. For four patients, such particles were the only low-density viral particles and they may also be present in unknown proportion in all patients. These particles could contribute to the high molar ratios of neutral lipid over apoB, assuming that they could be coimmunoprecipitated with apoB-positive LVPs; their presence would further increase the overall proportion of subviral particles. It should be stressed, however, that for some patients, all HCV RNA are immunoprecipitated by anti-apoB antibody.8

In conclusion, the HCV circulating viral particle populations are complex and include several forms, such as apoB-positive and -negative as well as nucleocapsid-positive and -negative LVPs that may contribute in different extent to the pathophysiology of chronic hepatitis C.


We acknowledge the contribution of the AniRA – Laboratoire L3/UMS platform of SFR Biosciences Gerland-Lyon Sud (UMS344/US8) for their help. We thank Patricia Barbot, Virobiotec, Center for Biological Resources, Hospices Civils de Lyon, and Claude Vieux for patient and sample management. We thank Vincenzo Vinzi (ESSEC, Cergy-Pontoise, F95000) for his help with statistical testing.