REVIEW OF THE LITERATURE

Enteroviruses

Enteroviruses form one of the most common groups of human pathogens, causing a wide range of diseases. The clinical features include poliomyelitis and other neurological diseases, meningitis, myocarditis, conjunctivitis, skin and mucosal eruptions, diarrhea, generalized infections of the newborn, and common cold-like illness. Many, if not most of the infections are subclinical, proceeding in a period of several weeks without significant symptoms in the host, but rare cases result in paralysis or even death. One of the peculiar features of enteroviruses is that there is both individual and temporal variation in the diseases caused by various enteroviruses. Most enterovirus serotypes, including coxsackievirus A9, may be associated with several disease patterns. Recently, prospective seroepidemiological studies have suggested a potential role of enterovirus infections in the etiology of insulin-dependent diabetes mellitus (Hyöty et al. 1995, Hiltunen et al. 1997) and atheroschlerotic heart disease (Roivainen et al. 1998)

The family Picornaviridae is formed by a large group of non-enveloped positive-stranded RNA viruses. The genus Enterovirus forms one of the nine genera in the family, the other eight being Aphthovirus, Cardiovirus, Erbovirus, Hepatovirus, Kobuvirus, Parechovirus, Rhinovirus, and Teschovirus. Genetically, enteroviruses cluster relatively closely with rhinoviruses, while the other genera are more distant (for example Pöyry et al. 1996, Hyypiä et al. 1997, Oberste et al. 1999).

Enterovirus taxonomy has recently been changed in accordance with the increasing genetic data (King et al. 1999). The current classification recognizes 87 serotypes in the genus Enterovirus, 64 of these being human viruses. The human enterovirus serotypes are divided into five species: the poliovirus species (PV), and the human enteroviruses A to D (HEV-A to HEV-D). Enteroviruses have been isolated from other animals as well, and some of these viruses form additional species.

The enterovirus mainly studied in this thesis is coxsackievirus A9 (CAV9) which is one of the most frequently isolated enteroviruses in clinical diseases (Hovi et al. 1996), and belongs to the species human enterovirus B (HEV-B) (King et al. 1999). In addition to CAV9, this species consists of all coxsackievirus B serotypes (1-6), all echoviruses (28 serotypes), and enterovirus 69. CAV9 is thus the only CAV serotype in this species, the 22 other ones belonging to species HEV-A (11 serotypes) and HEV-C (11 serotypes) (King et al. 1999, Oberste et al. 1998, 1999). The controversy between classification and nomenclature reflects the initially used primitive classification system that was based on replication in cell cultures and pathogenicity in experimental animals.

Table 1. Classification of human enteroviruses according to King et al. (1999).

The knowledge about enterovirus biology comes mostly from studies concerning poliovirus, information on other enteroviruses being much more limited, and rather sporadic. In discussing the biology of enteroviruses, the focus is therefore in poliovirus. Today, there is rather abundant genetic, and still, relatively narrow experimental data giving assistance in educated guesses concerning other enteroviruses.

Enterovirus structure

Enteroviruses are non-enveloped particles of about 30 nm diameter, composed of sixty copies of each of the four structural proteins VP1 to VP4, that surround a positive-stranded RNA genome of approximately 7500 nucleotides. The icosahedral capsid contains 12 pentagon-shaped pentamers of 5 protomers, each protomer being formed by one copy of each of the structural proteins. The basic organization of the capsid is shared by all picornaviruses.

To date, atomic-resolution structures of six enteroviruses have been resolved by x-ray crystallography. These structures include poliovirus 1 (PV1) (Hogle et al. 1985), PV3 (Filman et al. 1989), coxsackievirus B3 (CBV3) (Muckelbauer et al. 1995), bovine enterovirus (BEV) (Smyth et al. 1995), PV2 (Lentz et al. 1997), and coxsackievirus A9 (CAV9) (Hendry et al. 1999). In addition, structures of human rhinovirus 14 (HRV14; Rossmann et al. 1985, Arnold & Rossmann 1988), HRV1A (Kim et al. 1989), HRV3 (Zhao et al. 1996), and HRV16 (Hadfield et al. 1997) have been resolved. All these structures share a remarkable degree of overall similarity, while variation is seen on the surface and on the inner structures of the capsid. The major capsid proteins VP1 to VP3 are each folded into eight-stranded antiparallel b -sheets with a jelly-roll topology. These b -barrels of five copies of VP1 are located around the fivefold axis, while VP2 and VP3 are around the threefold axis. VP4 is much smaller than the other structural proteins, having a less ordered structure, and being located on the inner surface of the capsid, facing the RNA.

One of the most striking structural features in picornaviruses is the circular canyon around the fivefold axis, that was first seen in the structure of human rhinovirus 14 (HRV14) (Rossmann et al. 1985). The canyon hypothesis (Rossmann et al. 1985, Rossmann 1989) suggests that one strategy to escape the immune surveillance of the host organism would be to hide the receptor attachment site in a surface depression. This would sterically protect the attachment site from antibodies, still allowing recognition by the cell surface receptor that would be narrower than an antibody. The canyon is located roughly between VP1 on the 'north' side (the side closer to the fivefold axis), and VP2 and VP3 on the ‘south’ side. This organization leaves five copies of VP1 as a protrusion at the fivefold axis. The general topology of the canyon is variable, appearing more like five distinct depressions than a single canyon in CAV9 (Hendry et al. 1999).

The N terminal glycine of VP4 has a myristate, a saturated tetradecanoic fatty acid, covalently attached to it (Chow et al. 1987). In site-directed mutagenesis studies, myristoylation and myristate-protein contacts were found to be necessary in pentamer formation, RNA encapsidation (Moscufo et al. 1991), and for the stability of the virion (Moscufo & Chow 1992). It was also shown that the VP4 protein is necessary in uncoating or cell entry, as a nonviable VP4 mutant was capable of forming the structural intermediates thought to be necessary for infection, and when transfected to cells, the cDNA was capable of initiating infection that resulted in apparently mature virions (Moscufo et al. 1993).

All the known enterovirus and rhinovirus structures contain a hydrophobic pocket at the base of the canyon, covered by loops of VP1 (Hendry et al. 1999). This space is assumed to be normally filled by a natural pocket factor, a fatty acid with an aliphatic chain of variable length. These pocket factors, including sphingosine (PV1, PV3), palmitate (CBV3), and myristic acid (BEV), are believed to be present in the virion upon release from the infected cell. The pocket factors are seen as stabilizing agents in the capsid, and their presence is thought to prevent uncoating. It is therefore necessary that these molecules should leave the pocket when infection is initiated. This pocket is the site of action of several antiviral drugs and these are thought to bind to the pocket more tightly than does a natural pocket factor, thereby hindering uncoating (McSharry et al. 1979, Caliguiri et al. 1980, McKinlay 1985).

Enterovirus genome

Enteroviruses have a single molecule of positive-stranded RNA as their genome. The RNA is about 7500 nucleotides long, and there is a small virus-encoded protein VPg (or 3B) attached to the 5' end of the molecule. At both the 5' end and 3' end of the genome, there is an untranslated region (UTR). The 5' UTR makes up approximately 10 % of the genome. This 5' UTR is involved in the initiation of translation, directing ribosomes into the IRES, the internal ribosome entry site (Jang et al. 1988, Pelletier and Sonenberg 1988). The 3' UTR is shorter, only 70-100 nt in length, and it is followed by a poly-A tail. The 3' UTR is used in initiating the synthesis of negative-strand RNA (Rohll et al. 1995), but even deletion mutants lacking the 3' UTR have been shown to be infectious in cell culture (Todd et al. 1997).

The protein coding region encodes a single polyprotein, which is proteolytically cleaved into precursor proteins P1, P2 and P3, and thereafter into the structural proteins VP1 to VP4 (P1), and seven nonstructural proteins (P2 and P3). Many of the intermediate cleavage products are functional as well. Proteins 2A, 3C and 3CD are proteases, 3D is the RNA-dependent RNA polymerase, 2C is apparently a helicase, and has a function in encapsidation of RNA (Vance et al. 1997), and 2B, 2BC, 3A and 3AB have been associated with various functions in the replication of viral RNA.

Figure 3. The enterovirus genome. P1, P2, and P3 are the products of the initial proteolytic cleavages of the viral polyprotein.

Overview of enterovirus replication cycle

In order to initiate an infectious cycle, the virus must attach to a cell surface receptor on a susceptible cell. Upon binding to its receptor, an enterovirus undergoes structural rearrangements resulting in a particle that has lost the majority of the internal capsid protein VP4, has externalized the N terminus of VP1, and has a different sedimentation coefficient (135S versus 160S of the native virion), altered antigenic properties, and increased protease sensitivity. Some altered poliovirus particles are found in endosomes soon after infection (Fenwick & Wall 1973), but the entry site of a productive infection is still not known.

Before synthesizing complementary negative-stranded copies of the genome, the genomic RNA acts as a template for protein synthesis. The viral proteins need to be synthesized first to obtain the RNA-dependent RNA polymerase that the cellular machinery lacks. Translation is initiated from the IRES present in the 5' UTR (Jang et al. 1988, Pelletier and Sonenberg 1988). The complete viral protein coding region is translated as a single polyprotein, which is co-translationally cleaved, by virus-encoded proteases, into three precursor proteins P1, P2 and P3 (Kitamura et al. 1981). One of these, the capsid protein precursor P1, is co-translationally myristoylated. The P2 and P3 proteins are cleaved into seven nonstructural proteins, in addition to several forms of functional intermediates, acting in RNA replication and protein processing. The P3 cleavage products include the proteases 3C and 3CD, that are needed in the cleavage of other virus proteins from the precursors.

The same RNA molecule that has been used in translation, is then used as a template for synthesis of negative-stranded RNA molecules by the virus-encoded polymerase 3D. These strands are then used in synthesizing large amounts of positive-stranded copies that are used in translation of viral proteins, as well as being encapsidated into the assembling virions. The small VPg protein is linked to the 5' end of all transcripts (Flanegan et al. 1977).

The infection process is enhanced by cleavage of the eukaryotic initiation factor 4GII (eIF4GII) (Gradi et al. 1998, Svitkin et al. 1999) and apparently the poly(A)-binding protein as well (Kerekatte et al. 1999), by a mechanism that involves the protease 2A (Hellen et al. 1991, Wyckoff et al. 1992). These cleavages are among the known alterations in the host cell, resulting in cessation of cellular (cap-dependent) protein synthesis by poorly understood mechanisms. Enteroviruses use a different (cap-independent) method of translation initiation (Jang et al. 1988, Pelletier and Sonenberg 1988), being therefore still capable of performing protein synthesis in the infected cell.

The assembly of a new virion begins as the P1 precursor protein is cleaved into capsid proteins VP0, VP1 and VP3, and these proteins form a heterotrimer (Korant 1973). Thereafter, five of these heterotrimers associate into a pentameric structure, which is actually a 15-mer of five copies of each of the procapsid proteins VP0, VP1 and VP3 (Phillips et al. 1968). Again, 12 of these pentamers then associate together to form procapsids (VP013₆₀) that are filled with a copy of the positive-stranded genome. It is not known whether empty procapsids are first formed and then filled with the RNA molecule, or the procapsids associate around the RNA molecule. Poliovirus and foot-and-mouth disease virus (FMDV) particles have been reported to contain nonstructural proteins 2C, 3CD, 3C, and 3D, and although there is no evidence of their usefulness, it has been suggested that their presence in the virions might reflect a functional role in initiating replication in an infected cell (Newman & Brown 1997). During encapsidation of the RNA, the majority of the capsid protein VP0 is cleaved into VP2 and VP4. This is called the maturation cleavage, which is thought to lock the assembled capsid into a stable, mature virion. Poliovirus mutants that are unable to undergo this cleavage, are defective in assembly and RNA encapsidation (Compton et al. 1990, Ansardi & Morrow 1995), or can accumulate 150S particles that appear like procapsids filled with RNA, but are highly unstable (Hindiyeh et al. 1999). Maturation cleavage is known to be absent in parechoviruses 1 and 2, and in Aichi virus, demonstrating that this cleavage is not an absolute requirement for all picornaviruses (Hyypiä et al. 1992, Stanway et al. 1994, Yamashita et al. 1998). Empty capsids are also produced in natural infection, and these particles have intact VP0 proteins instead of VP2 and VP4, thus having the protein composition VP013₆₀ (Holland & Kiehn 1968, Jacobson et al. 1970). All the assembly intermediates described above can be demonstrated in an infected cell by sedimentation analysis: the heterotrimeric protomer (5S), the pentamers (14S), the empty capsid (approximately 75S), and the encapsidated mature virion (160S).

A single infected cell can typically produce 10⁴ to 10⁵ poliovirus particles, that are released by lysis of the cell. The lysis has been proposed to occur due to increased permeability of the cell membranes (Carrasco 1977), which results from increasing concentrations of nonstructural proteins 3A and 3AB (Lama & Carrasco 1992a, Lama & Carrasco 1992b), 2 BC (Aldabe et al. 1997), and 2B (Lama & Carrasco 1992b, van Kuppeveld et al. 1997). Recent data suggests that apoptosis might also be a significant factor in the symptoms that arise from persistent infections of the central nervous system (Girard et al. 1999), and that poliovirus 3C induces apoptosis by a caspase-dependent mechanism (Agol et al. 1998, Barco et al. 2000).

Cellular receptors for enteroviruses

Initiation of viral infection is dependent on attachment to a cellular receptor molecule. Attachment to these molecules is specific, and during evolution, viruses have adapted to use a variety of receptors. Several such receptors are known for various enteroviruses (Table 2). There are several examples of closely related viruses using different receptors for entry. On the other hand, some evolutionarily distant viruses can use the same receptor, as shown by the coxsackievirus-adenovirus receptor (CAR) (Bergelson et al. 1997a). The murine homologue of this molecule (mCAR) is also capable of functioning as a receptor for both coxsackie- and adenoviruses (Bergelson et al. 1998). Moreover, there are viruses that have been shown to use at least two different receptors. An example of such viruses is CAV9 (Roivainen et al. 1991a, Hughes et al. 1995, Roivainen et al. 1996), the virus mainly studied in this thesis. In addition to the primary receptor molecules, many if not all viruses need additional proteins, sometimes called coreceptors, for entry. CAV21 has been shown to require the presence of both ICAM-1 and the decay-accelerating factor (DAF) for cell entry (Shafren et al. 1997), and there is a post-attachment requirement for b 2-microglobulin in the entry process of CAV9 (Triantafilou et al. 1999).

When rhinoviruses were found to have a narrow canyon around the fivefold axis, these canyons were suggested to be the receptor recognition sites (Rossmann et al. 1985). Similar canyons have been found in all entero- and rhinovirus structures determined to date, and recently, structural studies have unambiguously resolved the precise location of binding. Structures of HRV14 and HRV16 complexed with domains of ICAM-1, and PV1 in complex with PVR have verified that the receptor binding site actually is in the canyon (Olson et al. 1993, Bella et al. 1998, Kolatkar et al. 1999, Belnap et al. 2000a, He et al. 2000).

There are at least two alternative receptors for CAV9, the integrin a _{vb 3}, which is the vitronectin receptor (Roivainen et al. 1991a, Roivainen et al. 1994), and an unidentified cell surface receptor (Hughes et al. 1995, Roivainen et al. 1996). The integrin a _{vb 3} recognizes the tripeptide arginine-glysine-aspartate (RGD), which is a known recognition signal for integrins (Ruoslahti & Pierschbacher 1986) containing the a _v subunit (Hynes 1992). The RGD sequence is used in some other animal viruses as a receptor recognition sequence, as exemplified by the echovirus 9 Barty strain (Zimmermann et al. 1997), FMDV (Fox et al. 1989, Berinstein et al. 1995), and human parechovirus 1, that was formerly known as echovirus 22 (Stanway et al. 1994). CAV9 has this tripeptide close to the C terminus of VP1, in a sequence of approximately 15 amino acids that is not found in other enteroviruses (Chang et al. 1989).

Integrin a _{vb 3} was found to act as a receptor for CAV9 in GMK cells (Roivainen et al. 1991a, Roivainen et al. 1994), but proteolytic cleavage (Roivainen et al. 1991a) and site-directed mutagenesis studies (Hughes et al. 1995) thereafter revealed that CAV9 lacking the RGD motif can still propagate in GMK cells, although displaying a small-plaque phenotype. Furthermore, in RD cells, removal of RGD from VP1 apparently had no effect in infectivity (Hughes et al. 1995, Roivainen et al. 1996). Regardless of the capability of CAV9 to survive in vitro without the RGD motif, all natural isolates contain this motif (Chang et al. 1992, Santti et al. 2000). These results show that although CAV9 does use the integrin a _{vb 3} as a receptor, it can also use an alternative receptor. This is one of the several examples of viruses that are not dependent on a single receptor, underlining the assumption that the identity of the receptor may not be as important for virus survival as has been thought earlier, although it can still have a role as a determinant of pathogenesis.

Judging by the phenomena described above, it seems that viruses are somewhat opportunistic in their receptor usage, being capable of adapting to different receptors during virus-host coevolution. This view is supported by the finding that there is variation in receptor specificity within a given CBV serotype, depending on the passaging history and between various isolates (Bergelson et al. 1997b). It has also been suggested that the ability to bind DAF may have originated more than once among enteroviruses (Powell et al. 1999).

Table 2. Current view of enterovirus receptors

Conformational transitions in the capsid

Upon attaching to the receptor, enteroviruses undergo a permanent reorganization of the capsid, characterized by a shift from a 160S sedimenting particle to a 135S particle. The 135S particle has lost the majority of the 60 copies of the capsid protein VP4, and the antigenicity and protease sensitivity are altered in comparison to the native virion (CBV3: Crowell and Philipson 1971; PV: Fenwick & Wall 1973, De Sena & Mandel 1977). One of the specific alterations is the exposure of the N terminus of VP1 (Fricks & Hogle 1990). This 135S altered particle (A particle) is thought to be an essential intermediate in initiating a new infectious cycle, while being itself noninfectious (PV: Joklik & Darnell 1961, Fenwick & Cooper 1962, Lonberg-Holm et al. 1975; CBV3: Crowell and Philipson 1971). Nevertheless, it has been shown that to a very small extent, it is capable of initiating infection in cells that are nonsusceptible to the intact 160S virion (Curry et al. 1996).

Poliovirus infection can proceed without accumulation of 135S particles, which could be interpreted as suggestive evidence of its lack of importance. In these studies, the infection was found to occur in the presence of a capsid-binding drug R78206 (Ofori-Anyinam et al. 1995), or the virus had been adapted to grow in HeLa cells at +25 ° C (Dove & Racaniello 1997), at which temperature the conformational transitions do not occur (Yafal et al. 1993). These data are not conclusive, since the results could be interpreted as showing that one kinetic barrier has been overcome in the conditions used, resulting in lack of accumulation of the transformed particles, rather than lack of transformation. Another kind of evidence was provided by a kinetic experiment in which the amounts of intact poliovirions, 135S particles, and 80S particles were followed during incubation with purified PVR. In this study, the 80S particles appeared to be transformed from the 160S particles, but not from the 135S particles (Arita et al. 1998). This seems to suggest that the accumulating 135S particles are not intermediate forms on the productive infectious pathway. However, there are several lines of evidence suggesting that the 135S particles are in fact an intermediate in the infection process. First, the 135S particle is the dominant virus form found inside the infected cells in the early phase of infection (Lonberg-Holm et al. 1975, Everaert et al. 1989), but a portion of the attached virus particles are also eluted from the cells as 135S particles, and can be purified (PV: Fenwick & Cooper 1962, CBV3: Crowell & Philipson 1971). Second, the same reorganization of the capsid is seen when incubating polioviruses with solubilized poliovirus receptor (PVR) (Kaplan et al. 1990), or when incubating enteroviruses with membrane extracts (CBV3: Roesing et al. 1975; PV: De Sena & Mandel 1976, 1977; Guttman & Baltimore 1977). Third, several antiviral drugs, such as arildone (McSharry et al. 1979, Caliguiri et al. 1980) and the WIN compounds (McKinlay 1985), function by binding to enteroviruses, thereby hindering the conformational transitions. Fourth, 135S particles bind to lipid bilayers, forming ion channels (Fricks & Hogle 1990, Tosteson & Chow 1997). Fifth, the 135S particle is moderately infectious (Curry et al. 1996). Sixth, mutant polioviruses overcome defects, caused by mutated receptor, by lowering the thermal energy required for the structural change into 135S particles (Wien et al. 1997).

For the above reasons, the 135S form of the virus particle is generally seen as an intermediate in the process where the stable virion transforms due to receptor binding, in order to finally release the genome into the cytoplasm of the susceptible cell. Empty capsids (VP123₆₀) can be purified from infected cells soon after initiation of infection, and by releasing the RNA, the particles have shifted from 135S particles to 80S particles.

Enteroviruses can transform into 135S particles without the presence of receptors, when incubated at elevated temperatures and hypotonic conditions (CAV13: Cords et al. 1975; PV: Lonberg-Holm et al. 1976, Wetz & Kucinski 1991). The resulting 135S particles are indistinguishable from those appearing during infection. The receptor therefore functions like an enzyme, by reducing the thermal energy required in transforming the stable capsid into an entry-competent form (Chow et al. 1997). Conversely, it has been shown that receptor defects can be overcome by viral mutations that lower the transformation energy (Wien et al. 1997).

Very recent data suggest that there is an intermediate structure in the transition pathway from 160S native virions to the 135S particles. These 147S particles contain VP4, they are infectious, and the N terminus of VP1 is apparently not exposed, as suggested by resistance to V8 protease digestion (Pelletier et al. 2000). In the conditions used, an alternative form of 135S particles was also detected, containing VP4.

In addition to the irreversible structural changes that occur in the capsid during uncoating, virus particles are known to undergo reversible transitions. In contrast to the static view of the virus particles that is given by the crystallographic studies, it is known that the capsids are actually flexible and dynamic entities. It has been known for a long time that polioviruses exist as two interconvertible forms that have different isoelectric points (pI 7 or 4; Mandel 1971). These two forms might reflect the apparent controversy observed between the crystal structures and various biochemical studies; namely, some of the regions that are inside the capsid in the crystal structures, are actually reachable by antibodies. The regions that have been shown to be involved in reversible exposure are the internal capsid protein VP4 (Li et al. 1994) and the N terminus of VP1 (Chow et al. 1985, Roivainen et al. 1993, Li et al. 1994).

The N termini of enterovirus VP1 proteins are very heterogeneous both in length and sequence, but there is a highly conserved region close to the N terminus of VP1 of all enteroviruses (Hovi & Roivainen 1993). This motif was first identified as a part of an immunodominant region of 20 amino acids (Roivainen et al. 1991b). As it was noticed that the immunogenicity was combined with a conserved sequence, it was shown that antibodies against this sequence can be used as a group reagent recognizing enteroviruses (Hovi & Roivainen 1993). Peptide antibodies against this motif (amino acids 42-52 in PV1/Mahoney) were shown to be capable of precipitating purified poliovirus particles, showing that this region is exposed in solution (Roivainen et al. 1993), while being on the inner surface of the capsid in the crystal structure. It was also conclusively shown that the neighboring region, amino acids 24 to 40 of VP1 of PV1/Mahoney is transiently exposed (Li et al. 1994), by demonstrating that antibody recognition of this motif is dependent on coincubation of the virus and the antibodies at +37 ° C. The transitions are thus reversible, occurring at physiological temperatures but not below that (Li et al. 1994). On the other hand, it has been shown that poliovirus can be immunoprecipitated at room temperature by overnight incubation with peptide antibodies against amino acids 42-52 (Roivainen, unpublished results). These apparently contrasting observations could be explained by a kinetic equilibrium of the two forms of the virion. The transiently exposed portions could bind to antibodies with kinetics determined by the incubation temperature, and then being locked in this conformation by the antibody.

Medium-resolution structures of the 135S and 80S particles (2.2 nm resolution) were recently revealed using cryo-electron microscopy and image reconstruction (Belnap et al. 2000b). The structures indicate that both 135S and 80S particles are approximately 4 % larger than the virion, therefore having thinner, as well as more angular capsids. They also indicate domain movements of up to 0.9 nm (3 % of the virion diameter), and gaps that are seen as plausible sites for the externalization of the VP4 and the N terminus of VP1. Other studies have suggested that VP4 and the N terminus of VP1 would exit through the fivefold axis (Hadfield et al. 1997, Hendry et al. 1999), or that VP4 would exit through a channel along the fivefold axis and the N terminus of VP1 through the pseudo threefold axis (Lentz et al. 1997). However, it was suggested that the exit site for both of these would be the base of the canyon, and the five extended N terminal arms of VP1 would follow the fivefold protrusion on the surface of the capsid, meeting each others at the fivefold axis, ideally positioned for membrane insertion (Belnap et al. 2000b). The fivefold vertex was not favored as the exit site for VP4 and the N termini of VP1, as there is a plug formed by the N termini of VP3 that is present in both 160S and 135S particles, seeming to prevent the use of this location as the exit site. It was suggested that despite the VP3 plug, the RNA could be released through the fivefold axis, as the plug might act as a float-valve (Belnap et al. 2000b).

Interactions of enteroviruses with lipid bilayers

Poliovirus 135S particles are capable of attaching to liposomes, artificial lipid bilayers (Lonberg-Holm et al. 1976), and this attachment is dependent on the exposed N terminus of VP1 (Fricks & Hogle 1990). On the other hand, poliovirus infection is known to permeabilize the infected cell to various compounds (Fernández-Puentes & Carrasco 1980). These observations have led to the view that as the infection of the cell is initiated, the enteroviruses form a pore on one of the cellular membranes (Fricks & Hogle 1990). The pore is believed to be composed of at least the five N termini of VP1 that are present near the pentamer fivefold axis, and the myristate-linked VP4 is also very likely to be involved (Moscufo et al. 1993). It is possible that the receptor or some other cellular compound is also involved.

Earlier it was thought that the conformational transition to the 135S particle is required before binding to the membrane can occur, but recently it was shown that also native poliovirus is capable of binding to a planar lipid membrane (Tosteson & Chow 1997). Furthermore, both 135S and 160S particles induce the formation of ion-permeable channels in such membranes (Tosteson & Chow 1997). This finding is in line with the observation that poliovirus infection changes the permeability of infected cells (Fernández-Puentes & Carrasco 1980), although the change in permeability allows entry of larger molecules than could be explained by these channels. The infected cells are permeabilized to compounds like luciferase, horseradish peroxidase (Otero & Carrasco 1987), and polysaccharides (Gonzáles & Carrasco 1987). The channels formed by the two forms of poliovirus were found to be different in nature, the ones formed by the 135S particle being larger and independent of temperature or voltage across the membrane. It can therefore be speculated that the two forms of channel either have a different function in the infection process, or that one (or both) of them has no role in it. Uncoating of poliovirus appears to be necessary for the permeabilization of infected cells, since uncoating inhibitors also block membrane permeabilization (Almela et al. 1991). Based on the observed conductance of the channels, it was estimated that the channel formed by the 135S particle should have a diameter of approximately 2 nm, which could in theory allow the RNA genome to pass through (Tosteson & Chow 1997). The diameter of a fully extended RNA molecule is approximately 1 nm. This data is consistent with the model that enterovirus capsid proteins would form a pore in one of the cellular membranes, allowing the RNA genome to enter the cytoplasm either through a pore or by lysis of an endosome. While the channels formed by the 135S particles might be an essential step in enterovirus infection cycle, it is unclear whether the channels induced by the native virions can be associated with the normal infection process, or whether they merely reflect properties of the virion that are necessary for the uncoating to occur.

It can also be speculated that the channels as such would not have a function in the normal infection, but that they are merely an artificial effect demonstrating the physical properties of the viral particles. Permeabilization of the newly infected cell is a specific phenomenon and does not occur in nonsusceptible cells (Otero & Carrasco 1987, Carrasco 1995). This would suggest that the virus binding and channel formation phenomena that occur nonspecifically in liposomes and other artificial membranes, do not necessarily occur similarly in a normal infection. Specifically, when the 135S particles are studied, they have already lost the majority of the VP4 protein that is needed in the cell entry (Moscufo et al. 1993), and may be important in the functional channel formation during this process.

Entry route

The productive infectious pathway of enteroviruses is ill-understood between attachment to a cell surface receptor and initiation of replication in the cytoplasm. It is known that both the 135S particles and even native virions can be internalized into the cell under appropriate conditions (Fenwick & Wall 1973, Lonberg-Holm et al. 1975, Willingmann et al. 1989, Kronenberger et al. 1992, Kronenberger et al. 1998). A major problem in trying to elucidate the infectious pathway is the very large particle/infectious unit ratio of all enteroviruses. Only one virus particle out of 100 or more actually initiates an infectious cycle, while all direct chasing methods follow the bulk of the virus population. It is therefore clear that any straightforward observation concerning the fate of the virus or its subunits is unavoidably clouded and potentially misled by the bulk of the virus population engaged in apparently futile events.

In order for the infection to be initiated, the viral genome must pass at least two obstacles: the stable virus capsid and a cellular membrane. The suggested models of infection propose three locations as the site of crossing the protein-membrane barrier between the viral RNA and cellular cytoplasm. Direct penetration through the plasma membrane was suggested when electron microscopic images showed apparently intact poliovirus particles in the cytoplasm 3 min after infection (Dunnebacke et al. 1969). The endocytic pathway was also supported by electron microscopy, since poliovirus particles were found in endocytic vesicles (Dales 1973). Most recently, EV1 was found to co-localize with caveolin-1 but not with markers of the clathrin endocytic pathway, suggesting caveoli as the entry site for this virus (Marjomäki et al. 2000). It has also been proposed that polioviruses are not dependent on a single infectious pathway and that several alternative pathways could exist together (Arita et al. 1998, Arita et al. 1999).

Numerous experimental setups have been used in order to find out whether acidification, that occurs in the clathrin-coated pits pathway, is necessary for the infection to occur. Some of the earlier results suggested that acidification of endosomes is needed for poliovirus infection (Madshus et al. 1984a, Madshus et al. 1984b), but increasing evidence argues the contrary (Gromeier & Wetz 1990, Pérez & Carrasco 1993, DeTulleo & Kirchhausen 1998).

Acidification appeared to be necessary when studying poliovirus infection with compounds that either dissipate proton gradients across membranes, or inhibit the acidification process of endosomes (Madshus et al. 1984a, Madshus et al. 1984b). These studies showed that both these types of compounds inhibited infection, and the inhibition was overcome by exposing the cells to low pH. Nevertheless, another study showed that use of these compounds resulted in a delay in infection, but not inhibition (Gromeier & Wetz 1990). The delay was due to side effects of these compounds (Cassell et al. 1984), and not the pH-elevating mechanism (Gromeier & Wetz 1990).

Bafilomycin A (BFLA) is an inhibitor of vacuolar proton-ATPases, enzymes that acidify endosomes as well as other vesicular organelles (Bowman et al. 1988). This drug has also been shown to have unwanted effects: it blocks endosomal transport in a cell line-dependent manner (Bayer et al. 1998). Therefore, inhibition of infection in the presence of BFLA does not necessarily imply that the virus is dependent on acidification, but a negative effect on inhibition apparently shows that the studied virus does not require acidification for infection. It has been shown that BFLA does not affect infection by poliovirus (Pérez & Carrasco 1993) or CAV9 (Roivainen, unpublished results), and these viruses therefore do not require an acidic environment for entry.

Further evidence against the necessity of using the clathrin endocytic pathway was obtained by using HeLa cell lines that expressed either dynamin^K44A, failing to load GTP, or dynamin^ts, failing to hydrolyze GTP at the non-permissive temperature (DeTulleo & Kirchhausen 1998). The phenotype of these cells appeared unaltered, except for the block in the formation of clathrin-coated pits and vesicles. Clathrin-dependent endocytosis is blocked in these cells, because dynamin functions in a GTP-dependent manner in the budding of a clathrin-coated pit. The results showed that poliovirus 1/Mahoney was able to infect the cells in a clathrin-independent manner, while the three other viruses studied were not. Two of these viruses had envelopes (Semliki Forest virus and Sindbis virus), but even human rhinovirus 14, an unenveloped picornavirus, appeared to be dependent on the clathrin-coated vesicle formation.