Abstract

1. Introduction

Picornavirus RNA genomes are mRNAs that also function as templates for viral RNA replication. Considering the RNA World hypothesis (Gesteland et al., 1999), picornaviruses represent modern day organisms with evolutionarily ancient replication strategies. Picornaviruses replicate via RNA intermediates within replication complexes assembled on lipid membranes in the cytoplasm of eukaryotic host cells. The replication of viral RNA within oligomeric viral protein particles or membranous replication complexes containing oligomerized viral proteins (Kopek et al., 2007) reveals common features of positive-strand RNA viruses, dsRNA viruses, and retroviruses (Ahlquist, 2006). Co-localization of viral proteins and viral nucleic acid templates within replication complexes facilitates interactions between viral proteins and viral RNA templates and also sequesters viral replication intermediates (like dsRNA) from innate antiviral host proteins. When picornavirus RNA genomes like that of poliovirus are introduced into cell-free reactions containing cytoplasmic extracts from uninfected host cells the viral RNA is translated, the viral polyproteins are processed, viral RNA replication complexes are assembled, viral RNA is replicated, and newly synthesized RNA genomes are packaged into progeny virus (Barton and Flanegan, 1993; Molla et al., 1991). Thus, all of the metabolic steps of picornavirus mRNA translation, RNA replication, and virus assembly are faithfully recapitulated in cell-free reactions containing cytoplasmic extracts. In addition to poliovirus, cell-free reactions containing cytoplasmic extracts support the translation and replication of EMCV (Svitkin and Sonenberg, 2003), rhinovirus 14 (Todd et al., 1997b), Aichi virus (Nagashima et al., 2005), and Coxsackievirus B3 RNAs (van Ooij et al., 2006). These data document the incredible ability of picornavirus RNA genomes, via cis-active RNA sequences and structures, to co-opt the cellular translation machinery of the cytoplasm, to express viral proteins, to build viral replication complexes anchored on host lipid membranes, to efficiently replicate viral RNA, and to assemble progeny virus.

Picornaviruses are a family of positive-strand RNA viruses in the order Picornavirales (Le Gall et al., 2008). There are nine genera in the Picornaviridae family: Enterovirus, Rhinovirus, Hepatovirus, Parechovirus, Kobuvirus, Erbovirus, Cardiovirus, Aphthovirus, and Teschovirus (Figure 1) (Stanway et al., 2002). Viruses from these genera include many notorious human and agricultural pathogens. Polioviruses, rhinoviruses, and hepatitis A virus are well known human pathogens. Foot and mouth disease virus (FMDV), encephalomyocarditis virus (EMCV), and Theiler's virus are notable non-human pathogens. As exemplified by the four human enterovirus species (HEV-A, HEV-B, HEV-C, and HEV-D) (Figure 2), each picornavirus genus often includes several species and there are often many individual serotypes of virus within each species. In addition to the viruses mentioned above, there are many new picornaviruses being discovered, including simian enteroviruses (Oberste et al., 2007b), a new genus of human rhinoviruses (Briese et al., 2008; Lau et al., 2007), and new human (Oberste et al., 2007a) and non-human picornaviruses (Hales et al., 2008).

Open in a separate window

Figure 1

The Picornaviridae family

Open in a separate window

Figure 2

Virus in the genus Enterovirus: human enterovirus species A-D

RNA recombination between more than one member of a particular picornavirus species leads to viable recombinants, providing a biologically meaningful relationship between individual viruses within each species (Brown et al., 2003; Oberste et al., 2004a; Oberste et al., 2004b; Oberste et al., 2004c). RNA recombination between individual viruses within a species occurs in the regions of the genome encoding for picornavirus replication proteins (Brown et al., 2003). Replication genes within this region of the genome have a high degree of identity (80-95% identity between viruses within a species) which allows for viable chimeric virus. Recombination between viruses in alternate species is restricted because the chimeric viral proteins are too unrelated to function effectively with the chimeric RNA genome. While RNA recombination can occur within intratypic capsid genes (i.e., the same serotype), the degree of variation between different (intertypic) serotypes, even those within a species, restricts the production of viable progeny. RNA recombination during viral RNA replication in co-infected cells is important in the poliovirus eradication campaign, where Sabin strains of the live-attenuated oral poliovirus vaccine recombine with non-polio group C enteroviruses in vaccinees and their contacts leading to progeny vaccine-derived polioviruses that have the capacity to circulate in human populations and cause paralytic poliomyelitis (reviewed in (Kew et al., 2005)). Thus, the production of viable chimeric progeny from RNA recombination helps define picornaviral species and is also biologically relevant in nature.

Although there are many different picornaviruses with various degrees of relatedness, all picornaviruses share several common features. All picornaviruses have single-stranded RNA genomes of positive polarity covalently linked to 5′ terminal viral proteins. The RNA genome contains a 5′ nontranslated region (NTR) with an internal ribosome entry site (IRES), an open reading frame encoding the viral capsid proteins and the viral replication proteins, a 3′ NTR and a 3′ poly(A) tail. The contribution of particular cis-active RNA elements to viral mRNA stability (Kempf and Barton, 2008), viral mRNA translation (Trono et al., 1988), and viral RNA replication have been determined for several representative picornaviruses. While some cis-active RNA elements like 3′ poly(A) sequences are functionally relevant during several steps of virus replication, including viral mRNA stability (Sachs, 1990), viral mRNA translation (Thoma et al., 2004), and viral RNA replication (Silvestri et al., 2006); this review will highlight the cis-active RNA elements of picornavirus RNA genomes involved only in RNA replication. Other chapters in this special edition highlight the role of Picornavirales IRES RNAs in viral mRNA translation.

2. Picornavirus RNA genomes and viral replication proteins

Picornavirus RNA genomes, including poliovirus (Figure 3), are composed of a covalently linked 5′ terminal protein called VPg (Viral Protein of the genome), a 5′ nontranslated region (NTR), an open reading frame (ORF), a 3′ NTR and a poly(A) tail of variable length (~ 20 to 150 adenosine residues long). The capsid proteins are encoded within the 5′ end of the ORF while the nonstructural viral proteins required for RNA replication are encoded within the 3′ two thirds of the viral ORF (Figure 3). The chemical formula for poliovirus (C_322,652 H_492,368 N_98,245 O_131,196 P_7,501 S_2,340), as reported by Eckard Wimmer (Wimmer, 2006), reveals the presence of 7501 phosphorus atoms in poliovirus. These phosphorus atoms correspond to the phosphodiester moieties of ribonucleotides in the 7,500 base long PV RNA which is packaged within PV particles (Figure 3).

Open in a separate window

Figure 3

Poliovirus and the poliovirus RNA genome

Poliovirus RNA genome (illustrated at the top of the figure) is packaged within virus particles (bottom left). The RNA genome is an mRNA encoding the capsid proteins and nonstructural proteins. Nonstructural proteins include the viral proteases and viral replicase proteins. Chemical formula for poliovirus from (Wimmer, 2006).

Translation of PV RNA in the cytoplasm of infected cells followed by polyprotein processing by viral proteases 2A^Pro and 3CD^Pro results in the synthesis of several PV proteins, all of which contribute to the formation and function of RNA replication complexes. 3D^Polymerase (3D^Pol) is a primer-dependent RNA-dependent RNA polymerase (Figure 3) (Flanegan and Baltimore, 1977; Neufeld et al., 1991). VPg (also called viral protein 3B) and the uridylylated form of VPg (VPgpUpU_OH) function as primers for PV RNA replication (Paul, 2002; Paul et al., 1998), thereby becoming covalently linked to the 5′ ends of PV RNA (Lee et al., 1977; Nomoto et al., 1977). 2C^ATPase (and its larger polyprotein precursor 2BC) has several critical roles associated with viral RNA replication including host membrane rearrangements associated with the formation of RNA replication complexes (Cho et al., 1994; Teterina et al., 1997), RNA binding domains that interact with viral RNA (Banerjee et al., 2001; Rodriguez and Carrasco, 1995), and ATPase activity required for the initiation of negative-strand RNA synthesis (Barton and Flanegan, 1997) as well as the conversion of VPg into VPgpUpU_OH (Lyons et al., 2001)(as discussed in more detail below). 2B is a porin that alters the permeability of host cell membranes (Madan et al., 2007; Nieva et al., 2003), however its precise role(s) in viral replication are unclear. Viral protein 3CD is a protease involved in viral polyprotein processing (Harris et al., 1992), specific host protein cleavages (Kundu et al., 2005; Sharma et al., 2004), as well as an RNA binding protein with important functions during RNA replication (Andino et al., 1993; Andino et al., 1990b; Gamarnik and Andino, 2000; Harris et al., 1994; Shen et al., 2007; Yang et al., 2004) (as discussed in more detail below). 2A^Pro functions during RNA replication (Li et al., 2001); however, its mechanistic role(s) are unclear.

3. Cis-active RNA elements required for viral RNA replication reside in the 5′ NTR, the 3′ NTR and the ORF of picornavirus RNA genomes

In the case of poliovirus, there are four distinct cis-active RNA elements required for viral RNA replication (Figure 4); a cloverleaf RNA structure at the 5′ terminus of PV RNA (5′ CL), a cis-replication element in the ORF (CRE), the 3′ NTR, and the 3′ poly(A) tail. The 5′ cloverleaf RNA forms ribonucleoprotein complexes (RNPs) containing poly r(C) binding protein (PCBP) and 3CD that are required for RNA replication (Barton et al., 2001; Herold and Andino, 2001). The CRE RNA located in the ORF templates the uridylylation of VPg to make VPgpUpU_OH and is the primary focus of this review (as discussed below). The 3′ NTR can be deleted from poliovirus without completely preventing replication (Todd et al., 1997a); however, the 3′ NTR is required for efficient RNA replication (Brown et al., 2005). The poly(A) tail is required for RNA replication (Sarnow, 1989). A poly(A) tail ≥20 bases long supports wildtype levels of viral RNA replication (Silvestri et al., 2006). When the poly(A) tail is reduced to 12 nucleotides in length viral negative-strand RNA synthesis is reduced and when the poly(A) tail is ≤ 8 bases long negative-strand RNA synthesis ceases (Herold and Andino, 2001; Silvestri et al., 2006). Comparable and/or related cis-active RNAs are present in the 5′ NTR, ORF, and 3′ NTR of other picornavirus RNA genomes. Together, these cis-active elements mediate the specific replication of viral RNA genomes via interactions with specific viral replication proteins. Furthermore, these cis-active RNAs mediate particular mechanistic steps of RNA replication, like the conversion of VPg into VPgpUpU_OH as discussed below.

Open in a separate window

Figure 4

VPg and cis-active RNA sequences and structures required for poliovirus RNA replication

VPg, a 22 amino acid long viral protein, is covalently linked by the tyrosine hydroxyl to the 5′ terminus of viral RNA via a phosphodiester (Ambros and Baltimore, 1978). There are four distinct cis-active RNA sequences and structures required for RNA replication: a cloverleaf RNA sequence and structure at the 5′ terminus (5′ CL), a cis-replication element (CRE) in the open reading frame, a 3′ nontranslated region (3′ NTR), and a 3′ poly(A) tail.

The IRES within the 5′ NTR of PV RNA is not directly required for viral RNA replication, and is therefore not illustrated in detail in Figure 4. When the PV IRES was replaced by heterologous IRES RNAs from EMCV and HCV, the chimeric viral RNAs replicated efficiently (Alexander et al., 1994; Lu and Wimmer, 1996; Zhao et al., 2000; Zhao et al., 1999). Furthermore, when IRES RNA sequences were deleted from PV RNA templates the viral RNAs were still efficient templates for RNA replication when PV proteins were provided in trans (Murray et al., 2004). These investigations revealed that the IRES RNA sequences could be substituted with heterologous sequences or deleted from PV RNA templates without preventing RNA replication. Together these experiments establish that IRES RNA sequences (and the proteins that interact with them) are not directly involved in viral RNA replication.

4. VPg and VPgpUpU_OH

There are two forms of VPg within infected cells, VPg and VPgpUpU_OH (Crawford and Baltimore, 1983). These two forms of VPg function as primers for the primer-dependent RNA-dependent RNA polymerase 3D^Pol. Much of the research in the area of picornavirus RNA replication over the past 10 years has been work designed to determine:

• how VPg is converted into VPgpUpU_OH,

• which of these two primers prime the initiation of negative-strand RNA synthesis, and

• which of these two primers prime the initiation of positive-strand RNA synthesis.

The goals of this research are to understand and to explain the mechanisms by which PV RNA is replicated in an asymmetric manner, wherein positive-strand RNA is made to great excess relative to negative-strand RNA (reviewed in (Paul, 2002; Sean and Semler, 2008)).

5. The discovery of CRE

While RNA elements in the 5′ NTRs (Andino et al., 1993; Andino et al., 1990a; Andino et al., 1990b) and 3′ NTRs (Pilipenko et al., 1992; Rohll et al., 1995) of picornavirus RNA genomes were known to be required for RNA replication for many years, until more recently no one had investigated whether there were RNA elements within the ORF of picornavirus RNA genomes which were required for RNA replication. Replication competent defective interfering particles with large in-frame deletions in the capsid encoding region of poliovirus suggested that capsid proteins and the RNA sequences encoding them were not required for RNA replication (Cole and Baltimore, 1973; Cole et al., 1971; Nomoto et al., 1979). Since deleting capsid protein encoding sequences also resulted in deletion of all RNA elements in this region, these studies suggested that there were no RNA elements in the capsid encoding region of the poliovirus ORF required for RNA replication. These observations inspired the construction of poliovirus RNA replicons without capsid encoding sequences that retained RNA sequences encoding the non-structural replication proteins and the necessary cis-active RNA sequences required for RNA replication (Collis et al., 1992; Hagino-Yamagishi and Nomoto, 1989; Kaplan and Racaniello, 1988). In contrast, when McKnight and Lemon deleted the capsid encoding region of the ORF of human rhinovirus-14 (HRV-14) in attempts to make RNA replicons like those for poliovirus, such rhinovirus RNAs were completely defective for RNA replication (McKnight and Lemon, 1996). It was subsequently discovered that there was a cis-acting replication element (CRE) within the HRV-14 VP1 capsid encoding region that was required for RNA replication (McKnight and Lemon, 1998).

Of the nine genera in the Picornaviridae family, a CRE like that found in HRV-14 has been identified in six of them: Enterovirus (Goodfellow et al., 2000; Paul et al., 2000), Rhinovirus (Gerber et al., 2001; McKnight and Lemon, 1998), Cardiovirus (Lobert et al., 1999), Aphthovirus (Mason et al., 2002), Parechovirus (Al-Sunaidi et al., 2007), and Hepatovirus (Yang et al., 2008) (Fig. 5). All but the Parechovirus and species C rhinovirus CREs have been shown to be required for RNA replication. The Parechovirus and species C rhinovirus CREs have yet to be experimentally tested and confirmed. A putative CRE has not been identified or reported for virus in the Kobu-, Erbo- or Teschovirus genera. While a CRE is presumably present within all picornaviral genomes, its location in the genomic RNA varies substantially depending on the picornavirus genus or species (Fig. 5). CRE is located in the 2C open reading frame of enteroviruses (Goodfellow et al., 2000; Paul et al., 2000), the 2A open reading frame of species A rhinoviruses (Gerber et al., 2001), the VP1 open reading frame of species B rhinoviruses (McKnight and Lemon, 1998), the VP2 open reading frame of species C rhinoviruses (Cordey et al., 2008), the VP2 region of cardioviruses (Lobert et al., 1999), just upstream of the IRES in the Aphthovirus foot-and-mouth disease virus (FMDV) (Mason et al., 2002), the VP0 region of parechoviruses (Al-Sunaidi et al., 2007) and the open reading frame of 3D for hepatoviruses (Yang et al., 2008). (Fig. 5A-H). FMDV is the only picornavirus thus far that has CRE located outside the viral open reading frame. The variable location of CRE within distinct picornaviruses suggests that its position within the genome may not be of particular importance and indeed several studies have demonstrated the positional independence of the CRE (Goodfellow et al., 2000; Goodfellow et al., 2003a; Mason et al., 2002; Yin et al., 2003) (Yang et al., 2008). Nonetheless, the natural location of CRE (as described above and in Fig. 5A-H) is conserved within the RNA genomes of particular picornaviral species (Cordey et al., 2008).

Open in a separate window

Open in a separate window

Figure 5

Location and secondary structures of CREs

The location in the viral genome and the secondary structures of representative CREs. Nucleotide numbers from full-length RNA genome sequences. (A) PV1 CRE, representative of CREs in human enterovirus species A-D (Goodfellow et al., 2000; van Ooij et al., 2006) (B) Human rhinovirus 2 (HRV 2) CRE, representative of species A rhinoviruses (Gerber et al., 2001). (C) HRV 14 CRE, representative of species B rhinoviruses (McKnight and Lemon, 1998). (D) HRV A2 CRE, representative of species C rhinoviruses (Cordey et al., 2008). (E) EMCV CRE, representative of Cardioviruses (Lobert et al., 1999). (F) FMDV CRE, representative of Apthoviruses (Mason et al., 2002). The FMDV CRE is located 218 to 275 bases downstream from the variable length poly(C) tract in the 5′ NTR. (G) Human parechovirus 1 (HPeV1) CRE, representative of Parechoviruses (Al-Sunaidi et al., 2007). (H) Hepatitis A virus (HAV) CRE, representative of Hepatoviruses (Yang et al., 2008). Only sequences from the apex of the 110 nt long HAV CRE are presented due to space constraints.

The CREs of human enteroviruses and rhinoviruses have characteristic 14 base loops where the 1st base is a purine, the 5th and 6th bases are A residues involved in templating the addition of uridine onto VPg (as discussed below), the 7th residue is a purine, and the 14th residue is a purine (Fig. 5A-D). Therefore, human enterovirus and rhinovirus CREs share a common sequence motif within the loop, RNNNAARNNNNNNR (Fig. 5A-D, conserved residues of loops in bold) (Cordey et al., 2008; Yang et al., 2002), which likely corresponds to a common structure (Thiviyanathan et al., 2004). The loop structures at the apex of Cardio-, Aphtho-, Parecho-, and Hepatovirus CREs are more variable in length (15 to 23 bases long) (Fig. 5E-H). Nonetheless, the AAAC sequence common in the loop of picornavirus CREs is evident (Fig. 5E-H, AAAC sequence in bold). The dsRNA stems of CRE RNA structures, with various internal loops and bulges, are more variable in sequence and length. Therefore, the more important feature of CREs is the loop structure containing the conserved adenosine residues. Below we discuss how CRE RNA sequences and structures participate in the conversion of VPg into VPgpUpU_OH.

6. Conversion of VPg into VPgpUpU_OH

Following the discovery of the HRV14 CRE by Kevin McKnight in the Lemon laboratory (McKnight and Lemon, 1998), Ian Goodfellow in the laboratory of David Evans discovered the CRE of poliovirus (Goodfellow et al., 2000). The poliovirus CRE is a stem-loop RNA structure of ~61 nucleotides in the 2C^ATPase portion of the ORF (Figure 5A). This CRE RNA structure is conserved in the 2C^ATPase portion of the ORF of all human enteroviruses (HEV-A, HEV-B, HEV-C, & HEV-D).

Aniko Paul in the Wimmer lab subsequently established that the poliovirus CRE RNA functioned as a template for the uridylylation of VPg, resulting in the synthesis of VPgpUpU_OH (VPg with two uridines covalently attached via a phosphodiester linkage to the tyrosine hydroxyl of VPg) (Paul et al., 2003a; Paul et al., 2000; Paul et al., 2003b). Conversion of VPg into VPgpUpU_OH was accomplished in defined reactions containing purified components including the CRE RNA template, 3D^Pol, VPg, and UTP (Figure 6) (Paul et al., 2003a; Paul et al., 2000; Paul et al., 2003b). Investigations using these defined reactions established that the viral RNA-dependent RNA polymerase 3D^Pol was responsible for catalyzing the addition of two uridine residues to the tyrosine residue of VPg and that the A₅ residue, in conjunction with the A₆ residue in the loop of the CRE RNA structure, functioned in a “slide-back” mechanism to template the addition of UTP onto the tyrosine residue of VPg (Figure 6 and reference (Paul et al., 2003b)). Each VPgpUpU_OH product probably remains associated with the 3D^Pol molecule involved in its synthesis (as illustrated in Fig. 6).

Open in a separate window

Figure 6

CRE-dependent VPg uridylylation

VPg is converted into VPgpUpU_OH in defined reactions containing CRE RNA templates, 3D^Pol, VPg, and UTP (Paul et al., 2003a; Paul et al., 2000; Paul et al., 2003b). Viral protein 3CD stimulates these reactions (Nayak et al., 2005; Nayak et al., 2006; Pathak et al., 2007; Pathak et al., 2002; Yang et al., 2004; Yin et al., 2003).

Viral protein 3CD stimulates the CRE-dependent conversion of VPg into VPgpUpU_OH; however, the mechanism by which 3CD stimulates the reaction is unclear (Figure 6). An RNA binding domain within 3C^Pro is required for stimulation of CRE-dependent VPgpUpU_OH synthesis and has been shown to bind to the upper portion of the CRE stem (Nayak et al., 2005; Nayak et al., 2006; Pathak et al., 2007; Pathak et al., 2002; Yang et al., 2004; Yin et al., 2003). While 3C^Pro alone can stimulate the reaction, it does not do so as well as 3CD^Pro, suggesting that the 3D^Pol portion of 3CD^Pro is also contributing to the uridylylation reaction (Nayak et al., 2006; Pathak et al., 2007). The involvement of the 3D^Pol moiety was confirmed by mutations within 3D^Pol of HRV14 that allow HRV14 to adapt to using a PV CRE for RNA replication and also by the presence of 3D^Pol in 3CD^Pro contributing to specificity of CRE binding by 3CD^Pro (Shen et al., 2007).

CRE-dependent VPg uridylylation was confirmed for several other picornaviruses using cognate 3D^Pol and CRE RNA templates from human rhinovirus 2 (Gerber et al., 2001), foot and mouth disease virus (Nayak et al., 2005; Nayak et al., 2006), Coxsackievirus B3 (van Ooij et al., 2006), and human rhinovirus 14 (Shen et al., 2007). Together, the experimental investigations using defined reactions have allowed for detailed analyses of the minimal components required for conversion of VPg into VPgpUpU_OH, however, these defined reactions do not support normal viral RNA replication wherein the VPg and/or VPgpUpU_OH primers support asymmetric RNA replication. Below we will discuss how CRE RNAs function coordinately with other cis-active RNAs in picornavirus RNA templates and the roles of VPg and VPgpUpU_OH primers during the asymmetric replication of picornavirus RNA.

7. PV RNA Replication

CRE RNAs do not function independently during viral RNA replication. Rather, CRE RNAs function coordinately with the other cis-active RNA sequences within picornavirus RNA and with the full complement of viral replication proteins within membranous RNA replication complexes (Bienz et al., 1992). Sequestering components of replication within membranous structures is an efficient way to keep multiple copies of viral proteins, like the membrane-associated precursors of VPg primers (Fujita et al., 2007), in close proximity to viral RNA templates. During RNA replication, each viral positive-strand RNA is copied into a complementary negative-strand RNA and the negative-strand RNA is then copied into 40 to 70 positive-strand RNA progeny (Novak and Kirkegaard, 1991). A VPg molecule is covalently attached to the 5′ end of each negative- and positive-strand RNA molecule (Pettersson et al., 1978). In addition, hundreds of excess VPgpUpU_OH molecules accumulate for each negative-strand RNA intermediate (Lyons et al., 2001). Therefore, each membranous RNA replication complex must accumulate many copies of VPg precursors before RNA replication commences. In contrast to the VPg molecules used in defined VPg uridylylation reactions (Figure 6), VPg is thought to accumulate within membranous RNA replication complexes as larger precursor polyproteins (3ABCD, 3BCD, 3ABC, and/or 3AB) (Towner et al., 1998). The actual composition of membranous RNA replication complexes is unknown; however, from the discussion above one can begin to imagine that many copies of particular viral replication proteins are likely present in each membranous RNA replication complex, along with one or more viral RNA templates.

Preinitiation RNA replication complexes (PIRCs) have proven to be a particularly useful tool to identify the coordinate interactions of distinct cis-active RNAs and viral proteins during poliovirus RNA replication. Guanidine HCl, a reversible inhibitor of viral protein 2C^ATPase, is used to allow PIRCs to accumulate within cell-free translation-replication reactions containing cytoplasmic extracts from HeLa cells (Barton et al., 1995). Removal of the reversible inhibitor guanidine from the accumulated PIRCs allows for the initiation of viral RNA replication. There are several advantageous features of this experimental system:

• viral RNA replication is synchronous and sequential, with negative-strand RNA being made before positive-strand RNA (Barton and Flanegan, 1997),

• viral RNA replication is asymmetric, with an excess of positive-strand RNA being made from each negative-strand template,

• viral RNA replication proteins can be provided in trans to viral RNA templates (Barton et al., 2002; Murray et al., 2004),

• lethal and non-lethal mutations in viral proteins and viral RNA templates can be phenotypically associated with defects in specific steps of viral RNA replication including the conversion of VPg into VPgpUpU_OH, the synthesis of negative-strand RNA, and the synthesis of positive-strand RNA.

The first two important insights made using PIRCs was that the 5′ cloverleaf RNA was required in cis for negative-strand RNA synthesis (Barton et al., 2001) and that the guanidine-inhibited 2C^ATPase was required immediately before or at the moment of initiation of negative-strand RNA synthesis (Barton and Flanegan, 1997). We and others interpreted these results to indicate that ribonucleoproteins (RNPs) at the 5′ and 3′ termini of viral RNA templates worked coordinately via proximal orientation to mediate PV RNA replication within membranous replication complexes and that 2C^ATPase functions were directly involved in mediating these long range RNP interactions. Cis-active RNA at the 5′ terminus of Aichi virus also appears to be required for negative-strand RNA synthesis, providing evidence that other picornaviruses with unrelated 5′ RNA sequences share common strategies for negative-strand RNA synthesis (Nagashima et al., 2005; Nagashima et al., 2008). Another discovery made using PIRCs was that the 5′ cloverleaf of poliovirus RNA templates was required in cis for CRE-dependent VPg uridylylation within membranous RNA replication complexes and that guanidine inhibited the conversion of VPg into VPgpUpU_OH (Lyons et al., 2001). These data suggested that viral 2C^ATPase activity, the target of guanidine inhibition (Pfister and Wimmer, 1999), is required immediately before CRE-dependent VPg uridylylation and that RNPs associated with the 5′ cloverleaf RNA may function coordinately with the CRE RNPs to help convert VPg into VPgpUpU_OH in the context of membranous RNA replication complexes.

The next important insight made using PIRCs was that CRE RNAs were not required for viral negative-strand RNA synthesis (Goodfellow et al., 2003b; Morasco et al., 2003; Murray and Barton, 2003). Mutations that destroyed CRE RNA structures within viral RNA templates did not prevent negative-strand RNA synthesis, however, positive-strand RNA replication was absolutely dependent upon CRE-dependent VPg uridylylation (Goodfellow et al., 2003b; Morasco et al., 2003; Murray and Barton, 2003). Because the tyrosine hydroxyl of VPg was required for negative-strand RNA synthesis, independent of CRE or VPgpUpU_OH, the Barton and Flanegan labs concluded that VPg could prime the initiation of negative-strand RNA synthesis and that VPgpUpU_OH was required to prime the initiation of positive-strand RNA synthesis (Morasco et al., 2003; Murray and Barton, 2003). This was theoretically satisfying because the asymmetric replication of picornavirus RNA demands that the mechanisms of positive-strand RNA synthesis be different from the mechanisms of negative-strand RNA synthesis (to account for the synthesis of more positive-strand RNAs relative to negative-strand RNAs). The Flanegan lab went on to show that two adenosines at the 3′ end of negative-strand RNA were the key cis-active RNA sequences allowing VPgpUpU_OH priming of positive-strand RNA synthesis (Figure 6, note complementarity of uridine residues in VPgpUpU_OH with the two A residues at the 3′ end of negative-strand RNA templates) (Sharma et al., 2005). Importantly, VPgpUpU_OH cannot prime the initiation of positive-strand RNA replication when the two adenosine residues are internal from the 3′ end of negative-strand RNA (Herold and Andino, 2000). Thus, 3D^Pol and VPgpUpU_OH, via unknown mechanisms, are translocated from the CRE RNA templates where they are made to the 3′ terminus of negative-strand RNA templates to prime the reiterative initiation of positive-strand RNA synthesis (Figure 7). Because the 3D^Pol/VPgpUpU_OH molecules are sequestered within membranous replication complexes they may function by mass action, where free diffusion within the confines of the replication complex favor interactions with the 3′ ends of viral RNA templates.

Open in a separate window

Figure 7

CRE-dependent VPgpUpU_OH synthesis and translocation to the 3′ termini of viral RNA templates for RNA replication

3D^Pol and VPgpUpU_OH must translocate from the CRE portion of viral RNA templates to the 3′ ends of positive- and negative-strand RNA templates to prime the initiation of RNA synthesis. These processes are supported within viral RNA replication complexes formed within cell-free translation-replication reactions (Goodfellow et al., 2003b; Lyons et al., 2001; Morasco et al., 2003; Murray and Barton, 2003; van Ooij et al., 2006).

Another important discovery made using PIRCs and tissue culture studies was that CRE RNAs with point mutations, particularly point mutations of the A₅ template residue, functioned as dominant negative inhibitors of viral RNA replication (Crowder and Kirkegaard, 2005). These mutations blocked both VPg uridylylation and negative-strand RNA synthesis (van Ooij et al., 2006). Thus, CRE RNP complexes inactivated by point mutations (especially in the templating A₅ residue) somehow prevent VPg from priming negative-strand RNA synthesis. Since these point mutations also block VPgpUpU_OH synthesis, some investigators interpret these studies to suggest that CRE-dependent VPgpUpU_OH is the normal primer for negative-strand RNA synthesis whereas others still consider VPg as a potential primer for negative-strand RNA synthesis. Despite these alternate interpretations, there is emerging consensus that CRE RNPs work coordinately with other cis-active RNAs at the 5′ and 3′ termini of picornavirus RNA templates to mediate PV RNA replication. These potential RNP interactions provide plenty of opportunity for mechanistic coordination between membrane-associated replication proteins and seemingly distal cis-active RNA elements.

8. CREs, in conjunction with VPgpUpU_OH, lower the concentrations of UTP required for viral RNA replication

De novo initiation of RNA synthesis by polymerases requires a relatively high concentration of the initiating NTP (NTPi) when compared with the concentration of NTPs required for the elongation of RNA synthesis (K_m of ~75 to 150 μM NTP for initiation and ~5-10 μM for elongation) (van Dijk et al., 2004). In contrast, poliovirus 3D^Pol requires relatively low concentrations of UTP (~4 μM K_m) to uridylylate VPg in reactions containing CRE RNA templates (Korneeva and Cameron, 2007). Furthermore, when CRE-dependent VPgpUpU_OH priming and CRE-independent VPg priming of poliovirus negative-strand RNA synthesis were compared a functional CRE RNA was found to lower the K_m of UTP required for negative-strand RNA synthesis (Steil and Barton, 2008). VPg primed the initiation of negative-strand RNA synthesis with a K_m of 103 μM UTP whereas (CRE-dependent) VPgpUpU_OH primed the initiation of negative-strand RNA synthesis with a K_m of 12 μM UTP (Steil and Barton, 2008). These results suggest that CRE RNAs, by virtue of their interactions with 3D^Pol, can lower the K_m of UTP required for de novo RNA polymerization (initiation), as compared to VPg priming by 3D^Pol on poly(A) templates. Thus, CRE-dependent VPg uridylylation may serve two important purposes: 1) providing VPgpUpU_OH primers for both negative- and positive-strand RNA replication and 2) overcoming the rate-limiting effects of NTPi concentrations, which in the absence of VPg uridylylation would otherwise restrict the initiation of RNA replication. In this manner, CRE-dependent PV RNA replication mechanisms circumvent the requirement of a high initiating NTP concentration which typically constrains RNA synthesis by prokaryotic and viral RNA polymerases (Amiott and Jaehning, 2006a; Amiott and Jaehning, 2006b; Gaal et al., 1997; Jia and Patel, 1997; van Dijk et al., 2004).

9. Unresolved aspects of RNA replication

There are several important unresolved aspects of RNA replication (Figure 7):

• When is VPg converted into VPgpUpU_OH (before, after, and/or during negative-strand RNA synthesis)?

• When are 3AB and/or other polyprotein precursors of VPg proteolytically processed into VPg (before, after, and/or during CRE-dependent uridylylation)?

• How are 3D^Pol proteins (and VPg) reiteratively delivered to CRE RNA templates within membranous replication complexes to allow for the synthesis of hundreds of VPgpUpU_OH molecules?

• What is the mechanistic contribution of 5′ cis-active RNPs to the initiation of negative-strand RNA synthesis and CRE-dependent VPg uridylylation?

• How is VPgpUpU_OH translocated from CRE RNA templates to the 3′ termini of viral RNA templates (Figure 7)?

• How do CRE RNA-VPg-3D^Pol interactions lower the concentration of UTP required for protein-primed RNA polymerization (initiation)?

The evidence at hand indicates that the processes of CRE-dependent VPg uridylylation must involve dynamic events where multiple copies of 3D^Pol and VPg are delivered to the CRE RNA portion of viral RNA templates within membranous RNA replication complexes. The mechanistic details of these dynamic processes probably involve interactions between RNP complexes containing 5′ cis-active RNA sequences, CRE and 3′ NTR/poly(A) sequences (as proposed in (Murray and Barton, 2003)). Because precursor proteins of VPg are membrane-anchored and substrates of 3CD^Pro, it is easy to speculate that 5′ cloverleaf, CRE, and 3′ NTR RNPs (which all contain 3CD^Pro) may work coordinately to proteolytically cleave membrane-associated VPg precursors and 3CD^Pro precursors to deliver both VPg and 3D^Pol to CRE RNA templates near membranes within the replication complexes.

10. Structural studies

Atomic structures of CRE (Thiviyanathan et al., 2004), 3D^Pol (Appleby et al., 2005; Thompson et al., 2007; Thompson and Peersen, 2004), VPg/3D^Pol co-crystals (Ferrer-Orta et al., 2006), and 3CD molecules (Marcotte et al., 2007) provide insights into the manner in which these molecules would interact with cis-active viral RNA structures and with other viral proteins. The Cameron lab has proposed models for how these proteins interact with CRE (Pathak et al., 2007). In this model, two 3CD molecules bind the CRE stem and stimulate VPg uridylylation, presumably through interactions of 3CD with 3D^Pol. Nonetheless, it remains to be determined how 3CD and other viral proteins interact coordinately and dynamically with the 5′ cloverleaf, CRE and 3′ NTR/poly(A) to initiate negative-strand RNA synthesis and VPg uridylylation. It will be particularly interesting to determine how CRE RNA templates and VPg molecules interact with the catalytic site of 3D^Pol and how such interactions are much more efficient at uridylylating VPg than 3D^Pol interactions with other templates such as poly(A), which under some circumstances can function as a template for VPgpUpU_OH synthesis (Paul et al., 1998).

11. Potential contributions of viral 2C^ATPase

Evidence suggests that CRE RNAs function coordinately with the other cis-active RNA sequences and structures at the 5′ and 3′ termini of picornavirus RNA templates, presumably via RNP interactions that may be facilitated by 2C^ATPase activity. The mechanistic function of 2C^ATPase activity is unclear. While it has long been considered a putative helicase (Kadare and Haenni, 1997), diligent efforts failed to detect 2C-mediated helicase activity (Pfister et al., 2000; Pfister and Wimmer, 1999) and the requirement of 2C^ATPase activity before and/or at the moment of initiation of RNA replication (Barton and Flanegan, 1997; Lyons et al., 2001; Murray and Barton, 2003) made us consider other possible mechanistic roles of 2C^ATPase activity. We currently favor the possibility that 2C^ATPase functions as a RNP chaperone to mediate the formation of functional RNP complexes on viral cis-active RNA structures. We consider this function to be similar to the RNP chaperone activities of SM-like proteins, which are required to assemble functional RNP complexes involved in RNA splicing (Pellizzoni et al., 2002b).

Dreyfuss et al. show that snRNP complexes can self-assemble by combining individual protein and RNA subunits in vitro (in the absence of SMN complexes) but that the ATPase activity of SMN complexes is required for the assembly and activity of snRNPs in cells (Pellizzoni et al., 2002b). snRNPs can self assemble inappropriately in vitro on U-rich RNAs other than snRNAs, leading to nonfunctional RNPs. Dreyfuss et al. show that SMN complexes are machines capable of preventing assembly of snRNPs on U-rich RNAs other than snRNAs (Pellizzoni et al., 2002b). The ATPase activity of particular subunits within SMN complexes is necessary for this ability to prevent illicit assembly of nonfunctional snRNPs (Baccon et al., 2002; Friesen et al., 2002; Gubitz et al., 2002; Pellizzoni et al., 2002a; Pellizzoni et al., 2002b; Wang and Dreyfuss, 2001a; Wang and Dreyfuss, 2001b). Thus, SMN complexes, through ATPase activity, function as RNP chaperones to prevent illicit assembly of snRNPs (Pellizzoni et al., 2002b). We hypothesize that ATPases within viral replication complexes function in a similar manner. In the case of poliovirus, we speculate that 2C^ATPase may function as a RNP chaperone within membranous replication complexes to coordinately assemble 5′ cloverleaf, 2C-CRE, and 3′NTR RNPs at the appropriate points in time with the appropriate protein subunits. This hypothesis accounts for some of the apparent paradoxes observed between experiments using purified subunits (reconstitution experiments) and work using preinitiation RNA replication complexes (Agol et al., 1999). Purified 3D^Pol can be used to make negative-strand RNA in vitro; however, 3D^Pol will use virtually any nonviral template RNA and any primer in vitro (Tuschall et al., 1982). Using purified 3D^Pol, VPg can be uridylylated in vitro in the absence of membranes or 2C^ATPase on various authentic or inauthentic templates (Paul et al., 2000; Paul et al., 1998; Rieder et al., 2000), However, within preinitiation RNA replication complexes, VPg uridylylation requires the 5′ cloverleaf and 2C^ATPase activity and VPg uridylylation occurs exclusively on CRE RNA sequences within PV RNA templates rather than on 3′ poly(A) sequences of PV RNA templates (Morasco et al., 2003; Murray and Barton, 2003). VPg uridylylation requires membranes when natural poliovirus replication complexes are isolated from infected cells (Takegami et al., 1983). Thus, preinitiation RNA replication complexes support authentic replication and 2C^ATPase activity is required for steps of RNA replication, including CRE-dependent VPg uridylylation, that require seemingly distal cis-active RNA elements (Barton and Flanegan, 1997; Barton et al., 2001; Lyons et al., 2001; Morasco et al., 2003; Murray and Barton, 2003). 2C^ATPase may act to prevent the formation of non-functional RNPs within membranous poliovirus RNA replication complexes in a manner analogous to that of ATPase subunits within SMN complexes. Studies of SV40 T antigen, an ATPase protein involved in managing RNP complexes at SV40 DNA origins of replication, may also provide insights into the regulation of dynamic RNPs associated with replication of viral nucleic acids (Li et al., 2003). The possible roles of 2C^ATPase and the manner in which picornaviral cis-active RNAs function coordinately to mediate RNA replication await further investigation.

12. CREs yet to be discovered?

CREs will likely be found in the genomes of the remaining picornaviruses where such elements have yet to be described; the Kobu-, Erbo-, and Teschoviruses. Furthermore, the advantages of CREs and nucleotidylylated protein primers described in this review suggest that other positive-strand RNA viruses, like those in the Dicistroviridae, Comoviridae, and Sequiviridae families, may employ CRE RNAs. Like picornaviruses, the viruses in the Dicistroviridae, Comoviridae, and Sequiviridae families have small VPg proteins covalently linked to the 5′ ends of their RNA genomes. CREs that are used as templates to nucleotidylylate the protein primers of these viruses have not been described, but may be discovered in the coming years.

13. Conclusions

Cis-active RNA sequences and structures within picornavirus RNAs coordinately interact with viral replication proteins to mediate the replication of viral RNA. One of the cis-active RNAs within the genomes of picornaviruses, the CRE, is a template for the conversion of VPg into VPgpUpU_OH. VPg and/or VPgpUpU_OH prime the replication of viral RNA, thereby becoming covalently linked to the 5′ ends of viral RNA. Progress in understanding the replication mechanisms of picornavirus RNAs will likely be informative and relevant to related viruses in the Dicistroviridae, Comoviridae, and Sequiviridae families of the Picornavirales order (Le Gall et al., 2007). We anxiously anticipate the new discoveries that will ultimately lead to a clear and theoretically satisfying explanation for the asymmetric replication of picornavirus RNA.

Acknowledgments

Footnotes

References