Article Text


Picornavirus uncoating
  1. M S Smyth1,
  2. J H Martin2
  1. 1Molecular Structure Solutions, MA Block, Wulfruna Street, Wolverhampton WV1 1SB, UK
  2. 2Division of Biomedical Sciences, School of Applied Sciences, University of Wolverhampton, Wolverhampton, WV1 1SB, UK
  1. Correspondence to:
 Dr M S Smyth, Molecular Structure Solutions, MA Block, Wulfruna Street, Wolverhampton WV1 1SB, UK;


Recently, much has been learned about the molecular mechanisms involved in the pathogenesis of picornaviruses. This has been accelerated by the solving of the crystal structures of many members of this virus family. However, one stage of the virus life cycle remains poorly understood: uncoating. How do these simple but efficient pathogens protect their RNA genomes with a stable protein shell and yet manage to uncoat this genome at precisely the right time during infection? The purpose of this article is to review the current state of knowledge and the most recent theories that attempt to answer this question. The review is based extensively on structural data but also makes reference to the wealth of biochemical information on the topic.

Statistics from

The family Picornaviridae is comprised of small, icosahedral, non-enveloped viruses with a single positive strand RNA genome. Its members are among the smallest known human and animal pathogens. The family is classified into six genera: the human rhinoviruses (HRVs), which are the causative agents of many common cold infections; the enteroviruses, which include poliovirus and bovine enterovirus (BEV); the aphthoviruses, which include foot and mouth disease virus (FMDV); the cardioviruses, including Mengo virus; the hepatoviruses, which include the causative agent of hepatitis A; and the parechoviruses, which are responsible for a wide range of human infections.


The picornavirus capsid, an approximately spherical structure of diameter 30 nm, is composed of 60 copies of each of four structural proteins VP1 to VP41 in icosahedral symmetry.2,3 The crystal structures of diverse representatives of the family have been solved.4,5 The fundamental capsid architecture is the same in all members. In each case, VP1, VP2, and VP3 are in the approximate range 240 to 290 residues, each taking the form of eight stranded antiparallel β sheet structures with a “jelly roll” topology. These proteins form the outer surface of the capsid (fig 1). The capsid shell varies in thickness but averages around 5 nm. In the case of enteroviruses and rhinoviruses, VP1 contains a cavity, or pocket, lined with hydrophobic residues and accessible from a depression on the outer surface of the virus capsid.2,6,7 VP4 is much shorter, around 70 residues, and is almost totally lacking in secondary structure. This protein lies across the inner surface of the capsid with its N-terminus close to the icosahedral fivefold axis and its C-terminal close to the threefold axis. The N-terminal residue of VP4 in all picornaviruses is covalently bonded to a myristic acid group,8 giving the capsid five symmetry related myristoyl moieties around the inner surface of the icosahedral fivefold axis. At this point, a channel runs through the capsid connecting the inner and outer surfaces.

Figure 1

Structure of bovine enterovirus (BEV). The colour scheme is: VP1, blue; VP2, green; VP3, red; and VP4, yellow. Only the main chain folding pattern is shown for clarity.

“VP1, VP2, and VP3 are in the approximate range 240 to 290 residues, each taking the form of eight stranded antiparallel β sheet structures with a jelly roll topology”

It is the seemingly minor structural variations on this theme that result in the diverse range of infectious agents, with their equally diverse modes of uncoating, known as the family Picornaviridae.


The first stage of picornavirus infection of susceptible cells is mediated by the interaction of the viral capsid with specific receptors on the cell membrane. No structurally distinct virus capsid features that are common to all members of the picornavirus family have been implicated in this process. However, it is known that the receptor binding sites in the human rhinoviruses and certain members of the enteroviruses lie in a deep cleft at the base of a depression, or “canyon”, which surrounds each fivefold axis.2 The residues in this cleft are both immunologically secluded and highly conserved within virus groups. Although this may be true for rhinoviruses9 and enteroviruses such as poliovirus,6 some members of this genus, such as coxsackievirus A9 (CAV9)7 and bovine enterovirus,3 do not possess a canyon, merely depressions at analogous positions on the capsid. This is caused by extended loops from the outer capsid proteins filling the canyon at various points. In the case of CAV9 there is a C-terminal extension to VP1 (of 15 residues), which contains an arginine–glycine–aspartate (RGD) motif10 that is known to bind integrin molecules. In BEV there is an insertion relative to other enteroviruses of five residues in VP3, which forms a loop that has been implicated in receptor binding. The BEV capsid proteins do not contain the RGD tripeptide, indicating that this virus uses yet another attachment mechanism. Aphthoviruses, which lack any appreciable surface depressions, do have an RGD motif in VP1, but in this case it is located in a long flexible surface loop,4 which is shared across all seven FMDV serotypes. Although the biochemistry of this attachment mechanism is apparently similar to that of CAV9, the FMDV case is unique among picornaviruses in that the attachment site is not only immunologically exposed but forms part of the major neutralising epitope.11

This attachment phase is the determinant of tissue tropism. In spite of the fact that atomic resolution structures have been solved for representatives of four of the six genera, there is comparatively little known about the mechanism of receptor binding. To date, the only structural information is for a human rhinovirus complexed with a fragment of its receptor, intracellular cell adhesion molecule 1.12 Cellular receptors for coxsackievirus and FMDV are known and all belong to the integrin family. The coxsackieviruses can use two different receptors, coxsackie and adenovirus receptor and decay accelerating factor,13–15 and the receptor for FMDV is the integrin αVβ3.16 The BEV receptor has not yet been identified, but the virus can readily adapt to many varied mammalian cells in vitro (MS Smyth, unpublished data, 1990). This may imply that the receptor present on susceptible bovine cells has close homologues in these other cell lines, or that the virus can use a second receptor binding site. A second site has been shown to be used by CAV9 because the RGD motif is lost when the VP1 C-terminal extension is removed by trypsin, yet the virus retains infectivity.17 Hence, the issue of receptor binding is a complex one, even in these apparently simple viruses.


The second phase, uncoating, remains even more poorly understood. The viral capsid must be stable enough for the virus to be transmitted between hosts while protecting the fragile RNA genome but must, at the appropriate signal, dissociate or alter in conformation sufficiently to allow genome release.

None of the electron density maps from the various picornavirus crystallographic data show an ordered structure for the genome, except for perhaps at most one or two bases stacked beneath tryptophan residues on the inner surface of the capsid.13,18–20 However, this is not necessarily evidence that there is true randomness in the packaging of the genome: if each virus particle in a crystal had its RNA molecule packaged in an identical conformation to the others, then because of the capsid symmetry there would be 12 possible orientations for those RNA molecules, thus giving apparent overall disorder. Even a crude calculation shows the genome packing density to be in the region of 1000 mg/ml. It seems unlikely that any biological system could efficiently achieve this with no ordered structure.

“ The myristate covalently linked to the VP4 N-terminus, which lies close to the inner opening of the fivefold channel, may interact with the host cell membrane, causing VP4 and the genome to leave the capsid through the channel”

It has been known for some time that the loss of VP4 is an early step in the uncoating process.21,22 Uncoating studies with poliovirus and HeLa cell membranes showed that the loss of VP4 was accompanied by a 10–20% reduction in S value, loss of infectivity, retention of a functional genome in an RNase resistant state, and sensitivity to proteolytic enzymes and detergents.23,24 The genome then appears to be injected into the cytoplasm from acidic endosomes in which the virus is internalised.25,26 Agents such as chloroquinine, monensin, tributylin, or metabolic inhibitors that raise endosomal pH, dissipate proton gradients, inhibit acidification, or deplete ATP inhibit entry of the genome into the cytoplasm. However, because VP4 is known to be completely internal and when isolated is completely unreactive with susceptible cells, the mechanism of its release is unclear. One possibility is that the myristate covalently linked to the VP4 N-terminus, and which lies close to the inner opening of the fivefold channel, somehow interacts with the host cell membrane, causing VP4 and the genome to leave the capsid through the channel.27 For this to occur, the capsid must first be destabilised because the diameter of the channel is insufficient to allow either the myristate–VP4 complex to leave or for some component of the membrane to enter. Information that also pre-dates any of the virus crystal structures is that the capsid undergoes a dramatic antigenic alteration just before uncoating.28 Loss of VP4 can be induced by acid conditions or, in the case of HRV16, by treatment of the virus with a preparation of its receptor. Associated with this loss of VP4 is the transition of the N-terminus of VP1 from its position on the interior of the capsid surrounding the bottom of the channel at the fivefold axis to an external location via the channel. It has been shown that mutations in VP1 of poliovirus can inhibit uncoating: the removal of residues 8 and 9 affects only RNA release, whereas removal of residues 1 to 4 also affects encapsidation.29 Virus particles that have lost VP4 have a buoyant density of 135S; they are termed A particles and are non-infectious. These A particles also have the N-terminus of VP1 externalised.30 Normal infectious particles sediment at 160S.


A channel exists at each of the fivefold axes that connects the outer surface of the capsid with the inner surface. The shape and dimensions of the channels of the picornaviruses are very similar. Furthermore, there is frequently a metal ion binding site in the channel. In the rhinoviruses, coxsackieviruses, and ϕX174 this is close to the outer surface, although in BEV it is deep in the channel and in CAV9 there are apparently five ions in the channel.

In BEV, the channel is a β cylinder approximately 1 nm in diameter, which is plugged at two places: by five symmetry related tyrosine side chains and beneath this by a metal ion (fig 2). This results in BEV having a lower caesium chloride buoyant density than its nearest phylogenic neighbours.31 An analogous channel exists at the fivefold axes of apparently diverse viruses, such as DNA bacteriophage ϕX174,32 the RNA nodaviruses,33 and tetraviruses.34

Figure 2

Electron density within the fivefold channel of bovine enterovirus (BEV). A discrete node of non-protein density consistent with a metal ion can be seen blocking the channel at this point.

The proposal that this channel is where the RNA exits from the capsid is consistent with the known exit mechanism of the DNA from phage ϕX174, where even the structure of the channel is similar.32 It has also been proposed that the channel is used for the exit of the RNA genomes of the tetraviruses34 and the nodaviruses.33

The N-terminal region of VP1, which becomes exposed just before uncoating, is located at the inner surface of the channel, but is not visible in most of the crystal structures. However, in the case of CAV9, the crystal structure of which was determined in complex with the antiviral compound WIN 51711,7 these residues are visible and their position is consistent with the hypothesis that the VP1 N-terminus exits via the channel. These residues are in an extended conformation in CAV9; however, in HRV16 they adopt a helix structure.20 In addition, five symmetry related helices exist in an analogous position in the tetraviruses.

The five symmetry related N-termini of VP3 are also located near the inner opening of the channel. It is possible that the myristoyl moieties of VP4 may stabilise the VP3 β barrel, which is known to be essential for assembly of some picornavirus capsids.30

Therefore it has been proposed that the first event in uncoating is the interaction of the myristoyl groups of VP4, through the channel, with the host cell membrane. This event would lead to the loss of VP4 (which is consistent with biochemical data), followed by the exposure, again through the channel, of the VP1 N-terminus. Together, these events somehow prepare the channel to serve as a port of exit for the RNA. To gain further insight into these mechanisms, many studies have been carried out using inhibitors of uncoating.


All of the polioviruses,18,35,36 coxsackieviruses,7,13 BEVs,3 and rhinoviruses2,37,38 whose structures have been solved have a hydrophobic pocket contained completely within VP1, and which is accessible through a pore on the virus surface. Many of the structures show that the pocket is occupied by a natural fatty acid “pocket factor”. This is thought to be obtained upon release of the virus from the previous host cell. Studies involving the incubation of viruses with an excess of pocket factor molecules have shown that their presence has an inhibitory but reversible effect on uncoating, suggesting that a dynamic equilibrium exists between full and empty pocket states. Ismail-Cassim et al showed that the uncoating of BEVs was inhibited by a range of saturated fatty acids varying in chain length in the range C12 to C15, whereas the attachment to cells was unaffected.39 These fatty acids have no effect on the uncoating of poliovirus or rhinovirus. Synthetic antiviral compounds, such as WIN 51711,40 which act by the inhibition of uncoating have been shown to bind tightly within this pocket, after displacement of the natural pocket factor. Taken together, these observations suggest that the pocket factor mediates the stability of the capsid and its release is a necessary prerequisite to uncoating.

In several enterovirus and rhinovirus crystal structures solved to date, the pocket within VP1 contains a node of electron density apparently commensurate with a fatty acid. These molecules are held in place by hydrogen bonds from their carboxyl oxygens to VP1 residues, which in the case of BEV are an arginine side chain and a main chain nitrogen.3 At the same end of the pocket as this head group is the pocket pore, which is constricted in the case of BEV with three bulky side chains: arginine, phenylalanine, and tyrosine. The closed end of the pocket is lined by hydrophobic residues. In BEV the pocket factor was modelled as myristic acid (fig 3). This refined well against the x ray crystallographic data, suggesting a high occupancy of all 60 pockets in the virion. It also agrees with the observation that the uncoating of BEV can be reversibly inhibited in the presence of a range of excess fatty acids. However, in BEV the electron density appears to fade rather than end discretely at the aliphatic end of the molecule. This suggests either that the aliphatic end of the fatty acid is less well ordered than the hydrogen bonded carboxyl end, or that the pockets are occupied by a range of fatty acids differing in length as above.

Figure 3

The hydrophobic pocket within VP1 of the bovine enterovirus (BEV) crystal structure contains electron density (blue) which has been modelled as myristic acid (white).

“The pocket factor mediates the stability of the capsid and its release is a necessary prerequisite to uncoating”

A range of antiviral compounds known as the Winthrop (WIN) compounds41 have been shown to inhibit the attachment of some picornaviruses to their cellular receptors. Structural studies have shown that the WIN compounds occupy the VP1 pocket, displacing and binding more tightly than the natural pocket factor. The crystallographic technique often used for such studies involves soaking a crystal of the virus in a solution of the antiviral compound. The fact that this technique is possible is further evidence for the dynamic equilibrium theory. In addition, it is often observed that the virus crystal disintegrates during the soaking process. This indicates that the entry of the antiviral compound into the pocket brings about structural changes that are manifested on the capsid surface, and which are sufficient to disrupt the fragile virion–virion interactions responsible for the crystal lattice.

Take home messages

  • Picornaviruses are small, icosahedral, non-enveloped viruses with a single positive strand RNA genome

  • The structure of all picornaviruses is similar, being composed of 60 copies each of four viral proteins, VP1–4, with a channel at each of the fivefold axes, which connects the inner and outer surfaces and has a myristoyl group close to the inner opening

  • The different picornaviruses attach to specific receptors on the host cell membrane by various methods

  • It is thought that the first event in uncoating is the interaction of the myristoyl groups of VP4, through the channel, with the host cell membrane, which leads to the loss of VP4, and is followed by the exposure, again through the channel, of the VP1 N-terminus

  • Together, these events somehow prepare the channel to serve as a port of exit for the RNA

  • The genome then appears to be injected into the cytoplasm from the acidic endosomes in which the virus is internalised

  • The polioviruses, coxsackieviruses, bovine enteroviruses, and rhinoviruses all contain a hydrophobic pocket in VP1 that is occupied by a natural fatty acid “pocket factor”, which is thought to mediate the stability of the capsid and which is released before uncoating

  • No pocket is present in the aphthoviruses, which may use a different method for uncoating

How does the occupation of the pocket affect events at the fivefold channel? Evidence from structural studies of rhinovirus in which the pockets are occupied with antiviral compounds indicate that various conformational changes occur in the vicinity of the channel in the presence of the antivirals.20 Residues at the N-terminus of VP3 that surround the interior of the channel become more rigid. Furthermore, the ion binding site in the channel becomes more occupied and the residues that coordinate with the ion also become more rigid. Evidence from the structure of CAV9 in the presence of an antiviral compound suggests that the VP1 N-terminus also becomes more ordered under these conditions. These observations are consistent with the necessity of flexibility for successful uncoating to take place. Additional evidence that the channel is the route of departure of the RNA comes from the acid treatment of HRV14, which induces uncoating: under these conditions, residues surrounding the ion binding site become more disordered.


Members of the aphthovirus genus, which includes the much studied FMDVs, have similar overall capsid architecture to the other picornavirus family members; however, important differences exist. The capsid shell has a smoother outer surface, lacking canyons or deep depressions, and there is no hydrophobic pocket within VP1.4 In addition, the channel at the icosahedral fivefold axis of FMDV is less constricted than that found in the rhinoviruses or enteroviruses, and is large enough in diameter to admit caesium ions, which accounts for FMDV having the highest buoyant density among all picornaviruses.42 The particles are highly acid labile, displaying a high concentration of histidine residues at the pentamer interfaces: the pK value of this residue being consistent with the pH of capsid dissociation. It is possible that, following internalisation, these viruses make sole use of pH dependent separation of the pentamers for uncoating.43 Indeed, it has been shown that the removal of calcium ions from the morphologically similar tomato bushy stunt virus has the same effect.44 This being the case, it is possible that structures such as the fivefold channel and the myristoyl moieties at the N-termini of VP4 are redundant, evidence that the aphthoviruses evolved from an ancestor common to the rhinoviruses and enteroviruses.

View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.