Viruses with single-stranded (ss) positive-sense (+) RNA genomes harbor various intra-genome RNA secondary structures and sequence motifs that play critical cis
-acting roles in their infection cycles [1
]. Among the best-known virus-encoded RNA secondary structures are the internal ribosomal entry site (IRES) elements found in many viruses that enable efficient translation of viral proteins by guiding ribosomes directly to the start codon [2
]. Different RNA sequence motifs or structures within the same or different viral genomic RNA (gRNA) are also known to engage in long-distance interactions in order to enhance the translation of viral genes, or to facilitate the synthesis of viral subgenomic RNAs (sgRNAs) [1
]. Additionally, many internally encoded stem–loop structures have been shown to exert diverse functions, including serving as the binding sites for viral RNA-dependent RNA polymerase (RdRP), as well as the initiation site of genome encapsidation [3
Despite their well-recognized roles in viral multiplication cycles, some of the RNA secondary structures are difficult to study because they often reside in coding sequences for important viral proteins, including RdRPs and auxiliary replication proteins (ARPs). For instance, many a (+) RNA virus encode RdRP by extending ARP at the C-terminus by avoiding the stop codon of the latter through either translational read-through or frame-shifting. Such read-through and frame-shifting events are tightly regulated through highly conserved RNA structures that simultaneously encode part of the RdRP [6
]. For RNA structures with roles in an early step of viral replication cycle, e.g., translation of ARP or RdRP, it is difficult to discern whether they also participate in later steps, like genome replication or assembly, because perturbing the early steps would in turn interfere with later steps.
Here we report an approach to uncouple the protein-coding capacities of viral gRNA from the RNA secondary structures it folds into, using turnip crinkle virus (TCV) as our model. TCV is a small (+) RNA plant virus that counts model plants Arabidopsis and Nicotiana benthamiana
as hosts [7
]. Its genome of 4054 nucleotides (nt) encodes five proteins (Figure 1
A). The 5′ proximal p28 and its read-through product (p88) are TCV-encoded ARP and RdRP, respectively, both needed for genome replication. The p8 and p9 movement proteins (MPs), and p38 capsid protein (CP), are translated from two sgRNAs produced during viral replication (Figure 1
A). CP is also the suppressor of RNA silencing-mediated host defense. However, it is not required for genome replication as long as the activity of RNA silencing is kept in check with a heterologous suppressor, such as p19 of tomato bushy stunt virus (TBSV) [8
Both p28 and p88 are translated directly from TCV gRNA, with the latter resulting from programmed translational read-through of the p28 stop codon. This translational read-through event is facilitated by a well-characterized, highly conserved RNA secondary structure known as a recoding stimulatory element (RSE), in coordination with another 3′ terminal structure through a long-distance kissing loop interaction [6
]. RSE also contains two internal stretches of sequences that interact with each other to form a pseudoknot structure necessary for its role in stimulating translational read-through [7
]. Earlier studies suggested that some of the sequences within RSE might also be important for TCV genome replication [13
]. However, this could not be easily examined because lower genome replication could also be caused by a lack of p88 resulting from RSE mutations that diminish the read-through translation.
In the current study, we assessed the translation-independent role of several previously identified structure/sequence elements within p28/p88 coding sequence by providing the p28/p88 proteins in trans
, thus obviating the need to maintain the protein-coding capacity and the structures that regulate protein translation within the p28/p88 coding region of TCV genome. This alternative approach allowed us to confirm a replicational role of a previously characterized structure, known as internal replication element (IRE) [5
]. More importantly, it permitted us to uncover a novel role of several RSE-resident structural elements in TCV gRNA accumulation. This novel approach should be easily adaptable to other viral RNAs, leading to further assessment of many known RNA structures and the identification of new structures.
We report here further examination of RNA secondary structures embedded in the coding sequence of two TCV replication proteins, p28 and p88. Some of these structures were previously shown to be essential for the translational read-through of the p28 stop codon in order to synthesize the p88 RdRP [6
]. However, it was unclear whether they had additional roles in TCV multiplication independent of translational read-through. Such roles were not easy to discern because mutations that disrupt the RNA structures needed for translational read-through (e.g., RSE) would block the translation of p88, hence abolishing replication, even if all RNA structures needed for replication remained undisturbed. Our alternative approach sought to provide p88, and later on, p28 as well, from separate replication-independent sources, in an attempt to bypass the need for translational read-through. This approach allowed us to make several important observations.
First, more than 1/3 of the 2328-nt p88-coding sequence can be removed without substantially compromising the replicability of the TCV genome relative to appropriate controls. This is evident from the notable replication levels of three deletion mutants: ΔFc (deleting 141 nt), Δ1208–2197 (deleting 990 nt), and Δsg1Pro (deleting 195 nt). Although the gRNA accumulation levels of these mutants were modest in comparison to the TCV_sg2R + p88 control, they were similar to another control replicon (813UAA + p88) incapable of producing any p88 by itself, suggesting a slight advantage of cis
-produced p88. Second, production of TCV sgRNA does not always depend on the synthesis of gRNA. In the presence of p88 in trans
, two deletion mutants, ΔFa* and ΔIRE, as well as several other mutants, failed to produce detectable levels of gRNA but exhibited normal levels of sgRNA. This observation indicates that minus-strand RNA synthesis is still active under these conditions, as sgRNA-sized minus-strands are required as templates for sgRNA transcription [18
]. Thus, the deleted RNA elements could be required for (i) completion of full-length minus-strand synthesis and/or (ii) efficient initiation and synthesis of the full-length plus-strands from a full-length minus-strand template. The uncoupling of sgRNA transcription from genome replication based on modifications to the RdRP has been reported for tombusviruses [19
]. Additionally, the synthesis of gRNA and sgRNA may occur in different microenvironments. This is consistent with recent observations of different sized replication organelles in cells replicating flock house virus that correspond to the different sized genomic and subgenomic RNAs [20
]. Conversely, additional cis
-acting elements in gRNA could allow replication proteins to discriminate against faulty templates.
Third, structures implicated in translational read-through can have additional roles in genome replication. This is best illustrated by the pseudoknot structure residing in the RSE of TCV, which was found earlier to be needed for efficient translational read-through. We found that it was also needed for efficient accumulation of TCV gRNA, probably through its participation in genome replication. It is possible that this dual function could be related to coordinating translational read-through with genomic minus-strand synthesis, which are directionally opposing processes on the genomic RNA. In this scheme, read-through would unfold the pseudoknot needed for full-length genome minus-strand synthesis, thereby inhibiting this competing process. The maintenance of sgRNA accumulation when the pseudoknot is disrupted suggests that the inhibition may be specific for synthesis of full-length genomic minus-strands. RNA-mediated coordination of these two processes occurs in tombusviruses, where formation of the long-range interaction between the RSE and 3′ UTR required for RdRP read-through concomitantly prevents formation of an alternative RNA structure in the 3′UTR needed for minus-strand RNA synthesis [6
]. More detailed studies are required to address possible regulatory effects of the RSE in TCV.
The integrity of the middle portion of the RSE (stem I) may also serve as an additional sentinel for quality control of TCV genome. It is possible that involvement of stem I in TCV replication could entail collaboration with p28 translated from the same RNA, because the reduction in gRNA levels caused by mA2 mutations was much less pronounced in the p28TS replicon backbone (in need of both p28 and p88 in trans
) than in the 813UAA backbone (needing only p88 in trans
). Finally, the RSE element could be under dynamic regulation through the folding of the alternative Basal structure that encompasses the Ff element within the p28 coding region, as mutating or deleting Ff appeared to also affect the robustness of TCV replication. Again, the Basal structure appears to be less critical when the replicon genome produces its own p28, as the m2-1 mutant would be expected to have disrupted the FfFa stem extensively, yet could replicate to 70% of its 813UAA parent (Figure 2
Finally, we wish to address why the replication of TCV mutants with trans
-provided p28 and p88 was inefficient compared to that of TCV replicons encoding their own p28 and p88. Multiple factors could have contributed to this inefficiency. First, both p28 and p88 were shown previously to repress TCV replication when overexpressed [10
]. Assuming the heterogeneity of their expression in different cells, it is possible that complementation might have occurred in a fraction of cells where the threshold for repression was not reached. It is also possible that efficient replication requires these two proteins to be present at specific intracellular concentrations and/or ratio, or be produced in a temporarily regulated manner. These conditions would be difficult to meet with our experimental setup. Lastly, cis
-production of these replication proteins may indeed be favored by the replication process [21
]. Nevertheless, our new approach did allow for the revelation of novel translation-independent roles of several RNA secondary structures within the p28/p88 coding sequence. This approach may prove valuable for the examination of similar RNA structures in other viruses, leading to a better appreciation of the role played by these structures.