The Association of Hepatitis C Virus Glycoproteins with Apolipoproteins E and B Early in Assembly Is Conserved in Lipoviral Particles*

Background: The mechanisms initiating the association between HCV and lipoprotein components to give lipoviral-particles remains elusive. Results: HCV E1E2 interacts with apolipoproteins B and E to produce a protein complex, which is found in mature viruses. Conclusion: The complex formation may be responsible for the initiation of lipoviral-particle morphogenesis. Significance: Obtaining clarity on how HCV particles assemble could open up crucial new treatment options. In patients chronically infected with hepatitis C virus and in the HCV cell culture system (HCVcc), it is known that highly infectious virus particles have low to very low buoyant densities. These low densities have been attributed to the association of HCV with lipoprotein components, which occur during the viral morphogenesis. The resulting hybrid particles are known as lipoviral particles (LVP); however, very little is known about how these particles are created. In our study, we used Huh7.5 cells to investigate the intracellular association between envelope proteins and apolipoproteins B and E (ApoB and ApoE, respectively). In particular, we were interested in the role of this association in initiating LVP morphogenesis. Co-immunoprecipitation assays revealed that ApoB, ApoE, and HCV glycoproteins formed a protein complex early in the HCV lifecycle. Confocal analyses of naïve, E1E2-transduced and HCVcc-infected cells showed that HCV glycoproteins, ApoB and ApoE were found strongly colocalized only in the endoplasmic reticulum. We also found that HCV glycoproteins, ApoB and ApoE were already associated with intracellular infectious viral particles and, furthermore, that the protein complex was conserved in the infectious viral particles present in the supernatant of infected Huh7.5 cells. The association of HCV glycoproteins with ApoE was also evidenced in the HCVpp system, using the non-hepatic HEK293T cell line. We suggest that the complex formed by HCV E1E2, ApoB, and ApoE may initiate lipoviral particle morphogenesis.

supplemented with recently approved NS3/4A protease inhibitors, are expensive and may be associated with drug-drug interactions and severe side effects. No therapeutic or prophylactic vaccine is currently available. New treatment options are urgently required, and their development will require improvements in our understanding of the mechanisms underlying virus assembly, maturation, and secretion from the cell.
Hepatitis C virus particles circulate in patients as hybrid particles, known as lipoviral particles (LVPs), consisting of a combination of HCV and very low-density lipoprotein (VLDL) components (1)(2)(3). LVPs have low densities (Ͻ1.06 g/ml) and are highly infectious. They are rich in triacylglycerol and total cholesterol and contain the viral RNA, the capsid protein, envelope glycoproteins, and apolipoproteins B, E, and C1 (ApoB, ApoE, and ApoC1). ApoB, ApoE, and ApoC1 are essential components of the ␤-lipoproteins LDL and VLDL. It is currently thought that the maturation and release of infectious HCV particles are closely associated with the VLDL synthesis pathway. Similar characteristics have been described for viruses produced in cell culture (HCVcc) by human Huh7.5 hepatoma cells (4 -7). The specific infectivity of HCVcc particles is significantly lower than that of HCV produced in vivo, with most of the viral RNA associated with particles of lower infectivity and a slightly higher density (8 -10).
However, the mechanisms of LVP morphogenesis and the specific structure of these hybrid viral particles remain poorly understood. Following an as yet unidentified triggering event, the nucleocapsid (NC) is translocated into the endoplasmic reticulum (ER), where it associates with glycoproteins, forming HCV RNA-containing particles. The initial viral structure associates with apolipoproteins along the VLDL maturation pathway, generating viral particles of low density, but high infectivity, which are secreted via the secretory pathway of the host cell (for review see Refs. 11,12). However, this interpretation of events is largely speculative, and conflicting data have recently been reported concerning HCVcc morphogenesis. Two studies have reported that microsomal triglyceride transfer protein (MTP) plays a key role in HCV assembly and secretion (4,7). The authors also reported that interfering RNAs directed against ApoB inhibited the release of infectious viral particles, indicating that ApoB may be a limiting factor for HCV assembly. Together, these results suggest that ApoB is required for HCV assembly. By contrast, other studies have indicated that ApoE is essential for the assembly and release of infectious particles, whereas ApoB and MTP have little or no effect on these mechanisms (13)(14)(15). In addition, recent studies have indicated that ApoE is required for HCV production in non-hepatic cell lines, whereas ApoB and MTP are dispensable (16,17). Finally, electron microscopy studies of the ultrastructure of purified HCVcc virions have shown that viral particles are associated with ApoE only (18) or ApoB and ApoE (19).
In this study, we aimed to unravel the mechanisms governing the initiation of interactions between the VLDL and HCV components and the incorporation of apolipoproteins into mature infectious particles. We also attempted to define the role played by ApoB and ApoE in viral infectivity and morphogenesis. We show that ApoB and ApoE form stable associations not only with each other, but also with the HCV glycoproteins. These proteins were found colocalized in the ER only. We show that this association is conserved, for both apolipoproteins and E1E2, in infectious intracellular or extracellular HCV particles. The association of HCV glycoproteins with ApoE was also evidenced in the HCVpp system, using the non-hepatic HEK293T cell line.
Our findings demonstrate that HCV glycoproteins, ApoB and ApoE have an intrinsic capacity to form a complex early in viral morphogenesis. This association could potentially trigger the formation of infectious hybrid particles (LVPs) containing viral and lipoprotein components.
Production of Retroviral Pseudoparticles-Pseudotyped retroviral particles were produced in HEK293T cells by standard methods (20,21). Briefly, retroviral particles were produced from three plasmids: a cytomegalovirus-Gag-Pol packaging construct, a murine leukemia virus HCV E1E2-transducing plasmid, and an expression vector encoding vesicular stomatitis virus G protein. The HCV E1E2-transducing plasmid was obtained by replacing the GFP sequence of the original plasmid with the HCV envelope protein sequences via a PCR-based strategy, making use of the AgeI and SalI restriction sites.
Immunofluorescence Labeling-Cells were grown on 12-mm glass coverslips for 72 h. Cells were fixed with 4% paraformaldehyde in PBS and then permeabilized with 0.1% Tween 20 -2% bovine serum albumin (BSA) in PBS. Cells were double-labeled by incubation overnight with the primary antibody in permeabilization buffer. They were then incubated with the secondary Ab in the same buffer. Confocal microscopy was performed with an Olympus Fluoview 500 confocal microscope.
Quantification of Colocalization-Colocalization was evaluated as previously described (24). The extent of colocalization of the two labels was measured with the "Colocalization" module of Imaris 7.2.3 ϫ 64 (Bitplane AG, Saint Paul, MN). The extent of colocalization was assessed by two different measurements: 1) The percentage of material colocalized, determined by the Imaris software. The program produces a measurement of colocalization for each of the labels. Two values (A and B) for the percentage were obtained, corresponding to each of the Abs used.
2) The Pearson's coefficient. The Pearson coefficient measures the correlation between the intensities of the two labels displaying colocalization in stacks of confocal sections acquired in two channels. This method provides a statistical evaluation of protein colocalization. Pearson's coefficient is a number between ϩ1 and Ϫ1, with positive values indicating a direct correlation, negative values indicating an inverse correlation, and values near 0 indicating no correlation.
Neutralization and Immunodepletion Assays-For the neutralization assay, triplicate samples of intracellular HCVcc or extracellular HCVcc (supernatant) were incubated for 1 h at room temperature with an anti-E2 AR3A (2 g), anti-ApoB 4G3 (4 g) or anti-ApoE (5 l serum) Ab before cell inoculation. For the anti-ApoE serum Ab, 1 l of serum contains ϳ3 g of Ab in total, according to the data supplied by the manufacturer. Intracellular HCVcc was prepared as previously described (6). For immunodepletion assays, protein G-Sepharose beads were incubated overnight at 4°C with the same amounts of the same Ab as for neutralization and then washed with PBS. Triplicate samples of intra-or extracellular HCVcc were then mixed with beads and incubated for 1 h at 37°C. The beads were discarded, and the infectious virus remaining in the supernatant was used to inoculate cells. For both assays, cells were stained 3 days later with an anti-HCV core protein Ab (C7-50) and viral foci were counted manually. The percentage inhibition was calculated by comparing levels with those for the irrelevant IgG.
Sequential Immunoprecipitations (Ip)-Protein G-Sepharose beads were incubated overnight at 4°C with anti-E2 AR3A (8 g), anti-ApoB Orb (16 g), anti-ApoE (4 l serum), or control IgG (Santa Cruz Biotechnology) (8 g) and then washed with PBS. Triplicate samples of extracellular HCVcc were then mixed with the beads and incubated for 1 h at 37°C (Step 1: Immunosubtraction). Beads were collected, and the amounts of HCV RNA or HCV core remaining in the supernatant were determined. The supernatant, containing unbound particles, was then mixed with a second set of beads prepared with the indicated Abs, in the same conditions as for step 1. Beads were collected, and the amounts of HCV RNA or HCV core remaining in the supernatant were determined (Step 2: Immunodepletion). The amounts of HCV RNA or HCV core bound to beads prepared with specific Abs were calculated by comparison with those on beads prepared with irrelevant IgG.
HCV RNA and HCV Core Quantification-HCV RNA was quantified with the routine Abbot m2000 sp -m2000 rt real-time PCR assay. HCV core was quantified in a fully automated microparticle chemiluminescence immunoassay (Architect HCV Ag; Abbot).
Pseudotyped Virus Infectivity Assays-Huh7.5 cells were used to seed 24-well plates at a density of 4 ϫ 10 4 cells/well and cultured overnight. The next day, 400 l of HEK293T cell supernatant containing HCVpp was combined with 5, 10, or 20 l of anti-ApoE serum, 2 g of anti-E2 AR3A , 2 g of anti-ApoB 1D1 Ab or corresponding amounts of isotypic Abs and incubated for 8 h at 37°C. Huh7.5 cells were subjected to infection with these mixtures for 3 days. We then analyzed GFP expression in 10 4 cells with a Gallios TM flow cytometer (Beckman Coulter).

HCV Glycoproteins, ApoB and ApoE, Form a Complex-In
serum samples from patients, HCV particles are associated with very low-density lipoprotein components (cholesterol and apolipoproteins: ApoB, ApoE, ApoCI). Indeed, HCV hijacks the VLDL pathway during its morphogenesis. The resulting complexes are known as lipoviral particles. VLDL morphogenesis is initiated in the endoplasmic reticulum (ER) before the delivery of the pre-VLDL to the Golgi apparatus (25). We investigated the role of HCV glycoproteins in the initiation of LVP formation by transducing Huh7.5 cells with E1E2 sequences only (Huh7.5-E1E2). These two proteins have been shown to form a heterodimer that is retained in the ER (26). The association of envelope proteins and apolipoproteins was also investigated in the context of a complete viral cycle (Huh7.5-HCVcc), with HCVcc-infected cells (JFH1 strain). Confocal immunofluorescence microscopy analysis and co-immunoprecipitation assays were performed to evaluate the association of ApoB, ApoE, and HCV E2 in Huh7.5, Huh7.5-E1E2, and Huh7.5-HCVcc cells. In addition, Giang et al. recently isolated anti-E1E2 Abs (AR4A and AR5A) recognizing different antigenic regions of the virus envelope glycoprotein complex and requiring correct folding of the E1E2 complex for binding (22). These Abs were also used in our immunoprecipitation experiments.
The co-immunoprecipitation with ApoB (anti-ApoB Rock Ab) was observed in all our cell systems for ApoE and in Huh7.5-E1E2 and Huh7.5-HCVcc cells for E2 (Fig. 1, A-C).
Following immunoprecipitation with anti-E2 AR3A , anti-E1E2 AR4A , or anti-E1E2 AR5A Abs, ApoB was systematically detected (Fig. 1, A and B). ApoE was strongly detected when anti-E1E2 AR5A was used (Fig. 1, A and B). Pull-down was less efficient when we used anti-E2 AR3A and anti-E1E2 AR4A Abs. The co-immunoprecipitation of ApoB and ApoE with an anti-E2 antibody was most efficient with anti-E1E2 AR5A Ab, the binding of which is dependent on the correct folding of the E1E2 complex. Of note, we calculated (using ImageJ) the efficiency of protein immunoprecipitation of our Abs when used as described above. We found that ϳ20, 50, 40, 30, and 20% of target proteins from a HCVcc cell lysate was pulled-down when anti-ApoB, Anti-ApoE, anti-E2 AR3A , anti-E2 AR4A , and anti-E1E2 AR5A was used, respectively. We observed weaker co-immunoprecipitation of E2 with anti-ApoE antibody, whereas ApoB was detected as a faint band (Fig. 1, A-C). One previous study showed an association of E1, but not E2, with apolipoproteins (28). However in our study, as the HCV E1 and E2 glycoproteins are believed to form a complex, either within cells or on infectious particles, the detection of HCV E2 glycoprotein presumably also indicates the presence of HCV E1 (referred to as "E1E2"). The apparent efficiencies of co-immunoprecipitation seemed to differ markedly between anti-ApoB, anti-ApoE, and anti-E2 antibodies. In particular, as described above, the anti-ApoE Ab failed to unambiguously pull down ApoB. Also we tested seven different anti-ApoB Abs to identify one (anti-ApoB Rock ) giving a positive pull-down (Fig. 1). It therefore seems highly likely that the quality of the Abs used for immunoprecipitation and the efficiency of the Abs used for detection of the precipitated proteins influence the results obtained. We hypothesize that the ambiguous result obtained with the anti-ApoE Ab may reflect the use of inadequate Abs for pulldown rather than an absence of interaction. In addition, we performed an immunoprecipitation on E1E2-transduced HEK293T cells lysate using the anti-ApoB Rock to demonstrate the specificity of this Ab. It has been demonstrated before that the HEK293T cell line does not express ApoB (27). When the anti-ApoB Rock was used, the immunoprecipitation of E2 was not detected. In contrast, an immunoprecipitation using AR3A on the same cells lysate yielded a positive result (data not shown). Thus, this experiment dem-onstrates that the anti-ApoB Rock does not cross-react with E1E2.
Nevertheless, these data indicate, overall, that HCV E1E2, ApoB and ApoE associate during viral morphogenesis either directly or through an unknown partner. This association occurs early during viral life cycle and is not dependent on the presence of other viral proteins since glycoproteins and apolipoprotein association was detected in Huh7.5-E1E2 cells.
HCV Glycoproteins, ApoB and ApoE Strongly Colocalize Only in the Endoplasmic Reticulum-The subcellular distributions of apolipoproteins and HCV E2 have been investigated before (25,29). Nevertheless, we wanted to establish in which subcellular compartment the interaction of ApoB, ApoE, and E2 may most likely be initiated. Thus the colocalization of these proteins was investigated in our Huh7.5 cells (Huh7.5-E1E2 or Huh7.5-HCVcc), in confocal immunofluorescence microscopy analysis using Imaris colocalization software. Double-immunolabeling experiments were carried out with anti-ApoB, anti-ApoE and anti-E2 Abs, together with antibodies specific for the ER (anti-calnexin) and Golgi apparatus (anti-GM130). ApoB and ApoE strongly colocalized with calnexin (Fig. 2, A and B) and the percentage of material colocalized in Huh7.5-E1E2 and Huh7.5-HCVcc cells was high, with a Pearson coefficient similar to or greater than that of the positive control (Fig. 2, C and  D). HCV E2 and calnexin were strongly colocalized in Huh7.5-E1E2 cells (Fig. 2, A and B) as shown by the percentage colocalization and the Pearson coefficient (Fig. 2C). However, we found that only a fraction of the E2 protein colocalized with calnexin in Huh7.5-HCVcc cells, and a non-negligible percentage of material colocalized with GM130 (Fig. 2, B and D), by contrast to previous reports. However, Pearson's coefficient was close to zero (Fig. 2D), indicating that the colocalization of HCV E2 with Golgi compartment's protein was barely detectable. In both cell systems, no significant colocalization of ApoB and ApoE with GM130 was observed (Fig. 2, A and B). In Huh7.5-E1E2 and Huh7.5-HCVcc cells, the percentage of material colocalized was high for HCV E2 and ApoB or ApoE, with a Pearson coefficient similar to or greater than that of the positive control (Fig. 1F). ApoB and ApoE were consistently found to be colocalized (Fig. 1, D and E). Altogether, these quantitative and statistical analyses showed that ApoB, ApoE, and HCV glycoproteins were found strongly colocalized only in the endoplasmic reticulum. These results are consistent with those of previous studies. This colocalization occurs early in LVP morphogenesis, because HCV E2 colocalization with ApoB or ApoE in ER was observed in Huh7.5-E1E2 cells. This colocalization is therefore not dependent on the presence of other viral proteins. Overall, our data suggest that there is an early interaction between HCV glycoproteins, ApoB and ApoE. This interaction occurs most likely in the ER.
In the HCV Cell Culture System, Most of the Infectious Particles Are Associated with E1E2, ApoB, and ApoE-We found that HCV glycoproteins were strongly associated with ApoB and ApoE within cells. If these results are biologically relevant, this association should also be observed on infectious particles present in the supernatant of infected cells and on newly formed intracellular infectious particles. We therefore performed neutralization and immunodepletion assays with anti-E2 AR3A , anti-ApoB 4G3 , and anti-ApoE Abs to demonstrate the presence of these proteins on the surface of intra-and extracellular LVPs (Fig. 3, A and B). The immunodepletion assay was carried out to determine whether the antibodies bound to the infectious particles before coming into contact with target cells. Anti-E2 AR3A , anti-ApoB 4G3 , and anti-ApoE Abs neutralized most of the infectious intracellular or secreted particles (Fig. 3A). Another anti-ApoB Ab (anti-ApoB 1D1 ) neutralized HCVcc infection as efficiently as anti-ApoB 4G3 Ab (data not shown). Immunodepletion with the anti-ApoB or anti-ApoE Abs was highly efficient in HCVcc infection. The immunodepletion resulted in a more modest (but statistically significant) decrease in infection when anti-E2 AR3A was used (Fig. 3B).
Our results, demonstrating very strong neutralizing or immunodepleting activity of monoclonal anti-ApoB 4G3 (or anti-ApoB 1D1 ) Abs are at odds with several studies suggesting that infectious LVPs produced in the cell culture system do not carry ApoB (16 -18). Taken together, our data suggest that most of the newly formed intracellular infectious LVPs carry ApoB, ApoE, and E1E2 and that this association is conserved on secreted LVPs. These data also indicate that ApoB may be involved in HCVcc entry.

In the HCV Cell Culture System, ApoE and E1E2 Are Found Associated on the Same Infectious or Non Infectious Physical
Particles-We found that ApoE, ApoB and HCV E2 were associated on the same infectious particles. However, if our theory is correct, these proteins should be associated on all physical particles (infectious or not infectious viral particles), and not only on infectious particles. We therefore investigated further whether ApoB, ApoE, and E1E2 were associated on the same physical particles by carrying out sequential immunoprecipitations. We first carried out an immunosubtraction on HCVccinfected cells supernatant with the antibody indicated. The supernatant was collected and subjected to immunoprecipitation with a second antibody (Fig. 4A). The presence of HCV RNA and core was then investigated in supernatants after the first and the second precipitations. Interestingly, when anti-ApoE Ab was used first, 82 and 65% of E2-associated RNA and core, respectively, were immunosubtracted (Fig. 4B). Conversely, the amounts of HCV RNA and core immunocaptured with anti-ApoE Ab were significantly lower after initial immunosubtraction with anti-E2 AR3A Ab (Fig. 4B). Thus, in HCVccinfected cell supernatants, most of the E1E2-associated HCV RNA and core molecules are also associated with ApoE. Together with previous results from neutralization tests, these findings confirm that E1E2 and ApoE are associated on the same infectious particles, but also on the same physical HCV particles (whether infectious or otherwise) in infected-cell supernatants. By contrast, we did not obtain interpretable results when the anti-ApoB Orb Ab was used. We have previously demonstrated that most of the HCVcc particles in an infected supernatant can be neutralized or immunodepleted by an anti-ApoB 4G3 Ab (see above), demonstrating the association of RNA and core-carrying infectious particles with this protein. It is therefore highly likely that the sequential immunoprecipitation results shown here indicate that the anti-ApoB Orb Ab is not an appropriate Ab for immunosubtraction, rather than that ApoB is not present on HCV particles. Additional studies are required to determine whether immunosubtraction can be made more efficient by careful selection of the most appropriate anti-ApoB Ab. Of note, an Ip on HCVcc-infected Huh7.5 supernatant using anti-E2 AR3A , anti-ApoE, and anti-ApoB Orb Abs pulled-down equivalent amounts of HCV RNA (data not FIGURE 2. HCV E2 glycoprotein, ApoB and ApoE strongly colocalize in the ER only. A and B, extent of colocalization of HCV E2, ApoB and ApoE with an ER marker (calnexin) and a cis-Golgi marker (GM130). A, Huh7.5 cells were transduced with HCV glycoproteins (Huh7.5-E1E2) or B, were infected with HCVcc strain JFH1 (Huh7.5-HCVcc). Cells were fixed 72 h after transduction or infection. Proteins were visualized by confocal immunofluorescence microscopy with anti-E2, anti-ApoB, anti-ApoE, anti-calnexin, or anti-GM130 antibodies (as indicated). Double-staining for ApoB/ApoB and for ApoE/core was used as a positive and a negative control, respectively. Confocal images were acquired from double-labeled fixed cells. The amount of colocalization was measured as the "percentage of material colocalized." Each pair of bars indicates the percentages of materials A and B (indicated under each bar pair, in that order) in the colocalized volume relative to the total material for that label above the threshold calculated by the program. The value below the graph indicates the extent of colocalization, measured as the Pearson coefficient in colocalized volumes, representing the statistical correlation between the intensities of the two labels. The extent of colocalization was measured in whole stacks of optical sections with Imaris Colocalization (see "Experimental Procedures"). The labeled proteins are indicated on the x axis. Bars are the mean Ϯ S.E. of values from 10 confocal images. Cal: Calnexin. C and D, visualization of HCV E2, ApoB, and ApoE colocalization. Proteins were visualized by confocal immunofluorescence microscopy with anti-ApoB, anti-ApoE, or anti-E2 antibodies. Double-staining for ApoB/ApoB and for ApoE/core was used as a positive and a negative control, respectively. Bar 2 m. E, extent of colocalization of HCV E2, ApoB, and ApoE. The extent of colocalization of the two labels was measured with the colocalization module of Imaris software and is indicated as the "percentage of material colocalized." Bars are the mean Ϯ S.E. of values from 10 confocal images. The value below the graph indicates the extent of colocalization, measured as the Pearson coefficient in colocalized volumes (see also A, B legend). FIGURE 3. ApoB, ApoE, and E2 are associated on infectious HCVcc particles. A, for the neutralization assay, triplicate samples of intra-or extracellular HCVcc were incubated with the indicated antibody before cell inoculation. B, for the immunodepletion assay, protein G-Sepharose beads were prepared with the indicated antibody and in the same conditions as for the neutralization assay. Samples (lysate or supernatant) containing intra-or extracellular HCVcc were then mixed with beads. Beads were discarded after incubation and the supernatant was used to inoculate cells. For both assays, cells were stained 3 days later with anti-HCV core Ab and viral foci were counted manually (presented as focus forming units/ml). Intra-cell, intracellular HCVcc. Extra-cell, extracellular HCVcc. **, p value Ͻ 0.01.
shown), indicating that ApoB is also associated with RNA-carrying particles.
ApoE and E1E2 Interact When Expressed Together in Nonhepatic Cells-Making use of the HCVpp system (in HEK293T: human kidney cell line), we wanted to determine if HCV glycoproteins had the intrinsic ability to interact with apolipoproteins or if this interaction was dependent on a hepatic environment. Several studies already demonstrated that human kidney cells produced ApoE (30,31). In addition, Tanner et al. showed, in a mass spectrometry study, that HEK293T cells produced small amounts of ApoE and no ApoB (27) (For more information, see the "MOPED" database). In our hand, ApoE was detectable in HEK293T cell lysate by conventional Western blotting (Fig. 5A), albeit much less efficiently than in the lysates of Huh7.5 or HepG2 cells. Interestingly, when HCVpp pro-duced in these cells were incubated with our anti-ApoE Ab before the inoculation of Huh7.5 cells, we detected a robust, dose-dependent neutralization of infection (Fig. 5B). This same Ab had no effect on VSVpp infection. Finally, anti-E2 AR3A Ab neutralized most HCVpp but anti-ApoB 1D1 Ab, which efficiently neutralized HCVcc, had no effect on these retroviral particles (Fig. 5C). These results suggest that ApoE associates with E1E2 during HCVpp morphogenesis and that this association is maintained on infectious particles present in the supernatant of HEK293T cells. No neutralization of VSVpp infection was detected with our anti-ApoE Ab, so it appears highly likely that this association of ApoE with HCVpp is linked to the presence of HCV E1E2 on these particles. Overall, these results confirm that ApoE and HCV glycoproteins spontaneously associate when expressed together. Our results suggest that this FIGURE 4. ApoE and E2 are associated on the same particles. A, immunosubtraction on triplicate HCVcc-infected cell supernatants was first performed with the indicated antibodies. The collected supernatant was then subjected to immunoprecipitation with a second antibody. B and C, amount of HCV RNA (dark bars, y axis on the left) or HCV core (light bars, y axis on the right) immunoprecipitated by the indicated antibody, with or without an initial immunosubtraction step (e.g. anti-ApoE Ab after anti-E2 AR3A or anti-ApoE Ab, respectively) was then calculated. Relative quantity tables: these tables show the percentage of RNA and core captured with the second antibody relative to that achieved with the same antibody but without initial immunosubtraction (antibody without an initial immunosubtraction step ϭ 100%). p values (Student's t test) were determined by comparing immunosubtraction ϩ immunoprecipitation values with immunoprecipitation-only values. **, p value Ͻ 0.01; *, p value Ͻ 0.1; ns, not significant. association is conserved on the surface of HCVpp, and that the limited amount of ApoE produced by HEK293T cells is sufficient to decorate most (at least 70%) infectious retroviral particles.

DISCUSSION
In patients, HCV particles have been shown to be associated with ␤-lipoprotein components (1,32), such as ApoB, ApoE and triglycerides. In the HCV cell culture system, the potential role of ApoB in the viral lifecycle is less clear. In addition, it is not clear how and where hybrid viral particles acquire exchangeable apolipoproteins, such as ApoE and ApoC1 (12,20). In co-immunoprecipitation experiments, we observed a strong association between ApoB and ApoE in Huh7.5 cells that was independent of the presence of HCV proteins. We also provide substantial evidence of early interplay between HCV envelope proteins and ApoB or ApoE during viral morphogenesis (Fig. 1). Recently, Steimann et al. showed that HCV E1E2 play an important role in membrane envelopment, which would depends on a genotype-specific interplay with additional viral factors (33). Our results showed that glycoproteins/apolipoproteins interaction was not dependent on the presence of other viral proteins. This association potentially precedes envelopment. Since they associate with envelope proteins, the apolipoproteins may also play a role in membrane envelopment. It is also conceivable that the E1E2 genotype may influence the efficiency of glycoprotein/apolipoprotein complex formation.
All our co-immunoprecipitation experiments were performed after careful target-cell lysis in the presence of detergent. It is therefore unlikely that these associations reflect the identification of vesicular or particulate structures containing these proteins, as described by Huang et al. (7). Our results instead indicate that these proteins form a complex that associates co-or post-translationally ApoB, ApoE, and E1E2. Moreover, we observed that ApoB and ApoE colocalize strongly with the HCV glycoproteins in the ER only. Surprisingly, we detected no colocalization of apolipoproteins with proteins of the Golgi apparatus. Studies of intracellular VLDL trafficking have shown that, after biogenesis in the ER lumen, lipoproteins are packaged into specialized transport vesicles and delivered to the Golgi lumen (25). Our results suggest that only a very limited proportion (beyond the limits of detection) of these apolipoproteins reach the Golgi apparatus, consistent with previous findings that VLDL secretion is not efficiently supported in Huh7.5 cells (34).
It is reasonable to speculate that the presence of a glycoproteins-apolipoproteins complex could trigger the formation of hybrid particles containing lipoproteins and viral components, according to the putative mechanism proposed in Fig. 6. The first step in this process is the lipidation of a nascent ApoB, accompanied by the simultaneous translocation of an HCV nucleocapsid into the ER lumen. These two precursors are pulled together by the association of the HCV glycoproteins and apolipoproteins into a protein complex. The process by which a monolayer (triglyceride-rich lipopoteins) and lipid bilayers fuse has been investigated in LVPs (35). The authors of this previous study highlighted the key role played by ApoB and the HCV envelope proteins, and indicated that this process occurred in a pH-dependent manner (acidic conditions). The pH of the organelles involved in the egress pathway gradually decreases as the cargo approaches the cell surface (36). Thus, nascent LVPs may eventually meet all the appropriate conditions on the secretory pathway for fusion to proceed, such as a physical association between a nascent VLDL and an HCV viral particle (via a protein complex involving ApoE, ApoB, and glycoproteins) and a low pH (step 2). During their progression along the secretory pathway, the LVPs may or may not acquire more triglycerides through fusion with a luminal lipid droplet FIGURE 5. ApoE and glycoprotein E2 spontaneously associate on HCVpp particles. A, Huh7.5, HepG2, and HEK293T cell lysates were separated by SDS-PAGE and analyzed by Western blotting with antibodies against the indicated proteins (left). B, HCVpp or VSVpp were incubated with various amounts of anti-ApoE or isotypic Ab before cell inoculation. Three days later, 10 4 cells were analyzed for GFP expression with a Gallios TM flow cytometer (Beckman Coulter). Percentage inhibition was calculated by comparison with the levels of GFP expression obtained with untreated pseudoparticles. C, HCVpp were incubated with the indicated antibodies before cell inoculation. The same protocol as in B was then followed. Iso. Ab (Goat, Human, or Mouse): control with specific isotypic Ab. **, p value Ͻ 0.01; ns, not significant.
(step 3), leading to the production of a population of viruses with heterogeneous densities. It has been suggested that the triglyceride content of the particle is the factor controlling LVP density (12).
The observation that HCV E1E2 interacts intracellularly with ApoB and ApoE led us to investigate the presence of ApoB and ApoE on infectious HCV particles produced in cell culture. This verification was necessary, because, as indicated above, several studies have suggested that infectious HCV particles produced in cell culture harbor ApoE but not ApoB (13,14,37). Our neutralization experiments demonstrated that intracellular and extracellular infectious particles carried ApoE and envelope proteins, as demonstrated before (for extracellular particles), but also ApoB. These results confirm that the proteins associate before the particles are released from the cell. Our results also provide the first demonstration that a monoclonal anti-ApoB antibody (4G3 or 1D1) can neutralize HCVcc infection to a similar extent to a well characterized anti-E2 AR3A Ab or an anti-ApoE Ab, suggesting that ApoB may play a role in HCVcc entry. As neutralization levels were close to 100% for all three Abs, it is reasonable to assume that most infectious particles carry HCV E1E2, ApoB, and ApoE. These results are in agreement with a recent ultrastructure analysis study showing that HCV particles incorporate ApoE and ApoB (19). In preliminary experiments, we tested five different commercial anti-ApoB Abs in our infection neutralization tests. Robust neutralizing activity was obtained for only one of these Abs. This low efficiency (in neutralization of infection or protein detection in Western blotting) of most anti-ApoB Abs may account for the discrepancy between our results and some published findings (13)(14)(15)(16)(17)(18). Two recent studies have shown that ApoE is required for efficient infectious particle production in non-hepatic lines, whereas ApoB is dispensable (16,17). These results appear to be at odds with our finding that infectious particles carry ApoB. Nevertheless, we showed in confocal and pull-down experiments that ApoB and ApoE associated spontaneously in naïve and infected Huh7.5 cells. Thus, even though ApoB is dispensable for HCV production, its association with ApoE may drive the transport of ApoB to the membranes of infectious particles.
The results obtained with HCVpp confirm those obtained with HCVcc. However, the association of ApoE with E1E2 on infectious HCVpp has never before been described. In a recent publication, Da Costa et al. described the reconstitution of the entire HCV lifecycle in HEK293T cells and showed that infectious particle production was dependent on the presence of ApoE and miR122 (17). They did not detect ApoE in HEK293T cells, although ApoE expression has been demonstrated in these cells before (27). Our results (displayed Fig. 5A) may thus be seen as conflicting with those of Da Costa et al. We had to use 200 g of cellular protein from these cells to obtain the modest positive result shown on the Western blot presented here. This discrepancy may therefore be explained by the extremely small amounts of ApoE produced in HEK293T cells, making detection challenging.
In summary, we have shown that envelope proteins associate with ApoB and ApoE within cells. We suggest that the resulting protein complex initiates LVP formation. This complex is also conserved on infectious particles found within infected cells or in the supernatant of infected cells. We demonstrate here that the HCV particles produced in cell culture and the viruses found in patients are more similar than previously believed. Finally, the presence of ApoE on infectious HCVpp has never before been reported and this finding should be useful for future studies. . LVP morphogenesis model. The first step in this process is the lipidation of a nascent ApoB, accompanied by the simultaneous translocation of an HCV nucleocapsid into the ER lumen. These two precursors are pulled together by the association of the HCV glycoproteins and apolipoproteins into a protein complex. The nascent LVPs gradually fulfill all the appropriate conditions on the secretory pathway for fusion to proceed, such as physical association between a nascent VLDL and an HCV viral particle (through a protein complex involving ApoE, ApoB, and glycoproteins) and a low pH (step 2). During their progression along the secretory pathway, the LVPs may or may not acquire more triglycerides through fusion with a luminal lipid droplet (step 3), leading to the production of a population of viruses of heterogeneous density. (See also the details provided in "Discussion").