Naked Viruses that Aren’t Always Naked: Quasi-enveloped Agents of Acute Hepatitis

Abstract

Historically, viruses are considered to be either enveloped or nonenveloped. However, recent work on hepatitis A virus and hepatitis E virus challenges this longheld tenet. While these human pathogens are shed in feces as naked nonenveloped virions, recent studies indicate both circulate in the blood completely enveloped in membranes during acute infection. These membrane-wrapped virions are as infectious as their naked counterparts, although neither expresses a virally-encoded protein on its surface, thus distinguishing them from conventional enveloped viruses. The absence of a fusion protein indicates that these “quasi-enveloped” virions have unique mechanisms for entry into cells. Like other enveloped viruses, however, these phylogenetically distinct viruses usurp components of the host ESCRT system to hijack host cell membranes and noncytolytically exit infected cells. The membrane protects these viruses from neutralizing antibodies, facilitating dissemination within the host, while nonenveloped virions shed in feces are stable in the environment, allowing for epidemic transmission.

INTRODUCTION

Almost a century has passed since it was recognized that rabies virus is made noninfectious by exposure to ether while poliovirus is resistant to similar treatment (1, 2). By the 1940s, these early observations had led to systematic studies that sought to classify pathogenic human viruses based on their sensitivity to ether inactivation (3, 4). Andrewes and Hostmann (4) recognized that those viruses that were sensitive to ether were also typically inactivated by bile salts (5), and proposed that ether sensitivity reflected the presence of lipid on the surface of the virus. Thus, prior to the much more recent availability of sequence phylogeny, the presence or absence of a lipid-containing envelope served as a basic attribute for classifying animal viruses. While many virologists have an ingrained notion that viruses are either enveloped or nonenveloped, current studies with two unrelated RNA viruses, hepatitis A virus (HAV) and hepatitis E virus (HEV), challenge that long-held tenet. This review focuses on what is known of the double life these viruses lead and how it contributes to the pathogenicity and survival of these human pathogens.

ON BEING ENVELOPED: ADVANTAGES AND DISADVANTAGES

Viral envelopes are closed lipid bilayers that comprise the outer surface of virions. In addition to providing a protective barrier, they play key roles in mediating the entry and exit of viral genomes from cells. The envelope is typically formed by budding from preformed cellular membranes, and generally contains one or more virus-encoded glycoproteins (peplomers) embedded within it (Figure 1). These peplomers face in an outward direction and mediate interactions with specific sets of cellular receptors. Receptor interactions define which cells become infected by a virus and thus contribute to tissue tropism. Proteins expressed on the surface of the envelope also facilitate the fusion of cellular and viral membranes following initial receptor interactions, resulting in the release of virion contents into the interior of the cell. An envelope may also prevent internal antigens from being recognized by neutralizing antibodies (6). Highly glycosylated envelope proteins, such as those present in hepatitis C virus (HCV) or human immunodeficiency virus (HIV), may also provide a ‘glycan shield’ that confounds neutralizing antibody responses (7). However, the most important function of the envelope is its facilitation of genome trafficking into and especially out of cells. The plasma membrane is a barrier that is not easily traversed by nonenveloped viruses in the absence of cell lysis.

Figure 1.

Distinguishing features of (a) naked, (b) “quasi-enveloped” and (c) conventional enveloped viruses. Quasi-enveloped viruses such as eHAV and eHEV are distinguished from other enveloped viruses by the lack of viral glycoproteins (peplomers) in the surrounding lipid bilayer, but may have unique internal proteins, such as VP1pX in the case of eHAV, that are not found in naked virions. The ORF3 protein of HEV is only found in virus derived from serum and is not present in virus shed in feces (not shown). Neither the pX extension of VP1 nor ORF3 are exposed on the surface of the quasi-enveloped particle.

On the other hand, having an envelope clearly limits the capacity of a virus to survive under harsh environmental conditions, including drying, or exposure to solvents and detergents. Thus, the transmission of enveloped viruses typically results from relatively close contact between infected and naïve hosts, and often requires inhalation of moist aerosols or exchange of secretions. Nonenveloped viruses have a clear advantage in terms of their capacity to spread through the environment. Food- and waterborne outbreaks of hepatitis A and hepatitis E provide good examples of this (8, 9), and show how stable, naked protein capsids can protect an otherwise labile RNA genome through both time and distance in the environment.

Recent studies suggest that HAV and HEV benefit from the best of these two worlds. Both of these viruses are hepatotropic and cause enterically-transmitted acute hepatitis in humans. They are both RNA viruses, but have very different genome architectures and are classified in different virus families. Both HAV and HEV are shed in the feces of infected individuals as stable, nonenveloped virions with their genome encapsidated in a naked protein shell (10, 11). However, HAV circulates during acute infection in a membrane-wrapped form in which the capsid is completely enveloped and sequestered from neutralizing antibodies, yet remains fully infectious (12). Circulating HEV appears to be similarly cloaked in membranes (13). These membrane-wrapped forms of HAV and HEV differ from ‘classical’ enveloped viruses in that the surrounding lipid bilayer appears to be devoid of virally-encoded proteins (Figure 1b). This allows extracellular virus to masquerade as host-derived exocytic vesicles, and is likely to facilitate dissemination of the virus within the host. However, it raises substantial questions about how cell entry occurs in the absence of virally-encoded receptor-binding and membrane fusion activities on the surface of the virion. Because of the absence of peplomers, these membrane-wrapped virions may be best considered to be ‘quasi-enveloped’ (Figure 1). For simplicity, we refer to them here simply as ‘enveloped’ HAV (eHAV) and HEV (eHEV).

THE ESCRT SYSTEM AND BIOGENESIS OF “CLASSICAL” ENVELOPED VIRUSES

To understand how these quasi-enveloped viruses become wrapped in membranes, it is useful to briefly review how more typical enveloped viruses acquire their membranes. The biogenesis of envelopes involves an energy-dependent process of budding at the plasma membrane, or in some cases at an internal cellular membrane such as membranes of the ER or Golgi (14, 15). The site of budding is linked to the secretory pathway of viral glycoprotein(s) as well as the trafficking of internal structural proteins that scaffold the envelope and in some cases assemble into an inner capsid. The phospholipid composition of the membrane also determines the precise location of budding, which typically involves microdomains within the membrane known as ‘lipid rafts’ that are enriched in sphingomyelin and cholesterol. These lipid rafts are more rigid than surrounding membranes, generating tension at their borders with the more liquid, disordered nonraft membrane (14). Virus particles budding at these regions possess envelope membranes that are similarly enriched in sphingomyelin and cholesterol. In addition, phosphatidylserine (PS), which is generally more abundant in the inner leaflet of the plasma membrane, is present on the surface of many enveloped viruses, including Ebola virus and dengue virus (16, 17).

Budding initially involves deformation and bending of the membrane, followed by formation of a neck-like structure and finally scission of the neck with resealing of the membrane on both sides (14, 15). Imparting a bend to the membrane requires an expenditure of energy, and with most enveloped viruses this derives from scaffolding proteins encoded by the virus that traffic to and interact with the cytoplasmic side of the membrane. Multimerization of these scaffolding proteins results in an outward bud in the membrane, away from the cytoplasm. The orientation of this nascent bud in the plasmid membrane is opposite to that occurring in endocytosis, and is analogous to the budding that occurs into endosomal membranes during the formation of multivesicular bodies (MVBs) (15). Thus, it is no surprise that many enveloped viruses usurp cellular machinery involved in MVB formation, the ESCRT (endosomal sorting complex required for transport) apparatus, to facilitate their budding and egress from cells (15, 18). While the actual process of membrane scission results directly from biophysical properties of the membrane when the neck of the bud reaches a threshold minimum diameter (14), ESCRT components facilitate the extreme membrane deformation required to trigger scission.

The ESCRT apparatus contributes to multiple aspects of cellular physiology, including cytokinesis and autophagy, but plays a primary role in MVB formation (19, 20). It is comprised of four distinct protein complexes that act sequentially to sort and load cargo into MVBs (19, 21). ESCRT-0, -I, and -II contain multiple ubiquitin-binding domains that function to identify ubiquitin-conjugated cargo. These are typically membrane-bound proteins but also cytosolic proteins (19). ESCRT-I and -II initiate the process of sequestering this cargo within MVBs by deforming the endosomal membrane and creating a bud into the endosome. ESCRT-III assembles into oligomers that pinch off the neck of the invagination, and recruits deubiquitinases to release the cargo. In a final step, VPS4 acts to dissociate ESCRT-III oligomers, allowing the system to recycle. Cargo sequestered within MVBs generally is destined for transport to lysosomes where it is degraded. However, MVBs can also be directed to the cell surface, where fusion of their outer membrane with the plasma membrane results in the release of exosomes: cargo-laden, single-membrane vesicles (22, 23).

HIV remains the best studied example of a virus that usurps the ESCRT apparatus, co-opting ESCRT components to bud through the plasma membrane (15, 24). Like many other enveloped viruses, HIV co-opts ESCRT machinery by expressing structural proteins with “late domain” (L domain) motifs that mediate interactions with specific components of the ESCRT system (25, 26). Several different L domain motifs have been identified. The HIV Gag-p6 protein contains a P(S/T)AP motif mediating interactions with TSG101, an early ESCRT complex-associated protein involved in sorting of cargo for loading into MVBs, as well as a YPX_(1or3)L motif that mediates interactions with ALIX (official name PDCD6IP, programmed cell death 6 interacting protein), a protein that bridges ESCRT-I and III complexes. The presence of multiple L domains in Gag-p6 likely provides functional redundancy in the process of budding and cellular egress. Multiple additional examples exist for L domain interactions with ESCRT-associated proteins in retroviruses, filoviruses, rhabdoviruses, arenaviruses, and other enveloped viruses (25, 26). While human cytomegalovirus (HCMV) exploits ESCRT proteins for envelopment at or near late endosomes, where they are normally engaged in MVB formation, others such as HIV co-opt these proteins at the plasma membrane where they also function in membrane scission required for cell division (27–29). Nonetheless, despite extensive use of the ESCRT apparatus in the budding and release of enveloped viruses, there are some, such as influenza virus, that appear to gain their release from the cell independently of ESCRT (14).

QUASI-ENVELOPED HEPATOVIRUSES: eHAV

eHAV, a picornavirus wrapped in membranes.

HAV is an ancient human pathogen and a common cause of enterically transmitted hepatitis (30). It is a positive-strand RNA virus, classified in the genus Hepatovirus of the family Picornaviridae, with a 7.5 kb genome containing a single large open reading frame like that of other picornaviruses (Figure 2A). Its capsid is comprised of 60 copies of each of 3 major proteins, VP1, 2, and 3 (or 1D, 1B, and 1C), and, possibly, a smaller polypeptide, VP4 (1A) (31, 32). While high-resolution X-ray structures are available for several pathogenic picornaviruses, none have yet been published for HAV. The capsid structure thus remains poorly defined, although previous studies suggest that its assembly differs significantly from other, better-studied picornaviruses for which the assembly of pentamer subunits is dependent upon N-terminal myristoylation of VP4 (33, 34). While HAV possesses a vestigial VP4 sequence, it does not contain a terminal myristoylation signal. Pentamer assembly is instead driven by sequence within an unusual 8 kDa C-terminal extension of VP1 known as pX (35–37) (Figure 2A). Previous studies indicate that pX is cleaved from VP1 at a late step in viral morphogenesis by a host proteinase, possibly cathepsin L (38–40). These features of the HAV capsid distinguish it from other well-studied picornaviruses. Similarly, the amino acid sequence of pX is unique, and does not align with other protein sequences in GenBank. A variety of cultured cell types, mostly of primate origin, are permissive for HAV replication, but efficient replication requires adaptive mutations in 2B, a large nonstructural protein, and within the internal ribosome entry site (IRES) in the 5’ untranslated RNA (5’UTR) (41–43). Even then, replication is slow and without shutoff of host macromolecular synthesis, and thus generally noncytopathic. As is the case with many positive-strand RNA viruses, RNA replication appears to occur within cytoplasmic vesicles, which are induced in part by 2B expression (44). Highly cell culture-adapted viruses induce apoptotic cell death (44, 45).

Figure 2.

Enveloped and nonenveloped hepatoviruses. (a) Genome organization of HAV. The 7.5 kb positive-sense RNA genome encodes a large polyprotein which is cleaved into 9 mature proteins. Two putative late domains in VP2 and three potential ubiquitylation sites within pX are highlighted. (b) Electron microscopic images of enveloped eHAV and nonenveloped HAV. (c) Immunoblots of gradient-purified eHAV and HAV using antibodies against VP1 and VP2. Note that most VP1pX in eHAV is not cleaved, whereas this has been fully processed to VP1 in HAV is naked particles. Panels b and c reproduced from Feng et al., Nature (2013) 496:367–71.

HAV infections are associated with a minimal intrahepatic type I interferon response (46). While studying host innate immune sensing of the virus, we observed that Huh-7 hepatoma cells infected with a low-passage, noncytolytic HAV variant, HM175/p16 (47), release two types of virions that band at distinct densities in isopycnic iodixanol gradients (12). Transmission electron microscopy (EM) revealed that the denser particles (~1.22 g/cm³) had the typical appearance of nonenveloped picornaviral virions with respect to both size (~27 nm diameter) and morphology (spherical), while the lighter particles (~1.08 g/cm³) were small, single membrane vesicles containing 1–4 (more typically 1 or 2) similar appearing capsids (Figure 2B). Remarkably, HAV capsid antigen could not be detected in fractions containing these light particles in the absence of prior detergent treatment, indicating that the capsid was completely occluded by membranes (12). Despite this, the light particles were infectious with a specific infectivity (genome equivalents/unit of infectious virus) nearly identical to that of the nonenveloped, denser particles. This infectivity was reduced approximately 100-fold by extraction with an equal volume of chloroform, which had no effect on the infectivity of the denser particle (12). While initially puzzling, given the resistance of the denser virions to chloroform, we determined that the loss of infectivity was due to partitioning of the membrane-associated virus into the organic phase from which it could not be back-extracted. The variable number of capsids present in each of the membrane-associated virions suggested they had become enveloped in membranes after their assembly. Consistent with their complete envelopment in membranes, these low-density enveloped HAV (‘eHAV’) particles were highly resistant to neutralizing anti-capsid antibodies in a quantal, infectious focus reduction assay (12).

Why were these infectious lipid-enveloped eHAV particles never observed previously? In fact, they were. We (S.M. Lemon and L. Binn) described a lipid-associated population of neutralization-resistant virions released from infected BSC-1 (monkey kidney) cells in 1985 (48). These particles banded in isopycnic CsCl gradients at a density of ~1.16 gm/cm³ and sedimented slowly and heterogeneously in rate-zonal sucrose gradients. In contrast, most HAV particles released from BSC-1 cells banded at 1.31 gm/cm³. The significance of the minor, lipid-associated virus fraction was uncertain at that time, but we noted that Provost and co-workers had reported earlier that infectious virus present in serum from acutely-infected marmosets also banded at two distinct densities in CsCl, 1.15 gm/cm³ and 1.34 gm/cm³ (49). In retrospect, it seems likely that in these early studies most eHAV released from infected cell cultures, or present in sera from infected marmosets, was at least partially disrupted during centrifugation in CsCl. This is suggested by the fact that the membrane-associated virus identified in 1985 remained largely, albeit incompletely neutralized by antibodies (49), indicating that it was not completely enveloped by membranes like the eHAV particles we have recently described (12). CsCl is a much harsher medium for isopycnic gradient centrifugation than the iodixanol we have used recently. In addition, many early efforts to characterize the biochemical and biophysical attributes of HAV incorporated purification schemes that included the use or detergents or extraction with organic solvents, reasonable steps for purifying a ‘nonenveloped’ picornavirus but both of which destroy or eliminate the eHAV particle (12). In multiple recent experiments, more than 80% of HAV particles released from infected Huh-7 cells band at the lighter density (12). In some harvests, little if any of the nonenveloped virus is found in cell culture supernatants, leading us to suspect that the nonenveloped virus is for the most part derived from eHAV particles that have lost their membranes. While we lack a precise understanding of the stability of the eHAV membrane, it is readily deformed and at times is found adjacent to what appears to be a ruptured vesicle on EM grids (12).

In our recent studies, only virus banding with the density of eHAV in iodixanol can be detected by RT-PCR in serum or plasma from acutely infected humans or chimpanzees (12). This confirms the presence of these unique picornaviral particles in vivo, and strongly supports the biological relevance of the cell culture studies described above. Virus shed in feces lacks membranes (12), however, providing the stability required for transmission of the virus through the environment, but raising many interesting questions about the biogenesis of these two extracellular forms of the virus.

Biogenesis of the eHAV envelope.

As described above, many enveloped viruses co-opt components of the host ESCRT pathway for budding. eHAV is no different (12). siRNA-mediated depletion of VPS4B, a component of the ESCRT machinery involved in the budding of viral envelopes, led to significant reductions in eHAV release while having no effect on replication of intracellular HAV RNA (12). In addition, a search for potential L domains within the HAV polyprotein identified two tandem YPX₃L motifs in VP2 (YPHGLL, residues 144–9, and YPVWEL, residues 177–182) (12). This motif is predictive of interactions with ALIX (26, 50, 51), a protein that bridges ESCRT-I and ESCRT-III complexes and that is involved in budding of the HIV envelope at the plasma membrane (52, 53). ALIX depletion also ablated eHAV release without negatively impacting viral RNA replication (12). We carried out a series of reverse molecular genetics experiments in an effort to define a role for the VP2 YPX₃L motifs in the envelopment of eHAV. Efforts to replace the leading Tyr residues in either or both motifs with Ala severely handicapped assembly of the capsid without significantly affecting first round replication of transfected viral RNA (12). Nonetheless, transfection of the VP2-Y177A mutant resulted in limited assembly of HAV capsids. Wild-type but not VP2-Y177A capsids could be immunoprecipitated with antibody to ALIX, providing an independent line of evidence in support of the involvement of ALIX in eHAV biogenesis (12). Additional depletion experiments suggest that other ESCRT-associated proteins, including TSG101 (ESCRT-0), HRS (ESCRT-I) and CHMP4A-C (ESCRT-III), are not essential for eHAV biogenesis. Thus, it appears that HAV usurps some but not all components of the ESCRT system to effect its release in membranes. Clearly, however, there is more work to be done here.

Other data point to a role for the unique C-terminal VP1pX extension in eHAV biogenesis. Surprisingly, eHAV particles contain only VP1pX and no fully processed VP1 (Figure 2c) (12). In contrast, the nonenveloped virus found in cell culture supernatant fluids or fecal samples contains little VP1pX and predominantly if not exclusively fully processed VP1. The 71 amino acid-long pX extension is fully protected from proteinases by the eHAV membrane, indicating that it is internal and not present on the surface of these particles (12). It is unstable following exposure of the eHAV particle to detergents, leading to its loss and the appearance of the mature ~274 amino acid VP1. Although the proteinase(s) responsible for this have never been identified, in vitro studies suggest that VP1pX can be processed to VP1 by trypsin or cathepsin L (40). While the N-terminal pX sequence is required for pentamer assembly and thus infectivity (36, 37), as discussed above, previous studies show that HAV will tolerate large deletions in the mid- and C-terminal parts of pX without loss of infectivity (37, 54). Most viable deletion mutants had a reduced replication phenotype, and one tested for virulence in marmosets demonstrated only a limited capacity to replicate in vivo, however (54). The effect of such deletions on envelopment of eHAV has yet to be studied. However, pX contains several conserved lysine residues that are predicted to be ubiquitylated with high confidence (Figure 2a). Ubiquitylation is a common signal for the recruitment of cargo to MVBs and ESCRT (15, 21, 26, 50), and ongoing studies in our laboratory are consistent with a role for these lysine residues in the production and release of quasi-enveloped eHAV (F. Hu, K.L. McKnight, and S.M. Lemon, unpublished data).

While additional studies are needed, the emerging data thus suggest there may be two signals directing the envelopment of HAV in host membranes: first, tandem late domains within VP2 that mediate interactions with ALIX, and second, possibly, ubiquitylation of the C-terminus of VP1pX by an unknown E3 ubiquitin ligase. We envision that these signals combine to mediate the recruitment of pre-assembled HAV capsids containing VP1pX to the site of envelopment (Figure 3). Ultrastructural studies of HAV-infected primate liver have demonstrated the presence of viral capsids within cytoplasmic vesicles (49, 55), suggesting that envelopment may involve budding into endosomes in a manner analogous to MVBs rather than at the plasma membrane. ALIX may play dual roles in this process, potentially contributing to the recruitment/sorting process as well as facilitating the early membrane deformation required for budding (15, 56). The extent to which ubiquitylation of pX is essential to this process remains to be determined. Importantly, ALIX is capable of directing ubiquitylation-independent sorting of the G-protein coupled receptor PAR1 to MVBs through its interaction with a YPX₃L motif in the protein (51).

Figure 3.

Biogenesis of quasi-enveloped eHAV and naked HAV particles. Assembled HAV capsids co-opt elements of the ESCRT system to bud either (a) into multivesicular bodies (MVB) or (b) at the plasma membrane, resulting in their release from cells wrapped in a single lipid bilayer. An alternative mechanism providing for release of membrane-wrapped picornavirus particles (AWOL), is shown in (c) wherein capsids are engulfed by autophagosomes, which then fuse to either MVBs or the plasma membrane giving rising to the release of single membrane particles (108). Available data do not support this pathway for eHAV. (d) Upon release into extracellular space, membranes may be shed from eHAV particles, resulting in extracellular naked virions.

Extracellular release of nonenveloped hepatoviruses.

eHAV is the only HAV particle type found in serum and plasma of infected humans and chimpanzees, but it is not found in fecal extracts that contain only nonenveloped, naked virions (12). Why is this? While an early study suggests that HAV might replicate within crypt cells of the small intestine (57), naked HAV virions produced within hepatocytes are shed into the gut via the biliary system in infected chimpanzees (55). These are generally considered to represent the source of virus found in feces. In collaborative studies with Dr. Lorne Tyrrell and colleagues at the University of Alberta, we confirmed that the virus present in bile and fecal pellets from infected Alb-uPA/SCID mice with chimeric human livers (58) was nonenveloped, while eHAV was predominant among virus circulating in the blood (D. Pang, M.A. Joyce, D.L. Tyrrell, Z. Feng and S.M. Lemon, unpublished data). The question then is why eHAV is found in serum and plasma, and only naked HAV particles in bile and feces.

Hepatocytes are highly polarized cells of epithelial origin (59). They are arrayed in plates with their basolateral membrane facing onto the space of Disse that is separated from the intravascular space by fenestrated sinusoidal endothelial cells (Figure 4a). A much smaller apical membrane, representing less than 15% of the hepatocyte surface, faces onto small biliary canaliculi that connect with the canals of Hering and thence to larger bile ducts that direct the flow of bile to the small intestine (Figure 4b). Bile is secreted from hepatocytes across the canalicular (apical) membrane. This membrane is highly specialized and metabolically very active, with vesicular secretion of phospholipids across it into the canaliculus associated with a rate of turnover of its outer leaflet estimated as high as 10%/min (59, 60). The canalicular membrane contains several ABC bile salt transporters, as well as flippase and floppase activities that help to maintain its integrity in the face of high concentrations of bile salts on its luminal side. The actual concentration of bile salts at the canalicular membrane where bile is formed cannot be determined, as bile is progressively diluted and buffered by phospholipid secretion as it flows toward the lower biliary track (61).

Figure 4.

Hepatocyte polarity. (a) Hepatocytes are organized as “plates” consisting of rows of polarized cells with basolateral membranes facing onto sinusoids and smaller apical membranes onto the lumen of the biliary canaliculi through which bile is secreted to the small intestine. Adapted from Treyer, A. and Müsch, A. (2013) Comprehensive Physiology 3:243–287. (b) Neighboring hepatocytes are connected through tight junctions (t.j.) and biliary canaliculi (b.c.). Also shown are Kupffer cells (k.c.), space of Disse (d.s.), microfilaments (m.f.), Golgi apparatus (g.a.), rough endoplasmic reticulum (r.e.r), and nucleus (n.). Adapted from Boyer, J.L. (2013) Comprehensive Physiology 3:1035–1078. (N.B., permission to reprint has been requested from Wiley).

Since only enveloped virions, eHAV, are found in serum or plasma (12), most if not all virus released across the basolateral membrane of hepatocytes into the perisinusoidal space can be assumed to possess an envelope. In contrast, only nonenveloped virus is released across the apical, canalicular membrane into bile or, alternatively, the envelope is shed from eHAV particles soon after their release into the canaliculus. What is clear is that release across both membranes occurs noncytopathically (that is, prior to any chemical or histologic evidence of hepatocellular injury) (46). Viral titers are typically >100-fold greater in feces than in serum, suggesting that most virus exits via the apical membrane (62). The possibility that there may be two distinct pathways for viral egress, with eHAV released from the basolataeral membrane and naked HAV particles from the apical membrane, would be consistent with the complexity of protein trafficking within hepatocytes (59). ABC bile salt transporters traffic directly from the ER and Golgi to the apical canalicular membrane, but many other apical membrane proteins traffic first to the basolateral membrane, and then move by transcytosis across the cell to the canalicular membrane (59). Thus far, however, studies in polarized cultures of CaCo2 colonic carcinoma cells and HepG2 hepatoma cells suggest that eHAV has a vectorial release pattern similar to the naked virus particle (A. Hirai-Yuki and S.M. Lemon, unpublished data). Thus, we favor the notion that the eHAV envelope is lost as virus exits the hepatocyte across the apical membrane into the canaliculus, where bile salt concentrations are particularly high. The eHAV membrane is stable in the presence of 98% porcine bile (12), but these conditions do not adequately recapitulate those in the canaliculus.

eHAV entry.

The apparent absence of virus-encoded peplomers in the quasi-envelope of eHAV is particularly interesting, since these particles are as infectious as the nonenveloped form of the virus (12). Although detailed proteomics studies have yet to be done, none of the capsid proteins of HAV possess predicted transmembrane domains or glycosylation sites, and there is no evidence at present that the eHAV particle contains any of the nonstructural proteins expressed by the P2 and P3 regions of the genome. Furthermore, as described above, biochemical evidence indicates that the capsid antigen is completely occluded and the pX extension of VP1 protected from proteinases by the membrane (12). This argues strongly for the absence of membrane fusion in the process of eHAV entry, providing a clear point of distinction from all known enveloped viruses. This makes sense, as fusion of the eHAV envelope with the plasma or endosomal membranes would deliver an intact capsid into the cytoplasm where it would have difficulty accessing its receptor to initiate uncoating.

Our current model for eHAV entry involves uptake of the virus into an endocytic pathway with movement to the late endosome/lysosome where its membrane is degraded and the viral capsid accesses its receptor (Figure 5). This model is based on the following, limited information. Both HAV and eHAV entry are sensitive to bafilomycin, but eHAV differs from the naked HAV particle in that it’s entry is inhibited by the lysosomal poison, chloroquine while naked HAV entry is not (12). eHAV thus enters via an endosomal pathway and traffics to an acidified, late endo/lysosomal compartment. Despite this, entry of both eHAV and HAV into GL37 monkey kidney cells is inhibited by antibody to TIM-1 (official name, HAVCR1), identified previously as a cellular receptor for HAV (12, 63). Importantly, although TIM-1 is expressed on the plasma membrane, it has scavenger activity and undergoes constitutive endocytosis with retrograde translocation to the lysosome (64). Membranes and lipids that enter the endolysosomal system via endocytosis or autophagy are degraded within the acidic environment of the lysosome via a very sophisticated program (reviewed in 65). Intra-lumenal lysosomal membranes sort and constitutively degrade these membranes in a process that involves the removal of cholesterol and conversion of sphingomyelin to sphingosine and fatty acids. Several lysosomal enzymes contribute to this degradative process, including acid phospholipase A2, acid ceramidase, and lysosomal hydrolases (65). Thus, the membrane enveloping eHAV is likely to be lost within the lysosome following endocytosis of the virus. The HAV capsid is extraordinarily stable, and with loss of the eHAV envelope would gain access to its receptor, TIM-1, on the lysosomal membrane. This would then initiate the uncoating process leading to the release of HAV RNA into the cytoplasm.

Figure 5.

Model proposed for eHAV entry and post-endocytosis neutralization. eHAV is taken up into cells by endocytosis, and moves rapidly to late endosomes and lysosomes where the membrane is degraded more slowly. This allows the capsid to interact with its receptor, TIM-1, which cycles from the plasma membrane to lysosomal membranes, resulting in uncoating and release of viral RNA into the cytoplasm. If antibodies are also taken up by endocytosis, and present in the lysosome, they will bind to the capsid and neutralize virus upon degradation of the eHAV membrane. Whether immunoglobulin receptors play a role in this process remains uncertain.

TIM-1 binds PS and via this mechanism facilitates the entry of a number of enveloped viruses in which the outer leaflet of the envelope is enriched in PS (16, 17). Although difficult to demonstrate, this likely to be the case with the eHAV membrane also. If so, TIM-1 could not only participate in capsid-receptor interactions leading to uncoating, but could also facilitate endocytosis of the enveloped eHAV particle. A role for other receptor-like interactions involving host-encoded proteins remains possible, but is entirely speculative at present. How exosomes deliver their cargo to recipient cells is not well established, and it is possible that there are some parallels. Nonetheless, endosomal uptake with lysosomal degradation of the eHAV membrane is an attractive hypothesis that explains several features of the entry process, including the slow uptake of eHAV, the inhibition of eHAV infection by chloroquine, and how entry occurs in the absence of a specific fusion activity in the eHAV envelope.

HEPEVIRUSES: A SECOND QUASI-ENVELOPED HEPATITIS VIRUS

Like HAV, HEV is an enterically-transmitted RNA virus that infects liver. It has been recognized for several decades as the cause of both sporadic disease and waterborne outbreaks of acute hepatitis in developing countries (66). However, in recent years cases unrelated to travel to endemic regions have been increasing recognized in many developed countries including the U.S. and Western Europe (67–69). Although similar to acute hepatitis A, hepatitis E has several distinguishing features, including a high mortality rate (up to 30%) in women in the third trimester of pregnancy, as well as persistent infections with progressive liver disease in immunocompromised individuals (67, 70, 71).

HEV is classified as the only member of the genus Hepevirus within the family Hepeviridae (70). Unlike HAV, it infects a wide variety of mammalian species. There are 4 distinct genotypes, and many human infections are likely zoonotic in origin. The genome is 7.2 kb in length, positive sense, and includes three discontinuous and overlapping open reading frames (ORF1–3) (72) (Figure 6). ORF1 encodes a large nonstructural polyprotein that is involved in genome replication. ORF2 and ORF3 partially overlap, and are translated from a single bicistronic subgenomic RNA. ORF2 encodes the solitary viral capsid protein (660 amino acids) while ORF3 encodes a small phosphoprotein (113 or 114 a.a.) that is essential for release of the virus from infected cells (73). Early EM studies revealed nonenveloped virions in fecal matter that were 27–32 nm in diameter (10, 74). Despite this, phylogenetic analyses indicate that HEV is most closely related to the Togaviridae, a large family of enveloped viruses (75).

Figure 6.

Hepevirus. (a) Organization of the positive-sense RNA genome HEV. The 7.2 kb genome contains three open reading frames, ORFs 1–3. ORF1 encodes a polyprotein that is involved in genome replication. ORF2 and ORF3 partially overlap, with ORF2 encoding a single large capsid protein containing both a putative signal sequence and possible glycosylation signals. ORF3 encodes a protein with a PTAP L domain required for production of membrane-wrapped viral particles. (b) Electron microscopic graphs showing HEV particles in the bile from an HEV (genotype 1) infected money (image kindly provided by Tian-cheng Li, National Institute of Infectious Diseases, Japan).

A detailed structural model of the HEV particle is not available, but crystallographic studies of virus-like particles (VLPs) assembled from a truncated ORF2 protein (amino acids 112–608) have provided useful information concerning its structure (76). This truncated capsid has three domains: S (shell), M (middle), and P (protruding). The S domain forms an internal scaffolding structure that is conserved among many types of viral capsids. On the other hand, the P domain of HEV differs significantly from other viral capsids, including that of the caliciviruses within which HEV was originally classified. The P domain, and particularly its apical surface, is likely involved in binding to cellular receptors and neutralizing antibodies (76, 77).

Although it has not been visualized by EM, virus circulating in the blood during acute hepatitis E bands at a density of ~1.15 gm/cm³ in sucrose density gradients and is associated with lipids (13, 78). This distinguishes it from the nonenveloped virions shed in feces that band at a density of ~1.27 gm/cm³. The ORF3 protein, which plays a pivotal role in release of the virus in cell culture (73), is found in serum-derived virus, but not in naked virions isolated from feces. Like eHAV, viral antigens appear to be completely masked by host membranes in the serum-derived particles. Monoclonal antibodies against ORF2 or ORF3 fail to capture serum-derived virions unless they are first treated with detergent (13, 79). Consistent with this, immune sera or monoclonal antibodies against ORF2 neutralize the infectivity of nonenveloped virions derived from feces, but have no effect on eHEV derived from serum (78, 80). Serum-derived virus becomes susceptible to neutralization by immune sera or monoclonal antibodies directed against ORF2 or ORF3 following detergent treatment. However, the detergent-treated virus appears as capable of infecting cultured cells as untreated virus (13). Circulating HEV particles are thus similar to eHAV in many respects, and likely to be similarly enveloped (hence “eHEV”).

This envelopment of HEV appears to be recapitulated in infected cell cultures, from which virus is released in what appears to be a membrane-occluded form that is of low density and not precipitated by antibodies to ORF-2 or ORF-3 without detergent treatment (13). Deletion of ORF3 ablates release of the virus without significantly affecting replication of HEV RNA (73). Consistent with a role in assembly and release, yeast two-hybrid studies have shown that ectopically-expressed ORF3 protein interacts with the ORF2 product, the large capsid protein, in a manner dependent upon phosphorylation of ORF3 at Ser-80 (81). However, little is known about the assembly of virus in HEV-infected cells. As with eHAV, its release in an apparently enveloped form involves components of the ESCRT system. The ORF3 protein contains a PSAP L domain motif that specifies interactions with TSG101, a host ESCRT-I protein involved in the budding of many enveloped viruses, including HIV and Ebola. Disrupting this late domain motif, or siRNA-mediated depletion of TSG101, significantly impairs release of eHEV consistent with involvement of the ESCRT system (82). In keeping with this, overexpression of dominant negative mutants of VPS4, a critically important ESCRT-related protein that recycles ESCRT complexes involved in membrane budding, also inhibited release of eHEV (83). ORF3 expression co-localizes with MVBs in infected cells, suggesting an endosomal or MVB site of budding. As with eHAV, however, relatively little is known about the composition of the eHEV membrane. A recent study suggests that TGOLN2, a protein expressed within the trans-Golgi network, is displayed on the surface of eHEV particles (84). This is consistent with eHEV budding into an intracellular vesicle, rather than at the plasma membrane, and represents another potential similarity with eHAV.

Most evidence suggests the eHEV membrane is devoid of virally-encoded proteins, like the eHAV membrane (13). However, the ORF2 protein has a typical signal sequence at its N-terminus and 3 potential internal N-glycosylation sites at Asn-137, Asn-310, and Asn-562. How these relate to the biology of HEV is uncertain. While ORF2 is glycosylated and transported to the cell surface when over-expressed (85), there is as yet no direct evidence this occurs in HEV-infected cells. The phosphorylated ORF3 product also preferentially interacts with non-glycosylated ORF2 protein (81). On the other hand, an Asn-to-Glu substitution at one of the putative glycosylation sites, Asn-562, ablated the infectivity of HEV despite continued production of virus particles (77). Similar substitutions at Asn-137 and Asn-310 resulted in a defect in virion assembly. Although the biological significance of the signal sequence and putative glycosylation sites in ORF2 remain uncertain, these features of ORF2, coupled with the phylogenetic relatedness of HEV to togaviruses, suggests that HEV may have evolved from an ancestral virus that possessed a canonical envelope.

The presence of enveloped eHEV particles in blood and the naked, nonenveloped virions shed in feces and bile (Figure 6B) directly parallels the distribution of eHAV and HAV in acute hepatitis A, as discussed above. As with HAV, it seems likely that the eHEV envelope is lost during the delivery of virus from the liver into the biliary canaliculus.

HOW DO ANTIBODIES PROTECT AGAINST QUASI-ENVELOPED VIRUSES?

As indicated above, current evidence suggests that there are no viral proteins displayed on the surface of the eHAV or eHEV particles. Both convalescent immune serum and anti-capsid antibodies fail to capture these quasi-enveloped particles unless they are first treated with detergent (12, 13). In the case of eHAV, high concentrations of proteinase K fail to degrade pX in the absence of detergent (12). The HEV ORF3 protein is similarly protected from protease digestion by its membrane (13). Most importantly, both eHAV and eHEV fail to be neutralized by anti-capsid antibodies in standard neutralization assays (12, 13), and both of these particles have been found to coexist with neutralizing antibodies in the blood during acute infections (13, 86). Collectively, it appears that these viruses have evolved a clever strategy for evasion of antibody-mediated neutralization that may facilitate their dissemination within the host. On the other hand, both infections elicit an antibody response, and at least in the case of hepatitis A, this plays a critical role in control of the infection (87). Small amounts of anti-HAV antibody provided by passive transfer of human immune immunoglobulin (IG) provide a high level of protection against symptomatic hepatitis, even when given as late as two weeks after exposure (87, 88). At this point in the infection, substantial quantities of virus are already present within the liver (46). How do antibodies provide protection against these quasi-enveloped viruses?

Current paradigms for antibody-mediated neutralization are largely based on classical concepts of enveloped versus nonenveloped viruses. These models cannot explain how antibodies neutralize quasi-enveloped viruses, in which the capsid epitopes targeted by neutralizing antibodies are completely occluded by membranes (Figure 1b). Nonetheless, the fact that eHAV is dependent on TIM-1 for its entry indicates that the HAV capsid becomes exposed at some point during the entry process, most likely in the lysosome after degradation of the membrane as discussed above (12). The transient exposure of the capsid following degradation of the eHAV membrane in the lysosome provides an opportunity for neutralization if anti-capsid antibody is present within the same compartment (Figure 5). This hypothetical mechanism provides an explanation for why antibody is able to neutralize the infectivity of eHAV when added to cultures several hours after a virus inoculum has been applied to cells, and then washed off after allowing for adsorption (12). Anti-capsid antibody added as late as 4–6 hrs after removal of an eHAV inoculum will reduce the amount of viral RNA present in cells 48–72 hrs after infection, consistent with a reduction in the number of virions initiating productive infection. Importantly, there is no effect when antibody is added 8 hrs after removal of the inoculum (12). Since endocytosed virus likely traffics rapidly to the lysosome, this suggests that it requires approximately 4–6 hrs for the eHAV membrane to be degraded. A monoclonal IgM anti-capsid antibody with potent neutralizing activity against nonenveloped HAV failed to neutralize eHAV in this type of experiment, most likely because it lost activity in the reducing environment of the lysosome. This provides evidence that antibody is not binding virus at the plasma membrane, as does the fact that there is no neutralization of naked HAV particles in similar experiments (12). There is no delay in the interaction of nonenveloped HAV with its receptor, and it quickly undergoes uncoating with transfer of the genome to the cytoplasm (Figure 5).

Hepatocytes have limited capacity for endocytosis of immunoglobulins and the mechanism that putatively delivers antibody to the lysosome is unclear (89–91). The asialoglycoprotein receptor (ASGPR) facilitates uptake and transcytosis of IgA, while the neonatal Fc receptor (FcRn) can mediate the uptake of IgG (91–93). FcRn is particularly interesting, as it has been implicated in neutralization of influenza virus within an acidic intracellular compartment (94). It has also been suggested to mediate the endocytosis of hepatitis B immune globulin by hepatocytes, resulting in inhibition of hepatitis B virus release (91). A partial knock-down of FcRn expression had no effect on post-endocytic neutralization of eHAV (Z. Feng and S.M. Lemon, unpublished data), but whether FcRn plays a role in the neutralization of these quasi-enveloped hepatitis viruses has not been excluded. It is possible that antibody enters the endocytic pathway via receptor-independent macropinocytosis (95).

Viral-antibody complexes entering the cytoplasm of cells are rapidly destroyed by the proteasome, to which they are directed by TRIM21 (tripartite motif-containing 21), a ubiquitously expressed, high-affinity, interferon-inducible cytosolic immunoglobulin receptor (96). While this antiviral mechanism appears to be operative within many different cell types, depletion of TRIM21 did not impair post-endocytic neutralization of eHAV (12). This is consistent with what we know about picornaviral entry: the intact capsid never gains access to the cytoplasm, but rather undergoes conformational changes after binding of its receptor that result in delivery of the encapsidated viral genome across the cell membrane (97).

Antibodies directed against the ORF2 product of HEV neutralize the infectivity of nonenveloped virions derived from feces, but have no effect on eHEV isolated from serum or produced in cell culture (13, 78). As with HAV (12), briefly treating cells with chloroquine prior to inoculating with eHAV, or adding antibodies shortly after removal of an eHEV inoculum reduces the amount of viral RNA present 72 hrs later (Z. Feng and S.M. Lemon, unpublished data). This suggests that eHEV may enter cells and be neutralized by antibodies by mechanisms similar to those proposed above for eHAV. However, there is less known about this and a role for TRIM21 is possible (96).

Finally, acute HAV infection is marked by a relatively lengthy incubation period during which there is little host immune response to the virus despite extensive replication within the liver (46). Virus-specific antibodies do not begin to accumulate until after 3–4 weeks after infection (8, 31). In addition to protecting circulating virus capsid from neutralizing antibodies, the envelopment of HAV may contribute on the front end to this delayed antibody response. Being sequestered in membranes, the relevant epitopes would be limited in their ability to drive B cell selection and proliferation. On the other hand, HAV RNA persists in the liver for many months following acute hepatitis A (46). Where and how this occurs is unknown, but it is interesting to speculate that this might represent neutralized eHAV-derived capsids remaining present within the endocytic compartment. Only small numbers of antibody molecules are required to neutralize other picornaviruses (1–6 molecules per virion), and antibody binding may both stabilize the capsid and prevent receptor engagement (98, 99). Whether or not the residual RNA represents viable virus is unknown. It declines in parallel with slow contraction of HAV-specific CD4+ T cell responses (46, 100).

IS QUASI-ENVELOPMENT OF VIRUSES A LIVER-SPECIFIC PHENOMENON?

The fact that the two viruses discussed above, HAV and HEV, are both hepatotropic agents that replicate in the liver may not be simple coincidence. The biliary tree provides a unique secretory pathway extending from hepatocytes to the external environment, and being able to survive the extracellular environment both with and without an envelope gives HAV and HEV some special advantages as described above. It seems likely that other non-hepatotropic viruses should have evolved similar survival strategies, but no clear examples of this have yet emerged. While many types of nonenveloped viruses, including reoviruses, papillomaviruses and parvoviruses, have exit strategies that are dependent upon vesicular transport in the absence of cell lysis (101, 102), there is no evidence that extracellular membrane-wrapped virions contribute to their dissemination within the host.

Other Picornaviruses.

Hepatoviruses constitute only one of 17 genera assigned to the family Picornaviridae (103). HAV is the only species in the genus, and multiple features distinguish it from other picornaviruses. Avian encephalomyelitis virus (AEV), genus Tremovirus, is phylogenetically the most closely related picornavirus to HAV and shares almost 50% amino acid identity with HAV in its capsid protein sequence (104). The literature provides no suggestion that AEV is released from cells enveloped in membranes (105), however, and the putative HAV VP2 L domains are not conserved in AEV. Moreover, the pX/2A sequences of the two viruses are very divergent. Much more data are available for other members of this virus family, but again, there is no clear evidence that any other picornavirus engages in the kind of membrane hijacking activity evident with HAV. As a family, these viruses replicate their RNA within or on cytoplasmic vesicles, and they induce substantial rearrangements of intracellular membranes within infected cells (106). These membranes are derived largely from the endoplasmic reticulum and secretory vesicles, but autophagosome formation promotes poliovirus RNA replication and may be required for maturation of infectious viral particles (107). Moreover, some data suggest that poliovirus present within double membrane vesicles of autophagosome origin could be released from cells wrapped in a single membrane after movement of the autophagosome to the cell surface and fusion of its outer membrane with the plasma membrane, a process termed AWOL (“autophagosome-mediated exit without lysis”) (Figure 3) (108). Virus released by AWOL would resemble enveloped eHAV particles, but autophagy is required neither for HAV replication nor eHAV release (12). Whether AWOL plays a significant role in poliovirus release in vivo remains to be established.

Nonetheless, it has been known for decades that some enteroviruses are difficult to neutralize with antibodies unless first treated with desoxycholate or chloroform, observations that suggest possible membrane association (109). In addition, an early ultrastructural study suggested that the picornavirus responsible for foot-and-mouth disease virus (FMDV, genus Aphthovirus), might be released from secretory epithelial cells of the bovine mammary gland by an exocytic mechanism involving membrane-limited vesicles (110). Given how long it required for the significance of similar findings to be recognized for HAV, it seems prudent to keep an open mind as to whether additional examples of membrane hijacking exist among picornaviruses. In line with this, a recent review has speculated on a possible role for microvesicles in the transmission of coxsackievirus B between cells (111).

Reoviridae.

A definitive example of another nonenveloped virus that is released from cells by a budding process is offered by bluetongue virus (BTV), the type species of the genus Orbivirus in the Reoviridae family (112). BTV is a widely distributed livestock pathogen that is transmitted by Culicoides spp. midges, making it one of very few known nonenveloped arthropod-borne mammalian viruses. Its multi-segmented, dsRNA genome is packaged within a large and complex double-shelled protein capsid. BTV infection is noncytopathic in insect cells in which the majority of progeny virus is trafficked to the plasma membrane in vesicles that facilitate its release. In contrast, while most mature virions are released by lysis from infected mammalian cells, some particles, particularly in early infection, appear to gain their release enveloped in membranes by budding through the plasma membrane (112, 113). Several features of the budding process echo mechanisms involved in the release of enveloped viruses. Most notably, the BTV nonstructural protein 3 (NS3) interacts with the human ESCRT-1-associated protein TSG101 and its insect homolog through a PSAP L domain motif near its N terminus (114). Mutations in this L domain alter normal viral egress patterns and leave nascent particles tethered to the cellular membrane (115). NS3 also binds the host S100A10/p11 protein, which is involved in membrane targeting, and co-localizes with the membrane lipid phosphatidylinositol (4,5) bisphosphate (112, 116). NS3 has additional interactions with the outer capsid protein, VP2, and thus functions as a bridge between the BTV capsid, cellular proteins involved in membrane trafficking, and ESCRT machinery at the site of budding (112, 116). While an enveloped form of the BTV particle was described in multiple early studies (113), the fate of the membrane surrounding the capsid after budding from the plasma membrane is not clear, nor is the role of the enveloped form of the virus in BTV pathogenesis. African horse sickness virus (AFSV), another member of the genus Orbivirus, appears to exit cells by a similar process (114). In addition, EM studies suggest that a nonoccluded insect reovirus, likely a member of the genus Cypovirus, that infects the parasitic wasp, Hyposoter exiguae, exits by budding through the plasma membrane of cells (117).

Rotaviruses represent another genus within the family Reoviridae. These nonenveloped viruses infect epithelial cells of the small intestine, and are released apically from polarized cultures of such cells in the absence of lysis (118). At an intermediate stage in their assembly, immature rotavirus capsids bud into the lumen of the ER through a process that is not well characterized (118, 119). The membrane enclosing the capsid is removed within the ER, allowing for assembly of the outer capsid protein, VP7, that is resident within the ER lumen. Thus, although rotaviruses exit infected cells by a process that involves budding of an assembly intermediate through the ER membrane, significant numbers of membrane-wrapped particles do not reach the extracellular environment making it quite different from eHAV and eHEV.

Poxviruses.

Vaccinia virus (VACV) is a large, enveloped DNA virus belonging to the genus Orthopoxvirus. Infection with it leads to the release of several distinct types of enveloped particles: intracellular mature virus (IMV) particles that are enveloped in a single lipid bilayer, ‘extracellular enveloped virus’ (EEV) particles that possess a second outer membrane, and cell-associated enveloped virus (CEV) that also have a double envelope but remain associated with the cell surface from which they are launched toward adjacent cells on actin tails (120, 121). IMV are produced within cytoplasmic virus factories by a unique process involving the de novo synthesis of a lipid bilayer around a previously assembled core (120). Most IMVs are released following cell lysis, but a proportion traffic to endosome and Golgi-derived membranes where they become further enveloped, resulting in ‘intracellular enveloped virus’ (IEV) with three surrounding lipid bilayers. IEV exits the cytoplasm following fusion of the outer, third lipid bilayer with the plasma membrane, leading to extracellular EEVs and CEVs. Both the outer and inner lipid bilayers of EEVs contain virally-encoded proteins, but only the inner membrane (equivalent to the only membrane enveloping IMVs) possesses the fusion activity required for cell entry (122). This poses a topological problem for EEV entry that is not unlike that faced by eHAV in interacting with its receptor, as EEV must first shed its outer envelope before its fusion protein can access the plasma membrane of a target cell. To overcome this hurdle, two VACV glycoproteins expressed on the outer envelope of EEV mediate the dissolution of the outer membrane upon contacting cell surface glucosaminoglycans (122). This ‘ligand-induced non-fusogenic dissolution’ of the outer EEV membrane appears to be unique to the poxviruses (122). A similar process is unlikely to be involved in eHAV entry given the apparent absence of HAV proteins within the eHAV envelope. Like eHAV, however, the additional outer membrane present on EEV renders these virions significantly more resistant to neutralization than extracellular IMVs, thereby facilitating dissemination of the virus within the host (120, 122).

CONCLUDING REMARKS

eHAV and eHEV provide a remarkable example of convergent evolution of two hepatotropic RNA viruses toward similar lifestyles wherein each is shed in feces in a stable, nonenveloped form yet circulates in the host during acute infection cloaked a quasi-envelope protecting the capsid within from neutralizing antibodies. The processes by which these viruses hijack cellular membranes to exit cells with a protective envelope are unique to each, but in both cases resemble aspects of how more classical ‘enveloped’ viruses bud from cells. These recent findings, coupled with earlier data from BTV (112), show that mechanisms of morphogenesis and release are not always that different for ‘enveloped’ and ‘nonenveloped’ viruses. Still to be answered are numerous questions, among them the extent to which similar mechanisms account for nonlytic release of many other ‘nonenveloped’ viruses as well as how quasi-enveloped viruses enter cells. Membrane hijacking appears to be an important part of the pathogenesis of hepatitis A and hepatitis E. Whether it will prove to be a strategy unique to these hepatitis viruses, or one shared more broadly among viruses in general remains to determined.