Cap-independent translation of proteins in Viruses

Some (+) ssRNA plant viruses, including those of the large families Potyviridae and Comoviridae, have replaced the cap by a small protein, VPg (Viral Protein genome-linked), which is covalently attached to the first nucleotide of the mRNA. Potyviruses and comoviruses closely resemble animal picornaviruses and caliciviruses in genome structure and in having a VPg. However, all these viruses possess a conventional poly(A) tail. Lacking a cap, they seem to rely on an internal initiation pathway of translation.

The internal ribosome entry site (IRES) is an RNA sequence that promotes direct binding of the 40S ribosomal subunit to internal regions of mRNA, usually upstream of the major ORF start site, thus skipping the requirement for the cap-binding protein, eIF4E (Hellen and Sarnow 2001). The IRESes of animal picornaviruses were the first to be discovered and remain the most extensively studied RNA elements that drive internal initiation. The RNA genomes of picornaviruses contain a very long structured leader with an IRES that promotes entry of the 40S ribosomal subunit near the AUG start codon for the major polyprotein. Direct binding of eIF4G to the encephalomyocarditis virus (EMCV) IRES is required for 40S ribosomal subunit entry via IRES/eIF4G/ eIF3/40S interactions (Pestova et al. 1996a, 1996b). A similar mechanism appears to operate in plants for at least two representatives of picorna-like potyviruses: tobacco etch virus (TEV, Carrington and Freed 1990) and turnipmosaic virus (TuMV, Basso et al. 1994).

The 5′-UTRs of TuMV and TEV differ from those of animal picornaviruses by their smaller length and less complex secondary structure. Zeenko and Gallie (2005) demonstrated that a pseudoknot element (PK1) of the TEV leader is a core structure of the IRES and is sufficient to promote cap-independent translation. However, under these conditions translation still required eIF4G (Gallie 2001). Binding of the 40S ribosome to the IRES is apparently mediated by eIF4F and eIFiso4F, since direct binding of eIF4G and eIFiso4G, as well as their complexes (eIF4F and eIFiso4F), to the TEV 143-nt 5′-leaderand PK1 was recently demonstrated, whereby eIF4G (or eIF4F) possesses stronger affinity to the TEV leader than does eIFiso4G (Ray et al. 2006).

The IRES preferentially recruits eIFiso4F to form the 48S preinitiation complex. At the same time, it is puzzling that the 5′-end of TEV genomic RNA can recruit eIFiso4E via VPg protein and that this interaction is strong enough to compete with cap binding in Arabidopsis (Miyoshi et al. 2006) despite the fact that VPg and cap bind to different sites of eIF4E (Léonard et al. 2000). Corresponding interactions between eIF4G, eIF4E and VPg have recently been demonstrated for lettuce mosaic virus (LMV; Michon et al. 2006) and TuMV (Miyoshi et al. 2006). Although the role of eIF4G/eIF4E/VPg complexes in regulation of translation initiation is not yet clear, their participation in IRES-mediated translation initiation has been proposed for TuMV (Khan et al. 2006). Indeed, in mammalian positive-stranded RNA caliciviruses, translation initiation is strictly dependent on an interaction between VPg and eIF4E, where VPg seems to substitute for the cap (Goodfellow et al. 2005). Whether VPg can substitute for cap in plants still needs to be clarified. The fact that PABP was found to be required for full IRES function invitro (Gallie 2001) suggested that TEV genomic RNA can be circularized via eIF4G, eIF4iso4G and PABP. The RNA polymerase (RdRp) of Zucchini yellow mosaic virus (ZYMV; Wang et al. 2000) and VPg-Pro of TuMV (Léonard et al. 2004) interact with PABP and can achieve circularization between 5′- and 3′- termini to facilitate replication. Although shutdown of the host translation machinery by plant viruses was not demonstrated, increasing concentrations of RdRp might interfere with the translation process.

VPg can recruit eIF4E (or eIFiso4E) for the benefit of the virus in different host plants: eIFiso4E recruitment by TEV VPg is required to infect Arabidopsis plants, while eIF4E recruitment is required for infection of solanaceous species (for a review, see Robaglia and Caranta 2006). The respective resistance genes of naturally occurring potyvirus resistant plant varieties have been shown to encode defective eIF(iso)4E or eIF4E (Lellis et al. 2002; Ruffel et al. 2002). Interestingly, involvement of a VPg/eIFiso4E complex in cell-to-cell movement or viral transport has also been suggested (Dunoyer et al. 2004; Gao et al. 2004).

Putative, unstructured IRESes of tobamoviruses

IRESes identified in plant tobamoviruses are unusual in that they are very short and unstructured. The RNA genome of tobamoviruses, of which tobacco mosaic virus (TMV) strain U1 is the type member, is a polycistronic capped RNA containing four large ORFs. The first ORF and its read-through ORF, which encode the two components of viral replicase, are translated directly from the genomic RNA. The virus produces two subgenomic (sg)RNAs: the dicistronic I2 RNA is used for translation of movement protein (MP), and a monocistronic RNA codes for CP (Palukaitis and Zaitlin 1986). Another member of the tobamoviruses, a crucifer-infecting virus (crTMV; Dorokhov et al. 1994), has a similar organization but its MP and CP ORFs overlap. crTMV harbours two unusual IRESes, the 148 nt  and the 75 nt IRESMP, which are thought to drive translation of the CP (Ivanov et al. 1997) and MP (Skulachev et al. 1999) ORFs, respectively. Both contain a purine- rich tract upstream of the AUG start codon. Interestingly, IRESCP can function in plants, yeast and HeLa cells (Dorokhov et al. 2002). However, the functional significance of these IRESes is obscure since both MP and CP canbe also translated from sgRNAs by a ribosome scanning mechanism (Ivanov et al. 1997; Skulachev et al. 1999). Although the IRESCP region is highly conserved between crucifer- infecting tobamoviruses, it differs from other tobamoviruses (Ivanov et al. 1997). Another very short purine-rich IRES element was identified upstream of the MP ORF of TMV U1 (Skulachev et al. 1999), which could be preferentially used for TMV U1 I2 RNA due to the possible lack of a 5′ cap in this RNA. The short purine-rich stretch was suggested as an IRES module that can mediate internal initiation in plants, mammals and yeast (Dorokhov et al. 2002).

Poly(A)-independent initiation

The positive-strand RNA genomes of Bromoviridae and Tobamoviridae have a 5′-cap and a non-polyadenylated 3′- end. These viruses use a strategy similar to that of cap- and poly(A)-containing RNA viruses but their 3′-UTRs can functionally substitute for a poly(A)-tail and work as translational enhancers in concert with the cap structure to ensure preferential translation of the viral genome. These 3′- UTRs contain a tRNA-like structure (TLS), either combined with a series of stem-loops [alfalfa mosaic virus (AMV)], or fused to a series of pseudoknots [(brome mosaic virus (BMV), turnip yellow mosaic virus (TYMV) and TMV]. These elements provide functions similar to that of the poly(A) tail in translation initiation by using different protein partners (AMV – Krab et al. 2005; BMV – Barends et al. 2004, TMV – Gallie 1991; TYMV – Matsuda and Dreher 2004). The TMV and TYMV 3′-UTRs enhance translation synergistically together with a 5′-cap structure (Matsuda and Dreher 2004), which suggests a 5′–3′ molecular bridge between their RNA termini.

All genomic and subgenomic TMV RNAs are co-linear and contain the same 3′-UTR, composed of a three RNA pseudoknot domain followed by a TLS, which can be specifically aminoacylated and can interact with eEF1A/GTP (Mans et al. 1991). Using a protoplast system, Leathers et al. (1993) revealed the critical role of the region upstream of the tRNA-like structure pseudoknot domain in enhancement of TMV RNA translation, where it acts synergistically with the 5’-cap. The heat shock protein, HSP101, binds specifically to this upstream pseudoknot domain and might mediate translation enhancement (Tanguay and Gallie 1996). The same pseudoknot cross-links specifically to eEF1A in an aminoacylation-independent manner (Zeenko et al. 2002). However, additional experiments are required to confirm the possible involvement of eEF1A or HSP101 in the synergy between TMV 5′- and 3′-UTRs. Although the 3′-UTR of TYMV, which comprises a TLS and a single upstream pseudoknot, acts synergistically with 5′-cap to enhance translation in a manner similar to that found in TMV, the TLS structure itself is required for translational enhancement (Matsuda and Dreher 2004). Moreover, only aminoacylation-competent TLS, which is able to tightly bind eEF1A-GTP, is active in translation activation (Dreher et al. 1999), while the pseudoknot seems to provide optimal spacing to present the TLS for aminoacylation (Matsuda and Dreher 2004).

In addition, eEF1A binding to the acylated TLS represses TYMV RNA replication by RNA-dependent RNA polymerase (Matsuda et al. 2004b). Thus, although the tRNA-like mimicry of TYMV RNA is clearly required for synergy in translational enhancement, the requirement for eEF1A in building a bridge between the 3′- and 5′-UTRs remains to be directly demonstrated. Direct 3′- and 5′-UTR interactions have been proposed in Alfalfa mosaic virus (AMV, family Bromoviridae), where they are mediated by coat protein (CP; Guogas 2004; Krab et al. 2005). The AMV RNA 3′-UTR adopts two alternative structures, of which only the one with several AUGC repeats separated by hairpins forms a strong interaction with at least two molecules of CP (Neeleman et al. 2004; Krab et al. 2005). CP enhances translation of AMV RNA in vivo 50- to 100-fold in the presence of the cap structure (Neeleman et al. 2004) and interacts with the eIF4G and eIFiso4G involved in formation of eIF4F and eIFiso4F, respectively (Krab et al. 2005). The complex between the AMV 3′-UTR and CP seems to mimic the PAPB/ poly(A) complex in that both can recruit eIF4F, thus converting the viral RNA into a closed loop structure.

Some 5′-UTRs of capped viral RNAs can enhance translation independently of the cap structure, using other host factors to recruit eIF4F or eIFiso4F or an internal initiation mechanism. The 68 nt TMV 5′-leader, the well known and extensively studied Ω sequence, acts as a translation enhancer in both plant and animal species, and is now widely used in biotechnological applications. The poly (CAA) region of Ω has been shown to mediate elevated translation via recruitment of the heat shock protein HSP101 (Wells et al. 1998b). The complex between Ω and HSP101 might mimic eIF4E/5′-cap or PAPB/poly(A) interactions in that it efficiently recruits eIF4F in order to increase translation (Gallie 2002). Genetic analysis has demonstrated that Ω-based enhancement also requires eIF3, which can be recruited via eIF4F (Wells et al. 1998b). The sequences of some plant viral leaders that can accomplish translation enhancement independently of the 3′-UTR show at least partial complementarity to the central region of 18S rRNA. These leaders bind to the 43S preinitiation complex in a cap- and eIF-independent manner (Akbergenov et al. 2004).