Main

Hepatitis C virus (HCV) is a major cause of acute and chronic liver disease, cirrhosis, hepatocellular carcinoma and liver failure, and it remains the leading indicator for liver transplantation1. A protective vaccine is not yet available, and the current therapies are suboptimal. HCV-specific, direct-acting antiviral compounds have recently been approved for use in clinical practice. However, their use is limited to patients infected with a single subset of HCV genetic types, they are mainly used to augment the standard therapeutic regimen of pegylated interferon-α and ribavirin, and viral resistance to these compounds arises predictably and rapidly2. Therefore, a better understanding of the HCV life cycle is essential if we are to develop additional strategies for antiviral intervention. The processes of virus entry and assembly are two important aspects of this life cycle that are being targeted for future antiviral development.

HCV is an enveloped, positive-strand RNA virus and is the type member of the genus Hepacivirus within the family Flaviviridae3. Other genera in this family are: Flavivirus, which includes many arthropod-borne human pathogens, including dengue virus, West Nile virus and yellow fever virus; Pestivirus, which includes several important pathogens of livestock; and Pegivirus, which includes blood-borne viruses that infect humans, non-human primates and other mammals and have varying disease associations4. All viruses in the family Flaviviridae have similarities in genome organization and overlapping aspects of viral replication.

The HCV RNA genome is 9.6 kb in length and encodes a single long ORF of 3,006–3,037 codons. The viral genome is directly translated to produce a polyprotein that is co- and post-translationally cleaved by viral and cellular proteases into ten gene products. The amino-terminal region of the polyprotein encodes the structural proteins, which are incorporated into the virus particle: the capsid-forming 'core' protein and the glycoproteins, E1 and E2. The carboxy-terminal two-thirds of the polyprotein encodes the non-structural (NS) proteins: p7, NS2, NS3, NS4A, NS5A and NS5B. By definition, NS proteins are expressed in virus-infected cells but are not incorporated into virus particles; they serve to coordinate the intracellular aspects of HCV replication, including RNA synthesis, modulation of host defence mechanisms and virus assembly3. For instance, the heterodimeric complex of NS3 and NS4A proteins, referred to as NS3-4A, contains a serine protease domain and an RNA helicase domain. The serine protease activity is responsible for cleaving the viral polyprotein and cellular antiviral signalling proteins, thereby abrogating the induction of type I interferon in infected cells5. The RNA helicase domain is presumably responsible for unwinding double-stranded forms of the viral genome or clearing proteins from the genome during RNA replication5. As described below, NS3-4A also has an essential role in the assembly of infectious virus particles.

This Review presents an overview of the currently known and unknown properties of infectious HCV particles, their entry and their assembly. Infectious forms of HCV have been historically difficult to study, so we first briefly review the state of the art in experimental systems used to study HCV. HCV particles exhibit unusual and surprising properties; here, we describe what is known about the structure of HCV particles and their entry, a multistep process that involves a surprising number of receptors and remains to be fully elucidated. We then review the other end of the virus life cycle: the assembly and release of HCV particles from infected cells. Emphasis is given to recent findings and current challenges to our understanding of HCV particle structure, entry and assembly. This Review does not describe in detail the intervening steps of RNA translation, modulation of host cell biology, RNA replication or subversion of innate cellular antiviral responses, and readers are directed to other reviews for discussions of these topics6,7,8.

Evolving experimental systems

Since HCV was first proposed to be a distinct infectious virus (in 1975)9 and formally identified (in 1989)10, the rate-limiting step in understanding this virus has been limitations in suitable experimental systems. Thus, our knowledge has grown in a step-wise manner, with intermittent advances in technology rapidly progressing to new insights into HCV biology. The first functional cDNA clones of HCV were developed in 1997 (Refs 11, 12) and allowed an initial molecular characterization of the virus, but these systems replicated only in chimpanzees. Early HCV replication systems made use of drug-selectable 'subgenomic replicons', which allowed the intracellular steps of HCV replication to be studied in cell culture, but infectious virus was not produced13. In the absence of cell culture-infectious virus, many groups began to characterize recombinant HCV-like particles produced from baculovirus expressing the HCV structural proteins14 and to study the HCV entry process by using retroviruses pseudotyped with the HCV glycoproteins (HCVpp viruses)15,16,17.

Complete, reverse-genetics HCV cell culture systems that produce infectious virus (called HCVcc particles) in cell culture have now been developed for a subset of viral genetic types18,19,20,21. It should be noted that most HCVcc studies are carried out in the Huh-7 hepatoma cell line, which efficiently supports HCV entry, replication and assembly. However, under current culture conditions, this cell line lacks many features of bona fide hepatocytes, such as the ability to form polarized sheets or to produce normal serum lipoproteins22. For instance, HCVcc particles produced in Huh-7 cells have slightly different biophysical properties to HCV particles produced in vivo or in primary human hepatocytes23,24. Ultimately, the gold standard for studying HCV biology remains the liver, and current efforts are underway to produce tractable and relevant cell culture and animal model systems25,26.

The enigmatic HCV particle

Physical properties of HCV particles. HCV particles are membrane enveloped and display two surface glycoproteins, E1 and E2, which mediate binding and entry into host cells (Fig. 1a). As described below, HCV entry requires acidification of the endosomal compartment27, presumably to induce rearrangement of the viral glycoproteins into a form that promotes fusion with the endosomal membrane. Beneath the envelope resides a nucleocapsid, which contains a single copy of the viral RNA genome in complex with multiple copies of the viral core protein.

Figure 1: Structural aspects of HCV and related viruses.
figure 1

a | Model of a hepatitis C virus (HCV) particle, showing the membrane bilayer envelope, E1–E2 surface glycoproteins, and the nucleocapsid containing core protein and viral RNA. HCV-associated lipoproteins are not shown. b | Representative electron micrographs of negatively stained cell culture-produced HCV particles captured on an affinity electron microscopy (EM) grid (left) or a magnetic bead recognizing the tagged E2 envelope glycoprotein (middle), and a cryo-EM image of an HCV virion (right). Scale bars represent 50 nm. These EM images demonstrate the lack of regular surface features and the pleiomorphic shape of HCV particles. c | Structure of the dengue virus 2 (DENV-2) E glycoprotein homodimer as viewed from the top and side views (Protein Data Bank (PDB) accession 1OAN). Domain I is coloured red, domain II is yellow, the fusion loop is green and domain III is blue. d | Structure of the bovine viral diarrhoea virus 1 (BVDV-1) E2 homodimer (PDB accession 4ILD), using the same scale and colouring as for DENV-2 E glycoprotein. Note that no fusion loop has been identified in BVDV-1 E2. e | Fitting of the E protein dimer structure into an electron density map of purified DENV-2 particles. For clarity, one dimer is highlighted. f | The two-particle model for lipoviral particle (LVP) structure, which proposes that HCV particles and serum lipoproteins transiently interact. Particles are illustrated with approximate relative scales. g | The single-particle model for LVP structure, illustrating an HCV particle sharing an envelope with a low-density lipoprotein (LDL) particle. Part b images courtesy of M.T. Catanese, The Rockefeller University, USA. apo, apolipoprotein; HDL, high-density lipoprotein.

PowerPoint slide

Full size image

Filtration and electron microscopy (EM) studies have indicated that HCV particles are 40–80 nm in diameter, pleiomorphic, lack obvious symmetry or surface features and contain electron-dense cores28,29,30,31,32 (Fig. 1b). In the absence of a clear model of HCV particle structure, inferences have been drawn by comparison to the related flaviviruses dengue virus, tick-borne encephalitis virus, West Nile virus and yellow fever virus, which are much better characterized than HCV at the structural level. Enveloped flavivirus particles are 50 nm in diameter and display an icosahedral scaffold of 90 antiparallel E glycoprotein homodimers, which lie flat against the viral surface33 (Figs 1c,e). During flavivirus entry, virus particles are internalized and trafficked to the endosome, where the low pH causes E protein dimers to dissociate and reform into trimeric spikes, inserting an internal fusion peptide into target cell membranes34. Remarkably, the fusion glycoproteins of alphaviruses (family Togaviridae), rubella virus (family Togaviridae) and Rift Valley fever virus (family Bunyaviridae) have similar structural folds to flavivirus E glycoproteins35,36,37 and are likely to undergo similar acid-induced rearrangement into their fusogenic trimeric forms, despite having distant evolutionary relatedness and an absence of sequence homology. Such structural and functional similarities have led to the classification of the flavivirus, alphavirus, rubella virus and Rift Valley fever virus glycoproteins as class II fusion proteins, distinct from the influenza virus, HIV and herpesvirus class I and class III viral fusion proteins, which exist as extendable, preformed trimeric complexes38,39.

On the basis of flavivirus and HCV relatedness, it has been suggested that HCV also uses a class II fusion mechanism40,41,42. However, this assumption has been challenged by recent studies43,44 suggesting that HCV and pestiviruses share an uncharacterized mechanism of membrane fusion. As mentioned above, flavivirus particles encode a single class II fusion glycoprotein that is primed for low-pH activation during viral egress45. By contrast, HCV and pestiviruses encode small E1 glycoproteins and large, immunodominant E2 glycoproteins that interact with host cell receptors46,47,48, and both HCV and pestiviruses require post-attachment priming steps before they can respond to low pH during viral entry27,49,50. The E2 glycoprotein of the pestivirus bovine viral diarrhoea virus 1 (BVDV-1) was recently shown by X-ray crystallography to exhibit a linear domain topology and to form tail-to-tail disulphide-linked homodimers43,44 (Fig. 1d), making it structurally dissimilar to known fusion proteins and indicating that it is unlikely to function as a class II fusion protein. One possibility is that pestivirus E1, rather than E2, is responsible for low-pH-induced fusion, although E1 is much smaller than canonical class II fusion proteins, indicating that it might use a unique mechanism of fusion. In this regard, it is notable that HCV E1 harbours a putative fusion peptide43,51,52. Clearly, structures of these viral glycoproteins in their pre- and post-fusion states are urgently needed to determine the mechanism by which HCV and pestiviruses mediate membrane fusion.

Virus plus lipoproteins makes lipoviral particles. The most striking feature of infectious HCV particles is their buoyant density, which is unusually low and heterogeneous for an enveloped RNA virus. Both serum-derived HCV and HCVcc particles with high specific infectivity (that is, low ratios of virus particles per infectious unit) have buoyant densities of ≤1.10 g per ml18,53, which is significantly lower than those of other enveloped RNA viruses (Table 1).

Table 1 Buoyant density of enveloped RNA viruses and serum lipoproteins
Full size table

The low buoyant density of HCV particles is due to their interaction with serum lipoproteins54,55 (Box 1). Consistent with this interaction, serum-derived HCV particles have been found to associate with the lipoprotein components apolipoprotein A-I (apoA-I), apoB-48, apoB-100, apoC-I and apoE54,56,57,58. Similarly, HCVcc particles interact with apoE and apoC-I, whereas interaction with apoB has been only variably detected31,32,59,60. Lipid profiling of highly purified HCVcc particles confirms that their lipid and cholesterol content is similar to that of serum lipoproteins, although similarity to a specific lipoprotein class could not be determined31. The interaction between virus particles and serum lipoproteins suggests that HCV forms hybrid 'lipoviral particles' (LVPs) that can facilitate virus entry into hepatocytes (see below) and protect the virus from antibody neutralization. However, the overall architecture of HCV particles and LVPs remains obscure. Key unresolved issues include how the HCV glycoproteins are arranged on infectious virus particles, what molecular interactions mediate HCV particle–lipoprotein associations, and whether LVPs represent either virus particles that are transiently interacting with separable serum lipoprotein particles (Fig. 1f), or hybrid particles that share a common envelope (Fig. 1g). In support of a two-particle model of transient, exchangeable interaction between HCV particles and serum lipoproteins, serum-derived HCV particles associate with apoB-48-containing chylomicrons58 (Box 1) and exhibit transient postprandial shifts in buoyant density57. Whichever model is correct, the production of HCVcc particles depends on cellular components of lipoprotein assembly and release (described below), suggesting that the interaction of virus particles with serum lipoproteins begins at an early step.

HCV entry

HCV circulates in the blood and thus has direct contact with the basolateral surface of hepatocytes, the cell type in which it replicates. This allows the virus to bind receptors on the surface of these cells and trigger virus entry. HCV is internalized via clathrin-mediated endocytosis. Clathrin is a cellular scaffold used for the budding of plasma membrane-bound vesicles, known as endosomes, into the cell cytoplasm. Following internalization, endosomes can be recycled back to the cell surface or progressively acidified and trafficked to lysosomes for ultimate degradation of their contents61. Many viruses use clathrin-mediated endocytosis for entry and typically require interaction with a small number of receptors to initiate endocytosis62. However, HCV entry poses an exception to this simplistic model of virus entry.

Not the usual suspects: HCV entry requires several factors. Initial attachment of HCV to cells occurs via low-affinity interaction with low-density-lipoprotein receptor (LDLR) and with glycosaminoglycans (GAGs) present on heparan sulphate proteoglycans (HSPGs), and both LDLR and GAGs can interact with virion-associated apoE63,64,65. LDLR is dispensable for HCV entry but might be needed for optimal replication of the virus within infected cells66. In addition to these attachment receptors, five cell surface molecules are essential for HCV particle entry: CD81, scavenger receptor class B member 1 (SRB1; also known as SRBI or SCARB1), claudin 1 (CLDN1), occludin (OCLN) and the cholesterol absorption receptor Niemann–Pick C1-like 1 (NPC1L1). As described below, defined subsets of these molecules determine both the hepatocellular tropism and host species tropism of HCV entry.

CD81 was the first HCV co-receptor to be identified, via expression cloning of a human cDNA that could confer binding of HCV E2 to a mouse fibroblast cell line46. CD81 is a ubiquitously expressed cell surface tetraspanin that directly binds HCV E2. Although the two proteins bind tightly67, time course studies with CD81-specific antibodies, which block HCV infection, indicate that CD81 mediates a post-attachment event in HCV entry68. Consistent with this, the CD81-binding region of E2 is masked by its hypervariable region 1 (HVR1), suggesting that native E2 needs to undergo a conformational change before this co-receptor is engaged69. Treatment of virus particles with soluble CD81 renders them sensitive to acid, indicating that interaction with CD81 might help to prime the HCV glycoproteins for low-pH activation during virus entry70. In addition, the function of CD81 in HCV entry requires interaction with CLDN1 (Ref. 71) (see below). Furthermore, CD81 helps to define the tropism of HCV for human cells72,73; specifically, HCV is capable of infecting mice that express human CD81 and OCLN in the liver72,73.

SRB1 was identified as a candidate co-receptor by its ability to bind recombinant HCV E2 on HepG2 cells, which lack CD81 expression48. SRB1 is highly expressed on hepatocytes, where it normally has important physiological functions in binding high-density lipoproteins (HDLs), very-low-density lipoproteins (VLDLs) and oxidized forms of LDLs, and in promoting the uptake of cholesterol from these lipoproteins74. Because other HCV entry factors are expressed by multiple cell types, the high expression level of SRB1 in the liver might help to define the hepatotropism of HCV entry. During viral entry, SRB1 seems to have multiple stepwise roles. First, SRB1 contributes to virus attachment through interaction with virus-associated lipoproteins. Second, the lipid transfer activity of SRB1 mediates a post-binding event that is important for productive virus entry75,76. Third, subsequent to lipid transfer, SRB1 interacts with the E2 HVR1 region48,75, and this exposes determinants in E2 that allow it to bind CD81 (Ref. 69).

CLDN1 was identified as an essential HCV entry factor through expression cloning of a cDNA that could confer HCVpp entry to the human kidney cell line, HEK293, which expresses CD81 and SRB1 (Ref. 77). CLDN1 is a multipass plasma membrane protein that localizes to tight junctions and, in lower levels, to the basolateral surface of hepatocytes and other polarized cells78. Although CLDN1 does not directly interact with the HCV glycoproteins, it contributes to the post-binding steps of HCV entry by interacting with CD81, which facilitates virus internalization71,77.

A similar cDNA expression cloning strategy was used to identify OCLN as a factor that could confer HCVpp entry to mouse cells engineered to express human CD81, SRB1 and CLDN1 (Ref. 79). Together with CD81, OCLN defines the species tropism of HCV entry; mice engineered to express human CD81 and OCLN specifically in the liver are permissive to HCV infection72,73. Similarly to CLDN1, OCLN is a tight junction protein that functions at a post-attachment step in HCV entry, although it is unclear whether OCLN directly interacts with HCV particles80. The precise role of OCLN in virus entry is still under investigation.

On the basis of the clinical observation that HCV patients often accumulate iron in the liver, one group found that the iron uptake receptor, transferrin receptor 1 (TFR1), acts as a cellular entry factor81. HCV infection is inhibited by TFR1 knockdown, TFR1-specific antibodies and a small molecule inhibitor of TFR1 recycling; time-of-addition studies with these inhibitors showed that TFR1 is important at a post-CD81-binding event81. However, the mechanism by which TFR1 contributes to HCV entry remains to be determined.

Finally, NPC1L1 was identified as an HCV entry factor. Ezetimibe, a specific inhibitor of NPC1L1, blocks HCV entry in cell culture and in mice bearing human liver grafts, and knockdown of NPC1L1 inhibits HCV entry in cell culture82. NPC1L1 localizes to the apical, canalicular surface of polarized hepatocytes, where it functions in the re-uptake of cholesterol from bile83 and can be internalized to endosomal compartments. The specific role of NPC1L1 in HCV entry is unknown, but it probably involves cholesterol uptake. Given that HCV enters hepatocytes from the basolateral side, NPC1L1 might mediate these effects within endosomes.

Hail to the thief: signals announcing HCV entry. Recent studies show that the CD81–CLDN1 axis of HCV entry depends on the integrity of specific signal transduction pathways. Inhibitors of protein kinase A (PKA) have been shown to disrupt the interaction between CD81 and CLDN1, leading to CLDN1 internalization from the cell surface and inhibition of HCV entry84, highlighting a key role for PKA in this process. Furthermore, HCV E2–CD81 engagement activates the RHO GTPase family members RAC1, RHOA and CDC42, which modify cortical actin filaments and allow lateral mobility of HCV–CD81 complexes to sites of cell–cell contact85. Thus, RHO GTPases might be necessary for HCV-bound CD81 to interact with the tight junction protein CLDN1 before virus internalization. Consistent with this, an inhibitor of RHOA was found to inhibit the endocytosis of CD81–CLDN1 complexes86.

In addition to the PKA and RHO signalling pathways, two receptor tyrosine kinases (RTKs) — epidermal growth factor receptor (EGFR) and ephrin type A receptor 2 (EPHA2) — were identified in a human kinome-specific siRNA library screen as host factors required for HCV entry87. Similar to PKA and RHO signalling, EGFR and EPHA2 signalling promotes CD81–CLDN1 interaction87. Moreover, HCV entry is blocked by the anticancer compounds erlotinib and dasatinib, which inhibit EGFR and EPHA2, respectively, as well as by transient or stable knockdown of EGFR or EPHA2 or by antibodies that block ligand–receptor interactions; by contrast, HCV entry is enhanced by the addition of EGFR and EPHA2 ligands87.

A follow-up study to clarify the role of EGFR in HCV entry found that binding of HCVcc particles to cells promotes the colocalization of CD81 with EGFR, which activates EGFR signalling, and that EGFR is required at a step between CD81–CLDN1 engagement and clathrin-mediated endocytosis88. HRAS, a membrane-bound GTPase that regulates the timing and intensity of multiple signal transduction pathways, including EGFR signalling, acts as a major integrator of EGFR-mediated HCV entry89. Indeed, the anticancer compound tipifarnib, which prevents the membrane localization of RAS proteins, was found to potently inhibit HCV entry89. Moreover, HRAS was found to associate with CD81 and promote its lateral diffusion into stable CD81–CLDN1 complexes89. Thus, it seems that EGFR and RAS proteins contribute to the functional association of HCV co-receptors, thereby playing a key part in HCV uptake. In addition to its role in HCV entry, EGFR signalling might render cells more permissive to viral infection by antagonizing the antiviral response of type I interferon90.

An integrated model of HCV entry. Inherent in the above discussion is that HCV entry uses numerous cellular entry factors in a temporally and spatially ordered manner (Fig. 2). Lipoprotein-associated HCV particles attach to the surface of hepatocytes by interacting with GAGs, LDLR and SRB1 (Refs 63, 64, 65, 66, 75). The cholesterol transfer activity of SRB1 might then serve to dissociate virus particles from their associated lipoproteins, and the interaction with SRB1 exposes the CD81-binding determinants on the HCV E2 glycoprotein48,69,75,76. A key step in HCV entry is the lateral movement of CD81-bound HCV particles to tight junctions and their interaction with CLDN1. This cell surface trafficking is dependent on several signal transduction pathways, including EGFR and downstream RAS GTPase signalling, as well as RHO GTPases, which remodel cortical actin85,86,87,88,89. The interaction of HCV–CD81 with CLDN1 then induces clathrin- mediated endocytosis86. Although the tight junction protein OCLN is also essential for HCV entry, its precise role in this process is currently unknown79,80.

Figure 2: HCV entry.
figure 2

Hepatitis C virus (HCV) lipoviral particles (LVPs) attach to the cell surface by interacting with heparan sulphate proteoglycans (HSPGs), low-density-lipoprotein receptor (LDLR) and scavenger receptor class B member 1 (SRB1). SRB1 might delipidate HCV-associated lipoproteins and induces conformational changes in the E2 glycoprotein, exposing the CD81-binding site (step 1). Transferrin receptor 1 (TFR1) has an unknown, post-CD81 role in HCV entry (not shown). Interaction of E2 with CD81 then activates signal transduction through epidermal growth factor receptor (EGFR) and HRAS, as well as through RHO GTPases (step 2). These signalling events promote lateral movement of HCV– CD81 complexes to sites of cell–cell contact (step 3), interaction of CD81 with claudin 1 (CLDN1), and HCV internalization via clathrin-mediated endocytosis (step 4). The low pH of the endosomal compartment induces HCV fusion (step 5). NPC1L1, Niemann–Pick C1-like 1; OCLN, occludin.

PowerPoint slide

Full size image

Following clathrin-mediated endocytosis, HCV– co-receptor complexes are trafficked to RAB5A- containing endosomal compartments86,91, where the lipid transfer activities of SRB1 (Ref. 75) and NPC1L1 (Ref. 83) might further modify the virus and its associated lipoproteins. The interaction of E2 with CD81 might then prime HCV glycoproteins to respond to the low pH in the endocytic compartment and induce fusion between the viral envelope and the bounding endosomal membrane70. Following fusion, the HCV genome is presumably released into the cytosol, where it is directly translated to produce viral proteins and initiate viral replication7.

Given the role of tight junctions in HCV entry, it should be noted that the above model is largely based on experiments carried out in cultured Huh-7 cells, which do not faithfully mimic the architecture of polarized hepatocytes. Moreover, cell-to-cell transmission of HCV (as opposed to binding and entry of blood-borne, cell-free virus) might occur in infected liver tissue and confer resistance to antibody-mediated neutralization92,93. Interestingly, compared with entry of cell-free virus, cell-to-cell transmission of HCV might be less dependent on CD81, SRB1 and TFR1 expression, although this is still controversial81,94,95,96,97. Ultimately, a comprehensive model of HCV entry will need to integrate information obtained from primary human hepatocytes in culture and in vivo.

Virus assembly and release

In the later steps of the viral life cycle, following RNA replication, nascent HCV particles form by budding into the ER, a process that must bring together the viral core protein, E1, E2 and the viral genome to be packaged in a temporally and spatially organized manner (Fig. 3). Nascent virus particles transit through the secretory pathway, where they undergo maturation and acquire their low density before exocytosis at the cell surface98,99. Several recent reports have revealed the role of host factors and the HCV NS proteins in coordinating these processes. Below, we briefly summarize the role of these viral and host factors, with an emphasis on recent discoveries.

Figure 3: HCV assembly and secretion.
figure 3

a | Early events in the hepatitis C virus (HCV) life cycle, including viral gene expression, recruitment of the viral RNA into an RNA replication complex, ER retention of the E1–E2 glycoprotein and p7–NS2 (non-structural protein 2) complexes, and cytosolic phospholipase A2 (cPLA2)-mediated trafficking of core dimers to cytosolic lipid droplets (cLDs), which are formed by the action of diacylglycerol O-acetyltransferase 1 (DGAT1). b | Late events in the HCV life cycle. Viral RNA is shifted out of replication and translation, and towards virus assembly. The interaction of p7–NS2 with NS3-4A recruits viral core protein to the site of virus assembly. Virus particles assemble by recruitment of E1–E2 complexes and budding into the ER. c | The pathways of very-low-density lipoprotein (VLDL) and HCV particle secretion, including maturation into low-buoyant-density particles (perhaps through specific lipidation steps and/or interaction with microsome-associated LDs) and the secretion of particles through the p7-buffered secretory pathway in HCV-infected cells.

PowerPoint slide

Full size image

Trafficking of the viral core protein. The mature core protein contains a positively charged N-terminal RNA-binding domain of unknown structure, and a C-terminal domain that facilitates peripheral membrane binding via two amphipathic helices and a palmitoylated cysteine residue100,101. Following synthesis on the ER, the core protein homodimerizes and is trafficked to cytosolic lipid droplets (cLDs), which are cellular lipid storage organelles102,103,104 (Fig. 3a). cLDs consist of a phospholipid monolayer, derived from the outer leaflet of the ER, surrounding a hydrophobic core of neutral lipids and cholesterol esters. The purpose of core cLD trafficking is unclear; perhaps it serves to sequester the core protein until it is needed for virus assembly. Consistent with this model, mutations that prevent trafficking of the core protein to cLDs strongly inhibit virus assembly105,106,107.

Trafficking of the core protein to cLDs requires the MAPK (mitogen-activated protein kinase)-regulated protein cytosolic phospholipase A2 (cPLA2)108. Moreover, a pharmacological inhibitor of cPLA2 potently blocks virus assembly, and this inhibition can be overcome by the addition of arachidonic acid, the product of cPLA2 (Ref. 108). As arachidonic acid can act as a secondary messenger of membrane-bound signalling, one interesting hypothesis is that a signalling pathway downstream of cPLA2 regulates core trafficking. Trafficking of the core protein to cLDs is also supported by diacylglycerol O-acetyltransferase 1 (DGAT1) (Fig. 3a), one of two DGAT enzymes required to synthesize the triglycerides that are stored within cLDs109. Indeed, inhibiting DGAT1 activity via genetic knockdown or a small-molecule inhibitor causes partial decreases in core trafficking and virus production109.

During virus particle assembly, the core protein must be retrieved from the surface of cLDs to the site of viral budding at the ER, and specific host factors have been shown to be involved in this retrieval process. For example, the core protein contains a conserved YXXø motif (in which X corresponds to any amino acid and ø corresponds to a hydrophobic amino acid) that interacts with AP2M1 (clathrin assembly protein complex 2 medium chain), which was shown to be essential for retrieval of core protein from cLDs and for virus assembly110. Consistent with these observations, the anticancer compounds erlotinib and sunitinib, which can inhibit RTKs that regulate AP2M1 activation, were reported to be potent inhibitors of HCV assembly110. However, other groups found that erlotinib inhibits HCV entry but not assembly87,88. Given that a key function of AP2M1 is to select cargo for clathrin-mediated endocytosis at the cell surface, its specific role in HCV assembly is not yet clear.

Recently, two groups have imaged live, virus-producing cells to obtain further insights into the trafficking of the viral core protein during virus assembly and release. One group developed an HCVcc variant encoding a core gene fused to a tetracysteine tag, allowing fluorescent labelling and imaging of functional core protein in live, virus- producing cells111. This study showed that the core protein is rapidly trafficked to cLDs and slowly recruited from cLDs into motile puncta that traffic on microtubules and probably represent virus particles within the secretory pathway; interaction between viral NS2 and NS3-4A is essential for this recruitment process111. By combining this imaging method with genetic and imaging analysis of host secretory components, another group provided further evidence that motile core puncta represent virus particles undergoing secretion112.

Synthesis and trafficking of the E1–E2 glycoprotein complex. Following their synthesis, E1 and E2 form ER-retained, non-covalent heterodimers113 (Fig. 3a). Folding of the HCV glycoproteins is interdependent, and the formation of native heterodimers is a slow process114,115. However, the E2 ectodomain can fold on its own into a native structure that binds cellular receptors and is recognized by conformation-specific antibodies116,117,118. Biochemical and biophysical characterization of the recombinant, folded E2 ectodomain has defined nine intramolecular disulphide bonds and has shown that E2 contains three β-sheet-rich domains separated by regions of random coils and β-turns41,119. Little is known about the structure of the E1 ectodomain, as it does not fold properly in the absence of E2. Further structural definition of the HCV E2 and E1–E2 ectodomains is eagerly awaited.

NS proteins as drivers of HCV assembly and egress. As mentioned above, NS proteins — p7, NS2, NS3, NS4A, NS5A and NS5B — have integral roles during virus assembly and egress, although they are not themselves packaged into virus particles. One of the first key steps in HCV assembly is the interaction of NS5A with the cLD-bound core protein106,120,121 (Fig. 3b), which is enhanced by DGAT1 (Ref. 122). NS5A is a homodimeric RNA-binding phosphoprotein that is important for RNA replication and virus assembly. It contains an N-terminal amphipathic helix that anchors it to membranes, followed by a structured Zn2+-binding domain that mediates homodimerization, and two long, flexible domains that are natively unfolded. NS5A has essential roles in virus particle assembly via determinants in its C-terminal unfolded domain120,123,124. Specifically, phosphorylation of a specific serine residue within this region by casein kinase IIα (CKIIα) is essential for virus assembly124. Interestingly, substitution of this serine residue with a phosphomimetic acidic residue renders HCV assembly insensitive to CKII inhibitors or genetic knockdown of CKIIα124. NS5A can also interact with apoE125,126, although the membrane topology of this interaction remains obscure, as well as with annexin A2, a trafficking protein that enhances virus assembly127.

Two additional NS proteins, p7 and NS2, have key roles in organizing the virus assembly complex128,129,130. NS2 is a polytopic membrane protein containing three N-terminal transmembrane domains (TMDs) and a C-terminal cysteine protease domain that causes NS2 to homodimerize131. Both the TMDs and protease domain of NS2 are required for the production of virus particles; although the cysteine protease activity is required for viral gene expression and replication, it is dispensable for virus assembly when decoupled from viral gene expression132,133. NS2 interacts with the small, membrane-bound NS protein p7, and this interaction is required to localize NS proteins and the core-containing cLDs to putative sites of virus assembly128,129.

NS2 (or the p7–NS2 complex) interacts with the NS3-4A enzyme, and this is a key step in the retrieval of the viral core protein from cLDs and into nascent virus particles. As mentioned above, NS3 contains an N-terminal serine protease domain that is required for viral gene expression and a C-terminal RNA helicase domain that is essential for viral genome replication5. NS3 interacts with NS4A, which anchors the complex onto membranes and functions as a cofactor for both the serine protease and RNA helicase activities of the NS3 enzyme5,134,135. Indeed, mutations in the C terminus of NS4A, which modulates NS3 helicase activity, cause profound defects in virus assembly134,136. Interestingly, mutations in the NS3 helicase domain that enhance RNA replication cause defects in virus assembly, implicating NS3 in virus assembly and suggesting that these two processes must be coordinated137. Furthermore, mutations in the NS3 helicase domain can act as genetic suppressors, overcoming virus assembly defects caused by genetic lesions in NS2, NS3, NS4A or the core protein128,138,139,140. One intriguing possibility is that the NS3-4A RNA helicase activity serves to package viral RNA during nucleocapsid formation.

Viral budding – a potential role for ESCRTs. The ESCRT (endosomal sorting complex required for transport) pathway is involved in budding and fission of membranous compartments that curve away from the cytosol and are used by many enveloped viruses to bud into extracellular compartments141. Three groups found that HCVcc particle release depends on components of the ESCRT pathway142,143,144, although surprisingly, the assembly of intracellular infectious virus particles was independent of the late ESCRT pathway that mediates membrane fission. One possibility is that ESCRTs are in fact involved in resolving the terminal membrane fission event during virus particle budding at the ER, but that incompletely budded virus particles were released during experimental preparation of cell extracts. Alternatively, the ESCRT pathway could be required for a post-assembly step during virus egress, such as trafficking into an intermediate secretory compartment. Consistent with this, HCV particles have been observed to traffic to recycling endosomes112,145.

HCV particle maturation. As the HCV particles transit through the secretory pathway, E1 and E2 are post- translationally modified. On purified HCVcc particles, E1 and E2 contain both high-mannose and complex N-linked glycans, indicating that virus particles transit through the Golgi50, where glycan modification is known to occur. Furthermore, virion-associated E1–E2 dimers are found in large, natively folded, disulphide-linked complexes50, which might contribute to the acid resistance of HCV particles27 and suggests that rearrangement of these disulphide bonds is necessary to prime HCV particles for low-pH-mediated fusion, similar to pestiviruses49.

In addition to glycoprotein modifications, HCV particles interact with serum lipoproteins as they transit through the secretory pathway (Fig. 3c). As mentioned above, Huh-7 cells produce VLDL-like particles that are under-lipidated, which might explain why HCVcc particles produced in these cells have a higher buoyant density and lower specific infectivity than HCV produced in bona fide hepatocytes23,24. Despite this caveat, several lines of evidence support a link between HCV maturation and lipoprotein secretion. First, nascent HCVcc particles were found to have an intermediate buoyant density and to acquire their low buoyant density during egress, much like VLDL particles98,99. Furthermore, HCV particle production is blocked by small-molecule inhibitors of microsomal triglyceride transfer protein (MTP), which is involved in transporting lipids to the ER lumen98,146,147,148 (Box 1). Although this was initially interpreted to mean that apoB expression is required for HCV assembly98,146,148, careful titration of an MTP inhibitor showed that HCV assembly and release depend on the small, exchangeable apolipoprotein apoE, but occur independently of apoB147. Similarly, another study found that nascent virus particles traffic in the secretory pathway with apoE, but not with apoB112. These findings are intriguing, as apoB is essential for VLDL assembly, suggesting that HCV assembly actually depends on apoE-containing microsome-associated LDs rather than on VLDL particles. Consistent with the essential role of apoE in HCV production, various cell lines were shown to produce infectious HCVcc particles in an apoE-dependent and apoB-independent manner149,150.

In addition to interacting with NS2 during virus assembly, p7 acts as a viroporin, oligomerizing to form hexameric and heptameric cation-specific ion channels that equilibrate pH gradients within the secretory pathway and thereby stabilize virus particles during maturation and egress151,152,153,154,155. Defects in p7 ion channel activity can be complemented by expressing the influenza virus viroporin protein M2 or by inhibiting the vacuolar-type proton ATPase with bafilomycin A1 (Ref. 154), arguing that p7 is a bona fide viroporin. NS2 also has an unknown role in viral egress, perhaps through its interaction with p7 (Ref. 156).

An integrated model of HCV particle assembly and release. HCV particles assemble by budding into the ER (Fig. 3a,b). An important early step in this process is the interaction between cLD-associated core and NS5A proteins, which might serve to shift RNA out of replication or translation and into virus assembly106,120,121. Similarly, p7–NS2 brings together the viral E1–E2 and NS3-4A complexes through the NS2 TMDs55,139. The interaction between p7–NS2 and NS3-4A recruits the viral core protein from cLDs into sites of virus assembly111, perhaps solely through protein–protein interaction or perhaps, intriguingly, by facilitating helicase-dependent RNA packaging and nucleocapsid assembly in concert with budding. Virus particles might complete the budding process independently of the ESCRT pathway, although this pathway does seem to be involved in virus release142,143,144.

Virus particles are released from cells by transiting through the secretory pathway (Fig. 3c). During this process, HCV particles acquire their characteristic low buoyant density by interacting with apoE-containing VLDL or HDL particles98,99. Furthermore, the viral surface glycoproteins are modified to contain complex glycans, and their disulphide bonds are reorganized50. As virus particles transit through the secretory pathway, they are protected from exposure to low pH by p7, which neutralizes intracellular compartments154.

Perspectives

The past several years have seen enormous growth in our understanding of key steps in the HCV life cycle. We now know that HCV virus particles exhibit a defined composition despite their pleiomorphic nature. Several cell entry factors have been identified, including signal transduction pathways that are important for trafficking of the virus to sites of endocytosis. Furthermore, we have learnt that virus particles assemble at the cLD–ER interface and require the concerted action of cellular enzymes and viral NS proteins, and then undergo maturation as they are trafficked through the secretory pathway.

Nevertheless, there are a number of key unanswered questions. We need to resolve the structure of the viral glycoproteins, understand how HCV particles are organized and define the molecular basis for the interaction between virus particles and serum lipoproteins. Although numerous HCV co-receptors and entry factors have been defined, understanding how they are used temporally and spatially remains a priority, particularly in cells that model the architecture of hepatocytes.

On the virus assembly front, we need a clearer picture of how nucleocapsids are assembled, how NS proteins contribute to this process and the relationship of virus particle assembly to lipoprotein secretion. Ultimately, answers to these questions must integrate our findings in the available cell culture systems with HCV biology in real hepatocytes and the liver.