tRNA-mimicry in IRES-mediated translation and recoding

ABSTRACT

Viruses maintain compact genomes that must be packaged within capsids typically less than 200 nanometers in diameter. Therefore, instead of coding for a full set of genes needed for replication, viruses have evolved remarkable strategies for co-opting the host cellular machinery. Additionally, viruses often increase the coding capacity of their own genomes by employing overlapping open reading frames (ORFs). Some overlapping viral ORFs involve recoding events that are programmed by the viral RNA. During these programmed recoding events, the ribosome is directed to translate in an alternative reading frame. Here we describe how the Dicistroviridae family of viruses utilize an internal ribosome entry site (IRES) in order to recruit ribosomes to initiate translation at a non-AUG codon. The IRES accomplishes this in part by mimicking the structure of a tRNA. Recently, we showed that the Israeli Acute Paralysis Virus (IAPV) member of the Dicistroviridae family utilizes its IRES to initiate translation in 2 different reading frames. Thus, IAPV has evolved an apparently novel recoding mechanism that reveals important insights into translation. Finally, we compare the IAPV structure to other systems that utilize tRNA mimicry in translation.

Ribosomes translate the mRNA code to synthesize protein via the addition of amino acids onto the growing nascent polypeptide chain. At the heart of this is the tRNA, which provides the structural and molecular link between the amino acid and the decoding anticodon that deciphers each codon within the open reading frame (ORF). Protein synthesis proceeds with high fidelity; maintenance of the reading frame as ribosomes translocate on the mRNA occurs through coordinated conformational changes as tRNAs traverse from the ribosomal A to P to E sites.¹ Recognition of the correct aminoacyl-tRNA is ensured through contributions from thermodynamic and kinetic quality controls.^2,3 Studies on viral mechanisms that alter translational reading frame, called programmed recoding, have been particularly illuminating as some RNA viruses with compact genomes have adapted recoding strategies to increase coding capacity of their genomes. Well-described mechanisms include ribosome readthrough, −1 and +1/−2 frameshifting and ribosome bypass.⁴ In many cases, these mechanisms involve cis-acting RNA structures that interact with the translating ribosome to alter the reading frame. Recently, we uncovered a novel mechanism within the Dicistroviridae family whereby an RNA element, called an internal ribosome entry site (IRES), adopts a tRNA-like structure to direct translation of 2 overlapping reading frames.⁵ In this RNA Biology POV, we briefly describe the current model of this IRES mechanism and highlight our major findings of a viral RNA that impacts ribosome function and fidelity.

Factorless IRES in dicistroviridae

Dicistroviruses are RNA viruses that mainly infect arthropods.⁶ There are currently 20 members of this viral family that include Cricket paralysis virus (CrPV) and Drosophila C virus (DCV), both of which can infect Drosophila melanogaster, Taura syndrome virus (TSV) that infects penaeid shrimp, and the honey bee viruses: Israeli acute paralysis virus (IAPV), Kashmir bee virus (KBV) and Acute paralysis bee virus, which have been associated with honey bee disease and colony collapse disorder.^7,8 CrPV and DCV have been useful models for uncovering fundamental virus-host interactions using virus infections in Drosophila.^9-11 Within their ∼9 kb positive-strand, monopartite RNA genomes, dicistroviruses contain 2 main open reading frames (ORFs) encoding the viral structural and non-structural proteins (Fig. 1A). Each ORF is driven by a distinct IRES within the 5′untranslated region and the intergenic region (IGR), thus allowing independent regulation and temporal control over each ORF to be translated.^12,13 An IRES is a cis-acting RNA element that in general adopts an RNA structure capable of recruiting the ribosome using limited translation factors.^14,15 The dicistrovirus IGR IRES mechanism is one of the most unusual as it possesses several unprecedented properties to drive translation.

Figure 1.

(A) Location of the dicistrovirus IGR IRES between ORF1 and 2, which code for the non-structural and structural proteins, respectively. VPg is viral protein, which is covalently linked to the 5′ end of the viral RNA. The expanded region shows the secondary structure of the IGR IRES, color-coded to emphasize the 3 different pseudoknot (PK) regions, the loop region L1.1, and the stem-loops SLIII, SLIV and SLV. VLR, variable linker region. (C) Diagram of the proposed IGR IRES factorless translation mechanism. The IGR IRES first extends the PKI domain into the A site of the ribosome. After eukaryotic elongation factor 2 (eEF2)-assisted pseudotranslocation of PKI into the P site, an amino-acyl tRNA^Gly is delivered by eEF1A (thick black arrows). Figure adapted from Journal of Virology⁵³.

Unlike other IRES mechanisms that utilize limited translation factors, the IGR IRES uses a highly streamlined mechanism that can directly recruit ribosomes without any factors or initiator Met-tRNA_i (Fig. 1B).¹⁶ Additionally, the IGR IRES initiates translation from a non-AUG codon.¹⁶ The ∼200 nucleotide IRES adopts a triple pseudoknot (PKI, PKII and PKIII) structure, which is conserved within the Dicistroviridae and is responsible for factorless translation; PKI is responsible for directing translation at a non-AUG codon, and PKII and PKIII form a compact core that directs ribosome recruitment (Fig. 1A).^17-23 Since the initial discovery of the CrPV IGR IRES in 2000,¹⁶ several groups have used biochemical, biophysical and structural approaches to investigate the mechanism of IGR IRES translation. The IGR IRES recruits the ribosome by making several specific contacts with each ribosomal subunit spanning all ribosomal E, P and A sites (Fig. 2). Within the PKII/PKIII domain, stem-loop (SL) IV and SLV both protrude to interact with the uS7 and eS25 proteins of the 40S subunit, whereas the L1.1 domain binds to the L1 stalk of the 60S subunit.^24-28 The PKI base pairing resembles an anticodon-codon base pairing, thus enabling it to occupy the ribosomal A site and in effect, the IRES acts as its own mRNA in cis.^25,26,29 The IRES then undergoes 2 rounds of translocation using elongation factors eEF1A and eEF2; The PKI, which is initially in the A site, is translocated to the P site by eEF2, leaving the A site empty for the delivery of the first aminoacyl-tRNA (Fig. 1B).^30-34 Another round of translocation results in movement of the aminoacyl-tRNA from the A to the P site. These two IRES-mediated translocation steps are called pseudotranslocation because translocation occurs without peptide bond formation. Recent kinetic and single molecule analyses show that an aminoacyl-tRNA can be delivered to the ribosome in the absence of eEF2, suggesting that translocation may occur spontaneously.^35,36 In general, structural, biochemical and biophysical approaches are all in agreement that eEF2 promotes robust translocation of PKI from the ribosomal A to P site to allow delivery of the first aminoacyl-tRNA.^5,32,33,36 In essence, the IRES uses its RNA structure to actively manipulate the ribosome by docking into its conserved core, setting the reading frame and driving factorless translation. Because the IRES commandeers a fundamental step of translation, it is not surprising that the IRES functions with ribosomes from yeast to mammals and possibly prokaryotes as well.^16,37-39 Thus, the dicistrovirus IRESs have served as excellent model systems, allowing researchers to explore fundamental aspects of ribosome function and ribosome reading frame selection.

Figure 2.

(A) Cryo-EM structure of the CrPV IGR IRES bound to the 80S ribosome (PDB ID 4V91)(Figure republished from Cell³² licensed under CC by 3.0. (B) Zoom-in view of the IRES (colored as in Fig. 1). The positions of the A, P and E site tRNAs (light gray) are superimposed.

IRES-mediated Reading frame selection

Using a bioinformatics approach, Sabath et al. and Firth et al. discovered a “hidden” alternate +1 frame ORF immediately downstream of a subset of dicistrovirus IGR IRESs, including those in the honeybee viruses and a fire ant virus.^40,41 Within these IRESs, we found that a subset of ribosomes recruited to the IGR IRES can direct translation in either of 2 overlapping reading frames, with the 0 frame encoding the viral structural proteins and a +1 frame, termed ORFx, encoding a protein of unknown function (Fig. 3A).⁴² A U-G wobble base pair immediately adjacent to the PKI anticodon-like domain is important for setting the ribosome in the +1 reading frame.⁴² We recently showed that other elements are also key for reading frame selection by the IGR IRES: mutations within PKI result in nearly exclusive 0 or +1 frame translation.^5,43 Our most recent findings demonstrate that the PKI domain mimics a complete tRNA, thus explaining how these IRESs can gain access to the ribosomal decoding center to direct reading frame selection (Fig. 3A).⁵ In honeybee and fire ant dicistrovirus IGR IRESs, the PKI domain contains an extra SLIII, thus forming a 3-helix junction. Using a SAXS/NMR hybrid approach, we found that the IAPV IRES adopts the same L-shape as a tRNA; the SLIII resembles the acceptor arm of a tRNA and the PKI and P3.2 helices mirror an anticodon stem (Fig. 3A). The PKI tRNA mimic docks within the ribosomal A and P sites, thus suggesting that the IRES can occupy the same tRNA sites within the ribosome. Despite appearing to mimic an anticodon:codon interaction, in isolation the PKI anticodon stem does not extend further than the anticodon stem loop (ASL) of a tRNA. However, once the PKI domain pseudotranslocates to the ribosomal P site, it is possible that the PKI domain flexes and extends further to mimic a tRNA-mRNA interaction to direct translation from the A site and set the reading frame. Indeed, recent cryo-EM structures of an eEF2-mediated translocated IRES within the ribosome supports this notion.^44,45 Unlike the IAPV IRES, the cryo-EM structure of the related TSV IRES bound to the ribosome shows that it binds to the ribosomal A site with the SLIII of the TSV IRES coaxially stacked in a more linear conformation on top of the PKI, rather than in a bent conformation resembling the elbow of a tRNA.⁴⁶ The different orientations observed for SLIII helices of the TSV and IAPV PKI domains may be due to sequence differences, dynamic motions, and/or interactions with the ribosome since the IAPV structure was solved in isolation whereas the TSV IRES structure was solved bound to the ribosome. Indeed, the orientation of the SLIII helix in the TSV IRES appears to be stabilized by an interaction with the rRNA helix h38.⁴⁶ It is possible that upon binding to the ribosome, the SLIII of the IAPV IRES adopts a similar conformation to that observed with the TSV IRES. Conversely, it is also possible that both the IAPV and TSV IRES adopt L-shaped structures during translocation. Given the tRNA-like mimicry of the PKI domain of the IAPV IRES, it seems probable that the IRES follows the path of a tRNA as it translocates through the ribosome. Another difference between the 2 IRESs is that the TSV IRES cannot support +1 frame translation (Qing S. Wang and Eric Jan, unpublished data), and thus may not adopt conformations that support translation in different reading frames. Of note, the length of the TSV SLIII helical stem (6 base pairs) is shorter than that of the IAPV SLIII (8 base pairs), suggesting that the IRESs may interact with the ribosomal core differently to direct translation in a reading frame. Given the recent advances in cryo-EM technologies that have yielded structures of the CrPV and TSV IRESs bound to the ribosome, it will be of interest to determine how the IAPV IRES interacts with the ribosome during translocation, and compare those results to the TSV IRES.

Figure 3.

(A) Secondary structure of the IAPV IRES. (B) NMR/SAXS structure of the IAPV IRES PKI domain (PDB ID 2N8V) superimposed over tRNA^Phe (PDB ID 4TNA, light gray spheres). (C) Secondary structure of tmRNA showing the tRNA-like domain (TLD), mRNA-like domain (MLD), Helices (H), and Pseudoknots (PK). (D) Structure of the tmRNA (PDB ID 4V6T, ref.⁵⁴). Reproduced from Frontiers in Microbiology⁴⁹ licensed under CC. (E) Secondary structure of Turnip yellow mosaic virus tRNA-like structure (TYMV TLS). F. Structure of the TYMV TLS (PDB ID 4P5J,⁵¹). Reproduced from Nature⁵¹ with permission.

How does the IGR IRES direct ribosomes in a specific reading frame?

Normally, the maintenance of a reading frame during translation is set by the anticodon-codon tRNA interaction as well as coordinated ribosome dynamics such as ratcheting as translocation proceeds via the eEF2 translocase.⁴⁷ In the case of the dicistrovirus IRESs, the IAPV IGR IRES supports +1 frame translation at a level of 20% of 0 frame translation, suggesting that a subset of ribosomes recruited to the IRES is set in one or the other frame.⁴² Extensive mutagenesis analyses of the PKI domains suggest that specific mutations can induce conformations that exclusively direct 0 or +1 frame translation,^5,42,43 thus leading to the hypothesis that the dynamic nature of the IRES allows it to adopt distinct conformations that direct the ribosome into the reading frame. The mutant IRESs show different reactivities to chemical probes, suggesting that they adopt distinct conformations responsible for 0 vs. +1 reading frames selection.^5,43 It is possible that the IRES samples different conformations or intermediate states that impact reading frame selection. The integrity and length of SLIII (which resembles the tRNA acceptor stem) are key for 0 frame translation but not +1 frame translation.⁵ Alterations in the length of the helical stem results in loss of 0 frame but not +1 frame translation, suggesting that the tRNA acceptor stem-like shape of SLIII may form interactions within the ribosome similarly to those of a normal tRNA to set the reading frame and direct 0 frame translation. However, the angle of SLIII relative to the rest of the PKI domain may be important for reading frame selection, as mutations of 2 unpaired adenosines in the 3 way junction of PKI affects 0 and +1 frame translation, depending on the mutation.⁵ In summary, it is likely that mimicry of a tRNA is important for occupancy in the ribosomal A and P sites to set the ribosome in the 0 frame and that the IAPV IRES directs the ribosome into a specific reading frame via distinct conformations.

Upon initial ribosome binding, the PKI domain first occupies the ribosomal A site and then translocates to the P site. To determine the step at which reading frame is selected, toeprinting analysis was used to monitor the precise position of the ribosome on the mutant IRESs that direct exclusive 0 or +1 frame translation.⁵ These data indicate that the reading frame may not be selected upon ribosome recruitment to the IRES (i.e. PKI in the A site) and occurs during or after translocation.^5,43 A recent study using single-molecule approaches and the CrPV IRES supports this model.³⁶ In this study, the authors found that the CrPV IRES can support +1 frame translation in vitro and that reading frame is determined by the arrival of the 0 or +1 frame aminoacyl-tRNA to the A site after translocation of PKI to the P site. Interestingly, the G of the first 0 frame codon of the CrPV IRES interacts with C1273 of the 18S rRNA. The significance of this interaction is not known but may contribute to reading frame selection. It will be important to determine whether the CrPV IRES can direct alternate reading frame translation in vivo in a manner similar to that of the IAPV IRES. The recent cryo-EM structures of translocated ribosomes on the CrPV and TSV IRESs also revealed that the tRNA-like PKI domains adopt several conformations, many of which resemble translocation of hybrid-state tRNAs through the ribosome, and that these conformations are associated with ribosome dynamics such as ratcheting and L1 binding to the E site tRNA.^44,45 It will be interesting to determine whether reading frame selection by the IAPV IRES occurs during these stages of IRES-mediated translocation. Finally, recent kinetic analyses of the first 2 pseudotranslocation steps show they proceed relatively slowly compared to subsequent translocation steps.³⁵ The significance of the slow translocation steps is not clear but they may involve contributions to reading frame selection. Given the findings that these dicistrovirus IRESs direct factorless translation initiation by manipulating and exploiting intrinsic ribosome conformations, we anticipate that studies on these IRESs will shed light into the detailed mechanisms that are fundamental to reading frame selection.

tRNA mimicry

tRNA (tRNA) achieves its L-shaped topology through tertiary interactions involving the D and T loops. The tRNA mimicry by the IAPV is unusual in that it is achieved with a topology that is completely different from the 4-way junction or “cloverleaf” structure of tRNA.⁵ The IAPV PKI domain also contains 4 helices, but they are derived from a 3-way junction plus a pseudoknot (Fig. 3A and B). In tRNA, the T and D loops interact via long-range base pairs to form the elbow of the L-shape conformation. In the IAPV PKI domain, the elbow region is formed by 2 unpaired adenosines wedged into the center of the 3-way junction. The adenosines are in close proximity but confirmation of a potential interaction should come from a higher resolution model of the IRES bound to the ribosome. Mutating the unpaired adenosines affects the structure of PKI and leads to alterations in reading frame selection.⁵ Given that tRNAs are heavily modified post-transcriptionally and that these RNA modifications can affect translational fidelity,⁴⁸ it would be of interest to investigate whether the IAPV PKI tRNA-like domain is modified.

The use of full or partial tRNA mimicry is also employed by other regulatory RNAs, most notably the tmRNA, which mediates trans-translation on “non-stop” mRNAs in quality control surveillance mechanisms in prokaryotes.⁴⁹ tRNA mimics are also found in the 3′untranslated regions of some plant RNA viruses that contribute to viral protein synthesis.⁵⁰ These 2 types of RNAs also participate in translation, and unlike the IAPV IRES, can be aminoacylated. The tmRNA only partially mimics the acceptor stem of a tRNA (Fig. 3C and D), and like the dicistrovirus IRESs, acts in cis- as its own mRNA to initiate translation. The tRNA-like structure (TLS) in some plant viral 3′UTRs mimics the full tRNA structure, similar to the IAPV PKI domain, but do so through a set of noncanonical intramolecular base pair interactions (Fig. 3E and F).⁵¹ Instead of a 4-way junction observed in tRNA, the TLSs uses a ‘linchpin’ interaction involving a pseudoknot near the 5′end of the TLS, which stabilizes the global structure and brings the D and T loop-like structures together. The topology of the TLS may allow regulatory folding and unfolding to permit dual functionality in viral translation and replication. In the case of the IAPV IRES, the tRNA-like shape manipulates the ribosomes to start translation in either of 2 reading frames. Future structural studies of the mutant IAPV IRESs that direct exclusive 0 or +1 frame translation, both in free and ribosome bound forms, should provide clues into the intra- and inter-molecular interactions that stabilize the tRNA-like PKI structure. tRNA mimicry is not only restricted to regulatory RNAs but also translation termination factors that mediate translation termination and ribosome recycling.⁵² Thus, given these examples of tRNA mimicry and the fact that tRNA plays a central and conserved role in the evolutionary link between RNA and protein synthesis, it is enticing to think that tRNA mimicry may be more widespread than initially thought and may contribute to the regulation of other RNAs.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Hilda Au for careful reading of the paper. SEB. is supported by an NIH grant (R35 GM118131) and E.J. is supported by a CIHR grant (MOP-81244) and an NSERC Discovery grant (RGPIN 341459-12).