Twenty years after its discovery, HCV (hepatitis C virus) still infects 170 million people worldwide and cannot be properly treated due to the lack of efficient medication. Its life cycle must be better understood to develop targeted pharmacological arsenals. HCV is an enveloped virus bearing two surface glycoproteins, E1 and E2. It only infects humans through blood transmission, and hepatocytes are its only target cells. Hepatic trabeculae are formed by hepatocyte rows surrounded by sinusoid capillaries, irrigating hepatic cells. Hepatocytes are polarized and have basolateral and apical poles, separated by tight junctions in contact with blood and bile respectively. In blood, HCV remains in contact with lipoproteins. It then navigates through hepatic microenvironment and extracellular matrix, composed of glycosaminoglycans and proteins. HCV then encounters the hepatocyte basolateral membrane, where it interacts with its entry factors: the low-density lipoprotein receptor, CD81 tetraspanin, and the high-density lipoprotein (scavenger) receptor SR-BI (scavenger receptor BI). How these molecules interact with HCV remains unclear; however, a tentative sequence of events has been proposed. Two essential factors of HCV entry are the tight junction proteins claudin-1 and occludin. Cell polarity therefore seems to be a key for HCV entry. This raises several exciting questions on the HCV internalization pathway. Clathrin-dependent endocytosis is probably the route of HCV transport to intracellular compartments, and the ultimate step of its entry is fusion, which probably takes place within endosomes. The mechanisms of HCV membrane fusion are still unclear, notably the nature of the fusion proteins is unknown and the contribution of HCV-associated lipoproteins to this event is currently under investigation.
- cell polarization
- hepatitis C virus (HCV)
- membrane fusion
Hepatocytes are the only targets of an insidious pathogen, HCV (hepatitis C virus). This enveloped virus strictly infects humans, and has the capacity to travel furtively in the blood from its initial point of entry in the body, due to its association with lipoproteins. At the approach to the liver, it diffuses through the hepatic microenvironment, where it comes into contact or interacts with several molecules of the extracellular matrix. Hepatocyte invasion involves a set of HCV entry factors or receptors that most likely act in a combination to accomplish virus cell entry. HCV is internalized into endosomal compartments, and ultimately fuses its lipid envelope with intracellular membranes to deliver its genetic material to the cell cytosol, launching virus replication. In the present review, we first place HCV in the context of the hepatic microenvironment it has to cross, and of the cell it invades. The human hepatocyte is a polarized cell, organized into a basolateral pole and an apical pole. This delimits two membranes, separated by TJs (tight junctions) composed of specific proteins. Since two of the recently described entry factors of HCV are components of the TJ, it appears that the polarity of the hepatocyte is a key for HCV entry. We then summarize recent studies on HCV early steps of entry, demonstrate a crucial role for a quartet of proteins in a finely tuned process, and we address the question of HCV hepatotropism. Finally, we describe our current knowledge about HCV membrane fusion.
THE HEPATITIS C VIRUS: FACTS AND FIGURES
HCV infects an estimated 3% or 170 million of the world's population . In Egypt, the HCV prevalence rate is the highest of the world, with approx. 25% of the population positive for HCV. This is the unfortunate result of campaigns of intravenous mass administration of tartar emetic against schistosomiasis, without disposable needles and syringes . HCV is transmitted mainly through exposure to contaminated blood or blood derivatives, i.e. by transfusion or intravenous drug administration in drug users or by pricking with HCV-soiled needles. Some cases of sexual transmission have also been reported [3,4]. HCV is a major cause of chronic hepatitis, liver cirrhosis and hepatocellular carcinoma, and hepatitis C is also the most frequent indication for liver transplantation. A protective vaccine is not available yet, and therapeutic options are still limited. Current treatment consists in the administration of pegylated recombinant IFN-α (interferon-α) associated to ribavirin. However, sustained virological response is achieved in roughly 50% of patients and toxic side effects can be extremely severe [5,6]. There is therefore an urgent need to develop more effective and better tolerated therapies for chronic hepatitis C. However, a more detailed understanding of the viral life cycle is necessary for such an achievement.
HCV was identified in 1989 by immunoscreening of an expression library with serum from a patient with post-transfusion non-A, non-B hepatitis . HCV is the only member of the Hepacivirus genus. This genus belongs to the Flaviviridae family, which includes human and animal pathogens (yellow fever, dengue, West Nile and tick-borne encephalitis, human flaviviruses, cattle pestiviruses and simian GB-virus B) . HCV infects only humans and chimpanzees, which sets this virus apart from the other Flaviviruses. Another peculiarity of HCV is the exceptional low density of the virus particles resulting from the association of the virus with serum lipoproteins . Indeed, the majority of HCV circulating in blood was found associated with β-lipoproteins and VLDL (very-low-density lipoproteins) and LDL (low-density lipoproteins) [9–11] (see below). A final peculiarity of HCV is that it targets the liver, and among liver cells, HCV infects exclusively hepatocytes, the polarized cells where the virus replicates. However, HCV RNA has been observed in other human tissues and cells, including B- and T-lymphocytes, monocytes and dendritic cells. In spite of these observations, no HCV replication could be detected in these cells. It is thus very likely that blood cells do not support the full HCV infection cycle . It has been suggested that B-cells might constitute a reservoir for HCV and help the virus to persist and to be transmitted to the liver . The question therefore arises about the specific tropism of HCV towards liver cells.
HCV isolates from the serum of patients fall into three major categories, depending on the degree of sequence divergence of the HCV RNA: genotypes, subtypes and isolates. There are six major HCV genotypes differing in their nucleotide sequence by 30–35%, and a seventh genotype has been discovered in 2008 [14–17]. Within an HCV genotype, several subtypes (designated as a, b, c etc.) can be defined that differ in their nucleotide sequence by 20–25%.
HCV is an enveloped virus which contains a positive strand RNA genome composed of a 5′-NCR (non-coding region) includ-ing an IRES (internal ribosome entry site), an open reading frame encoding structural and non-structural proteins, and a 3′-NCR. The structural proteins include the capsid protein C (core), compacting viral RNA, and the envelope glycoproteins E1 and E2, involved in viral entry and membrane fusion (Figure 1). The non-structural proteins include the p7 ion channel, the NS2–3 protease, the NS3 serine protease and RNA helicase, the NS4A polypeptide, the NS4B and NS5A proteins, and the NS5B RNA-dependent RNA polymerase (RdRp) (for reviews, see [15,18]). All HCV proteins are anchored to or associated with cell membranes, implying that the whole viral life cycle deals with membranes, from the cell entry to the budding [19,20]. Structural proteins compose the viral particle, whereas non-structural proteins are involved in viral replication and assembly after cell invasion. To date, no three-dimensional structure of the structural proteins is available, which seriously hampers our comprehension of the molecular mechanism by which the virus infects its target cells, how the nucleocapsid is assembled and, consequently, of how the virion is formed. Similarly, no high-resolution images of HCV are available to date, which would allow a three-dimensional reconstruction of the whole virus. However, HCV has been visualized by conventional techniques in transmission electron microscopy, revealing viral particles of approx. 60 nm in diameter [21–25] and the presence of the envelope glycoproteins on its surface [21,26]. HCV envelope glycoproteins E1 (∼31 kDa) and E2 (∼70 kDa) are anchored in the viral membrane by their C-terminal transmembrane domains. They form a heterodimer stabilized by non-covalent interactions. This oligomer is most likely to be the structure present at the surface of HCV particles which is involved in viral entry [27–30]. Although the exact role of E1 still remains unclear, E2 was shown to bind a number of entry factors such as glycosaminoglycans, the ubiquitous tetraspanin CD81 or the scavenger receptor SR-BI (scavenger receptor BI; reviewed in [31–34]). Therefore virus-associated E2 is likely to be directly involved in the interactions important for virus attachment and the initiation of productive infection.
As mentioned above, the density of HCV circulating in the blood of patients is very heterogeneous, with very-low-, low- and high-density virus fractions. Very-low- and low-density fractions of HCV are associated with TRL (triacylglycerol-rich lipoproteins), positive for apolipoproteins B, CII, CIII and E [11,35]. This heterogeneity, together with low viral titres in the blood of patients, limited if not precluded the use of plasma-isolated viruses for biochemical analyses that would require a high number of particles. The advent of HCV pseudotyped particles, HCVpp, in 2003 was a milestone on the difficult pathway towards the comprehension of HCV infection and particularly of mechanisms of HCV entry [36–39]. HCVpp are composed of a lipid envelope harbouring HCV E1–E2 proteins assembled on to a retroviral nucleocapsid. We have visualized HCVpp at high resolution by cryo-transmission electron microscopy (Figure 2) . They appear as regular 100-nm spherical structures containing the dense nucleocapsid surrounded by the lipid bilayer. E1–E2 glycoproteins were not readily visible on the membrane surface. Although this model system allowed major breakthroughs in the field of HCV entry research (reviewed in e.g. [31,32,41]), it only helped to understand the early steps of HCV internalization, since HCVpp are built around a retroviral core. Another major milestone came in 2001 with the discovery and isolation by the group of Wakita of a genotype 2a HCV clone from a patient with a fulminant hepatitis . This clone, called JFH1 (for Japanese fulminant hepatitis 1), was able to replicate and produce infectious viral particles in cell culture, the so-called HCVcc (cell-culture produced HCV) [21,43,44]. The introduction of this model in 2005 therefore made it possible to study the whole HCV life cycle in cell culture. From these milestones, the comprehension of the HCV life cycle entered a new era, and studies dealing with biochemistry and cell (patho)biology of HCV infection could be conducted.
After a brief introduction to the architecture of the liver and more specifically of hepatocytes, we will focus on the most recent studies of the complex and subtle interplay between HCV, hepatocytes and their microenvironment during virus internalization, with an emphasis on three main aspects: receptor recognition in the context of polarized cells such as hepatocytes, early steps of HCV entry and, finally, the biochemical aspects of its membrane fusion.
HCV EN ROUTE TO THE LIVER: A TRIP TO AND ACROSS THE HEPATIC MICROENVIRONMENT
Since the target cell of HCV is the hepatocyte, HCV has to first travel in the blood from the initial site of contamination. HCV circulating in the blood can be separated on gradients into high- and low-density fractions, the latter demonstrating the association of elements of the virus with the serum β-lipoproteins VLDL and LDL [9–11,35]. These light fractions were found to be extremely infectious to chimpanzees . Similarly for HCVcc, the highest infectivity was observed for low-density fractions , positive for apolipoproteins B and E , which are components of LDL and VLDL . The notion that highly infectious HCV represents in fact a ‘lipo-viro-particle’ (LVP or low-density fractions containing HCV RNA) has therefore emerged [15,35,48,49]. As we will see below, this might be of relevance for driving HCV entry into hepatocytes.
HCV approaches the liver in sinusoid capillaries, which form a fenestrated endothelium composed of endothelial cells and macrophages, present at the border and the Küpffer cells (Figure 3A). Leaving the circulation through the fenestrae of these capillaries, HCV must then cross the space of Disse lined with hepatic stellate cells (or Ito cells), which play a major role in the production of the hepatic ECM (extracellular matrix) [50,51] (Figure 3A). A direct interaction was observed between a soluble form of HCV E2 and the tetraspanin CD81 present at the surface of hepatic stellate cells, leading to the activation of the matrix metalloproteinase MMP-2 . This proteolytic enzyme is involved in ECM degradation and remodelling, and its HCV-mediated activation could explain subsequent liver injuries (inflammation, acute and chronic tissue damage, fibrogenesis). The space of Disse is therefore a zone of active exchange between blood and hepatocytes. Hepatic ECM, although forming a very limited compartment within the normal liver (less than 3% of the relative area of a normal liver section ), is of major importance in liver physiology through its scaffolding effect and its role in biological functions such as cell proliferation, migration, differentiation and gene expression. It is mainly composed of collagens , of glycoproteins such as laminin, fibronectin, tenascin and nidogen, and of GAGs (glycosaminoglycans) and proteoglycans (formed by a ‘core’ protein on to which several GAG chains are anchored) such as heparin, heparan-, dermatan-, chondroitin-sulfates, perlecan, hyaluronic acid, biglycan and decorin [55,56]. The overall density of the highly hydrated gel formed by the ECM is low, which allows easy diffusion between blood and liver cells. Interestingly, several viruses were found to interact with elements of the ECM of their target tissue [57,58]. Among Flaviviridae, the Dengue virus and the classical swine fever virus bind to heparan sulfates [59–62]. Molecules of the ECM were also shown to play a role in HCV adhesion to the cell membrane, although with somewhat controversial results. Pietschmann and co-workers showed a deleterious effect of heparin on HCVcc infection of hepatoma Huh-7 cells . This agrees with the inhibition of the binding of plasma HCV to Vero cells after treatment of this plasma with heparin , and with the inhibition of HCVpp binding and infection of Huh-7 cells by heparin [65,66]. Heparin was also shown to block HCVcc entry into hepatoma cells at the early step of their membrane binding . Using purified HCV glycoproteins, both E1 and E2 were shown to bind to heparin [65,66,68]. Thus heparin was used to purify plasma HCV  and HCVcc  efficiently by affinity chromatography. Conversely, Wakita and co-workers showed a limited effect of the pretreatment of viral particles with heparin on HCV infectivity, in spite of an inhibition by heparin of HCVcc binding to Huh-7 cells. A similar conclusion was reached by treatment of cells with heparinases . HS (heparan sulfates; in particular highly sulfated HS) were also reported to play a role in HCV entry [65,68]. Indeed, synthetic peptides from E2 displayed a similar affinity towards heparin, HS and dextran sulfate in a solid-phase heparin-binding assay . These controversial results might come from the diversity of models used, parameters observed and technologies applied to the considered study.
LPL (lipoprotein lipase) is present in the blood and can be bound to microvilli at the surface of hepatocytes (Figure 3B) . It hydrolyses triacylglycerol from TRLs, and is involved in the hepatic clearance of TRL-rich lipoproteins from the circulation. LPL acts by forming a bridge between the lipoprotein and HS at the cell surface . The membrane-bound (via HS) and catalytically active form of LPL was found to enhance plasma HCV internalization into hepatoma cells, but to inhibit HCV infection [74,75]. It was suggested that by interacting with the lipoprotein moiety of the viral particle, LPL would stimulate a non-productive route of HCV entry, leading to virus degradation.
HCV AND HEPATOCYTES: A FINELY TUNED TOXICITY
After its journey across the hepatic microenvironment, HCV finds access to the hepatocyte, its target cell for productive infection. Hepatocytes are the ‘chief’ functional cells of the liver and perform an astonishing number of metabolic, endocrine and secretory functions. Hepatocytes contribute to roughly 80% of the mass of the liver. They are epithelial cells with an unique architecture as compared with other epithelial cells. In three dimensions, hepatocytes are arranged in plates that anastomose with one another in trabeculae (Figure 3A). The cells are polygonal in shape and their sides can be in contact either with sinusoids (through their sinusoidal face) or neighbouring hepatocytes (through lateral faces). A portion of the lateral faces of hepatocytes is modified to form bile canaliculi . Microvilli are present abundantly on the sinusoidal face and project sparsely into bile canaliculi. This biliary pole forms the apical membrane delimited by TJs and is equivalent to the apical pole of simple epithelial cells (Figure 3B). Owing to the particular structure of their apical pole, at least two faces of hepatocytes are in contact with blood sinusoids and form the basolateral or sinusoidal pole. This pole does not sit on a basal lamina, in contrast with simple epithelia (Figure 3C), but it is in contact with the hepatic ECM. At their basolateral pole, hepatocytes capture nutrients, information from their microenvironment and substances to be transported to the bile, and several receptor or effector molecules are concentrated at this membrane, such as integrins (receptors to elements of the ECM), transporters of nutrients, transferrin or lipoprotein receptors and tetraspanins. At the apical pole, hepatocytes secrete bile components and re-absorb other molecules. Membrane proteins such as transporters of bile acids and multi-drug-resistance proteins (ATPase pumps for various substrates) are localized at this pole . The formation of TJs is contemporary with that of the apical pole, and the development of membrane polarity is integral to the process of hepatocyte differentiation through cell–cell contact, cell–ECM contact or both [78,79]. Importantly, the formation of such specialized membrane domains implies different protein compositions at the apical and basolateral poles.
In this context, HCV probably ‘lands’ on the hepatocyte basolateral membrane, where several molecules will ensure its ‘guidance system’ into the cell.
RGE/RGD is a well known cell adhesion motif for integrin recognition . This motif is also present in the sequences of flaviviral envelope proteins and involved in virus entry [81,82]. It is also found in the HCV E2 glycoprotein and conserved among various virus genotypes. However, its role in HCV entry through the recognition of an integrin molecule could not be demonstrated .
LDL-R (LDL receptor)
Since HCV is found associated to serum β-lipoproteins, the LDL-R has long been considered a candidate molecule involved in HCV entry [9,84]. Physiologically, LDL-R transports cholesterol-rich LDL particles from the extracellular medium into cells by clathrin-dependent endocytosis. The endocytosis of plasma HCV mediated by LDL-R was established by Agnello et al. , relying on the association of the virus to LDL and VLDL, but not to HDL (high-density lipoprotein). However, this was performed with hepatoma cells where HCV replication was not fully supported. Primary cultures of highly differentiated human hepatocytes are therefore likely to represent the most physiologically relevant model to investigate serum-derived HCV infection. Using plasma HCV and primary human hepatocytes freshly isolated from HCV-negative liver biopsies, Maurel and co-workers clearly demonstrated that the LDL-R is involved in an early stage of HCV infection, through elegant approaches based upon competition with LDL, peptides from LDL and antibodies to the LDL-R . However, LDL-R is not present only on hepatocytes, and expressing this receptor ectopically does not restore cell responsiveness to HCV. This indicates that other ‘players’ are required in the ‘toxic game’ to mediate HCV entry.
The HDL receptor, SR-BI
SR-BI is mainly present at the basolateral pole of hepatocytes, but is also found at the apical membrane [87,88]. It was found to interact directly with cholesterol by a photolabelling approach, suggesting its localization in cholesterol-enriched domains of the membrane, also called ‘rafts’ . It behaves as an authentic receptor for HDL. After binding of its ligand, SR-BI could undergo a rapid internalization into early endosomes by clathrin-dependent endocytosis, leading to removal of cholesterol and recycling of the protein part of HDL into recycling endosomes . However, another pathway of cholesterol uptake that did not involve endocytosis has also been reported . Recent data indicate that SR-BI plays a role in the metabolism of VLDL, and could be a receptor for these lipoproteins as well . SR-BI is highly expressed in steroidogenic cells as well as in the liver. Its implication in HCV entry was first suggested by the group of Vitelli in 2002, which showed that a soluble form of E2 (sE2) could bind to human hepatoma cells, and the isolated receptor molecule was SR-BI . The binding of sE2 to human hepatoma cells could also be blocked by an anti-SR-BI serum , and direct binding of sE2 to cells non-permissive for HCV infection but expressing SR-BI was recently demonstrated . Concerning the whole virus, conflicting results have been reported, some suggesting a glycoprotein-dependent SR-BI–HCV interaction , other reporting an interaction between HCV and SR-BI through the lipoprotein part of the viral particle, namely apolipoprotein B . This apparent discrepancy might come from the use of viruses from different origins, HCVcc derived from the plasma of mice transplanted with human hepatocytes in the former case, HCV derived from the plasma of infected patients in the latter.
Infection of Huh-7 cells  and of primary human hepatocytes  by HCVpp was prevented by anti-SR-BI antibodies and SR-BI gene extinction strategies. Similar conclusions were reached using HCVcc and human hepatoma cells or primary human hepatocytes [67,96]. Altogether, these results defined SR-BI as an essential factor for HCV entry. The establishment of cell lines where ectopic expression of SR-BI restored HCV infection allowed the demonstration that the lipid transfer function of SR-BI should be intact to convey HCV entry . Since SR-BI is mainly considered an HDL receptor, several investigations dealt with the relationship between its capacity to bind HDL and its involvement in HCV entry. Interestingly, HDL was found to enhance HCV infectivity [96,98–101], although no direct interaction between HCV and HDL could be observed [98,101] (for a specific review, see ). Anti-SR-BI antibodies abrogated this enhancing effect, through a subtle interplay between inhibition of HCV binding to SR-BI, inhibition of HDL binding to SR-BI/cells and/or inhibition of the SR-BI lipid transfer function [67,93,96]. This led to the conclusion that SR-BI could mediate HCV infection in the absence of lipoproteins, which would then uncouple its HDL transport function from its ability to act as a receptor for HCV. How HDL enhances HCV infectivity still remains unclear. A role for apolipoprotein C1, a minor component of HDL , has been suggested [100,102,103]. ApoC1 would play a role through its association with the HCV particle, that could subsequently enhance its membrane fusion properties.
The tetraspanin CD81
This tetraspanin was the first molecule described to interact with a soluble form of HCV E2 , and therefore proposed as an HCV receptor candidate. Ectopic expression of CD81 in CD81-negative hepatoma cells non-permissive to HCV infection restored permissiveness [36,105,106], or increased HCV infectivity in permissive cell lines . Recently, plasma HCV entry into primary human hepatocytes was shown to be inhibited by antibodies directed to CD81 or by CD81 gene silencing . Together these data established that CD81 is an essential factor for HCV entry. CD81 belongs to the vast family of tetraspanins, structurally related proteins with four membrane-spanning domains, and is an ubiquitous molecule present, e.g., on various epithelial cells, blood cells and myocytes . The reader is referred to reviews dedicated to their structure and function [109–113]. These proteins are organized in the so-called tetraspanin web, where tetraspanins interact with themselves, other tetraspanins and/or integrins forming a net with a fine mesh. The involvement of this web in downstream signalling cascades by recruiting other protein partners intracellularly is one of its key functions. CD81 has no known endogenous ligand and does not possess any intracellular internalization motif . This could explain its very slow endocytosis process . In the liver, CD81 is expressed on sinusoidal endothelium and on hepatocytes, where its localization is mainly basolateral, although some CD81-specific staining has been observed at the apical pole . CD81 ectodomain is formed by two extracellular loops, a large (LEL) and a small (SEL) one. The LEL was shown to be the reception platform for HCV , involving specific regions of E2 [115–119]. A high density of cell-surface-exposed CD81 appeared to be crucial for productive HCV entry into human hepatoma cells . However, the cascade of events triggered after HCV interaction with CD81 remains unclear. Owing to the slow internalization process of CD81, three hypotheses could be raised that are not mutually exclusive: (i) HCV internalization is mediated by other cell surface molecules; (ii) other surface molecules promote CD81/HCV entry; (iii) HCV binding to CD81 induces a rapid internalization of the virus–receptor complex. We have previously seen that other molecules play a role in HCV entry. Recently, the idea emerged that HCV internalization could involve a subtle interplay between CD81 and SR-BI. Both molecules were found to act in a co-operative manner to govern internalization of HCVcc in Huh-7 cells . Further investigation showed that HCVcc bound to cells non-permissive for HCV infection only when SR-BI was expressed, but not CD81 . Using antibodies against CD81 and SR-BI, Baumert and co-workers showed that the kinetics of the inhibition of HCV infection were comparable for both antibodies, suggesting that HCV entry steps mediated by CD81 and SR-BI are linked to each other . Taken together, these data suggest the following sequence of interaction: (i) SR-BI; (ii) CD81. However, further investigation will be needed to determine the exact role of each molecule in the early steps of HCV recognition at the molecular level; it also remains to be determined whether one of these molecules, both or none is(are) endocytosed along with HCV.
Intriguingly, none of the ‘landing platform’ molecules for HCV described above explains the hepatotropism of this virus. Indeed, all of them are ubiquitous and involved in essential functions for various tissues or cell types. A further level of complexity came in 2007 with the astonishing discovery that claudin-1, a protein of the cellular TJ, acted as a co-receptor for HCV entry at a late post-binding step . As described above, TJs in the hepatocytes delimitate the apical membrane (Figure 3B), which is not readily accessible for an element such as HCV coming from the lumen of sinusoid capillaries. This discovery raises several questions. (i) How does HCV find its way to the apical membrane to encounter claudin-1? (ii) Why and how does HCV use such a tedious pathway to enter hepatocytes? (iii) Is the polarity of hepatocytes critical to explain HCV entry? Conversely, one can wonder how specific claudin-1 localization is to TJs, how mobile it is or could be, and whether HCV itself could trigger intracellular signals leading to re- or mis-localization of claudin-1 to the basolateral membrane.
TJs can be seen as seals between neighbouring cells, forming discontinuous ridges in electron microscopy [122,123]. They are essential in maintaining different lipid and protein compositions on the apical and basolateral poles in polarized cells . The backbone of the TJ is organized as strands composed mainly of claudins . Claudins form a superfamily of proteins, all related to functions in cell–cell contact, differentiation and proliferation. They are integral membrane proteins with four transmembrane passages and two extracellular loops, EL1 and EL2. Among claudins, claudin-1 is expressed in the liver, but is not restricted to this organ . This molecule is therefore not the explanation of HCV hepatotropism. In addition to claudins, TJs are composed of the cytosolic protein ZO-1 (zonula occludens-1) and of the integral membrane protein occludin . Owing to their function, TJs are largely viewed as static structures under steady-state conditions. The group of Turner recently demonstrated that this view could be revised at the light of their data obtained by FRAP (fluorescence recovery after photobleaching) on TJs of polarized kidney cells. Indeed, fluorescently tagged proteins of the TJ displayed distinct behaviours towards FRAP, with claudin-1 remaining stably localized at the TJ, occludin diffusing rapidly within the TJ, and ZO-1 displaying a highly dynamic behaviour and exchanging with intracellular pools . It therefore appears that binding interactions at the TJ are much more dynamic than thought. However, claudin-1 and occludin either remained localized at or moved within the TJ, suggesting a somewhat constrained distribution.
Interestingly, in sections of normal liver, claudin-1 is found at the apical/canalicular TJ region as well as at the basolateral/sinusoidal membrane of hepatocytes . Human hepatoma HepG2 cells transfected with CD81 displayed a similar pattern, and the pharmacological alteration of TJ integrity had only minimal effects on cell polarity and HCV infection . Altogether, these studies support the notion that a pool of claudins present at the basolateral pole (and therefore non-junctional) could be involved in HCV recruitment to the membrane. At the basolateral membrane, claudin-1 co-localizes with SR-BI, but at the apical pole it is located close to CD81 . This would be consistent with SR-BI acting before CD81 for HCV membrane recruitment. A proposed sequence of events could then be SR-BI, then CD81 and claudin-1, all present at the basolateral pole. Interestingly, claudin-1 expression is enhanced at the basolateral membrane in HCV-infected liver cells, where it then co-localizes with CD81 . Conversely, the pattern of expression of SR-BI and CD81 remains unchanged . This suggests that HCV could induce re- or mis-localization of claudin-1 to the basolateral membrane, thereby enhancing the physical number of molecules available for interaction with the virus. The hypothesis of an initial interaction between viral and cellular partners at the basolateral pole does not contradict the view that HCV could ‘travel’ from the basolateral to the apical membrane to reach cell–cell contact zones enriched in TJ proteins, notably claudin-1. Different forms of claudin-1 could be involved in HCV entry [88,121]; the structure and localization of these forms are unknown at present, and one could hypothesize that different forms would be expressed in TJs and in non-junctional zones. It also remains to be determined where effective internalization occurs. Also, the nature of the interaction between claudin-1 and HCV is still a matter of debate, but increasing evidence indicates that neither E1 nor E2 interact directly with any region of claudin-1 [121,130]. Further studies are needed to clarify these points, in particular in terms of binding and internalization kinetics.
A major breakthrough came most recently with the discovery of occludin, a component of the TJ, as an HCV entry factor [131,132]. Cells non-permissive to HCV infection became permissive to HCVpp and HCVcc infection by expressing all four molecules at their surface. These four molecules, SR-BI, CD81, claudin-1 and occludin, present together on hepatocyte membranes (but not necessarily on the same membranes nor interacting together), seem therefore to be the ultimate quartet in the finely tuned partnership HCV establishes with the hepatocyte at the onset of infection. This again places hepatocyte polarity and TJs at the center of the chessboard . Occludin depletion did not perturb claudin-1 expression or localization, suggesting that both proteins function separately for HCV entry, but it severely impaired HCV entry [131,134]. The initial binding steps of HCV to the cell membrane were not affected by occludin gene silencing, suggesting that occludin plays a role in a late entry event . At the TJ, occludin binds ZO-1 which in turn binds claudin-1 . A close interaction between occludin and claudin-1 at the TJ was suggested . At present, a fine localization of occludin is not available yet; the question remains whether it could be localized outside the TJ, e.g. at the basolateral pole in a similar manner to what was observed for claudin-1. In the current context, and considering the fact that claudin-1 and occludin are mainly confined to the TJ , a model for HCV sequential recruitment of these four molecules at the hepatocyte membrane is proposed (Figure 4).
IN SEARCH FOR THE ‘HEPATO-TROPISM MOLECULE’ OR EWI-2WINT, THE ABSENT PLAYER
At this point, we are still left with a major question: what defines the propensity of HCV to infect hepatocytes and not other cell types? None of the four molecules described above is specific to the liver or to the hepatocyte. Is it a subtle (and finely tuned) combination of each of these molecules, possibly recruited by the virus itself that defines HCV hepatotropism? Or a specific combination between these molecules in terms of interaction kinetics? These questions remain entirely open at present. Addressing the question in different terms, the group of Dubuisson recently introduced another molecule, or more precisely the absence of this molecule, to the ‘toxic game’ . It was indeed demonstrated that HCV infection of Huh-7 cells could be blocked by expressing the protein EWI-2wint (EWI-2 without its N-terminus) at the surface of these cells. EWI-2wint is a cleavage product of EWI-2, which is known to be a partner of the tetraspanin web , interacting with CD81 in a highly stoichiometric manner [137,138]. It contributes to a variety of CD81 functions seen in different cell types and tissues. EWI-2 was detected in human liver at the surface of hepatocytes, and is associated with CD81 on freshly isolated hepatocytes . EWI-2wint is absent from hepatocytes , but present on other cells harbouring EWI-2. It is able to inhibit the interaction between CD81 and HCV glycoproteins. This is the first description of a factor that could confer very specific properties to the membrane of a cell, the hepatocyte, by being absent from that membrane. This might therefore be the very first step toward the thorough comprehension of the strict hepato-tropism of HCV.
EARLY STEPS OF HCV INTERNALIZATION
After its interaction with several molecules, HCV is sent to intracellular compartments. Biochemical approaches using chemical inhibitors of endosome acidification showed that HCV entered cells via a pH-dependent pathway [36,39,140,141]. Strategies blocking clathrin polymerization, silencing the clathrin heavy chain gene or using dominant-negative mutants of proteins involved in endocytosis resulted in the inhibition of HCVpp or HCVcc entry into hepatoma cells [141–143]. This points to a prominent role for clathrin-dependent endocytosis in HCV internalization. However, it was suggested that additional low pH-dependent events occurring after internalization would be required for productive HCV entry . The nature of these events remains unclear. It is also unclear whether any of the ‘landing platform’ molecules, and in particular the quartet of molecules used by HCV to enter cells, would be internalized concomitantly and, if internalized, by which pathway. Several viruses have developed various strategies to enter their target cells (for a recent review, see ). Concerning HCV, not only the involvement of receptor molecules in its endocytosis remains to be determined, but also further investigation is clearly needed to define whether other pathways of entry could exist. A recent study demonstrates that intact and dynamic microtubules are required for HCV entry into cells, and for post-fusion steps as well . Clathrin-coated vesicles transporting HCV may therefore migrate along microtubules to ensure an efficient delivery of the nucleocapsid at appropriate intracellular sites.
HCV MEMBRANE FUSION
On its pathway to productive infection, HCV has now arrived in an intracellular compartment considered to be early endosomes. The virus will have to get rid of its envelope, which will release the viral nucleocapsid into the cytosol. This corresponds to the fusion step, where the envelope of the virus will form one continuum with the endosome membrane. Envelope proteins are thought to play a key role in HCV membrane fusion.
Viral membrane fusion has been the subject of a plethora of reviews, bringing out several common features from different viruses and structurally unrelated fusion proteins (for recent reviews, see [146–151]). Indeed, for viral fusion to occur, two minimal partners are required: a fusion machinery and lipids. These two partners must act in concert and in a co-operative fashion so that the fusion process can be completed. Fusion machineries have therefore two things in common: (i) they interact with lipids, so they possess hydrophobic segments (fusion peptides or loops) or are able after rearrangements to enter into hydrophobic interactions with membranes; (ii) they adopt specific conformations related to the fusogenic and non-fusogenic states, since fusion is limited in space and time. Dissimilar viruses could therefore infect their host cells by very similar mechanisms at the molecular level of proteins and lipids. Interestingly, similar mechanisms can be brought about by structurally different fusion machineries . Concerning HCV, as already mentioned, no three-dimensional structure of its envelope proteins is currently available, which renders it difficult any study of the HCV fusion mechanism at the molecular level. In this context, the first questions to be addressed were the following. (i) Are HCV envelope proteins involved in membrane fusion? (ii) Is fusion pH-dependent? Does it depend on specific lipids? (iii) Which protein is the fusion protein and where is the fusion peptide/loop?
A convenient way to assay viral membrane fusion directly is to use artificial target bilayers or liposomes whose sizes can be controlled, and in the membrane of which fluorescent lipid probes have been incorporated. Fusion between virus and liposomes is then measured as an increase in the fluorescence signal of the probe . Using this elegant in vitro fusion system, we could demonstrate that HCV fusion relies on the presence of the functional envelope glycoproteins at its surface, for both HCVpp  and HCVcc . Indeed, viral particles devoid of the envelope proteins did not display any fusion. This was corroborated in a cell–cell fusion assay where the E1–E2 complex was expressed at the surface of HEK-293T cells [human embryonic kidney 293 cells expressing the large T-antigen of SV40 (simian virus 40)] . In vitro fusion occurred only at low pH, but at a broad range which would correspond to the pH of early to late endosomes inside the cell [153–155]. This suggests that HCV could fuse its envelope with the membrane of different endosomal compartments along the endocytic pathway. Cholesterol of the target membrane was found to play a critical role in the fusion process. Indeed, fusion was enhanced when cholesterol was incorporated into liposomes ; conversely, cholesterol depletion from the cell membrane inhibited HCV infection . Cholesterol of the viral membrane was also reported to play a key role in fusion, since HCV depleted from its membrane cholesterol displayed only minimal infectivity . Interestingly, sphingomyelin had a strong enhancing effect on HCV fusion when target membranes also contained cholesterol , and sphingomyelin from the viral membrane played a crucial role in HCV cell infectivity . This suggests that microdomains composed of cholesterol and sphingomyelin, the so-called rafts, could play a role in HCV entry, specially at the step of membrane fusion.
As mentioned above, HCV has the ability to associate with β-lipoproteins and light fractions of HCVcc were found to be positive for apolipoproteins B and E. Indeed, we recently showed that these fractions display the highest infectivity, which could be correlated with their higher fusogenicity as compared with fractions of high density . This is a strong indication that the lipids associated with HCV, whatever this association may be , play a key role in the process of HCV membrane fusion.
In the absence of structural data, several studies aimed at determining which HCV glycoprotein is the fusion protein. These studies were based upon indirect strategies analysing the effect of point mutations in E1 or E2 on HCV infectivity and fusion. From the postulate that a fusion protein contains (a) fusion determinant(s), being either a fusion peptide or a fusion loop, we introduced carefully selected mutations at specific sites of E1 or E2, expressed in the context of HCVpp. This strategy proved useful, since it revealed the existence of three regions defined as potential fusion peptide candidates . One region is in E1, the others in E2. Interestingly, three point mutations in a similar region of E1 reduced HCVpp infectivity and severely perturbed HCVcc production in hepatoma cells . This region of E1 was also pointed out as a fusion peptide candidate in a study by the group of Poumbourios . Recently, we demonstrated that one residue in E2 defined in the HCVpp context as essential for infectivity and fusion also played a similar key role in the HCVcc context . This residue is a glycine at position 418 of the sequence of E2, nearby a highly conserved glycosylation site (Asn417). Therefore this region is probably not directly involved in membrane destabilization, but it could act as a fusion ‘helper’ or facilitator through (low pH-induced?) structural rearrangements at the onset of fusion. These results also pointed to a crucial role played by E2 in HCV membrane fusion. However, the question of which protein is the fusion protein remains largely open at present.
Recently, the group of Majano showed that occludin was required for HCV glycoprotein-dependent cell–cell fusion . This raises the elegant hypothesis that at least one protein shown as a key factor for HCV entry could play a role in the fusion mechanism. This would be consistent with the involvement of occludin in late post-binding steps of HCV entry. However, we are left with the question of how it would play this role. By priming the fusion protein? Outside and/or inside the cell? In a pH-dependent manner?
CONCLUSIONS AND PERSPECTIVES
Further studies are needed to understand the molecular aspects of HCV entry and membrane fusion. Although HCV belongs to the family of Flaviviridae, which includes viruses with fusion proteins lying flat on the surface of their lipid envelope (class II fusion proteins), HCV in several molecular details of its life cycle does not behave as other members of this family. HCV is a strict human pathogen, it is found associated with β-lipoproteins and displays a high heterogeneity in density (and therefore probably in morphology). The ectodomains of HCV E1 and E2 are small compared with those of envelope proteins of other viruses of the Flaviviridae family , and E1 or E2 share only little sequence homology with these latter envelope proteins. This structural difference probably translates into functional aspects that are peculiar to HCV. However, the structure of the virus and of its envelope proteins is currently unknown. HCV internalization requires a quartet of receptors necessary and sufficient to allow productive infection, but none of these molecules explains HCV hepatotropism. Surprisingly, an explanation to this hepato-tropism could come from the absence of a factor from the surface of hepatocytes, which is the first description of a cellular property conferred by the absence of a molecule.
In conclusion, the path to a thorough comprehension of HCV infection is still long, and understanding HCV peculiarity is essential for the development of an efficient and targeted pharmacological arsenal aimed at combating HCV toxicity.
This work was supported by the CNRS (Centre National de la Recherche Scientifique) and by the Agence Nationale de Recherche contre le SIDA et les hépatites virales (ANRS) (to E.-I. P.).
E.-I. P. thanks Jean-Francis Michel for the expert drawing of a schematic HCV particle in Figure 1, Agata Budkowska for critical reading, Elodie Teissier for thoughtful comments and Dick Hoekstra for just being Dick Hoekstra.
Abbreviations: ECM, extracellular matrix; EWI-2wint, EWI-2 without its N-terminus; FRAP, fluorescence recovery after photobleaching; GAG, glycosaminoglycan; HCV, hepatitis C virus; HCVcc, cell culture-produced HCV; HCVpp, HCV pseudotyped particle(s); HDL, high-density lipoprotein; HS, heparan sulfate(s); LDL, low-density lipoprotein; LDL-R, LDL receptor; LEL, large extracellular loop of CD81; LPL, lipoprotein lipase; NCR, non-coding region; SR-BI, scavenger receptor BI; TJ, tight junction; TRL, triacylglycerol-rich lipoprotein; VLDL, very-low-density lipoprotein; ZO-1, zonula occludens-1
- © The Authors Journal compilation © 2009 Biochemical Society