ABSTRACT
Tobacco necrosis virus, strain D (TNV-D), is a positive-strand RNA virus in the genus Betanecrovirus and family Tombusviridae. The production of its RNA-dependent RNA polymerase, p82, is achieved by translational readthrough. This process is stimulated by an RNA structure that is positioned immediately downstream of the recoding site, termed the readthrough stem-loop (RTSL), and a sequence in the 3′ untranslated region of the TNV-D genome, called the distal readthrough element (DRTE). Notably, a base pairing interaction between the RTSL and the DRTE, spanning ∼3,000 nucleotides, is required for enhancement of readthrough. Here, some of the structural features of the RTSL, as well as RNA sequences and structures that flank either the RTSL or DRTE, were investigated for their involvement in translational readthrough and virus infectivity. The results revealed that (i) the RTSL-DRTE interaction cannot be functionally replaced by stabilizing the RTSL structure, (ii) a novel tertiary RNA structure positioned just 3′ to the RTSL is required for optimal translational readthrough and virus infectivity, and (iii) these same activities also rely on an RNA stem-loop located immediately upstream of the DRTE. Functional counterparts for the RTSL-proximal structure may also be present in other tombusvirids. The identification of additional distinct RNA structures that modulate readthrough suggests that regulation of this process by genomic features may be more complex than previously appreciated. Possible roles for these novel RNA elements are discussed.
IMPORTANCE The analysis of factors that affect recoding events in viruses is leading to an ever more complex picture of this important process. In this study, two new atypical RNA elements were shown to contribute to efficient translational readthrough of the TNV-D polymerase and to mediate robust viral genome accumulation in infections. One of the structures, located close to the recoding site, could have functional equivalents in related genera, while the other structure, positioned 3′ proximally in the viral genome, is likely limited to betanecroviruses. Irrespective of their prevalence, the identification of these novel RNA elements adds to the current repertoire of viral genome-based modulators of translational readthrough and provides a notable example of the complexity of regulation of this process.
INTRODUCTION
RNA plant viruses expand their coding capability by employing a variety of expression strategies. Recoding mechanisms, such as translational frameshifting and readthrough, provide a means to produce distinct proteins using the same initiation site (1). In translational frameshifting, the ribosome shifts frames before encountering its customary stop codon, often resulting in a C-terminally extended protein. Translational readthrough also produces elongated proteins; however, this process involves the normal stop codon being decoded by a near cognate tRNA, thereby allowing for C-terminal polypeptide extension (1). Many different factors, such as RNA sequences and higher-order RNA structures (1–11), RNA remodeling (12), and viral (13, 14) or host (15, 16) proteins, have been implicated in regulating recoding in eukaryotic viruses. Among these, the RNA elements that modulate translational readthrough in plus-strand RNA viruses have been the focus of many studies (1, 9).
Members of family Tombusviridae are plus-sense RNA viruses that express their RNA-dependent RNA polymerase (RdRp) by either −1 frameshifting or readthrough (17). In both cases, there is an extended stem-loop (SL) RNA structure located immediately downstream of the slippery sequence or stop codon that is required for efficient recoding (5, 6, 12, 18, 19). In addition, for maximal recoding to occur, a bulge in this RNA structure must base pair, via a long-range RNA-RNA interaction, with a sequence located 3′ proximally in the viral genome (5, 6, 18, 19). It has been proposed that this RNA-based communication with the 3′ end also assists in coordinating the directionally opposing processes of translation of the RdRp and minus-strand genome synthesis (5, 18, 19).
Tobacco necrosis virus, strain D (TNV-D), is a member of the family Tombusviridae (17), and the type member of the genus Betanecrovirus (20). Currently, three members of this genus have been sequenced: TNV-D (21), beet black scorch virus (BBSV) (22), and leek white stripe virus (LWSV) (23). These viruses possess monopartite, single-stranded, ∼3.8-kb plus-sense RNA genomes that lack both a 5′ cap and a 3′ poly(A) tail. Instead, they use a 3′-cap-independent translational enhancer (3′CITE) located in their 3′ untranslated regions (3′ UTRs) to recruit eukaryotic translation initiation factors (24–26). The TNV-D genome encodes five open reading frames (Fig. 1A) (21). Accessory replication protein p22 and the p82 RdRp share the same initiation codon, the latter being a readthrough product of the former (Fig. 1A). Readthrough in TNV-D, as for most tombusvirids, requires an extended RNA secondary structure, the readthrough stem-loop (RTSL) (Fig. 1B), located just 3′ to the p22 stop codon, which must interact with sequences in the 3′ UTR, via proximal and distal readthrough elements (PRTE and DRTE, respectively) (Fig. 1A) (6). Other 3′-proximal proteins, involved in virus movement (p7a and p7b) and packaging (p29, capsid protein), are translated from two subgenomic (sg) mRNAs that are transcribed during infections (27–31). A small amount of p29 capsid protein is also translated in vitro from the full-length TNV-D genome via a putative internal ribosome entry site (30). The relevance of this occurrence is currently unknown; however, a similar activity in another tombusvirid, the pelargonium flower break virus (genus Carmovirus), was shown to be important for efficient infection of plant hosts (32).
The TNV-D genome and mutational analysis of the RTSL. (A) Linear representation of TNV-D RNA genome with encoded proteins represented as gray boxes. Initiation sites for subgenomic (sg) mRNA1 and -2 are indicated below the genome, as well as the relative position of the 3′ cap-independent translational enhancer (3′CITE). Hatched black lines represent p22 protein and its readthrough product p82. Green double-headed arrows denote the long-range base-pairing interaction between the proximal readthrough element (PRTE) and distal readthrough element (DRTE). (B) Mfold-predicted RNA secondary structure of the readthrough stem-loop (RTSL) with the p22 stop codon in red and the PRTE in green. The gray boxes show wt and mutant forms of the PRTE bulge. Red arrows indicate the position of deleted nucleotides, with the corresponding amino acid deletions shown below and indicated by red lines on the RNA structure. (C) In vitro translation assay of TNV-D genomic RNA in wheat germ extract. The names of the TNV-D mutants are listed above their respective lanes, and the identities of the protein products (p82, p29, and p22) are shown on the left. The mock lane consists of a reaction performed with no viral RNA, and TNV represents a reaction completed with wt TNV-D genome RNA. Relative readthrough (Rel. RT) was calculated as the ratio of p82/p22, with that for wt TNV set as 100%. The corresponding means (± the standard errors where applicable) were calculated based on at least three independent assays. (D) Northern blot analysis of TNV-D RNA accumulation in plant protoplasts incubated for 22 h at 22°C. Shown above each lane is the name of the viral genome analyzed. The positions of the viral genome (g) and sg mRNAs (sg1 and sg2) are indicated to the left of the blot. The relative genome levels (Rel. g) are provided below, with the corresponding means (± the standard error, where applicable) calculated from at least three independent protoplast infections. (E and F) Substitutions in the apical loop and upper stem of RTSL are shown as red nucleotides. Relative readthrough and genome levels are provided below each mutant.
In this study, we investigated the RTSL and whether RNA sequences and structures that flank the known regulators of readthrough in TNV-D genome, i.e., RTSL and DRTE, could modulate the production of the p82 RdRp. Our results identified additional RNA structures, proximal and distal to the readthrough site, that are required for optimal readthrough in cell extracts and efficient viral genome accumulation in protoplasts. These findings expand the current documented assortment of RNA-based regulators and show that readthrough can be modulated by multiple discrete RNA elements.
RESULTS
The PRTE is critical for efficient readthrough and virus viability.To assess the importance of the size of the PRTE-containing bulge, viral genomic mutants containing deletions in multiples of three nucleotides were tested for their ability to produce p82 in wheat germ extract and to accumulate in protoplast transfections (Fig. 1B). The triplet nucleotide deletions were designed to precisely remove complete amino acids in p82 (i.e., methionine, glycine, and arginine). Removal of 3 or 6 nucleotides (nt) from the bulge in mutants Δ3 and Δ6, respectively, decreased translational readthrough to ∼10% that of wild type (wt), and eliminated genome accumulation in protoplasts (Fig. 1C and D, left panels). Although the Δ3 construct contained the complete PRTE, the 5′-proximal guanosine in the PRTE was predicted by Mfold to pair with the adjoining stem (Fig. 1B), and this could reduce its accessibility to the DRTE. To investigate this idea, a silent substitution of U to G was made in the nucleotide just 3′ to the PRTE, creating Δ3G (Fig. 1B). In this mutant, the introduced G was predicted to pair with the stem, thus allowing for presentation of the entire PRTE in the bulge. For Δ3G, there was an ∼2-fold increase in p82 accumulation levels over Δ3, suggesting that accessibility was enhanced; however, replication in protoplasts was not rescued (Fig. 1C and D, right panels). Complete deletion of the 9-nt bulge in mutant ΔPRTE, which generated an uninterrupted portion of helix in RTSL and a more stable RTSL (i.e., ΔG for wt RTSL = −50.22 kcal/mol and ΔPRTE RTSL = −57.64 kcal/mol, at 30°C), also yielded low levels of p82 production (∼8%) and no genome replication (Fig. 1C and D). Thus, all deletions in the PRTE bulge dramatically reduced p82 translation in vitro and eliminated genome accumulation in infections.
Structural features of the upper portion of the RTSL moderately affect viral genome levels but not readthrough.A previous study confirmed the importance of the stability of the RTSL stem for efficient readthrough (6). However, the potential role of other structural features in the RTSL, such as its apical loop and an AG mismatch in its stem, was not investigated (Fig. 1B). Substitutions in the apical loop that maintained the p82 amino acid sequence in mutants Lm1, Lm2, and Lm3 did not notably affect either readthrough or infectivity levels at 22°C (Fig. 1E). Elevating the incubation temperature of protoplast infections, which also increases the stringency of base pairing, can sometimes reveal defective phenotypes (33, 34). Testing the terminal loop mutants at the elevated temperature of 29°C resulted in decreases in genome levels to ∼57 to 75% that of the wt (Fig. 1E).
TNV-D, BBSV, and LWSV all contain a single mismatch in the upper portion of their RTSLs. However, only the AG mismatch in TNV-D (Fig. 1B) could potentially be replaced with a CG canonical pair without altering the p82 amino acid sequence; nonetheless, the virus does not implement this option. When the above-mentioned substitution was introduced into mutant USm1, wt levels were observed for readthrough and genome accumulation at 22°C, but at 29°C the genome levels dipped by ∼20% (Fig. 1F).
A conserved RNA hairpin upstream of the RTSL affects readthrough minimally and facilitates viral genome accumulation.To further explore RNA elements that could be involved in the readthrough production of p82, the region upstream of the RTSL was examined. A small GC-rich stem-loop structure, here termed Pre-RTSL, was identified 10 nt upstream of the p22 stop codon. Comparable RNA hairpins are also present in other members of the genus Betanecrovirus (Fig. 2A), and similarly positioned stem-loops are predicted in all tombusvirids (12). Silent mutations designed to disrupt the stem of Pre-RTSL in mutants Pre1 through Pre3, (Fig. 2B) reduced readthrough by ∼10 to 20% (Fig. 2C). In protoplasts incubated at 22°C, there were various degrees of genome accumulation, ranging from ∼55 to 97%, whereas at 29°C the level of all mutants was reduced by half (Fig. 2D and E). These results suggest a relatively minor role for Pre-RTSL in readthrough under our in vitro conditions, but a noteworthy role during infections.
Mutational analysis of Pre-RTSL. (A) Mfold-predicted secondary structure of Pre-RTSL element in betanecroviruses TNV-D, beet black scorch virus (BBSV), and leek white stripe virus (LWSV). The bottom of the stems for TNV-D and BBSV are separated from their cognate UAG stop codons by 10-nt spacers, whereas this spacer length is 3 nt in LWSV. (B) TNV-D Pre-RTSL and RTSL structures. Pre-RTSL mutants are shown in the gray box with substituted nucleotides in red. (C) In vitro translation analysis of Pre-RTSL mutants. (D and E) Northern blot analysis of protoplast infections with Pre-RTSL mutants.
RNA structures downstream of the RTSL.In TNV-D, there is a predicted tertiary RNA structure, here termed Post-RTSL (or Con-1), located immediately downstream of the RTSL (Fig. 3A). Compared to corresponding segments in other betanecrovirus genomes, regions of conservation included a base-paired stem structure (blue), and a potential pseudoknot (PK) forming interaction (yellow). The stems in BBSV and LWSV are extended by three base pairs and contain substitutions that maintain the helices (Fig. 3A, upper panel). The PK interaction is also well conserved and, notably, includes covarying base pairs (Fig. 3A, lower panel). In TNV-D, but not the other betanecroviruses, this region can adopt an alternate fold, Con-2 (Fig. 3B), which was predicted in 6 of the top 10 most thermodynamically stable structures when the complete TNV-D genome was analyzed by Mfold (Fig. 3C). Thus, either or both structures could be functionally relevant in TNV-D.
Comparative structural analysis of the Post-RTSL element in betanecroviruses. (A) Mfold-predicted structures for conformation 1 (Con-1) of the Post-RTSL for TNV-D, BBSV, and LWSV. Conserved stems and pseudoknots (PKs) are highlighted, respectively, in blue and yellow. Naturally occurring substitutions in BBSV and LWSV are shown in orange, and those that maintain base pairing are boxed. (B) Mfold-predicted structure for Con-2 of the Post-RTSL in TNV-D. The positions of nucleotides involved in the conserved stem and PK of Con-1 are highlighted in blue and yellow, respectively. (C) Mfold predictions of Con-1 or Con-2 for wt TNV-D genome. Number 1 in the list corresponds to the predicted most stable (or optimal) structure, while those below represent predicted suboptimal structures listed in order of decreasing stabilities.
To further clarify the formation of Con-1 and/or Con-2, in vitro selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE) analysis was performed on transcripts of the full-length TNV-D genome. In SHAPE probing, nucleotides that are flexible are more readily modified and are predicted to correspond to single-stranded residues (35). All residues in regions 1 and 3 (blue), corresponding to the stem of Con-1, exhibited low flexibility (Fig. 4A), a finding consistent with the formation of this stem (Fig. 4B). Regions 2 and 5 (yellow), which mapped to the proposed pseudoknot of Con-1, exhibited lower levels of flexibility at the 3′ end of region 2 and the 5′ end of region 5, compared to the rest of the sequences in these two regions (Fig. 4A). This suggests that, under the assay conditions, the proposed pseudoknot may be partially formed and less stable. The presence of highly reactive residues within predicted double-stranded regions in Con-2 and of poorly reactive nucleotides in proposed single-stranded regions (Fig. 4C) indicates that Con-2 is likely not a dominant structure in solution. Accordingly, both comparative structural analysis (Fig. 3) and chemical probing (Fig. 4) more strongly support Post-RTSL forming Con-1.
SHAPE analysis of the Post-RTS element. (A) The relative reactivity of each nucleotide is plotted graphically, where red, green, and black bars represent highly reactive, moderately reactive, and poorly reactive residues, respectively. Regions of the graph corresponding to the stem and pseudoknot of Con-1 are highlighted in blue and yellow, respectively, and numbered. (B) Mfold-predicted structure of Con-1 with color-coded stem and PK. Nucleotides in red, green, and black convey relative SHAPE reactivity (see key). (C) Corresponding Mfold-predicted secondary structure for Con-2, color-coded as described in panel B.
Con-1 of Post-RTSL is important for readthrough and virus viability.To determine whether the formation of the stem portion of Con-1 was important for p82 production, compensatory mutational analysis was used to disrupt and then restore one of the base pairs in the helix (Fig. 5A). The downstream substitution in mutants 1B and 1C resulted in a conservative amino acid change of threonine to serine; however, this same replacement is present naturally in BBSV. Neither the in vitro translation assay nor protoplast infections incubated at 22°C showed compelling deviations from the wt of either p82 or genome levels, respectively (Fig. 5B and C). However, when the protoplast incubation temperature was raised to 29°C, both disruptive mutants showed a decline in genomes to ∼50 to 60% of the wt, and the restorative mutant resulted the in recovery of accumulation to ∼86% (Fig. 5D). Thus, the stem of Con-1 appears to contribute to virus viability.
Mutational analysis of the Post-RTSL Con-1 stem. (A) Wt and mutant stem interactions are shown with nucleotide substitutions in red. The nucleotide substitution in mutants 1B and 1C results in an amino acid change from threonine to serine in the p82 product. (B) In vitro translation analysis of pseudoknot mutants. (C and D) Northern blot analysis of protoplast infections with Pre-RTSL mutants.
To investigate the importance of the proposed Con-1 pseudoknot formation, three base pairs were targeted with silent compensatory mutations in PKA, PKB, and PKC (Fig. 6A). In wheat germ extract, the disruptive mutants resulted in a decline in readthrough to ∼52 and ∼57%, and the regenerative mutant resulted in a recovery of p82 production to ∼99% (Fig. 6B). In protoplasts incubated at 22°C there was no decline in genome accumulation for all mutants (Fig. 6C), but at 29°C the levels correlated well with those observed for readthrough (Fig. 6D). Mapping the substitutions in PKC, which yielded wt performance, onto Con-2, revealed that the changes would destabilize the predicted structure (Fig. 6E). Accordingly, Con-1 of Post-RTSL appears to be the functionally relevant structure.
Mutational analysis of the Post-RTSL Con-1 pseudoknot. (A) Wt and mutant pseudoknot interactions are highlighted in yellow with compensatory mutations indicated in red. (B) In vitro translation analysis of pseudoknot mutants. (C and D) Northern blot analysis of protoplast infections with pseudoknot mutants. (E) Substitutions from mutant PKC mapped onto Con-2.
The silencer/3′-end interaction modulates readthrough and genome accumulation.The 3′-terminal nucleotides in tombusvirid genomes are complementary to a nearby upstream sequence, termed the silencer element (36). Silencer/3′-end interactions have been shown to be important for viral genome replication in several family members, including tombusviruses (34, 37), carmoviruses (38), and aureusviruses (39). TNV-D has a predicted silencer/3′-end interaction (Fig. 7A); however, its functional importance has not been assessed. This 3′-proximal region is also relevant to translational readthrough since the sequence that intervenes the silencer/3′-end interaction includes the DRTE, which must pair with the PRTE for efficient p82 production (Fig. 7A) (6). Consequently, the silencer/3′-end interaction could influence the presentation or accessibility of the DRTE.
Mutational analysis of the silencer/3′-end interaction. (A) Predicted secondary structure of the 3′ UTR of TNV-D with the silencer/3′-end interaction highlighted in gray and the DRTE sequence in green. Wt and mutant base pairing interactions for the interaction are shown, with nucleotide replacements in red. (B) In vitro translation analysis of silencer/3′-end mutants. (C and D) Northern blot analysis of protoplast infections with silencer/3′-end mutants.
To address this possibility, compensatory mutations were introduced at the UA base pair in the silencer/3′-end interaction (Fig. 7A). Replacement with a CG base pair was chosen, because these residues are present at corresponding positions in the BBSV genome. The CA mismatch in Sil-A produced lower levels of p82 (∼37%) and reduced genome accumulation in protoplasts incubated at 22°C (∼70%), while milder negative effects were seen for Sil-B containing the UG wobble base pair (Fig. 7B and C). The presence of a canonical CG base pair in Sil-C restored both activities to wt levels (Fig. 7B and C). Increasing protoplast incubation temperature to 29°C resulted in negligible genome levels for Sil-A and Sil-B (Fig. 7D), which were recovered to ∼85% of the wt level in Sil-C. Therefore, the silencer/3′-end interaction modulates readthrough and is crucial for genome accumulation.
The apical loop of SLII facilitates readthrough and genome accumulation.Two stem-loop structures, SLI and SLII, flank the DRTE (Fig. 8A). To investigate their possible effects on readthrough and genome levels, the terminal loop sequences in SLI and SLII were replaced with superstable tetranucleotide loops, GAAA (Fig. 8A). When the loop of SLI was replaced, readthrough and replication levels remained close to those for wt (Fig. 8B and C). In contrast, loop replacement in SLII led to a significant decrease in readthrough (∼15%) and genome (∼30%) levels at 22°C (Fig. 8B and C). To confirm the latter result, the SLII terminal loop (AACA) was replaced with a different class of superstable tetra-loop (UUCG) or a modified loop sequence (GACG) (Fig. 8D). In both cases, there was a decrease in p82 production and genome accumulation to below 20% of wt, reinforcing the functional relevance of the loop. Strikingly, these strong phenotypes were observed at the less stringent 22°C incubation temperature of protoplasts, further underscoring the significance of this structural element.
Mutational analysis of SLI and SLII in the 3′ UTR. (A) Base pairs in SLII targeted with compensatory mutations are boxed in black. Nucleotide substitutions in apical loops are shown in red. GNRA-type tetraloops are more stable when closed by a CG base pair. SLI-m1 is a control to determine the effect of changing the wt GC closing base pair to CG. (B) In vitro translation analysis of SLII and SLI apical loop mutants. (C) Northern blot analysis of protoplast infections of SLII and SLI apical loop mutants. (D) Analysis of additional SLII apical loop mutants. Relative readthrough and genome levels are provided below each mutant. (E) Analysis of sets of compensatory mutants in SLII. The position for each mutant set in SLII is indicated in panel A.
Due to the critical role of the terminal loop in SLII, the possible influence of its immediate context was examined using a series of compensatory mutations targeting base pairs in the upper (SLII-1, -2, and -3 series), middle (SLII-4 series), and lower (SLII-5 series) regions of the stem in SLII (Fig. 8E). All disruptive mutations in the upper portion of the stem had various negative effects that were partially rescued upon restoration of the base pairs; the exception being SLII-1C, where the regenerative GC pair did not lead to recovered readthrough (Fig. 8E). Assessment of the middle and lower stem regions (SLII-4 and -5 series) also supported the importance of stem stability in regions for both readthrough and genome accumulation (Fig. 8E). Accordingly, the entire SLII structure is relevant to the activities that were monitored.
One possible mode of function for the terminal loop of SLII is to interact via base pairing with a complementary sequence elsewhere in the TNV-D genome. Indeed, there is ample precedence for intragenomic interactions modulating various activities among tombusvirids (26, 40) and other plus-strand RNA viruses (7, 41, 42). Several candidate partner sequences for the terminal loop of SLII were selected for analysis based on their: (i) conservation between TNV-D, BBSV, and LWSV; (ii) potential to extend base pairing beyond the AACA loop sequence; and (iii) proximity to previously characterized functional regions of the genome (Fig. 9A). Most of the putative partner interactions investigated involved canonical base pairing (Fig. 9B); however, two did not (Fig. 9C). In all cases, silent modifications to the candidate partner sequence did not notably affect either readthrough or genome accumulation, thereby ruling out their possible involvement in the activity associated with the terminal loop in SLII (Fig. 9B and C).
Testing potential base pairing partners of SLII. (A) Cartoon representation of the TNV-D genome with the relative positions of substitutions. Mutant names indicate the TNV genome coordinate for the first nucleotide position of each potential pairing interaction. (B) Substitutions introduced into potential canonical base pairing partners of SLII loop are shown in red. The effects of the substitutions on relative readthrough and relative genome accumulation are shown below each mutant. (C) Analysis of substitutions introduced into potential noncanonical base pairing partners of SLII loop. Lines above and below the sequence define areas of potential base pairing with ACAA.
DISCUSSION
Translational readthrough represents an important gene expression strategy utilized by different classes of viruses (9). Our analysis of TNV-D suggests that RNA elements beyond those previously identified can markedly influence this process and the robustness of viral infections. These features and their possible roles in translational readthrough and virus reproduction are discussed below.
RTSL.The RTSL in TNV-D contains a PRTE that interacts with the 3′-proximal DRTE to enhance readthrough production of p82 (Fig. 1A and B) (6). Currently, how this and similar interactions in other tombusvirids (5, 12) promote readthrough is unknown. One possibility is that binding of the DRTE to the PRTE-containing bulge in RTSL stabilizes the RTSL helix. This is a reasonable option, because the stability of RNA structures 3′ proximal to readthrough sites, including that in TNV-D (6), is known to positively correlate with readthrough efficiency (1). However, our results indicated that deletion of the bulge in mutant ΔPRTE, which would stabilize the RTSL helix, inhibited readthrough activity (Fig. 1C). This suggests that the PRTE-DRTE interaction does more than simply stabilize the RTSL. Instead, the specific structure formed by the interaction could be important for function, or the interaction could recruit other required 3′-proximal RNA elements or their associated proteins. The readthrough defects observed in vitro in mutants with deletions in the PRTE-bulge could be responsible for the corresponding lack of infectivity in protoplasts (Fig. 1D), but it is also possible that the deletion of encoded amino acids rendered the RdRp inactive.
The upper region of the TNV-D RTSL was also investigated for function. In turnip crinkle virus (TCV) (genus Carmovirus, family Tombusviridae), an apical portion of its RTSL is essential for efficient readthrough, because it forms a local pseudoknot with the 3′ end of the RTSL (12). When the upper portion of the TNV-D RTSL was investigated, the results indicated that neither the terminal loop nor a base pair mismatch affected readthrough; although moderate reductions in genome accumulation were observed at 29°C (Fig. 1E and F). Therefore, unlike in TCV, the upper region of the RTSL in TNV-D does not appear to play a major role in readthrough under our assay conditions.
Pre-RTSL.GC-rich RNA hairpins are predicted just 5′ of frameshift and readthrough sites in tombusvirids (12); however, they do not share any noteworthy sequence identity and their positions relative to their cognate stop codons vary, i.e., ∼0 to 10 nt (unpublished data). Functional RNA hairpins immediately upstream of recoding sites have been described and studied previously. The RNA hairpin in pea enation mosaic virus (genus Umbravirus, family Tombusviridae) does not influence frameshifting in the wt context, but can act as an inhibitor when tested in a modified viral context (11). Conversely, the upstream hairpin in barley yellow dwarf virus (genus Luteovirus, family Luteoviridae, but related to tombusvirids) (43) was found to enhance frameshifting (18). In a third example, severe acute respiratory syndrome coronavirus, the upstream hairpin downregulated frameshifting activity (10). Thus, the effects of such RNA structures on frameshifting can vary significantly.
For readthrough, disruption of the upstream hairpin present in TCV did not affect readthrough in vitro; however, genome accumulation was reduced in infections to 28% that of wt (12). Our results with Pre-RTSL in TNV-D are similar to those for TCV in that infectivity was reduced (∼48% of wt), but, in contrast, readthrough levels were also negatively affected, albeit to a lesser degree (79 to 88% of wt) (Fig. 2). Consequently, the TNV-D Pre-RTSL appears to moderately affect readthrough under our in vitro assay conditions. The variability in distance of upstream tombusvirid hairpins from their stop codons suggests that, if some of them also function in readthrough, precise spacing may not be important. Also, since sequence identity is not conserved among the different hairpins, the formation of a stable helix is likely to be a key feature for their function. Theoretically, the Pre-RTSL could influence ribosome readthrough activity (i) when the ribosome first encounters the RNA structure, which may stall it, (ii) if the structure reforms when the ribosome is engaging the stop codon, and/or (iii) if the structure functions as an “insulator” that sequesters proximal sequence to guard against their interference with proper folding of the RTSL. Additional studies will be required to determine the role of these conserved structures in readthrough and/or other aspects of virus replication.
Post-RTSL.Unexpectedly, an RNA element located just 3′ to the RTSL was found to be important for efficient readthrough in vitro. SHAPE, mutational, and comparative RNA structural analyses were all consistent with Con-1 being the functional structure of the Post-RTSL. The Post-RTSL is predicted to form a pseudoknot structure that is located just 9 nts downstream from the base of the RTSL. This would position it close to ribosomes engaging the stop codon and could allow for the structure, or possible associated proteins, to interact with the ribosome or release factors and cause increased readthrough. Alternatively, as considered for the Pre-RTSL, the Post-RTSL could act as a type of structural insulator to promote proper folding of the RTSL.
As a distinct local structure, positioned immediately downstream from the RTSL, the Post-RTSL represents a novel type of readthrough-modulating element. In TBSV, a similarly positioned (i.e., 3 nt from the RTSL), but structurally different local RNA secondary structure was shown to contribute to efficient genome accumulation (44); however, its influence on translation or readthrough was not assessed. Likewise, aureusviruses have a distinct conserved local secondary structure 4 to 6 nt downstream from their RTSL that has not been functionally characterized (unpublished). The Post-RTSLs in these three genera do not share any obvious structural similarities, but they all reside in RdRp coding regions, and the RdRps of betanecroviruses are most closely related to those of tombusviruses and aureusviruses (20). Thus, although the structures have taken different evolutionary paths, they may maintain their original function which, based on our results with TNV-D, may be, in part, to facilitate readthrough.
3′ UTR.The 3′-proximal ∼80 nt of the TNV-D genome harbors RNA elements important for both genome replication and readthrough (6, 25). Our findings show that the silencer/3′-end interaction is critical for genome accumulation in protoplast infections (Fig. 7D), which is the first demonstration of this for a betanecrovirus and consistent with results from other studies showing its importance for tombusvirid genome replication (36–39). The same interaction also modestly modulated readthrough under our assay conditions (Fig. 7B). In contrast, the loop of SLII in the 3′-proximal region was found to be a major contributor to readthrough efficiency because three different loop modifications led to an average reduction in readthrough of ∼7-fold (Fig. 8B and D) and a corresponding average drop in genome accumulation of ∼6-fold (Fig. 8C and D). Collectively, these results suggest that the identity of the nucleotides in the loop, AACA, is important for its function. In contrast, data from compensatory mutational analysis of SLII indicated that base pairing of the stem region, not sequence identity, is more important (Fig. 8E). Although most of the stem may act to properly present the loop sequence, the inability to recover readthrough activity in compensatory mutant SLII-1C suggests a possible identity-dependent role for the wt CG closing base pair (Fig. 8E).
One way that the AACA loop sequence could function is by interacting with a complementary sequence proximal to SLII, the RTSL, or other regions of the viral genome. Accordingly, possible base pairing partner sequences were selected and assessed by introducing substitutions into them that were predicted to disrupt potential interactions with AACA (Fig. 9). In all cases, no major effects on either readthrough or genome accumulation in infections were observed. It is, however, possible that the AACA sequence has a base pairing partner sequence other than those tested. Conversely, the sequence could engage other genomic sequences through noncanonical interactions or bind to a required protein factor. Also of significance is that when the PRTE-DRTE interaction is formed, it positions SLII proximal to the readthrough site, which would allow for it or any bound cargo to engage the ribosome or translation release factors. Interestingly, in the two other sequenced betanecroviruses, LWSV and BBSV, the corresponding SLII loop sequences are cAACAg and cUCCUAUg, respectively (loop sequences are capitalized). Hence, the sequence in LWSV is identical to that in TNV-D, whereas the one in BBSV is markedly different in both size and nucleotide content. It remains to be determined whether either of these sequences is able to function in a manner similar to that observed for TNV-D.
MATERIALS AND METHODS
Plasmid construction.All TNV-D clones created for this study were derived from wild-type (wt) TNV-D cDNA in a pUC19 plasmid, supplied by Robert Coutts (21), which was modified to introduce a SmaI site at the viral 3′ terminus (31). Standard PCR-based site-directed mutagenesis was used to create TNV-D mutants, using a Velocity DNA polymerase PCR kit (Bioline). Sequencing confirmed the correctness of all mutant constructs and the changes introduced into the TNV-D genome are shown in the accompanying figures.
Preparation of viral RNAs.Uncapped in vitro transcripts of SmaI-linearized TNV-D constructs were produced using the AmpliScribe T7-Flash transcription kit (Epicentre Technologies) as described previously (45). RNA transcript concentrations were quantified by spectrophotometry, and the quality of transcripts was verified by agarose gel electrophoresis.
RNA secondary structure prediction.RNA secondary structures were predicted at 37°C using Mfold version 3.6 using default settings (46, 47).
In vitro SHAPE RNA structure probing.Selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE) was performed as previously described (5). Briefly, full-length in vitro transcribed TNV-D genomic RNA was refolded, treated with 1-methyl-7-nitroisatoic anhydride, and then subjected to reverse transcription using Superscript IV reverse transcriptase (Invitrogen). Fluorescently labeled primers complementary to TNV-D nt 1079 to 1108 were used to evaluate the region of interest. Raw fluorescence intensity data were analyzed using ShapeFinder (48). The top 10 peak intensities were averaged, and all raw nucleotide reactivities were divided by this average. The average relative reactivities from two in vitro SHAPE experiments were plotted graphically and mapped onto Mfold-predicted conformations.
In vitro translation.The levels of protein accumulation were measured by incubating RNA transcripts in nuclease-treated wheat germ extract (Promega). Uncapped viral genomic RNA (0.5 pmol) was incubated in the extract for 1 h at 25°C in the presence of [35S]methionine, as described previously (6). Products were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis in a 12% polyacrylamide gel. Viral proteins produced were detected using a Typhoon TRIO+ variable mode imager (GE Healthcare) and quantified using QuantityOne software (Bio-Rad). All trials were repeated at least three times, and the averages were calculated with the standard errors.
Protoplast infection.Cucumber cotyledon protoplasts were prepared and transfected with uncapped in vitro transcribed viral genomic RNA (45). Specifically, 3 × 105 protoplasts were transfected with 3 μg of viral genomic RNA and incubated under constant light for 22 h at either 22 or 29°C. The total nucleic acids were extracted as described previously (45) and separated in nondenaturing 2% agarose gels. Viral RNAs were detected via Northern blotting using three 32P-radiolabeled probes complementary to the 3′ end of TNV-D: nt 2821 to 2840, 3520 to 3532, and 3643 to 3663 (31). Viral RNA accumulation was monitored using a Typhoon TRIO+ variable mode imager and quantified using QuantityOne software. All trials were repeated at least three times, and averages were calculated with the standard errors.
ACKNOWLEDGMENTS
We thank members of our laboratory for reviewing the manuscript and Robert Coutts for providing the TNV-D clone.
This research was funded by an NSERC Discovery Grant. L.R.N. was supported by an NSERC Graduate Scholarship.
FOOTNOTES
- Received 19 December 2016.
- Accepted 27 January 2017.
- Accepted manuscript posted online 1 February 2017.
- Copyright © 2017 American Society for Microbiology.
