Introduction

Poliovirus is undoubtedly the best-known enterovirus and one of the most studied viruses. Since its identification in 1908, poliovirus has been intensively investigated to understand its life cycle and to control its spread and the debilitating disease it causes — poliomyelitis. The finding that poliovirus can be propagated in cultured, non-neural human cells not only led to the development of polio vaccines but also hugely contributed to the establishment of molecular virology as a new field. In fact, many important breakthroughs and concepts in molecular virology stem from studies on poliovirus. Poliovirus was the first animal RNA virus for which the complete genome sequence was determined1 and for which a reverse genetics system was established2. Furthermore, poliovirus3 and rhinovirus4 were the first animal RNA viruses for which the three-dimensional structure was solved by X-ray crystallography.

The genus Enterovirus of the Picornaviridae family consists of 13 species, of which seven contain human pathogens. Besides three poliovirus serotypes, these seven species contain more than 250 other types (Box 1). These non-polio enteroviruses (NPEVs) include many important pathogens, such as coxsackieviruses, echoviruses, numbered enteroviruses and rhinoviruses. Enteroviruses spread either through the faecal–oral route or via respiratory transmission. From their primary sites of replication in the gastrointestinal or respiratory tract, they can disseminate and infect other tissues and organs, including the central nervous system. Although most infections go unnoticed, NPEVs can cause a wide range of disorders with varying presentation and severity, most often in infants, young children and immunocompromised individuals. Coxsackieviruses, echoviruses and numbered enteroviruses are the main viral cause of aseptic meningitis. Furthermore, they can cause neonatal sepsis-like disease, encephalitis, acute flaccid paralysis (AFP), non-specific febrile illness, hand-foot-and-mouth disease (HFMD), herpangina, pleurodynia, pericarditis and myocarditis5. Coxsackieviruses B and echoviruses are also implicated as environmental factors in the aetiology of type 1 diabetes by persistently infecting and causing inflammation of pancreatic β-cells6. Rhinoviruses cause the common cold but can also trigger severe respiratory tract disease and exacerbations of asthma and chronic obstructive pulmonary disease (COPD)7. Despite their name and transmission route, enteroviruses are not associated with gastrointestinal illnesses. Apart from two inactivated enterovirus (EV)-A71 vaccines that were recently marketed in China, there are currently no effective measures to prevent or treat NPEV infections.

Some specific NPEVs have emerged as serious public health threats, particularly EV-A71 and EV-D68 (Box 2). EV-A71 is a major cause of HFMD, a disease that is usually mild and self-limiting, but serious and life-threatening neurological complications, such as brainstem encephalitis, meningitis and poliomyelitis-like paralysis, may occur. In the past 2 decades, large outbreaks of EV-A71 have occurred in Southeast Asia8,9. EV-D68 is an atypical enterovirus that replicates in the respiratory tract. EV-D68 has long been considered a rare pathogen, but it is now being increasingly detected during respiratory disease outbreaks worldwide. EV-D68 usually causes mild respiratory disease but can also cause severe bronchiolitis or pneumonia. In 2014, a nationwide outbreak of EV-D68 in the United States was associated with severe respiratory disease and a cluster of AFP and cranial nerve dysfunction in children10,11. Another well-known NPEV is coxsackievirus A24 variant, the major cause of large outbreaks and pandemics of acute haemorrhagic conjunctivitis, which affects millions of people5.

In this Review, we discuss our current understanding of enterovirus structure, host–receptor interactions, mechanisms of uncoating and the recent discovery of a universal enterovirus host factor that is involved in viral genome release. Moreover, we briefly explain the mechanism of viral genome replication and assembly and describe potential targets for antiviral therapy. We reflect on how these recent discoveries may help the development of antiviral therapies and vaccines.

The enterovirus life cycle

Enteroviruses are non-enveloped viruses of ~30 nm in size with a single-stranded RNA genome of positive polarity (+) RNA. The life cycle of all enteroviruses (Fig. 1) starts with binding to one or multiple cell surface receptors, resulting in receptor-mediated endocytosis. Enteroviruses can use different endocytic routes, depending on the serotype and cell type. Receptor binding and/or pH changes in the endosomal system induce virus uncoating, which is the release of the viral genome from the capsid into the cytoplasm via a pore in the endosomal membrane. Although different enteroviruses may use different receptors and entry pathways, many post-entry steps are highly conserved. After delivery to the cytosol, the viral RNA is translated into a single large polyprotein, which is proteolytically processed by viral proteinases 2Apro, 3Cpro and 3CDpro into ten proteins (capsid proteins VP0, VP1 and VP3 and replication proteins 2A–2 C and 3A–3D) and some stable and functional cleavage intermediates. Enterovirus genome replication takes place on virus-induced membrane structures termed replication organelles (ROs)12. Genome replication by the RNA-dependent RNA polymerase 3Dpol starts with the synthesis of a negative-strand copy of the incoming viral genome to generate a double-stranded RNA replication intermediate. The negative strand serves as a template for the synthesis of new positive strands. Newly synthesized viral RNAs either serve as a template for further translation and replication or are encapsidated into new virions. Enterovirus particles form by assembly of the structural capsid proteins VP0, VP1 and VP3 into protomers and pentamers. Together with a nascent viral RNA, pentamers form the provirion — a process that is intimately coupled with viral RNA replication. Finally, RNA-induced processing of VP0 into VP2 and VP4 yields mature virions.

Fig. 1: Schematic overview of the enterovirus life cycle.
figure 1

After binding to its receptor, or receptors, and endocytic uptake in the cell, the virion delivers its positive-strand (+) RNA genome across the endosomal membrane into the cytoplasm. The viral genome is covalently linked to the viral protein VPg (3B), which is required as a primer for replication. Genome translation yields a single polyprotein that is proteolytically cleaved into replication proteins (2A–2C and 3A–3D) and capsid proteins (VP0, VP1 and VP3). Genome replication by the viral RNA-dependent RNA polymerase (3Dpol) starts with synthesis of a negative-strand (−) RNA that serves as a template for synthesis of new (+) RNA molecules. Replication takes place on membranous replication organelles, in which a favourable lipid environment is created by viral proteins 2BC and 3A, aided by the host proteins acyl-CoA-binding domain-containing protein (ACBD3), phosphatidylinositol 4-kinase-β (PI4KB) (which synthesizes phosphatidylinositol-4-phosphate (PtdIns4P)) and oxysterol-binding protein (OSBP) (which recruits cholesterol). Nascent (+) RNA molecules can either enter a new round of replication or be packaged into progeny virions. Capsid proteins self-organize into protomers and pentamers and, in concert with replication machinery and genomic RNA, assemble into provirions that are converted into infectious, mature virions upon the genome-induced cleavage of VP0 into VP4 and VP2. Mature virions exit the host cell via non-lytic release of extracellular vesicles or via cell lysis in a late stage of infection. dsRNA, double-stranded RNA.

Besides cleaving the viral polyprotein, viral proteinases 2Apro and 3Cpro cleave several host proteins to optimize virus translation, replication and virus spread and to suppress antiviral cellular responses. For example, they induce host shut-off (by cleaving cellular factors involved in transcription and cap-dependent translation of host mRNAs), disrupt nucleocytoplasmic transport (by cleaving nucleoporins), interrupt type I interferon and stress pathways (by cleaving RNA sensors and signalling proteins) and disrupt the cellular cytoskeleton (reviewed in refs13,14,15). Collectively, these changes induce dramatic changes in cellular morphology (that is, a cytopathic effect) and culminate in death and lysis of the cell (reviewed in ref.16). Although enteroviruses are described in textbooks as obligate lytic viruses that are released from ruptured cells, evidence is now accumulating that enteroviruses can also egress before cell lysis in membrane-bound structures that can accommodate multiple virions (reviewed in ref.17).

Targeting the viral life cycle

There is a great need for broad-range antiviral drugs to treat NPEV infections. Such drugs may also be instrumental in the global poliovirus eradication initiative to treat so-called chronic shedders18. Several enterovirus inhibitors have been identified and some have been tested in clinical trials, mainly to treat rhinovirus infections, but all have failed owing to lack of efficacy or toxicity issues19 (Table 1). Generally, antiviral compounds can target either viral proteins or essential host factors. Drugs targeting viral proteins usually have limited side effects, whereas the advantage of drugs that target host factors is that viruses are less likely to develop resistance. For many years, drug discovery has relied heavily on screens for antiviral activity in cell-based infectivity assays using large compound libraries, followed by stepwise chemical optimization. An advanced understanding of the structure and function of enterovirus proteins and essential host factors will aid broad-range antiviral drug development by allowing the implementation of dedicated high-throughput in vitro assays and the structure-guided development and optimization of drug candidates. Drug repurposing (that is, the use of compounds that have been developed to treat another condition) is an emerging strategy that can speed up antiviral drug development because safety and pharmacokinetic studies have already been performed. In the following sections, we discuss the various steps of the enterovirus life cycle and highlight examples of inhibitors at each stage.

Table 1 Selection of pharmacological inhibitors of the enterovirus life cycle in humans

Enterovirus structure

Rhinovirus 14 and poliovirus were the first animal RNA viruses for which three-dimensional structures were resolved3,4. Since then, numerous high-resolution structures have been determined, revealing that enterovirus particles share a similar architecture (Fig. 2a). Enterovirus particles are constructed of 60 repeating protomers, with each protomer consisting of the four structural proteins VP1, VP2, VP3 and VP4. Together they form the icosahedral shell with a pseudo T = 3 arrangement that encapsidates the viral genome. The small protein VP4 is myristoylated and located on the inside of the virion, whereas the surface of the particle is composed of subunits of VP1, VP2 and VP3, which each adopt the typical eight-stranded, antiparallel β-barrel fold. The surface of the capsid has several highly variable loops, which are readily accessible to the host immune system and account for the high antigenic diversity of enteroviruses. Most enteroviruses have a deep, circular surface depression or canyon encircling each fivefold axis of symmetry that frequently serves as the receptor binding site20.

Fig. 2: Picornavirus structure.
figure 2

a | Schematic representation of a picornavirus particle, showing the asymmetric unit (outlined in blue), the different symmetry axes (red) and the location of the canyon (green). Sixty protomers consisting of the surface proteins VP1, VP2 and VP3 and the internal VP4 constitute the capsid shell. b | Comparison of capsid surfaces between different picornaviruses belonging to the genus Enterovirus (poliovirus, enterovirus A71, enterovirus D68 and rhinovirus C15a); the genus Hepatovirus (hepatitis A virus); and the genus Kobuvirus (Aichi virus). Surfaces are coloured according to their distance from the centre of the virion, as shown in the colour bar. RCSB Protein Databank entries for the virion structures in part b: poliovirus (PDB ID 2PLV), enterovirus A71 (PDB ID 3VBS), enterovirus D68 (PDB ID 4WM8), rhinovirus C15a (PDB ID 5K0U), hepatitis A virus (PDB ID 5WTE) and Aichi virus (PDB ID 5GKA).

Despite these similarities, some enteroviruses display unique structural features. The capsid of rhinovirus C (RV-C) contains protrusions on its surface, contrasting with most enterovirus particle surfaces, which are smoother and spherical21 (Fig. 2b). As a consequence, RV-C particles have shallow, narrow and non-continuous canyons, much resembling those of EV-D68 (ref.22). The biological consequences of such atypical canyons are yet to be determined. The floor of the canyon of all enteroviruses (with the exception of RV-C21) harbours a small, hydrophobic ‘pocket’ that is ordinarily filled with a lipid moiety or pocket factor, which is involved in regulating particle stability. Although not much structural variation exists within the Enterovirus genus, a high number of unique and distinctive structural features can be found among other picornavirus genera. For example, the Aichi virus contains poly-l-proline type II helices on its capsid23, whereas hepatitis A virus (HAV) is more angular, and its surface is devoid of any canyon or depression24. Similarly, Ljungan virus lacks a depression on its capsid surface but instead possesses marked protrusions that arise from the fivefold axes25. Owing to developments in structural biology (high-resolution cryo-electron microscopy (cryo-EM) and high-speed fixed-target X-ray crystallography at X-ray free-electron lasers)26, it is likely that the number of high-resolution structures will increase in the near future and enhance our understanding of the structural diversity within the picornavirus family.

One of the earliest explored strategies to inhibit enterovirus infections is to target the viral capsid proteins using so-called capsid binders. These molecules occupy the hydrophobic pocket and displace the pocket factor, thereby increasing particle stability and preventing receptor binding and/or genome release. The first series of capsid binders, known as the WIN compounds27, were chemically optimized to improve their antiviral and drug-like properties, which eventually resulted in the development of pleconaril. Several clinical trials have been performed with pleconaril and other classes of capsid binders; however, insufficient clinical benefits were observed. Additionally, the rapid emergence of drug-resistant viruses and the unusual structure of RV-C, with a canyon that lacks a pocket that can be targeted by capsid binders22, pose obstacles for the development of broad-range capsid binders.

Enterovirus entry receptors

One of the key determinants of viral tropism and pathogenesis is the availability of a specific cell surface receptor that can be used for viral attachment and entry. In 1989, poliovirus receptor (also known as CD155) (ref.28) and intercellular adhesion molecule 1 (ICAM1) (ref.29) were the first enterovirus receptors that were identified for poliovirus and major rhinoviruses, respectively. Since then, many NPEV receptors have been identified, of which the majority belong to the cell surface immunoglobulin-like or integrin receptor family (Table 2). The observation that enteroviruses can bind to a large repertoire of different receptors provides a likely explanation for the diverse spectrum of diseases that are caused by these viruses.

Table 2 Enterovirus receptors

In this Review, we categorize receptors as either uncoating or attachment receptors. Uncoating receptors induce conformational changes in the viral capsid that mediate the uncoating process, whereas attachment receptors facilitate cell attachment and may promote virus uptake. Although uncoating receptors are usually proteins, it was recently shown that EV-D68 binds sialylated glycans via its canyon and that these cause the conformational changes that initiate uncoating30. This finding represents the first example of an enterovirus that can bind to carbohydrates to trigger uncoating. For influenza viruses, tropism is determined by the distribution of different sialic acids on the respiratory epithelium. Whether a similar correlation exists between the expression pattern of different sialic acids and EV-D68 tropism still needs to be explored. In addition, ICAM5, which is enriched in telencephalic grey matter, was identified as a second uncoating receptor, possibly explaining why EV-D68 is occasionally associated with AFP31. The reason why this virus engages two uncoating receptors is currently unknown. Besides uncoating receptors, attachment receptors can also be of importance for infection. For instance, binding to the attachment receptor complement decay-accelerating factor (DAF; also known as CD55) by group B coxsackieviruses activates signalling events in polarized epithelial cells that liberate the uncoating receptor (coxsackievirus–adenovirus receptor; CAR) from tight junctions, making CAR accessible to the virus32. Moreover, attachment factors can be determinants of viral tropism, which was highlighted by a recent study on coxsackievirus A24 variant, in which it was shown that adaptation to the attachment receptor sialic acid was associated with the emergence of acute haemorrhagic conjunctivitis pandemics33.

For EV-A71, two different receptors have been identified: scavenger receptor class B member 2 (SCARB2)34 and P-selectin glycoprotein ligand 1 (PSGL1)35. SCARB2 was shown to be the receptor of all EV-A71 strains studied thus far34,36, whereas PSGL1 was found to be the receptor for only some strains (EV-A71-PSGL1-binder (EV-A71-PB))35. In contrast to SCARB2, which was shown to induce uncoating36, it is still unclear how PSGL1 supports EV-A71-PB infection. Autopsies of fatal EV-A71 infection cases revealed viral antigens in the brainstem and spinal cord, but no specific cell type that is targeted by the virus has been identified thus far37. SCARB2 is expressed on neurons and glial cells38, and transgenic mice expressing human SCARB2 develop similar pathological features as humans39, suggesting that SCARB2 is the main determinant for the development of neurological disease. PSGL1 is primarily expressed on leukocytes, although expression in human neuronal tissue was also reported38. Despite this information, the role of PSGL1 in EV-A71-induced disease remains unclear.

Receptor identification is not only useful to better understand viral pathogenicity but also to develop suitable cell and/or animal model systems to study the biology of a virus and to identify antiviral drugs. This is exemplified by RV-C, a major cause of severe respiratory tract infections (Box 2). Despite its discovery more than 10 years ago40, our knowledge of this virus was greatly hampered owing to its inability to grow in standard cell culture systems41. The recent discovery of the respiratory epithelium-expressed cadherin-related family member 3 (CDHR3) as the protein receptor and the generation of a CDHR3-overexpressing cell line that supports RV-C infection can thus be considered important milestones in RV-C research42. In differentiated human bronchial epithelial cell cultures, RV-C was found to preferentially infect CDHR3-expressing ciliated cells, providing the first insight into RV-C tropism43. Intriguingly, a coding SNP in CDHR3 that was previously associated with an increased risk of severe childhood asthma exacerbation was shown to enhance RV-C infection in cell culture by increasing CDHR3 expression at the cell surface44.

Although many NPEV receptors have been identified, the interplay between different receptors and the exact mechanism by which these receptors mediate infection are still poorly understood. Further studies are necessary to improve our understanding of virus tropism and pathogenesis and to define new targets for intervention. Several strategies targeting virus–receptor interactions have been reported, including the use of soluble receptor fragments45 and antibodies that bind to receptors or to receptor-binding epitopes on the viral capsid46. However, these strategies will be limited to a narrow range of viruses that share a common receptor.

Enterovirus uncoating

Uncoating stages

The entry mechanism by which enveloped viruses fuse with the cellular membrane is relatively well understood; however, it is less clear how enteroviruses deliver their genome into the host cell. Although enteroviruses utilize a highly diverse set of receptors and endocytic routes, the uncoating mechanism of poliovirus and NPEVs follows a number of conserved steps, which gradually destabilize the virus particle and lead to genome release at the correct time and location. Early studies on poliovirus identified three distinct particle types that occur during the entry process (Fig. 3a) on the basis of differences in their sedimentation coefficients and antigenic properties: the 160S, 135S and 80S particles47. The 160S particle is the native, infectious form of the virion. The 135S particle (altered particle, or A-particle) is a destabilized, infectious particle type that is formed in vitro, but it remains unclear whether this intermediate also exists during the uncoating process within infected cells. The 80S particle is the empty virion, which has released its genome and is a non-infectious end product of the uncoating process. The structure of the native particle is dynamic and undergoes ‘breathing’, which is the recurrent expansion and shrinkage of the virion, accompanied by transient exposure of VP4 and the N-terminal extension of VP1 (ref.48). Cellular uncoating cues (that is, receptor binding or low endosomal pH) lock the virion in its expanded state, forming the A-particle. Unlike the native particle, this uncoating intermediate is hydrophobic owing to the irreversible externalization of the VP1 N terminus, which contains an amphipathic helix that allows the virion to interact with membranes49 (Fig. 3b). The A-particle is no longer bound to its receptor, lacks the internal protein VP4 (ref.50) and has a ~4% increased radius compared with the native virion47, probably to allow rearrangement of the genomic RNA in preparation of its release.

Fig. 3: The enterovirus uncoating mechanism.
figure 3

a | Particle types observed during uncoating of enteroviruses in vitro, with their respective sedimentation coefficients between brackets. Uncoating cues (receptor binding or low pH) induce changes that convert the native virion into the altered particle (A-particle). These changes include capsid expansion, externalization of the capsid protein VP1 N terminus and release of myristoyl–VP4. The subsequent release of the RNA genome yields the empty particle. b | During cell entry, the first destabilizing event after receptor binding is the release of the pocket factor (a lipid that stabilizes the virus particle) from the virion. Subsequently, the externalized VP1 N terminus anchors the virion to the endosomal membrane, and myristoyl–VP4 forms hexameric membrane pores. The viral RNA genome is translocated through these pores into the cytoplasm, aided by the host protein phospholipase A2 group XVI (PLA2G16) via an unknown mechanism.

Uncoating cues

Although all enteroviruses undergo similar structural changes during uncoating, the cues driving these destabilizing events are diverse. For most enteroviruses, the first uncoating cue encountered during cell entry is receptor binding. These uncoating receptors bind in the canyon, where they engage the GH loop of VP1, causing collapse of the hydrophobic pocket and expulsion of the pocket factor51. While proteinaceous uncoating receptors directly engage the VP1 GH loop, a recent study describing the crystal structure of EV-D68 in complex with its sialic acid receptor30 revealed that sialic acid indirectly displaces the GH loop via a series of long range conformational changes. Although this unconventional mode of receptor binding triggered pocket factor expulsion, sialic acid did not induce A-particle formation, suggesting that additional uncoating cues are required to complete the uncoating process. In addition to receptor binding, some enteroviruses rely on endosomal acidification as a cue to initiate uncoating. As a low pH is restricted to specific cellular compartments, this chemical cue allows a virus to control the timing and site of genome release. Whereas endosomal acidification is the only known uncoating cue for minor-group rhinoviruses52, some other enteroviruses, like EV-A71 (ref.53), require simultaneous engagement with their receptor and a low pH to fully uncoat.

Pore formation

Numerous structural studies that revealed enterovirus uncoating intermediates have provided extensive information about the molecular events that occur between receptor binding and A-particle formation, but the subsequent process of RNA release is still poorly understood. Because enteroviruses lack an external membrane, the viral genome must be translocated across a membrane to reach the cytoplasm. Early studies on poliovirus genome delivery showed that virions interact with lipid membranes and induce the formation of ion-permeable channels54, a capacity that was found to be abrogated by mutations in VP4 (ref.55). Indeed, recombinant myristoylated VP4 of RV-16 was shown to independently induce membrane permeability56 by forming multimeric pores57. These pores have a channel size consistent with the size that is required for translocation of the single-stranded viral RNA genome57. Whether the virion is directly associated with this transmembrane channel during genome release has been unclear for a long time. Yet, an electron tomography study of poliovirus bound to receptor-decorated liposomes revealed approximately 50 Å-long ‘umbilical’ connectors linking the virion to the membrane58. It was proposed that these connectors represent genomic RNA and viral proteins, which might shield the RNA during translocation. Consistent with this observation, a recent study showed that the poliovirus genomic RNA is protected from degradation by co-endocytosed RNase during the entry process59, supporting a scenario in which a discrete channel connects the virion to the host cell membrane.

Genome release

Other major questions in studies of viral uncoating concern the exit site and orientation of the viral RNA during egress from the virus particle. So far, our understanding of the directionality of RNA release is based solely on a study of RV-A2, which revealed that the 3' end of the genome exits the virion before the 5' end60. The site of RNA release from the virion was originally proposed to be located at the fivefold axis61, but cryo-EM structures of uncoating intermediates (A-particles or empty particles) revealed holes in the capsid at the twofold axis and near the quasi-threefold axis51,62. The opening at the quasi-threefold axis was found to serve as a channel for externalization of the VP1 N terminus63. Indications of the site of RNA release are based on asymmetric cryo-EM reconstructions of heated poliovirus particles that are in the process of genome release64. Although the resolution was low (~50 Å), these structures implicated a region spanning the twofold and quasi-threefold axes. A limitation of most cell-free uncoating studies performed to date is that viral uncoating is triggered by numerous cues (for example, heat, acid or soluble receptor molecules) that act on the entire virion (that is, global stimulation), whereas uncoating during a natural infection is probably asymmetrical, because only one side of the virion interacts directly with the membrane. In a recent study, asymmetric coxsackievirus B3 uncoating was induced with receptors embedded in lipid bilayer nanodiscs, generating the first high-resolution structures of an enterovirus uncoating intermediate attached to a membrane65. Unlike particles formed upon global stimulation, this structure revealed an opening at the threefold axis. Together, these findings show that more studies employing lipid membranes, as well as high-resolution cryo-EM, are required to determine the exact site of RNA exit from the virion.

The pan-enterovirus host factor PLA2G16

Although receptor binding is the only known uncoating cue for many enteroviruses, genome release does not take place immediately after receptor binding on the cell surface but after a period of internalization into the cell66. For that reason, several studies have suggested the existence of an additional unknown uncoating cue that controls the moment at which the virion releases its genome65,67. A plausible candidate for this role is the host factor phospholipase A2 group XVI (PLA2G16), a lipid-modifying enzyme of which the physiological function remains poorly understood. PLA2G16 was identified as a universal enterovirus host factor that facilitates the displacement of viral genomes from virus-containing endocytic vesicles68. It is likely that PLA2G16 is involved in the formation, expansion or maintenance of the membrane pore, either by directly interacting with viral proteins or RNA, or indirectly, by creating a hospitable lipid environment that favours pore formation. For instance, lipids modified by PLA2G16 may affect membrane fluidity or interact with pore components, thereby facilitating their insertion into the endosomal membrane. Interestingly, some (non-Enterovirus) picornaviruses possess a non-structural 2A protein with homology to PLA2G16. Although these 2A proteins have never been found in association with virions, the possibility exists that these proteins perform a similar function in virus entry as PLA2G16. Analogously, most parvoviruses have a capsid protein that contains a phospholipase A2 domain, which is thought to modify the membrane of endocytic vesicles to allow viral genome release into the cytoplasm69. These findings suggest that various non-enveloped viruses have independently evolved to employ lipid-modifying enzymes for virus entry via mechanisms that are not yet fully understood.

Targeting the uncoating stage of enterovirus infection could be a promising strategy in the development of antiviral drugs against NPEVs. In addition to pocket factor analogues, which stabilize the virion, molecules that prevent genome translocation by targeting viral proteins or host factors that are involved in pore formation might be promising candidates. The pan-enterovirus host factor PLA2G16 could be an excellent drug target, because all enteroviruses tested to date require this factor and because the activity of PLA2G16 is dispensable for the survival of mice68. Because no selective PLA2G16 inhibitors have been reported thus far, developing specific in vitro assays to screen for inhibitors of its phospholipase activity could be a promising strategy to develop broadly acting anti-enterovirus drugs.

Translation and genome replication

Following its delivery to the cytosol, the viral genome is directly translated by ribosomes, yielding capsid proteins and viral proteins that mediate genome replication. Translation is mediated by an internal ribosome entry site (IRES) and requires a subset of the canonical translation machinery as well as several IRES trans-acting factors (ITAFs) (reviewed in refs70,71). The mechanism of RNA replication is well conserved among all enteroviruses (reviewed in ref.72). The core enzyme of the replication machinery is the viral RNA-dependent RNA polymerase 3Dpol. Replication is initiated at higher-order RNA elements in the 5′ and 3′ UTRs and requires a primer, the viral peptide 3B (also known as VPg) coupled to two uridines. VPg uridylylation requires a structural RNA element located in the coding region of the genome. 3Dpol operates in conjunction with several viral proteins, including 3CDpro and 3AB, which together bind to structural RNA elements involved in priming, and 2C, which has ATPase-dependent helicase and ATPase-independent RNA chaperone activities. Besides viral proteins, several host RNA-binding proteins are involved in genome replication, including poly(rC)-binding protein 2 (PCBP2), polyadenylate-binding protein 1 (PABP1) and heterogeneous nuclear ribonucleoprotein C (HNRNPC).

Replication occurs on virus-induced, tubulovesicular ROs, which are thought to originate from endoplasmic reticulum (ER) and/or Golgi apparatus membranes. RO formation requires the viral membrane-associated proteins 2BC and 3A73 and a specific set of host factors, but the exact composition and mechanism of formation of ROs remains unknown. 3A interacts with acyl-CoA-binding domain-containing protein 3 (ACBD3) to recruit the lipid kinase phosphatidylinositol 4-kinase-β (PI4KB) to membranes, leading to an accumulation of phosphatidylinositol-4-phosphate (PtdIns4P) lipids. PtdIns4P attracts oxysterol-binding protein (OSBP) to form membrane contact sites between the ER and ROs and mediate a PtdIns4P-driven accumulation of cholesterol at ROs (reviewed in refs12,74). Increased levels of PtdIns4P and/or cholesterol are important for RO formation and efficient genome replication, possibly by facilitating proper processing of the viral polyprotein75,76,77,78. 3A also binds Golgi-specific brefeldin A-resistance guanine nucleotide exchange factor 1 (GBF1), but its exact role in replication is unclear. The roles of 2BC, 2B and 2C in membrane reorganization are less well understood. It has been suggested that 2BC contributes to PI4KB recruitment79. 2B has viroporin activity80,81, but its exact function in RO formation is unknown. 2C binds reticulon 3, an ER protein that promotes membrane curvature, but the importance of this interaction for RO formation has not been investigated82. Some host factors appear to be important for a subgroup of enteroviruses. For example, valosin-containing protein (VCP) was identified as a host factor required for poliovirus replication but not coxsackievirus B3 replication83. Of note, several other host proteins have been implicated in enterovirus replication, but their roles in replication are often poorly understood. Additionally, enteroviruses trigger activation of autophagy. Although autophagosomes have been suggested to have a role in genome replication84, it is currently thought that virus-induced autophagosomes are primarily involved in non-lytic virion release (see below).

The mechanisms of genome translation and replication are highly conserved between enteroviruses and are therefore attractive targets for the development of broadly acting antiviral drugs, which can target viral enzymes, specific host factors or essential cellular processes (Table 1). Several inhibitors targeting viral enzymes 3Cpro and 3Dpol have been described. 3Cpro inhibitors include peptidomimetics that target its active site, such as rupintrivir. 3Dpol inhibitors include nucleoside analogues (for example, the anticancer drug gemcitabine (Gemzar)85 and non-nucleoside inhibitors (for example, GPC-N114 (ref.86)), which target a conserved RNA template channel in 3Dpol). Additionally, a number of 2C inhibitors have been reported, but their mode of action is poorly understood. Among them is the antidepressant drug fluoxetine (Prozac)87,88, which inhibits enterovirus B and enterovirus D members. Fluoxetine was successfully used to treat an immunocompromised child with chronic enteroviral encephalitis89, underscoring the potential of drug repurposing in antiviral therapy. The recently elucidated structure of EV-A71 2C90 will likely encourage soaking or co-crystallization studies and provide more insight into the mode of action of 2C inhibitors, allowing optimization through structure-guided drug design. In addition, inhibitors of host factors, including compounds targeting PI4KB and OSBP, inhibit a broad range of enteroviruses. Unfortunately, PI4KB inhibition was found to be associated with adverse side effects91,92. OSBP inhibitors include the antifungal drug itraconazole (Sporanox)93, which was shown to have antiviral activity in mice94.

Virion assembly and release

Assembly

The enterovirus life cycle ends with the formation of new infectious progeny, which is a highly complex, stepwise process that is poorly understood95. In a first step, heat shock protein 90 (HSP90) associates with the myristoylated capsid precursor P1, catalysing the proteolytic processing by 3CDpro to yield VP0, VP1 and VP3 (ref.96). This yields a protomeric particle that subsequently self-oligomerizes to form pentameric particles. These pentamers are stabilized by their myristate moieties97,98 and, for some enteroviruses, by association with glutathione99,100. The current model is that in a next step, pentameric particles condense around replicating viral RNA to form the provirion. Concurrently with encapsidation and as a last quality control step, the viral RNA mediates VP0 cleavage into VP2 and VP4, generating the mature virus particle101.

Although many RNA viruses encode at least one RNA packaging signal to provide specificity and precision during viral assembly, packaging signals have not been identified thus far in the genomes of enteroviruses. For enteroviruses, it is currently believed that encapsidation specificity is governed by direct protein–protein interaction between 2C, as part of the replication complex, and the viral capsid protein VP3 without an apparent involvement of an RNA packaging signal102. Intriguingly, parechovirus 1 and foot-and-mouth disease virus (FMDV) were recently shown to contain multiple short regions of ordered RNA structures that bind capsid proteins, dispersed throughout their genomes103,104. Whether enterovirus or other picornavirus genomes also contain such structures needs to be investigated.

Enterovirus assembly can be inhibited by geldanamycin105, an HSP90 inhibitor, but this inhibitor has cytotoxic effects and has mainly been investigated for its antitumour activity. Glutathione levels can be manipulated in several ways, but this is unlikely to be an effective antiviral strategy because only a subset of enteroviruses depends on glutathione for efficient assembly99,100,106.

Virion release

Enteroviruses are typically considered cytolytic viruses that kill and lyse their host cell for release. However, several studies have revealed that enteroviruses can also be released in a non-lytic manner in vesicles that mediate en bloc transmission of virions17,107,108,109,110. The current model is that enteroviruses trigger autophagy and upregulate the formation of double-membrane autophagosomes that engulf clusters of virions84. Fusion of these autophagosomes with lysosomes is inhibited by the virus to prevent degradation of their contents. Instead, fusion of the outer membrane of virion-containing, double-membrane autophagosomes with the plasma membrane releases single-membrane vesicular structures filled with virions to the extracellular milieu. The membranes of these extracellular vesicles are enriched in the lipid phosphatidylserine, which assists uptake in cells, likely via phosphatidylserine receptors109. Nevertheless, the bona fide virus receptor is still needed for infection, implying that at some point, virions are released from the extracellular vesicles109. Non-lytic release may be a much more widely adopted transmission strategy of picornaviruses than previously thought and has also been implicated for HAV, which evades neutralizing antibodies by cloaking itself in exosome-like structures111,112. Thus, non-lytic release may enhance infection efficiency, but its exact role in enterovirus dissemination and pathogenesis remains to be explored. The seemingly widespread role of autophagy in virion release by enteroviruses and other picornaviruses makes this a potential broad-range antiviral target. However, given the central role of autophagy in cellular and organismal homeostasis, it seems unlikely that autophagy inhibitors are sufficiently safe for clinical application.

Advances in enterovirus vaccine development

Besides antiviral drugs, vaccines are of great importance in combating enterovirus infections. Both inactivated poliovirus vaccines (IPV) and live, attenuated oral poliovirus vaccines (OPV) have been instrumental in preventing poliomyelitis. The major advantages of OPV are that less virus is needed to induce protective immunity and that OPV limits virus spreading by inducing mucosal immunity. However, a major drawback is that OPV can revert and/or recombine with closely related Enterovirus C members, leading to vaccine-associated paralytic poliomyelitis and circulating vaccine-derived poliovirus113. Currently, safe and effective OPVs are being developed, involving various attenuating traits, including destabilizing mutations in the IRES114, mutations in 3Dpol that improve proofreading activity115,116 and lower RNA recombination capacity117, and synonymous mutations in the capsid-coding region that alter codon pair bias118.

Theoretically, inactivated or live, attenuated vaccines can be developed against any NPEV, as exemplified by the EV-A71 vaccine that is currently marketed in China9. Development of a polyvalent, broad-spectrum vaccine against many different enterovirus serotypes was considered unfeasible for a long time. However, recently it was shown that combining 50 inactivated rhinovirus types into a single vaccine elicited neutralizing antibodies against 49 types, suggesting that universal, polyvalent, inactivated enterovirus vaccines will be developed119. Another promising approach is the use of virus-like particles (VLPs)120, which assemble upon co-expression of capsid proteins and 3CDpro, allowing the manufacture of vaccines of viruses that are difficult to culture, for example, Rhinovirus C members or viruses that are restricted to high-containment facilities (poliovirus in the post-eradication era). Enterovirus VLPs can differ slightly from native virions in their structure, antigenic properties and stability, owing to the absence of a stabilizing genome. However, this problem can be circumvented by introducing specific stabilizing capsid mutations121. Recently, stabilized poliovirus VLPs were efficiently produced in plants — which represent a cheap, safe and high-level expression system — and immunization of mice with these VLPs provided protection against a poliovirus challenge122. In conclusion, several strategies can be used to develop potent vaccines against NPEVs.

Conclusions and future perspectives

The large group of NPEVs contains many human pathogens that cause a wide spectrum of illnesses, including severe neurological and respiratory diseases. In this Review, we described the current knowledge of the different stages of the life cycle of enteroviruses, most of which is derived from studies on poliovirus. Although these studies have provided valuable insights into the enterovirus life cycle, many important questions need to be addressed. For instance, future studies should elucidate the mechanism of viral genome translocation across the endosomal membrane and the contribution of PLA2G16 to this process; clarify the early post-entry events and identify where the viral RNA is first translated and replicated; and establish the exact origin and function of ROs. In addition, how the viral RNA is encapsidated and which viral and cellular factors are involved in this process, how the viral RNA is organized inside the virion, and how virus-containing extracellular vesicles are formed and what their physiological function is are all important questions that need to be addressed. Recently, we have seen major advances in technologies that may be instrumental in solving these questions. Several new microscopy techniques are becoming increasingly available, for example, three-dimensional scanning electron microscopy approaches that allow the reconstruction of entire cells and the visualization of virus-induced intracellular changes at nanoscale resolutions. Moreover, in situ cryo-electron tomography, an imaging modality that allows structural analysis of macromolecular complexes in their physiological microenvironment, may be applied to gain more insight into viral replication complexes and their association with ROs. CRISPR–Cas9 gene knockout screens can be used to identify essential host factors, and nuclease-inactive Cas9 can be used to bind specific viral RNA sequences, allowing RNA tracking in living cells. Novel advances in mass spectrometry, including phospho-proteomics and crosslink-proteomics, lipidomics and metabolomics, may shed new light on our understanding of virus-induced changes in host cell biology and signalling that are essential for virus replication and/or propagation. Future NPEV research should focus on receptor usage and tropism, pathogenesis, transmission dynamics, epidemiology and evolution. Evolution studies are also needed to better understand how poliovirus evolved from EV-C coxsackievirus123, which is an essential question in view of the polio eradication campaign.

Intensified efforts should be undertaken to develop preventive and therapeutic options to control NPEV infections. New insights into the viral life cycle, combined with rapid developments in (computational) drug discovery, are expected to spur the development of novel, broad-range enterovirus inhibitors. The successful development of highly effective antiviral drugs against another (+) RNA virus — hepatitis C virus — implies that it should be possible to develop antiviral drugs against enteroviruses, provided that sufficient investments are made. Besides treating NPEV infection, broad-range drugs may also contribute to the eradication of poliovirus. Research on this devastating human pathogen has laid important foundations for the field of molecular virology. Although poliovirus may soon be eradicated, fundamental research on NPEVs is likely to have far-reaching implications that go beyond the field of picornaviruses.