key: cord-325230-3kg4oe4g
authors: Agol, Vadim I.; Gmyl, Anatoly P.
title: Viral security proteins: counteracting host defences
date: 2010-11-09
journal: Nat Rev Microbiol
DOI: 10.1038/nrmicro2452
sha: 
doc_id: 325230
cord_uid: 3kg4oe4g

Interactions with host defences are key aspects of viral infection. Various viral proteins perform counter-defensive functions, but a distinct class, called security proteins, is dedicated specifically to counteracting host defences. Here, the properties of the picornavirus security proteins L and 2A are discussed. These proteins have well-defined positions in the viral polyprotein, flanking the capsid precursor, but they are structurally and biochemically unrelated. Here, we consider the impact of these two proteins, as well as that of a third security protein, L(*), on viral reproduction, pathogenicity and evolution. The concept of security proteins could serve as a paradigm for the dedicated counter-defensive proteins of other viruses. SUPPLEMENTARY INFORMATION: The online version of this article (doi:10.1038/nrmicro2452) contains supplementary material, which is available to authorized users.

All bona fide RNA viruses encode at least two proteins, a capsid protein and an RNA-dependent RNA polymerase (RdRP). In this article, we do not consider 'abnormal' RNA viruses such as hepatitis delta virus, which does not possess its own replication machinery and borrows a capsid from another virus 1 , or the capsid-less narnaviruses, which encode only an RdRP 2 ; hepatitis delta virus and narnaviruses are closer to viroids and plasmids, respectively, than to fully fledged RNA viruses. RNA viruses encoding only a capsid and an RdRP are known to infect protists such as Giardia spp. 3 . Some RNA phages and fungal RNA viruses express only three proteins. However, most RNA viruses do not rely on such a scant protein repertoire. The proteomes of the majority of RNA viruses comprise up to 12 different proteins, with the genomes of some coronaviruses encoding nearly 30 proteins.

To infect, viruses must reach an appropriate intracellular environment and, if necessary, adapt this environment to their own requirements. Thus, viral infection requires not only replication but also interactions with host defences. To carry out these tasks, RNA viruses have only a few proteins at their disposal, with the available protein arsenal being limited by the genome size. As the proteomes of these viruses are strikingly small and specific viral functions generally require more than one protein, most proteins encoded by RNA viruses are multi functional. At the same time, however, there is a certain division of labour between viral proteins. We consider some aspects of this problem using, as an example, the picornaviruses, a large family of small animal viruses with a medium-sized, single-stranded, positive-sense RNA genome (BOX 1) . Although they share a common genome organization, these viruses exhibit sufficient genetic variation to be separated into at least 12 distinct genera (BOX 2) . The family includes important human and animal pathogens that cause a range of disorders, including poliomyelitis, foot-and-mouth disease, the common cold, gastroenteritis, hepatitis, meningitis, myocarditis and uveitis. In addition to acute diseases, these viruses can also cause chronic persistent disorders.

Of the 12 'mature' (fully processed) picornaviral proteins (BOX 1) , the most ancient group includes the capsid proteins and a set of conserved proteins considered to be picornavirus signature proteins 4 , comprising the RdRP (3D pol ), the primer for RNA synthesis (VPg), a protease (3C pro ) and an ATPase (2C ATPase ). The capsid and signature proteins are indispensable for viral viability. Two less conserved proteins, 2B and 3A, are also essential for viability but are less directly involved in viral reproduction. The main functions of these two proteins are to target replicative proteins to the correct destinations and aid in the creation of a suitable replicative niche.

Finally, two other non-structural proteins that flank the capsid precursor in the polyprotein molecule, the leader (L) and 2A proteins, constitute a distinct group, although they have no common structural or biochemical features. They are the most variable among the picornavirus proteins, and some viruses even lack L altogether (FIG. 1; TABLES 1, 2) . This group also includes the so-called L* protein, which is encoded in a different reading frame of the RNA of certain cardioviruses. Functionally, these proteins (L, L* and 2A) have been characterized in some detail for only a few picornaviruses, and in these cases their major function has been shown to be counteracting host defences, thereby ensuring optimal conditions for viral reproduction. In this Review we argue that picornaviral L and 2A proteins that have yet to be characterized are likely to fulfil the same biological role.

We have proposed that this group of proteins should be called security proteins 5 . They are sometimes also referred to as virulence factors 6 ; such a designation, although it may be more familiar, is somewhat less preferable to us for the reasons explained below. Viruses have no special 'desire' to be virulent, that is, to harm, even less to kill, their hosts. The pathogenic properties of a virus are not a prerequisite for viral fitness. In fact, the most severe harm in a viral infection comes not from viral reproduction but from the (sometimes miscalculated) host defence response. Host damaging or even suicidal defensive reactions include RNA degradation, inhibition of translation, and the induction of endoplasmic reticulum stress, apoptosis, autophagy and inflammation, and these reactions are aimed at limiting viral reproduction and spread. As a natural response, viruses have evolved tools directed not only at overcoming the specific innate and adaptive immune responses but also at inhibiting the general host metabolic functions on which these specific defences are based -that is, processes such as transcription and translation, cell signalling and intracellular trafficking. Inhibiting these processes is the major function of the security proteins. Accordingly, the capacity to withstand host defences, rather than virulence (the ability to harm the host) per se, is the property that is selected for during viral evolution. The long-term co-evolution of a virus and its host will probably result in their mutual adaptation accompanied by a decrease in viral virulence. Thus, in our opinion, the term 'security protein' is a better reflection of the evolutionary origin of the relevant proteins than the term 'virulence factor' . moreover, the possession of efficient security proteins does not necessarily make a virus particularly virulent.

The aim of this Review is to consider the properties and biological significance of picornavirus security proteins and, on the basis of this knowledge, to put forward a general view of security proteins as dedicated counterdefensive proteins that evolved primarily to overcome various mechanisms of host resistance.

Both L and 2A are extremely heterogeneous with respect to size, sequence and biochemical properties (FIG. 1; TABLES 1,2). The length of L varies from ~70 amino acid residues in cardioviruses to ~450 residues in some sapeloviruses, and the variation is striking even in a single genus; for example, in sapeloviruses, the length varies from ~80 residues (one L protein) to ~450 residues (two L proteins). Although certain sapeloviruses seem to express two L proteins (Supplementary information S1 (figure)), approximately half the picorna virus genera lack L altogether. The length of 2A varies ~30-fold (from 9 residues in senecaviruses to ~300 residues in avihepatoviruses and some sapeloviruses), and it can comprise one to three separate polypeptides. The L proteins of cardioviruses and klasseviruses are strongly acidic and markedly basic, respectively, whereas the 2A proteins of these viruses exhibit the reverse ratio of acidity. The overall charge of the proteins will influence the choice of their interaction partners. The primary structures of L and 2A are, in most cases, totally unrelated when compared across different genera, and even intra-genus conservation among seemingly homologous proteins can be low, as is the case, for example, with the L proteins of kobuviruses and 2A proteins of cardioviruses and sapeloviruses.

With regard to their biochemical activities, only a few security proteins have been characterized in any detail. 2A of enteroviruses (2A pro ) is a protease 7 with a chymotrypsin-like fold in which the catalytic serine residue has been replaced by cysteine 8 . 2A pro performs an essential function in viral reproduction through its involvement in the so-called 'primary' co-translational cis-cleavage of the polyprotein between the amino-terminal amino acid of 2A pro and the carboxy-terminal residue of the capsid protein VP1. Aphthovirus L is also a protease (L pro ) with a papain-like fold 9 . Translation of aphthovirus RNA can start at two in-frame AuG codons, generating distinct L proteins, Lab (~200 residues) and Lb (~170 residues) 10 , with Lb predominating in vivo. L pro cleaves the bond at the L pro -VP4 boundary 11, 12 . erbovirus L pro exhibits similar

Picornaviruses possess a single-stranded positive-sense RNA genome (see the figure) comprising approximately 7,000-9,000 nucleotides. The RNA is packaged into an icosahedral capsid usually composed of four distinct proteins (VP1-VP4), but in some viruses VP4 and VP2 are fused (forming VP0). After productive contacts with specific cellular receptors, which differ between viruses, the genome is uncoated and enters the cytoplasm, where the main steps of viral reproduction occur. The viral RNA, which contains a single ORF (with one exception described in the main text), is translated in a cap-independent manner into a 2,200-2,500 amino acid polypeptide, which is eventually processed by limited proteolysis into a dozen 'mature' proteins. These proteins include: capsid proteins; an RNA-dependent RNA polymerase (3D pol ); a protein (VPg, or 3B) that serves as a primer for the initiation of RNA synthesis; an ATPase with a conserved superfamily 3 helicase motif (2C ATPase ) and an essential but poorly defined role in viral RNA replication; a chymotrypsin-like protease (3C pro ), which, as the mature protein or as a precursor, is a major factor in polyprotein processing; two hydrophobic membrane-binding proteins (2B and 3A) that participate in the generation of a virus-friendly environment; and, flanking the precursor of the capsid proteins, one or two highly variable proteins (L and 2A), the structure and functions of which are the subject of this Review. The entire coding region of the RNA is surrounded with untranslated regions (UTRs) harbouring regulatory elements that control viral translation (for example, the internal ribosome entry site (IRES)) and replication (the origins oriL and oriR), and this entire RNA sequence is flanked by the covalently linked viral protein VPg and a poly(A) tract at the 5′ and 3′ termini, respectively. There are four structurally and functionally distinct types of IRES 131 and many types of oriL and oriR.

The mode of translation for which initiation is dependent on the presence of the so-called cap structure (m 7 GpppN) at the 5′ end of an mRNA. Specific cap-binding proteins (translation initiation factors) recruit the ribosome to the 5′ end of the mRNA, and the ribosome scans the template until it encounters the initiation codon. This translation mode is exploited by most mRNAs in eukaryotic cells.

A protein that binds to the 3′ poly(A) tail of eukaryotic mRNAs on the one hand and to cap-dependent translation initiation factors on the other hand. This dual binding results in a non-covalent circularization of the mRNA template, which is accompanied by a significant increase in translation efficiency. enzymatic activity 13 . Owing to the presence of a characteristic motif, the 2A proteins of some sapeloviruses have also been proposed to be proteases 14 .

2A of aphthoviruses is a short peptide that promotes co-translational interruption of polyprotein synthesis just downstream of its own coding sequence (and before the 2B moiety); cleavage at the aphthovirus VP1-2A boundary is accomplished by 3C pro . Interruption of polyprotein synthesis (also known as ribosome skipping) is proposed to be caused by an interaction between the 2A peptide and the exit tunnel of the ribosome, preventing the interaction of the ribosome with Pro-tRNA (proline is the amino-terminal residue of aphthovirus 2B) 15, 16 . To achieve this, a short peptide harbouring an NPG(P) motif (the parentheses mark the interruption site) is sufficient. Short 2A peptides from erboviruses 17 , teschoviruses 18 , senecaviruses 19 and cosaviruses 20 also possess NPG(P) motifs and seem to be involved in polyprotein processing at the 2A-2B border. Separation of these peptides from the VP1 capsid protein has been assumed but not yet clearly demonstrated. As these peptides possess no other known activities, their assignment as security proteins is not warranted; we refer to them here as 2A sp (2A short peptide).

The ribosome-skipping NPG(P) motif is also present in some 'composite' 2A proteins, such as those of Ljungan virus 21 (a parechovirus), avihepatoviruses 22 and seal picornavirus type 1 (REF. 23 ), in which they seem to ensure self-separation from the downstream 2A moieties. In cardiovirus 2A, the NPG(P) motif is located at the C terminus and interrupts polyprotein synthesis at the 2A-2B boundary 24 , which corresponds to the polyprotein 'primary' cleavage site in this genus 25 .

Apart from the NPG(P)-containing peptides, 2A proteins of cardioviruses, avihepatoviruses, Ljungan virus and seal picornavirus type 1 contain distinct but poorly characterized sequences. The cardiovirus encephalomyocarditis virus (emCV) encodes a 2A protein with RNAbinding affinity 26 . The bipartite 2A of Ljungan virus, the tripartite 2A of avihepatoviruses and the NPG(P)lacking 2A proteins of kobuviruses, tremoviruses and some parechoviruses all have a ~140-150-residue H-NC domain, which contains a histidine with a downstream asparagine-cysteine dipeptide and a putative transmembrane domain 21, 27 . The observation that a similar H-NC domain is present in certain cellular tumour suppressors 21 suggests that these viral proteins are also involved in the control of host activities. Between the NPG(P) and H-NC motifs, 2A of the avihepatovirus duck hepatitis A virus possesses an additional moiety that contains the so-called AIG1 domain 22 , which is also found in representatives of the Ras-like GTPase superfamily.

Cardiovirus L proteins 28 , which are devoid of any enzymatic activity and exhibit noticeable intra-genus variability (TABLE 1) , contain a non-classical but functional zinc finger (Cys-His-Cys-Cys) motif 29,30 and a downstream acidic motif 31 that, in some of these L proteins, contains potential phosphorylation sites 31, 32 . Certain strains of Theiler's murine encephalomyelitis virus (TmeV) are unique among picornaviruses in expressing a functional protein, L*, that is encoded in an alternative translation frame 33,34 that starts within the L coding sequence, goes through the VP4 coding sequences and terminates in the VP2 coding sequence of the main reading frame. The 2A and L proteins of other picornaviruses neither contain easily identifiable amino acid motifs nor have known specific biochemical activities.

Effects on general host metabolism Security proteins contribute substantially to the shut-off of host macromolecular synthesis that occurs in response to infection with many picornaviruses. One could perhaps argue that such effects of security proteins contradict the key proposal of this Review that these proteins are dedicated to counter-defensive functions. However, as already mentioned, inflicting harm on their hosts does not bring viruses any benefits per se. The only reason (or at least, the main reason) why viruses evolved the ability to damage infected cells is their need to incapacitate the cellular defensive machinery. This machinery includes several specific mechanisms (such as innate and adaptive immunity), the implementation of which requires general cellular functions such as translation, transcription and controllable nucleocytoplasmic trafficking. Therefore, virus-induced impairment of these all-purpose metabolic functions can be regarded as a component of the viral counter-defensive strategy.

The effect of security proteins on cap-dependent translation of cellular mRNA is particularly important. 2A pro from diverse enteroviruses [35] [36] [37] [38] and L pro from aphthoviruses 12,39 cleave eukaryotic translation initiation factor eIF4G. Interestingly, erbovirus L pro does not seem to cleave eIF4G and does not trigger translational shutoff 13 . Poly(A)-binding protein 1 (PABP1; also known as

Members of the Picornaviridae family are small (~30 nm), non-enveloped animal viruses with similar genome organizations and mechanisms of reproduction (BOX 1) . This family contains important human and animal pathogens, studies of which have made crucial contributions to our understanding of the molecular biology of viruses and the mechanisms of virus-cell interactions. The first animal virus (foot-and-mouth disease virus; FMDV) and the first human virus (poliovirus) to be discovered belong to this family.

The current official classification of picornaviruses (see the International Committee on Taxonomy of Viruses) includes 12 genera: Aphthovirus (for example, FMDV), Avihepatovirus (for example, duck hepatitis A virus), Cardiovirus (including important model viruses such as encephalomyocarditis virus, mengovirus and Theiler's murine encephalomyelitis virus, as well as human and animal pathogenic viruses such as the newly recognized Saffold virus), Enterovirus (a large genus that includes poliovirus and many other human and animal pathogenic viruses, such as coxsackieviruses, echoviruses and other 'numbered' enteroviruses, as well as rhinoviruses, the causative agents of the common cold), Erbovirus (equine rhinitis B virus), Hepatovirus (hepatitis A virus), Kobuvirus (for example, Aichi virus, which induces gastroenteritis in humans, as well as some animal pathogens), Parechovirus (the causative agents of various human diseases, particularly in children; also isolated from rodents), Sapelovirus (including an avian, a porcine and a simian pathogen), Senecavirus (Seneca Valley virus, a candidate oncolytic agent), Teschovirus (a porcine pathogen) and Tremovirus (avian encephalomyelitis virus). In addition, there are two candidate genera, not yet officially approved, that are associated with human diarrhoea: Cosavirus and Klassevirus. Finally, seal picornavirus type 1 is a distinct virus not assigned to any of the above genera. Several new picornavirus genera have been identified recently, and it seems plausible that this number will grow, particularly with the breakthroughs in sequencing techniques and the achievements of metagenomics. cytoplasmic PABP), a host protein that is also involved in translational control, is another target of enterovirus 2A pro (REFS 40, 41) .

Security proteins can inhibit host translation by mechanisms other than proteolysis. mutations of cardiovirus 2A (which is not a protease) alleviate virus-induced translational shut-off 42, 43 . It is thought that association of cardiovirus 2A with ribosomes 44, 45 , which seems to take place in the nucleolus 46, 47 and possibly occurs through the RNA-binding activity of 2A 26 , might contribute to preferential use of the internal ribosome entry site (IReS)-dependent viral templates. Cardiovirus L was also reported to mediate translational shut-off 48 . However, this effect could largely be due to L-mediated inhibition of nuclear export of mRNA rather than to inhibition of translation per se 49, 50 . On the basis of ectopic expression experiments, hepatitis A virus 2A has also been implicated in inhibition of cap-dependent translation 51 , but the relevance of this observation is uncertain, as hepatitis A virus does not exert translational shut-off.

The effects of security proteins on cellular transcription are less well studied, although synthesis of the mRNA for cytokines and chemokines is inhibited in certain cases (see below). Individually expressed poliovirus 2A pro was reported to cleave the general transcription factors TATA-box-binding protein 52 and cyclic AmP-responsive element-binding protein 1 (REF. 38 ), but the biological significance of these effects is debatable 52, 53 . Poliovirus 2A pro cleaves GemIN3 (also known as DDX20), a protein that is involved in the formation of spliceosomes 54 . Cardiovirus 2A has also been implicated in virus-triggered inhibition of host transcription 47 , but this effect was not investigated further. Foot-and-mouth disease virus (FmDV) L pro (REFS 55, 56) and human parecho virus 2A 57 accumulate in the nucleus during the course of infection and as a result of ectopic expression, respectively, but their nuclear effects are unknown. enterovirus 2A pro cleaves some cytoskeletal proteins, such as cyto keratin 8 (REF. 58 ) and dystrophin, a protein that connects the cytoskeleton to the plasma membrane 59 .

The security proteins of several picornaviruses profoundly affect nucleocytoplasmic transport in infected cells. The targets of enterovirus 2A pro include nucleoporins, which are nuclear pore components that control nucleocytoplasmic exchange 60, 61 . As a result, bi directional passive diffusion through the nuclear envelope is facilitated. Another manifestation of the perturbation of intracellular trafficking is the suppression of nuclear export of mRNAs, ribosomal RNAs and U spliceosomal small nuclear RNAs 62 . Passive nucleocytoplasmic diffusion of proteins is also facilitated by the cardiovirus L protein 63, 64 , and this effect seems to be caused by L-triggered phosphorylation of nucleoporins 50, 65, 66 .

Effects on innate immunity and viral pathogenicity One major consequence of the effects of security proteins on host metabolism is the downregulation of innate immunity. FmDV L pro suppresses interferon production 56,67 and action 68, 69 , largely through the inhibition of nuclear factor-κB-dependent transcription that is caused by the degradation of the p65 subunit of this transcription factor 55, 56 . The transcription of genes encoding various cytokines and chemokines (including tumour necrosis factor (TNF; also known as TNFa), T cellspecific protein RANTeS (also known as CCL5), myxovirus resistance protein 1 and interferon regulatory factor 7) is also suppressed by FmDV L pro (REF. 56 ).

Poliovirus reproduction, which is largely resistant to the effects of interferon, becomes interferon sensitive in 2A pro mutants. Introduction of the poliovirus 2A pro gene into interferon-sensitive emCV facilitated its replication in cells pretreated with interferon 70 . The ability of rhinovirus 2A pro to cleave mitochondrial antiviral-signalling protein (mAVS; also known as VISA, CARDIF and IPS1), an intermediate in the interferon generation pathway, might contribute to the insensitivity of these viruses to interferons 71 . 2A pro also cleaves the catalytic subunit of DNA-dependent protein kinase, which, among other activities, is involved in the induction of pro-inflammatory cytokines 72 . Cardiovirus L also suppresses interferon production by affecting the activation of interferon regulatory factor 3, but the exact mechanism of this interference has not yet been identified 32, 50, 73, 74 . TmeV L* was proposed to suppress the antiviral cytotoxic T cell response in TmeV-infected mice 75, 76 .

In line with these observations, the functions of security proteins are less crucial for viral 'well-being' in hosts with innate immunity defects. mutations in cardiovirus L 48,77,78 or 2A 79 ) ), rather than one L protein. For TMEV, an alternative L protein, L*, is encoded in an alternative reading frame that starts within the L coding sequence. ‡ According to the data in GenBank. § Values were calculated using ProtParam 132 . || For the viral genera with more than one species, the level of interspecies amino acid identity (and similarity, in parenthesis) was calculated with the aid of CLUSTAL_X2 alignments, using utilities implemented in BioEdit and the BLOSUM62 similarity matrix 133 . To focus on the core protein sequences, the internal insertions of >15 amino acid residues, as well as terminal insertions, were not taken into account. ¶ The peptidase_C28 motif was revealed by BLAST searches in the NCBI Conserved Domain Database 134 . # The conserved non-canonical zinc (Zn) finger domain (with a CHCC motif) was found to be present.

reproductive potential, but such mutations are less detrimental for viral growth in BHK-21 cells, which are deficient in interferon production, than in immunecompetent cells. Similarly, cardiovirus mutants lacking L grow better in mice that have a deficient interferon system than in wild-type mice 50, 74 . For the L*-expressing TmeV strains, functional L* is important for the ability of the virus to infect macrophages or microglia 80, 81 .

One of the components of the innate immune response is apoptosis, which can potentially limit viral reproduction and spread, although certain viruses can subvert the apoptotic machinery to their benefit. Conflicting results have been reported on the relationship between the security proteins and the apoptotic machinery. ectopic expression of enterovirus 2A pro triggers an apoptotic response 38,82,83 and enhances the sensitivity of cells to the apoptogenic activity of tumour necrosis factor 84 . However, in the context of the whole genome, 2A pro possesses anti-apoptotic activity 85, 86 . Similarly, expression of TmeV L in macrophage-like cells induces apoptosis, as occurs during TmeV infection of these cells 87 , whereas emCV and mengovirus L are anti-apoptotic in the context of the whole virus (at least in HeLa cells, in which the virus itself elicits necrotic death) 5 . Whether this apparent discrepancy is a result of the use of different assays or hosts or of the intrinsic peculiarities of cardiovirus L proteins remains unknown. TmeV L* has also been reported to exhibit anti-apoptotic activity 88 .

Several illuminating examples strongly suggest that security proteins can markedly modulate viral pathogenicity. FmDV lacking L is highly attenuated 89 , 132 . || For the viral genera with more than one species, the level of interspecies amino acid identity (and similarity, in parentheses) was calculated with the aid of CLUSTAL_X2 alignments, using utilities implemented in BioEdit and the BLOSUM62 similarity matrix 134 . To focus on the core protein sequences, the internal insertions of >15 amino acid residues, as well as terminal insertions, were not taken into account. ¶ These are manually recognizable motifs. # One of the strains is reported to harbour an NPR(P) motif instead of NPG(P). and both L 73,90 and L* (REFS 33, 81, 88) 

Notwithstanding the low level of conservation of L proteins between emCV and TmeV (TABLE 1) , these proteins can be functionally exchanged with respect to their ability to inhibit interferon formation and to 'open' nuclear pores 90 . The replacement of L in the full-length cardiovirus genome with FmDV L pro generates a virus that can overcome host defences more efficiently than its leaderless counterpart, although it has lower fitness 92, 93 . Such interchangeability of structurally and biochemically distinct proteins attests to the similarity of their biological functions. It is hardly by chance that viruses that encode long or multiple Ls tend to have a short 2A or lack 2A altogether (assuming that 2A sp is not a security protein), and vice versa (FIG. 1) . For example, simian and porcine sapeloviruses harbour a long 2A and a short L, whereas avian sapelovirus encodes the longest L identified to date and the predicted 2A ORF is very short, if it encodes a protein at all. This tendency is consistent with the notion that L and 2A have similar roles in virus-host interactions. As already mentioned, some picornaviruses, for example, cosaviruses, encode no security proteins. even viruses that do encode security proteins can, under certain circumstances, survive and replicate after these proteins have been inactivated or eliminated. Notably, cardiovirus L 31, 48, 77, 78 and L* (REFS 34, 81, 94) and FmDV L pro (REF. 95) are not essential for viral viability, and extended deletions in 2A of cardioviruses 42, 79 and hepatitis A virus 96, 97 do not kill these viruses. Furthermore, although some data suggest that poliovirus 2A pro has an essential replicative function 98 , recent experiments have demonstrated its dispensability for viral viability 86 .

Although this Review focuses on the counter-defensive functions of security proteins, it should be kept in mind that other picornavirus proteins are also often engaged in similar functions. The targets of 3C pro (or its proteolytically active precursors) might include proteins that are involved in innate immunity [99] [100] [101] [102] . 2B and 3A are also important players in the virus-host struggle, being involved in the rearrangement of cytoplasmic membranes and in suppression of trafficking, secretion and antigen presentation on cellular plasma membranes 84, [103] [104] [105] . 2C can also participate in some of these activities 103 . even specialized proteins such as VPg 106 and capsid proteins 107 can sometimes assist in overcoming host defences. In all these cases, however, the 'security' functions are neither the main nor the conserved roles of these proteins.

Conversely, as well as representing a dedicated counterdefensive system, security proteins can be directly involved in viral reproduction. The products of the cleavage of eIF4G, and possibly of some other host proteins, by FmDV L pro and enterovirus 2A pro can stimulate IReSdependent cell-free translation 108, 109 ; the effect of 2A pro is partly the result of stabilization of the viral RNA 110 . The physiological relevance of these phenomena is unclear, however. The poliovirus IReS dysfunction that is caused by some mutations can be compensated for by mutations in 2A in a host-dependent manner 111 , suggesting the participation of host proteins. Cardiovirus L was also implicated in control of IReS activity, because emCV RNA with deletions in the L-coding sequence exhibited a decrease in cell-free translatability 31 .

Poliovirus 2A pro was reported to stimulate strand initiation of negative-strand RNA, and this was seemingly independent of its effects on RNA stability and translation 110 . Deletion of a carboxy-terminal sequence from this protein does not substantially affect its protease activity but does inhibit RNA replication 112 . Deletion of the amino-terminal region notably, but incompletely, suppresses replication of the relevant replicon 113 . Similarly, deletions in 2A of Aichi virus (a kobuvirus) inhibit viral RNA replication 114 . However, the mechanisms responsible for these effects have not been elucidated, casting doubts on whether these 2A proteins affect viral replication directly or by modulating host cell activities.

A role for hepatitis A virus 2A in virion assembly and maturation has been established 97, 115 . The primary cotranslational cleavage of the viral polyprotein takes place between 2A and 2B and is accomplished by the viral proteinase 3C pro (or its precursor), leaving the 2A sequence fused to the carboxyl terminus of capsid protein VP1 (REFS 116, 117) . In immature virions, VP1 retains this 2A extension, which is eventually cleaved off by an unidentified enzyme 118 . The involvement of 2A in the maturation of some other picornaviruses cannot be excluded therefore. Notably, the primary co-translational scission of the cardiovirus polyprotein generates a fusion between 2A and the capsid protein precursor 25 .

Origins and evolution of security proteins Obviously, it is not possible to construct a genealogical tree of either L or 2A, because they are unrelated. These proteins are not considered at all in the hypothesis on the origin of picorna-like viruses 4 , and only the short, translation-interrupting 2A peptides (that is, the 2A sp peptides) are briefly mentioned in the proposal on picornavirus classification 119 . There is no correlation between the nature (or even the presence) of security proteins and either the type of IReS that lies upstream of the L protein or the RdRP lineage (which is generally accepted as the most reliable indicator of viral relatedness) (FIG. 2) .

The diversity of the security proteins and their absence from many picornaviruses suggest that they are independent and late evolutionary acquisitions 5, 120 . It can be speculated that the most ancient of the 2A molecules are the 2A sp peptides, as hinted by their presence in most picornavirus genera, including those belonging to different lineages (FIG. 2) . Interruption of polyprotein synthesis after translation of the capsid proteins might be advantageous for viral reproduction. It might, for example, facilitate proper protein folding, ensure optimal kinetics of protein synthesis or control the ratios of structural to non-structural proteins, which could theoretically be achieved by incomplete translational re-initiation at the second proline residue of the NPG(P) motif. It is worth noting that there is a difference in the translation factor requirements for translation of the emCV polyprotein upstream and downstream of the 2A-2B boundary 121 . Strikingly, 2A sp peptides -or more accurately, the DXeXNPG(P) motifs (where X is any amino acid) that are characteristic of these peptides -are found in some other picorna-like and unrelated viruses, where they seem to serve the same function [122] [123] [124] . Assuming that the acquisition of NPG(P) was an early event, this motif might have been lost by some parechoviruses (and other picornaviruses) at a certain step of evolution 21 . In contrast to picornaviruses, the acquisition of 2A sp by members of some other families of RNA viruses has been proposed to have occurred at a late stage 122 . The idea that these peptides might have originated in picornaviruses is an attractive hypothesis. The NPG(P) motif can also be found in several cellular proteins, but the current data do not allow researchers to determine whether the recoding ability of this motif was a viral or cellular invention.

The non-NPG(P)-containing regions of the 2A proteins are unrelated acquisitions. Cardiovirus 2A possesses these additional moieties in its amino-terminal region, whereas other viruses of this subset (parechoviruses, 

avihepatoviruses and seal picornavirus type 1) contain these moieties in their carboxy-terminal regions (FIG. 1) . This may suggest that these moieties were acquired after the NPG(P) motif. In avihepatoviruses and some parechoviruses, these newly acquired sequences harbour the H-NC motif, which is also present in the NPG(P)-lacking 2A proteins of other parechoviruses, kobuviruses and tremoviruses (that is, viruses of different RdRP lineages) (FIG. 2) . A plausible hypothesis is that H-NC-containing proteins from cellular organisms were hijacked by certain picornaviruses on several independent occasions, but the possibility that 2A sp peptides were formed by deletions from larger 2A proteins 120 cannot be ruled out.

Taking into account the chymotrypsin-like fold that is found in enterovirus 2A pro , it is reasonable to assume that this protein was derived from picornavirus 3C pro or a cellular protease 125 . A similar assumption could perhaps be made for the putative sapelovirus 2A protease. There are no obvious clues to the possible origins of the 2A proteins of hepatoviruses and klasseviruses or of the non-NPG(P) moieties of the 2A proteins of cardioviruses and seal picornavirus type 1.

With regard to the picornavirus L proteins, one hypothesis is that the papain-like L pro proteases of aphtho viruses and erboviruses originated from cellular enzymes. No obvious relatives of other L proteins can currently be identified among cellular or viral proteins. Only the origin of cardiovirus L* seems certain: the first L* probably came into being accidentally, by translation of an alternative reading frame, and was then shaped by mutations preserving the functional integrity of L, VP4 and VP2. It is worth noting that recently identified human TmeV-like cardioviruses lack this alternative reading frame 126 .

Theoretically, several mechanisms could underlie the acquisition of security proteins. For example, one possibility is that viral genes were duplicated and subsequently substantially modified (such a scenario can be imagined for the origin of enterovirus 2A pro (REF. 125) ). Other scenarios are: recombination with viral or cellular RNAs encoding related proteins such as proteases; conversion of non-coding RNA sequences into coding sequences (which could occur through different mechanisms, such as interspecies recombination 127 (V.I.A., A.P.G., e. V. Khitrina and W. J. melchers, unpublished observations), or introduction or activation of an upstream inframe AuG codon); frame shifting 120 (or double frame shifting, in the case of 2A proteins); and using an alternative reading frame. unfortunately, we can only pinpoint a definite mechanism for the case of L*, for which the last mechanism has obviously been operative.

The acquisition of security proteins, by whatever mechanism, required that the new 'additions' did not interfere with the function of the adjacent viral proteins -VP4 (or VP0) in the case of L, and VP1 and 2B in the case of 2A. The new proteins should either be separated from these neighbours (by self-proteolysis, the action of other viral proteases or translation interruption) or, if they remain fused, they should not impair the functions of these neighbours (as is the case with the hepatitis A virus VP1-2A fusion).

A separate issue is the evolution of the security proteins themselves. It was proposed that the L proteins of TmeV and emCV diverged during evolution to adapt to the different replication fitnesses of these viruses 90 . The data are too scarce, however, for any generalizations at this point.

L and 2A constitute a distinct and remarkable set of picornavirus proteins. They exhibit striking structural and biochemical diversity but (with the exception of the 2A sp peptides) accomplish similar biological functions by counteracting host defensive reactions. However, their abilities to solve similar problems might involve fundamentally different molecular mechanisms. This is illustrated in FIG. 3 , which summarizes the properties of the three best studied security proteins. The biological functions of enterovirus 2A pro and cardiovirus L are strikingly similar: inhibition of host macromolecular synthesis, permeabilization of the nuclear envelope and inhibition of active nucleocytoplasmic transport, suppression of specific innate immunity mechanisms and control of the apoptotic machinery of the host cells. No less striking is the difference in the mechanisms by which the two proteins achieve these goals. Conversely, aphthovirus L pro can solve only a subset of these problems. An intriguing Figure 3 | Major biological functions of the best studied but unrelated security proteins. There is a striking similarity between the functional activities of the enterovirus 2A protease (2A pro ) and the cardiovirus leader protein (L), but the underlying mechanisms by which these functions are carried out are fundamentally different. By contrast, the known functional activities of aphthovirus L, which is a protease (L pro ), seem to be more limited. IRF3, interferon regulatory factor 3; NF-κB, nuclear factor-κB. question is how, and whether, the diverse security functions exhibited by a given protein (FIG. 3) are related to each other. The fact that mutations inactivating one function usually also impair other functions suggests the existence of a common upstream target. most of the picornavirus L and 2A proteins still await researchers' attention. Nevertheless, the data discussed here allow us to provisionally assign security functions even to those L and 2A proteins that have not yet been characterized. Indeed, L and 2A do not definitely belong to the set of essential reproductive proteins that ensure translation and replication of the viral genome (although 2A of hepatitis A virus assists virion maturation).

We propose that the concept of security proteins is of general relevance and can be applied to viruses other than picornaviruses. The hallmarks of these proteins are as follows: structural and biochemical unrelatedness or even absence in related viruses; the dispensability of the entire protein or its functional domains for viral viability; and, for mutated versions of the proteins, fewer detrimental effects on viral reproduction in immune-compromised hosts than in immune-competent hosts. Possessing one of these features would make a viral protein a good candidate security protein, whereas a combination of these features would probably confirm this designation.

Viruses with large DNA genomes possess impressive arsenals of security proteins. The complement of security proteins is much more limited in RNA viruses but is sufficient for their evolutionary success. There are also tentative examples of security proteins in non-picornavirus RNA viruses. Coronaviruses have several so-called accessory proteins, which are neither conserved nor essential and exhibit the capacity to suppress host innate immunity by a range of mechanisms 128 . Some, but not all, flaviviruses use ribosome frame shifting to express a non-essential non-structural protein, NS1′, mutational inactivation of which results in viral attenuation 129 . The NSS protein of the Rift Valley fever phlebovirus suppresses the interferon system, and its inactivation does not kill the virus but attenuates its pathogenicity 130 . This list can readily be extended.

The spectrum of picornavirus-induced diseases is extraordinarily broad. The reasons for this variability are poorly understood, although receptor compatibility and effects on viral protein and RNA synthesis that are caused by differences in the availability of host factors are surely important contributors. However, the interaction between host defences and viral counterdefence is certainly one of the key factors underlying the pathogenicity of picornaviruses and other viruses. This interaction cannot be fully understood without elucidation of the roles of the security proteins. Treatment and prevention of viral diseases may also markedly benefit from such elucidation. The study of security proteins is therefore an underdeveloped but highly promising research area.

RNA replication without RNA-dependent RNA polymerase: surprises from hepatitis delta virus

Persistent yeast single-stranded RNA viruses exist in vivo as genomic RNA·RNA polymerase complexes in 1:1 stoichiometry

Giardiavirus double-stranded RNA genome encodes a capsid polypeptide and a gag-pol-like fusion protein by a translation frameshift

The Big Bang of picorna-like virus evolution antedates the radiation of eukaryotic supergroups

The proposal to consider L and 2A proteins of picornaviruses as a distinct class of counter-defensive 'security proteins

Evading the host immune response: how foot-and-mouth disease virus has become an effective pathogen

A review considering, among other topics, counter-defensive functions of aphthovirus L protein

A second virus-encoded proteinase involved in proteolytic processing of poliovirus polyprotein

Viral cysteine proteases are homologous to the trypsin-like family of serine proteases: structural and functional implications

Putative papain-related thiol proteases of positive-strand RNA viruses. Identification of rubi-and aphthovirus proteases and delineation of a novel conserved domain associated with proteases of rubi-, alpha-and coronaviruses

Structure of the FMDV translation initiation site and of the structural proteins

The identification of the L protein in FMDV

A second protease of foot-and-mouth disease virus

The two species of the foot-and-mouth disease virus leader protein, expressed individually, exhibit the same activities

Conservation of L and 3C proteinase activities across distantly related aphthoviruses

Genomic evidence that simian virus 2 and six other simian picornaviruses represent a new genus in Picornaviridae

Cleavage of foot-and-mouth disease virus polyprotein is mediated by residues located within a 19 amino acid sequence

Site-specific release of nascent chains from ribosomes at a sense codon

Equine rhinovirus serotypes 1 and 2: relationship to each other and to aphthoviruses and cardioviruses

Sequence analysis of a porcine enterovirus serotype 1 isolate: relationships with other picornaviruses

Complete genome sequence analysis of Seneca Valley virus-001, a novel oncolytic picornavirus

A highly prevalent and genetically diversified Picornaviridae genus in South Asian children

Molecular analysis of three Ljungan virus isolates reveals a new, close-to-root lineage of the Picornaviridae with a cluster of two unrelated 2A proteins

Molecular analysis of duck hepatitis virus type 1

A highly divergent picornavirus in a marine mammal

The cleavage activity of aphtho-and cardiovirus 2A proteins

The position of polypeptide G on the encephalomyocarditis virus polyprotein cleavage map

RNA-binding properties of nonstructural polypeptide G of encephalomyocarditis virus

The 2A proteins of three diverse picornaviruses are related to each other and to the H-rev107 family of proteins involved in the control of cell proliferation

The discovery of cardiovirus L protein, the first leader protein to be found for picornaviruses

The leader peptide of Theiler's murine encephalomyelitis virus is a zinc-binding protein

NMR structure of the mengovirus Leader protein zinc-finger domain

Leader protein of encephalomyocarditis virus binds zinc, is phosphorylated during viral infection, and affects the efficiency of genome translation

The mengovirus leader protein suppresses a/β interferon production by inhibition of the iron/ferritin-mediated activation of NF-κB

Alternative translation initiation site in the DA strain of Theiler's murine encephalomyelitis virus

A picornaviral protein synthesized out of frame with the polyprotein plays a key role in a virusinduced immune-mediated demyelinating disease

The demonstration that enterovirus 2A pro cleaves a translation initiation factor

Poliovirus proteinase 2A induces cleavage of eucaryotic initiation factor 4F polypeptide p220

Purification of two picornaviral 2A proteinases: interaction with eIF4γ and influence on translation

Coxsackievirus B3 proteases 2A and 3C induce apoptotic cell death through mitochondrial injury and cleavage of eIF4GI but not DAP5/p97/ NAT1

Leader protein of foot-and-mouth disease virus is required for cleavage of the p220 component of the cap-binding protein complex

Cleavage of poly(A)-binding protein by enterovirus proteases concurrent with inhibition of translation in vitro

Cleavage of poly(A)-binding protein by coxsackievirus 2A protease in vitro and in vivo: another mechanism for host protein synthesis shutoff?

Genetic analysis of mengovirus protein 2A: its function in polyprotein processing and virus reproduction

Rapamycin and wortmannin enhance replication of a defective encephalomyocarditis virus

Virus-specific proteins associated with ribosomes of Krebs II cells infected with encephalomyocarditis virus

Cardiovirus 2A protein associates with 40S but not 80S ribosome subunits during infection

Encephalomyocarditis viral protein 2A localizes to nucleoli and inhibits cap-dependent mRNA translation

Encephalomyocarditis virus (EMCV) proteins 2A and 3BCD localize to nuclei and inhibit cellular mRNA transcription but not rRNA transcription

Mengovirus leader is involved in the inhibition of host cell protein synthesis

A picornavirus protein interacts with Ran-GTPase and disrupts nucleocytoplasmic transport

Inhibition of mRNA export and dimerization of interferon regulatory factor 3 by Theiler's virus leader protein

Inhibition of cap-dependent gene expression induced by protein 2A of hepatitis A virus

Poliovirus-encoded protease 2A Pro cleaves the TATAbinding protein but does not inhibit host cell RNA polymerase II transcription in vitro

The effect of poliovirus proteinase 2A pro expression on cellular metabolism. Inhibition of DNA replication, RNA polymerase II transcription, and translation

Inhibition of U snRNP assembly by a virus-encoded proteinase

Degradation of nuclear factor kappa B during foot-and-mouth disease virus infection

A conserved domain in the leader proteinase of foot-and-mouth disease virus is required for proper subcellular localization and function

Intracellular localization and effects of individually expressed human parechovirus 1 nonstructural proteins

2A proteinase of human rhinovirus cleaves cytokeratin 8 in infected HeLa cells

Enteroviral protease 2A cleaves dystrophin: evidence of cytoskeletal disruption in an acquired cardiomyopathy

Bidirectional increase in permeability of nuclear envelope upon poliovirus infection and accompanying alterations of nuclear pores

Differential targeting of nuclear pore complex proteins in poliovirus-infected cells

RNA nuclear export is blocked by poliovirus 2A protease and is concomitant with nucleoporin cleavage

The leader protein of Theiler's virus interferes with nucleocytoplasmic trafficking of cellular proteins

Nucleo-cytoplasmic traffic disorder induced by cardioviruses

Mengovirus-induced rearrangement of the nuclear pore complex: hijacking cellular phosphorylation machinery

Leader-induced phosphorylation of nucleoporins correlates with nuclear trafficking inhibition by cardioviruses

Ability of foot-and-mouth disease virus to form plaques in cell culture is associated with suppression of alpha/beta interferon

Inhibition of L-deleted foot-and-mouth disease virus replication by a/β interferon involves double-stranded RNA-dependent protein kinase

The leader proteinase of foot-and-mouth disease virus inhibits the induction of beta interferon mRNA and blocks the host innate immune response

Proteinase 2A pro is essential for enterovirus replication in type I interferon-treated cells

Cleavage of IPS-1 in cells infected with human rhinovirus

Proteolytic cleavage of the catalytic subunit of DNA-dependent protein kinase during poliovirus infection

The leader protein of Theiler's virus inhibits immediate-early alpha/beta interferon production

The mengovirus leader protein blocks interferon-a/β gene transcription and inhibits activation of interferon regulatory factor 3

A Theiler's virus alternatively initiated protein inhibits the generation of H-2K-restricted virus-specific cytotoxicity

CD4 T cells are important for clearance of the DA strain of TMEV from the central nervous system of SJL/J mice

Involvement of cardiovirus leader in host cell-restricted virus expression

This work demonstrates that cardiovirus L is dispensable

The leader polypeptide of Theiler's virus is essential for neurovirulence but not for virus growth in BHK cells

Protein 2A is not required for Theiler's virus replication

L* protein of the DA strain of Theiler's murine encephalomyelitis virus is important for virus growth in a murine macrophage-like cell line

Influence of the Theiler's virus L* protein on macrophage infection, viral persistence, and neurovirulence

Poliovirus 2A protease induces apoptotic cell death

Infection with enterovirus 71 or expression of its 2A protease induces apoptotic cell death

Poliovirus protein 3A inhibits tumor necrosis factor (TNF)-induced apoptosis by eliminating the TNF receptor from the cell surface

Bypass suppression of smallplaque phenotypes by mutation in poliovirus 2A that enhances apoptosis

2A protease is not a prerequisite for poliovirus replication

Theiler's murine encephalomyelitis virus leader protein is the only nonstructural protein tested that induces apoptosis when transfected into mammalian cells

A protein critical for a Theiler's virus-induced immune system-mediated demyelinating disease has a cell type-specific antiapoptotic effect and a key role in virus persistence

Pathogenesis of wild-type and leaderless foot-and-mouth disease virus in cattle

Cardiovirus leader proteins are functionally interchangeable and have evolved to adapt to virus replication fitness

An attenuating mutation in the 2A protease of swine vesicular disease virus, a picornavirus, regulates capand internal ribosome entry site-dependent protein synthesis

Construction of a chimeric Theiler's murine encephalomyelitis virus containing the leader gene of foot-and-mouth disease virus

Differential IFN-a/β production suppressing capacities of the leader proteins of mengovirus and foot-and-mouth disease virus

Theiler's murine encephalomyelitis virus (TMEV) L* amino acid position 93 is important for virus persistence and virus-induced demyelination

The foot-and-mouth disease virus leader proteinase gene is not required for viral replication

An article showing that aphthovirus L is dispensable

Hepatitis A viruses with deletions in the 2A gene are infectious in cultured cells and marmosets

Analysis of deletion mutants indicates that the 2A polypeptide of hepatitis A virus participates in virion morphogenesis

Studies on dicistronic polioviruses implicate viral proteinase 2A pro in RNA replication

Molecular Biology of Picornaviruses

Proteolytic cleavage of the p65-RelA subunit of NF-κB during poliovirus infection

Disruption of innate immunity due to mitochondrial targeting of a picornaviral protease precursor

RIG-I is cleaved during picornavirus infection

Molecular Biology of Picornaviruses

Inhibition of cellular protein secretion by poliovirus proteins 2B and 3A

Coxsackievirus B3 inhibits antigen presentation in vivo, exerting a profound and selective effect on the MHC class I pathway

Role of nonstructural proteins 3A and 3B in host range and pathogenicity of foot-and-mouth disease virus

Early phosphatidylinositol 3-kinase/ Akt pathway activation limits poliovirus-induced JNK-mediated cell death

Foot-and-mouth disease virus Lb proteinase can stimulate rhinovirus and enterovirus IRES-driven translation and cleave several proteins of cellular and viral origin

The C-terminal domain of eukaryotic protein synthesis initiation factor (eIF) 4G is sufficient to support capindependent translation in the absence of eIF4E

2A pro is a multifunctional protein that regulates the stability, translation and replication of poliovirus RNA

Coding changes in the poliovirus protease 2A compensate for 5′NCR domain V disruptions in a cell-specific manner

The C-terminal residues of poliovirus proteinase 2A pro are critical for viral RNA replication but not for cis-or trans-proteolytic cleavage

Replication of poliovirus RNA and subgenomic RNA transcripts in transfected cells

Aichi virus 2A protein is involved in viral RNA replication

Intrinsic signals for the assembly of hepatitis A virus particles

Morphogenesis of hepatitis A virus: isolation and characterization of subviral particles

Identification and site-directed mutagenesis of the primary (2A/2B) cleavage site of the hepatitis A virus polyprotein: functional impact on the infectivity of HAV RNA transcripts

The unique role of domain 2A of the hepatitis A virus precursor polypeptide P1-2A in viral morphogenesis

Picornavirales, a proposed order of positive-sense single-stranded RNA viruses with a pseudo-T = 3 virion architecture

Origin and Evolution of Viruses

Translational barrier in central region of encephalomyocarditis virus genome. Modulation by elongation factor 2 (eEF-2)

Occurrence, function and evolutionary origins of '2A-like' sequences in virus genomes

An article discussing viral translation-interrupting 2A peptides

A case for ''StopGo'': reprogramming translation to augment codon meaning of GGN by promoting unconventional termination (Stop) after addition of glycine and then allowing continued translation (Go)

Euprosterna elaeasa virus genome sequence and evolution of the Tetraviridae family: Emergence of bipartite genomes and conservation of the VPg signal with the dsRNA Birnaviridae family

Molecular Basis of Viral Evolution

Genomic features and evolutionary constraints in Saffold-like cardioviruses

Mechanisms of severe acute respiratory syndrome pathogenesis and innate immunomodulation. Microbiol

NS1′ of flaviviruses in the Japanese encephalitis virus serogroup is a product of ribosomal frameshifting and plays a role in viral neuroinvasiveness

NSs protein of Rift Valley fever virus induces the specific degradation of the doublestranded RNA-dependent protein kinase

Divergent picornavirus IRES elements

The Proteomics Protocols Handbook

Clustal W and Clustal X version 2.0

CDD: specific functional annotation with the Conserved Domain Database

We thank F. van Kuppeveld for critical reading of this manuscript. Recent research in the authors' laboratory was supported by grants from the Russian Foundation for Basic Research (RFBR), the Scientific School Support Program and The Netherlands Organization for Scientific Research-RFBR (NWO-RFBR).

The authors declare no competing financial interests. 

See online article: S1 (figure)All liNkS Are ActiVe iN the oNliNe Pdf