|Home | About | Journals | Submit | Contact Us | Français|
The mechanisms by which the respiratory syncytial virus (RSV) RNA-dependent RNA polymerase (RdRp) initiates mRNA transcription and RNA replication are poorly understood. A previous study, using an RSV minigenome, suggested that the leader (Le) promoter region at the 3′ end of the genome has two initiation sites, one at position +1, opposite the 3′ terminal nucleotide of the genome, and a second site at position +3, at a sequence that closely resembles the gene start (GS) signal of the RSV L gene. In this study, we show that the +3 initiation site of the Le is utilized with apparently high frequency in RSV-infected cells and yields small RNA transcripts that are heterogeneous in length but mostly approximately 25 nucleotides (nt) long. Experiments with an in vitro assay in which RSV RNA synthesis was reconstituted using purified RdRp and an RNA oligonucleotide showed that nt 1 to 14 of the Le promoter were sufficient to signal initiation from +3 and that the RdRp could access the +3 initiation site without prior initiation at +1. In a minigenome assay, nucleotide substitutions within the Le to increase its similarity to a GS signal resulted in more-efficient elongation of the RNA initiated from position +3 and a reduction in RNA initiated from the NS1 gene start signal at +45. Taken together, these data suggest a new model for initiation of sequential transcription of the RSV genes, whereby the RdRp initiates the process from a gene start-like sequence at position +3 of the Le.
Respiratory syncytial virus (RSV) is the leading cause of pneumonia and viral deaths in infants worldwide and is increasingly recognized as a significant pathogen of the elderly and immunocompromised (1). Treatment for RSV infection is limited to supportive care, and at present there are no approved vaccines. RSV is a member of the family Paramyxoviridae in the order Mononegavirales, the nonsegmented negative-strand (NNS) RNA viruses. Like other viruses in this order, the RSV genome serves as the template for mRNA transcription and genome replication, carried out by the viral RNA-dependent RNA polymerase (RdRp) (reviewed in reference 2).
The RSV genome is 15.2 kb and encodes mRNAs from 10 sequentially arranged genes (reviewed in reference 3). Two short extragenic regions border the genome, a 3′ 44-nt leader region (Le) and a 5′ 155-nt trailer (Tr) region (4, 5). The Le region contains promoter sequences that are required for transcription of all RSV genes and genome replication (4). Transcription occurs in an obligatorily sequential, polarized manner to generate 10 subgenomic mRNAs (6). During this process, the RdRp is controlled by cis-acting gene start (GS) and gene end (GE) sequences that flank each of the genes (7, 8). The GS sequences direct initiation of RNA synthesis. They are highly conserved in RSV, with 9 of 10 GS sequences being identical and the tenth having only slight differences (9). In related viruses, the GS signals have been shown to be multifunctional, directing the RdRp to cap and methylate the nascent transcript (10–12). Capping is thought to allow the RdRp to enter into a stable elongation mode in which it can extend the RNA to the GE signal, where the RdRp polyadenylates and then releases the mRNA (13). The RdRp is then able to scan the template to locate the next GS signal and reinitiate mRNA synthesis (14). RNA replication involves synthesis of a replicative intermediate, the antigenome, which is a full-length positive-sense complement of the genome. The trailer complement (TrC) region at the 3′ end of the antigenome contains a strong promoter for synthesis of genome RNA (4, 15). In contrast to the viral mRNAs, antigenome and genome RNAs are encapsidated in viral nucleoprotein (N) as they are synthesized. Each N monomer binds 7 nucleotides (nt), and the N-RNA complex assembles into a helical nucleocapsid structure (16), which can act as a template for the next round of RNA synthesis. Encapsidation increases the processivity of the RdRp and is believed to enable the RdRp to read through GE signals (17, 18).
Because transcription and RNA replication are both initiated from within the Le promoter region, the mechanism by which the RdRp is coordinated between the two processes is poorly understood. Fundamental to understanding this is determining how the RdRp initiates transcription. The mRNA for the first RSV gene (NS1) is initiated at nt +45 on the genome, but it is unclear how the RdRp accesses this site. There are basically two models for how this can occur. The first model is based largely on studies of other paramyxoviruses and suggests that the RdRp initiates transcription opposite nt +1 of the Le, in the same way as replication initiation. However, during transcription, the RdRp does not form a stable elongation complex and so releases the nascent Le transcript and scans to locate the first GS signal and reinitiate transcription (19). The second model is based on studies with vesicular stomatitis virus (VSV) and posits that there are two pools of RdRp, with the transcriptase form of the RdRp able to enter the template directly at the first GS signal (20, 21). It has also been proposed that both mechanisms might come into play during infection and that the entry site of the RdRp to begin transcription is governed by the level of available N protein (19).
Studies to investigate the mechanisms of transcription and replication initiation in RSV have mapped the sequence elements involved in these processes (reviewed in reference 22). It was found that nt 1 to 13 of the Le are sufficient to recruit both transcription- and replication-competent RdRp complexes and that nt 1 to 13 can signal initiation of RNA synthesis (23). Single nucleotide substitutions at positions 3, 5, 8, 9, 10, or 11 resulted in ablation of almost all detectable transcription, indicating that the RdRp has to engage with nt 3 to 11 of the Le to begin mRNA synthesis from the first GS signal (24). Efficient transcription also required a U-rich sequence lying at the end of the Le and the first (NS1) GS sequence at positions 45 to 54 (18, 25). Nucleotides 3, 5, 8, 9, 10, and 11 were also shown to be critical for replication (24), and the Le sequence up to 34 nt was required in addition for full-length antigenome synthesis, suggesting that Le nt 1 to 34 might contain a signal for RNA encapsidation (18). These findings are most consistent with the first model for transcription initiation described above and suggest that transcription is initiated from a signal located at the 3′ end of the Le region. However, the data do not eliminate other possible scenarios. For example, the L protein of other paramyxoviruses has been shown to oligomerize (26–28), raising the possibility that the RdRp can contact the Le 3′ end and simultaneously have an active site positioned at or near the NS1 GS signal. Another possibility is that the RdRp is able to bind to nt 3 to 12 and then has some mechanism to scan to the first GS signal without first initiating RNA synthesis. Thus, despite having identified the cis-acting sequences required for transcription, the mechanism by which the RdRp is recruited to the template and accesses the NS1 GS signal remains elusive.
Recent studies using the RSV minigenome system identified RNA initiated from position +1 of the Le and also at position +3 (29). Initiation at position +1 was expected, as this is the initiation site for antigenome synthesis, but identification of RNA initiated at position +3 was a surprising finding. The goal of the current study was to determine if the +3 site is utilized during RSV infection, characterize the RNA that is generated from it, and investigate the mechanism by which this site is accessed.
HEp-2 cells (ATCC) were used for all experiments and were grown in Opti-MEM reduced serum medium (Invitrogen) supplemented with 2% fetal bovine serum. Wild-type (wt) RSV was strain A2.
For analysis of RSV RNAs from infected cells, 80 to 90% confluent HEp-2 cells were infected with RSV A2 at a multiplicity of infection (MOI) of 5. Cells were incubated at 37°C and harvested at 17 h postinfection. Cells were collected by centrifugation and resuspended in TRIzol (Invitrogen). Total intracellular RNA was prepared according to the manufacturer's instructions, except that following isopropanol precipitation it was resuspended in 2 M NTE (2 M NaCl, 40 mM Tris [pH 7.4], 1 mM EDTA), extracted with phenol and chloroform, and precipitated with ethanol.
The minigenomes used in this study have all been described previously (15, 24). Each minigenome plasmid (0.1 μg) was transfected into HEp-2 cells along with 0.4 μg N, 0.2 μg P, 0.1 μg M2-1, and 0.1 μg codon-optimized L plasmids using a method described previously (30) and coinfected with vaccinia virus strain MVA-T7 for expression of T7 polymerase (31). Forty-two to 48 h posttransfection, cells were harvested and RNA was isolated as described above. For identification of nuclease-resistant RNA, infections or transfections were performed in duplicate, and one-half of the samples were treated with micrococcal nuclease to digest unencapsidated RNA, as described previously (30).
Viral RNA 5′-ends were analyzed by reverse transcription of cDNAs using the Sensicript RT kit (Qiagen), with end-labeled primers that corresponded to leader and NS1 sequences (see panel A of each figure), as described previously (32). The primer extension products were analyzed by electrophoresis in a 6% urea-acrylamide gel. End-labeled oligonucleotides corresponding to initiations from positions +1, +2, and +3 were used as markers, and gels were visualized by autoradiography. Phosphorimager analysis was carried out with a Bio-Rad Personal Molecular Imager System and Quantity One quantification software.
RNA isolated from RSV-infected or minigenome-transfected cells was subjected to electrophoresis in 1.5% agarose-formaldehyde gels and transferred to nitrocellulose membranes using the Whatman TurboBlotter downward capillary transfer system (Sigma-Aldrich), as described previously (30). The methods used to detect RNA with a riboprobe have been described previously (30). For detection of RSV RNA with an oligonucleotide probe, blots were prehybridized for 1 h in 5× Denhardt's solution, 6× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 0.1% SDS, and 0.01% sodium pyrophosphate (NaPPi) at 62°C and hybridized for 12 to 18 h with either a [32P] end-labeled locked nucleic acid modified oligonucleotide specific to nt 7 to 34 (relative to the 5′ terminus) of antigenome sense RNA (5′-TTTATGCAAGTTTGTTGTACGCATTTTT-3′) in the same buffer at 62°C. For detection of NS1 mRNA, blots were prehybridized for 1 h as above at 56°C and hybridized for 12 to 18 h with a [32P] end-labeled oligonucleotide specific to either nt 56 to 75 (relative to the 5′ end of antigenome RNA) (5′-AAGTGGTACTTATCAAATTC-3′) or nt 504 to 540 (5′-GCTAGTTGATATTAATTATAATTTATGGATTAAGATC-3′). End-labeled oligonucleotide-probed blots were washed 2 × 15 min at room temperature, and 2 × 10 min at the hybridization temperature with 6× SSC, and the RNA was detected by autoradiography. To detect low-molecular-weight RNAs, RNA was analyzed by gel electrophoresis in 6% urea-acrylamide gels alongside a low-range single-stranded RNA (ssRNA) ladder and a microRNA (miRNA) ladder (NEB). The ladder lanes were excised from the gel prior to transfer and stained with ethidium bromide for visualization with UV light. The remainder of the gel was transferred to a Nitran-N positively charged Nylon membrane (Sigma-Aldrich) and probed as described above for detection with an oligonucleotide probe. The autoradiogram was aligned with an image of the excised ladder to determine the size of the RNA transcripts.
The methods for expressing the RSV L/P complex have been described previously (33). Briefly, a codon-optimized version of the RSV (strain A2) L protein open reading frame (ORF) was chemically synthesized (GeneArt), and the mutant LN812A was generated by site-directed mutagenesis. Wild-type L or LN812A was expressed in insect cells together with the RSV A2 P ORF, which was tagged with a six-histidine sequence, separated from the ORF by a tobacco etch virus (TEV) protease cleavage site. RSV L/P protein complexes were isolated from cell lysates by affinity chromatography, TEV protease cleavage, and size exclusion. Isolated L/P complexes were analyzed by SDS-PAGE and PageBlue staining, and the L protein concentration was estimated against bovine serum albumin reference standards. The bands migrating between 160 and 250 kDa and ~35 and 50 kDa were confirmed to be RSV L and P polypeptides by mass spectrometry, and the band migrating between 70 and 80 kDa was determined to be HSP70/HSC70 by Western blotting using a specific antibody (33).
RSV RNA synthesis was reconstituted in vitro, as described previously (33). RNA oligonucleotides corresponding to nt 1 to 14 of the Le promoter or its complement (LeC) (Dharmacon) were PAGE purified and used as the templates. RNA (10 μM) was combined with the purified L/P complex (containing ~100 ng of L protein) in RNA synthesis buffer (50 mM Tris HCl, pH 7.4, 50 mM NaCl, 5 mM MgCl2, 5 mM dithiothreitol [DTT]), with nucleoside triphosphates (NTPs) included at the concentrations indicated in the legend, and 1 μl of either [α-32P]GTP or [γ-32P]GTP in a final volume of 50 μl. Reaction mixtures were incubated at 30°C for 3 h, followed by incubation at 90°C for 3 min to inactivate the RdRp, and cooled on ice. Reaction mixtures were combined with 7.5 μl 10% SDS, 0.5 μl 500 mM EDTA, and 10 μg proteinase K and incubated at 45°C for 45 min before phenol-chloroform extraction and ethanol precipitation. The RNA was analyzed by electrophoresis on a 20% polyacrylamide gel containing 7 M urea in Tris-borate-EDTA buffer, followed by autoradiography. The nucleotide lengths of the RNA products were determined by comparison with a molecular weight ladder generated by alkali hydrolysis of a [32P] end-labeled RNA oligonucleotide representing the anticipated 14-nt LeC RNA product. The bottom of each gel was cropped to eliminate the nonspecific signal from unincorporated radiolabeled NTPs that were not always efficiently removed during RNA purification and electrophoresis.
As described in the introduction, in a recent study we found evidence that in the minigenome system, the RSV RdRp initiated RNA synthesis from the +3 site of the Le promoter, in addition to the expected initiation site at +1. To determine if this was representative of what occurs during RSV infection, intracellular RNA from RSV-infected cells was examined by primer extension analysis using a genome sense primer that corresponded to nt 15 to 39 of the Le sequence (g15-39) (Fig. 1A). By using a primer that hybridized so close to the 5′ end of Le-complement (LeC) RNAs, the initiation sites of RNA products from the 3′ end of the Le promoter could be determined precisely. This analysis indicated that RNA was initiated from two sites within the Le: the +1 initiation site, which is the expected initiation site for antigenome synthesis, and position +3 (Fig. 1B, panel i, lane 4). Surprisingly, using this primer, significantly more RNA initiated at position +3 could be detected than RNA initiated from position +1 (Fig. 1C). It is not possible to determine the relative frequency of initiation at +1 versus +3 based on levels of intracellular RNA because of potential differences in RNA stability in the cellular environment; however, antigenome RNA initiated at +1 would be expected to be very stable due to encapsidation. Therefore, this finding indicates that the RSV RdRp could initiate at the +3 site relatively frequently during infection. To investigate if the RNA initiated from the +3 site was efficiently elongated, primer extension analysis was performed using a primer that corresponded to nt 45 to 68 of the RSV genome, which lie at the beginning of the NS1 gene (g45-68; Fig. 1A). RNA initiated at +3 could also be detected with the g45-68 primer, but it was less abundant relative to the RNA initiated at +1 than was observed with the g15-39 primer (Fig. 1B, panel ii, lane 5, and C), indicating that the RNA initiated at +3 was not elongated as efficiently as RNA initiated from +1. Although the g15-39 and g45-68 primer extension reactions could not be directly compared (due to possible differences in primer hybridization efficiency or secondary structure in the target RNA), it was possible to compare the relative levels of +3 and +1 initiation products. Quantification and normalization to the RNA initiated from the +1 site showed that there was a 7.8-fold decrease in the level of RNA initiated at +3 that could be detected with the g45-68 primer compared to the level detected with the g15-39 primer (Fig. 1C). Thus, this result shows that RNA synthesis can be initiated from the +3 site of the RSV Le region during viral infection and that while some RNA initiated at +3 was extended beyond the end of the Le region, the bulk was terminated within a short distance of the initiation site.
To further characterize the RNAs generated from within the Le promoter region, RNA isolated from RSV-infected cells was analyzed by Northern blotting with a negative-sense oligonucleotide probe corresponding to Le nt 7 to 34 (g7-34). The RNA was also hybridized with two negative-sense NS1-specific probes (g56-75 and g504-540), to detect NS1 containing mRNAs (Fig. 2A). The Northern blot analysis was performed using two different electrophoresis conditions: formaldehyde agarose gel electrophoresis, designed to detect RNAs in the range of ~100 to ~20,000 nt in length (Fig. 2B), and polyacrylamide gel electrophoresis, to allow resolution of small RNAs in the range of 17 to 500 nt in length (Fig. 2C and andD).D). Northern blot analysis of larger RNAs using agarose gels showed that the Le and NS1 probes detected the expected RNA species. The g7-34 probe detected a high-molecular-weight band corresponding to antigenome RNA, and the NS1 probe identified bands of appropriate sizes to be NS1 and NS1-NS2 readthrough mRNAs. Long exposures were required to detect antigenome RNA with this probe, reflecting the relative scarcity of antigenome RNA compared to NS1 mRNA (data not shown). Analysis of the same RNA samples using polyacrylamide gel electrophoresis showed that the g7-34 probe was also able to detect a band of small RNAs that were somewhat heterogeneous in size but predominantly approximately 25 nt, with some transcripts extending up to ~50 nt in length (Fig. 2C, panel i, lane 2). Under these electrophoresis conditions, the NS1 oligonucleotide probe detected full-length NS1 mRNA [532 nt excluding the poly(A) tail] near the top of the gel, as well as some prematurely terminated NS1 RNAs that could be detected with probe g56-75 but not g504-540 (Fig. 2C, panels ii and iii, lanes 2). However, no ~25-nt RNA species were detected with the NS1 probes, indicating that the small RNAs detected with the Le-specific probe were not antigenome RNA that had become degraded during the RNA purification procedure. These results show that the RSV Le generates small RNA transcripts that are mostly ~25 nt in length, in addition to the expected full-length antigenome RNA. To determine if the ~25-nt RNA was encapsidated, cell extracts from mock- and RSV-infected cells were treated with micrococcal nuclease prior to RNA purification and Northern blot analysis. Analysis of RNA with probe g56-75 showed that the NS1 mRNA was efficiently digested (Fig. 2D, panel ii). Analysis with probe g7-34 showed that nuclease treatment resulted in a reduction in the level of the ~25-nt RNA, suggesting that it was not encapsidated (Fig. 2D, panel i). However, the nuclease treatment did not completely eliminate the 25-nt RNA, indicating either that the nuclease digestion of this very small RNA was incomplete or that the ~25-nt RNA that could be detected was comprised of a mixture of unencapsidated and encapsidated RNA. Taken together with the primer extension data shown in Fig. 1, these results indicate that the RSV RdRp can initiate RNA synthesis at +3 of the Le region to generate small RNAs ~25 nt in length, at least some of which are unencapsidated.
We had previously noted that nt 3 to 12 of the RSV Le sequence contains a C/U-rich sequence with strong similarity to the conserved GS sequence found at the start of the each RSV gene (29). This similarity was even more pronounced if the promoter sequence was aligned with the L GS signal, which differs slightly from the canonical GS sequences but appears to function in a similar way (7) (Fig. 3A). Eight of 10 nucleotides from positions 3 to 12 of the Le sequence are identical to the L GS sequence (underlined in Fig. 3A), with the first residue of the GS sequence (C) coinciding with position +3C of the Le promoter. The similarity between the L GS signal and Le nt 3 to 12 indicates that there is a common initiation element within these two signal sequences, and this might explain how RNA synthesis could be initiated from the +3 position. One of the models for NNS RNA virus transcription initiation is that the first RNA synthesis initiation event occurs opposite nt +1 of the Le, regardless of whether the RdRp is going to engage in transcription or replication. According to this model, the RdRp accesses internal GS signals by releasing the transcript initiated at +1 and then reinitiating RNA synthesis. If this model were correct for RSV, initiation at position +3 could be a secondary initiation event performed by an RdRp that had previously initiated at position +1. Alternatively, it was possible that initiation at position +3 could occur independently of prior initiation at +1. In recent work, we showed that RSV RNA synthesis can be reconstituted in vitro using purified L-P complexes and an RNA oligonucleotide template representing the TrC promoter (33). Although the template was not encapsidated in this assay, we found that the RdRp demonstrated specificity for RSV promoter sequence and that its activities at the TrC promoter were representative of those occurring in RSV-infected cells (33). Therefore, this assay was employed to determine if initiation at +3 of the RSV Le promoter was dependent on prior initiation at +1 or if the +3 site could be accessed directly.
The RNA synthesis assay was performed using an RNA oligonucleotide template corresponding to nt 1 to 14 of wt Le RNA (Fig. 3B). The RNA was incubated with purified L-P complexes (Fig. 3C) in a reaction mixture containing ATP, CTP, UTP, and GTP and supplemented with [α-32P]GTP. Following the reaction, the RNA products, containing incorporated [α-32P]GTP, were analyzed by denaturing gel electrophoresis. Two forms of L-P were used, either wt RdRp or a mutant RdRp control containing an amino acid substitution in the catalytic GDNQ motif of the L protein, which would inhibit RNA synthesis (33, 34). Analysis of the RNA generated in the reactions showed that the reaction mixture containing wt RdRp yielded RNA products ranging from 7 to 12 nt in length. These products were not detected in reaction mixtures containing no RdRp or mutant RdRp, indicating that they were generated by the RSV RdRp (Fig. 3D, panel i, compare lanes 2 and 3 with lane 4). Similarly, no RNA products were detected in a reaction mixture containing a template RNA consisting of the complement of the Le promoter (LeC), confirming that the RdRp had template specificity (Fig. 3D, panel i, lane 5). These results are similar to those obtained previously in studies of the RSV TrC promoter sequence (33) and show that the isolated RSV RdRp was functional and had specificity for RSV promoter sequence.
The fact that the longest RNA transcript that was synthesized in this assay was 12 nt long could indicate that the RNA either initiated at position +1 and extended to nt 12 on the template or initiated at +3 and extended to the end of the template. The smaller RNAs could either have been terminated prematurely or have initiated from downstream sites. To distinguish between these possibilities, the RNA synthesis reaction was performed with ATP, CTP, and GTP and supplemented with [α-32P]GTP, but with UTP omitted. Under these conditions, RNA synthesis would not be able to proceed beyond the first site of UTP incorporation at position 12 of the Le 1 to 14 template, and so the sizes of the RNA products would indicate their likely initiation site. Under these conditions, the 11- and 12-nt bands were not present, and the 8- and 9-nt bands increased in intensity. The intensity of the smaller bands remained unchanged (Fig. 3D, panel ii, compare lanes 4 and 7). This indicated that the RNAs were initiated at +3 and either extended to the end of the template or terminated prematurely. Importantly, there was no evidence for an 11-nt RNA, indicative of initiation at +1. To further characterize the RNA products, the reaction was performed using [γ-32P]GTP as a label. This label would be incorporated only into RNA initiated with a 5′-“G” residue and so would detect RNA initiated at position +3 of the template but not at +1. In this case, the banding pattern of RNA products was similar to what was observed in reaction mixtures containing [α-32P]GTP (Fig. 3D, panel iii, compare lanes 2 and 3 and lanes 6 and 7). Taken together, these data indicated that the RNA generated from the Le promoter under these reaction conditions was initiated at position +3.
Because there was no evidence for RNA initiated from the +1 position, these findings suggested that initiation at +3 occurred independently of initiation at +1; however, the possibility remained that the RdRp initiated at +1 to yield very short transcripts (e.g., 2 to 3 nt in length) and then reinitiated at position +3. To eliminate this possibility, a mutant template containing a 2G-to-A substitution was also tested. In reaction mixtures containing this template, omission of UTP from the RNA synthesis reaction mixture would prevent incorporation of the second NTP into RNA initiated at position +1, thus inhibiting the +1 initiation event. Labeling of RNA with [γ-32P]GTP showed that the Le 2G-to-A template yielded the same levels of RNA products as the wt Le-14 RNA, regardless of whether UTP was present or absent in the reaction mixture (Fig. 3D, panel ii, compare lanes 4 and 5 and lanes 7 and 8; panel iii, compare lanes 3 and 5 and lanes 7 and 9). These results strongly indicated that RNA synthesis was initiated at position +3 even when extension through position +2 was prevented. Thus, the major initiation site for the RSV RdRp on a Le nt 1 to 14 template under these particular reaction conditions was at position +3, and the RdRp could access this initiation site without prior RNA synthesis from position +1.
The data illustrated in Fig. 1 and and22 show that the RSV RdRp could generate short transcripts from an L GS-like sequence within nt 3 to 12 of the Le region. These findings raise the question of why RNA initiated at position +3 was truncated after such a short distance, whereas the L mRNA can be extended many hundreds of nucleotides. One possible explanation is that there are features of a GS signal that determine RdRp processivity and that this is impacted by the two-nucleotide differences between Le nt 3 to 12 and the L GS signal. To investigate if this possibility was correct, nucleotide substitutions were introduced into Le nt 3 to 12 to increase its similarity to the L GS signal. Two mutant minigenomes were created, containing either a G-to-C substitution at position 4 (4C), such that Le nt 3 to 12 differed from the L GS signal by a single nucleotide, or two nucleotide substitutions of 4G-to-C and 7U-to-G (4C7G), such that Le nt 3 to 12 were identical to the L GS signal (Fig. 3A). The mutations were introduced into a minigenome that was restricted to the antigenome step of replication, so that the levels of template RNA available for transcription and RNA synthesis from position +3 would not be affected by mutations that inhibited RNA replication. RSV transcription and RNA replication were reconstituted in cells using the minigenome templates, and total intracellular RNA was extracted and analyzed.
As a control, the levels of negative-sense input template were examined by Northern blotting and were found to be similar for each minigenome (Fig. 4D, panel iii). Primer extension analysis of the positive-sense RNA products of the minigenomes was performed using the g15-39 and g45-68 primers (Fig. 4A) to examine the relative efficiency of elongation of the RNA, similar to the experiment described in Fig. 1. Analysis using primer g15-39 showed that the ratio of +3 to +1 initiation products generated from the minigenome with wt Le was significantly less than that observed in RSV-infected cells (compare Fig. 4C, left panel, with Fig. 1C). This suggested that either the frequency with which the initiation sites were used or the stabilities of the RNA initiated at +1 and/or +3 differed in transfected versus infected cells. Nonetheless, the RNA initiated at +3 was detectable, allowing the effect of the Le substitution mutations to be determined. The 4C and 4C7G minigenomes generated higher levels of RNA from the +3 initiation site than the minigenome with wt Le, indicating either a higher level of initiation and/or greater stability of this RNA (Fig. 4B, panel i, compare lane 2 with lanes 3 and 4). The 4C minigenome also showed a higher level of RNA initiated from +1, whereas there was no detectable RNA from the +1 site of the 4C7G minigenome. Analysis with the g45-68 primer showed that RNA initiated at +1 could be detected for both the wt and 4C minigenomes, as expected, and that the relative amounts of RNA initiated at +1 from these two minigenomes were similar regardless of which primer was used (Fig. 4B, compare panels i and ii, lanes 2 and 3). This indicated that although the 4C minigenome yielded more RNA from +1 than the wt minigenome, the efficiency of elongation of this RNA was similar in both cases. However, the efficiency of elongation of the RNA initiated at +3 was apparently different. Analysis of the wt minigenome showed that compared to RNA initiated at +1, there was a decrease in RNA initiated at +3 that could be detected by the g45-68 primer compared to the g15-39 primer (Fig. 4B, compare panels i and ii, lane 2; Fig. 4C, left panel). This indicated that the RNA initiated at +3 of the wt minigenome was not efficiently extended, similarly to what was observed in RSV infected cells. In contrast, analysis of the RNA from the 4C minigenome showed that the ratio of RNA from +1 and +3 was similar for the g15-39 and g45-68 primers (Fig. 4B, compare panels i and ii, lane 3; Fig. 4C, left panel). Likewise the g45-68 primer was able to readily detect RNA initiated at +3 from the 4C7G minigenome (Fig. 4B, panel ii, lane 4). These data indicate that the RNA initiated from position +3 of the 4C and 4C/7G minigenomes was more efficiently extended into the NS1 gene than RNA initiated from +3 of the wt minigenome.
The RNA was also examined by Northern blotting. Northern blots probed with oligonucleotide g7-34, to detect RNA initiated within the Le region, showed that the wt minigenome generated antigenome RNA but no other long RNAs containing LeC-specific sequence (Fig. 4D, panel i, lane 2), consistent with the results obtained with RNA isolated from RSV-infected cells (note that the antigenome comigrates with a background band, which can be detected in the −L lane). In contrast, the 4C and 4C7G minigenomes generated readily detectable LeC-containing subgenomic RNA, including a distinct band of the appropriate size to be Le-chloramphenicol acetyltransferase 1 (CAT 1) readthrough mRNA and a smear of smaller RNAs ~200 nt in length (Fig. 4D, panel i, lanes 3 and 4). This result is consistent with the hypothesis that alteration of Le nt 3 to 12 to increase their similarity to a GS signal can increase the ability of the RdRp to extend RNA transcripts initiated at position +3 to the GE signal of the CAT 1 gene. For comparison, duplicate Northern blots were probed with a negative-sense CAT-specific riboprobe to detect total levels of full-length antigenome and mRNA, including mRNA generated from the NS1 GS signal (Fig. 4A and andD,D, panel ii). This analysis showed that minigenome 4C produced a greater level of antigenome than the wt minigenome, whereas minigenome 4C7G produced no detectable antigenome. Both the 4C and 4C7G minigenomes produced mRNA at lower levels than the wt minigenome. These results are consistent with previously published data (24).
Given that the substitutions in Le nt 3 to 12 increased the ability of the RdRp to elongate the RNA transcripts initiated from +3, it was of interest to investigate the effect of the substitutions on initiation from the NS1 GS signal at position +45. Analysis of the RNA with primer g56-75 showed that all three minigenomes were able to initiate RNA synthesis from the NS1 GS signal (Fig. 4B, panel iii). However, the level of initiation at +45 was decreased approximately 2-fold in the case of the 4C and 4C7G minigenomes (Fig. 4C, right panel), indicating that changing the way that the RdRp recognized and responded to the initiation signal within nt 3 to 12 had consequences for mRNA initiation from nt +45. Longer exposures of the gel, shown in Fig. 4B, panel iii, allowed direct comparison of the abundance of RNA generated from the +1, +3, and +45 initiation sites (Fig. 4B, panel iv). This analysis showed that even though the 4C and 4C7G mutations allowed more efficient elongation of the RNA initiated from +3 and inhibited initiation from +45, the RNA initiated from the +45 site remained the most abundant in each case, indicating that this initiation site was used more frequently and/or that the RNA produced from this site was more stable.
The data described above indicated that in the case of the 4C and 4C7G minigenomes, on some occasions the RdRp that initiated at +3 of the Le could extend the RNA to the first GE signal at the end of the CAT 1 gene. To determine if this was a bona fide transcriptase, we examined if the RdRp that completed LeC-CAT 1 mRNA synthesis was able to engage in sequential transcription and transcribe the CAT 2 gene of the minigenome. This analysis was performed using minigenome templates with either wt Le (i.e., with a G at position 4) or mutant Le with a C residue at position 4 (4C). The minigenomes were modified to remove the NS1 GS signal, so that the only initiation sites near the 3′ end of the template were in the Le region (Fig. 5A, upper minigenome, termed −GS). A minigenome with a wt Le (i.e., 4G) and an intact NS1 GS signal was included in the experiment as a marker to show the sizes of the CAT 1 and CAT 2 mRNAs (Fig. 5A, lower minigenome, termed +GS). In this experiment, the minigenomes contained an intact Tr region and so were capable of multicycle replication, to enhance the signal of the CAT 2 mRNA. As shown in Fig. 5B, the −GS 4C minigenome produced mRNA similar in length to the CAT 1 mRNA produced from the marker minigenome, consistent with it being a LeC-CAT1 readthrough mRNA, whereas the −GS wt minigenome did not (Fig. 5B, lanes 2 to 4). The −GS 4C minigenome also produced an abundant level of RNA of appropriate size to be CAT 2 mRNA. A similarly sized RNA was also produced from the −GS wt minigenome, but at a significantly lower level (Fig. 5B, lanes 3 and 4, and C). To confirm that the LeC-CAT 1 RNA was not encapsidated, a series of transfections were performed in parallel and the cell lysates were treated with micrococcal nuclease prior to RNA extraction to digest unencapsidated RNA. This eliminated the LeC-CAT 1 readthrough mRNA, while the antigenome RNA remained intact, indicating that the LeC-CAT 1 RNA was not encapsidated (Fig. 5B, lanes 7 and 8). Taken together with the results shown in Fig. 4, these data suggest that mutation of the Le region to increase similarity of nt 3 to 12 to a GS signal allowed the RdRp to extend mRNA initiated within the Le region to the GE signal and then to continue sequential transcription.
The data presented in this paper confirm the existence of a position +3 initiation site in the RSV Le region. The RdRp that initiated at this site yielded short RNA transcripts that could be readily detected in RSV-infected cells. By using an in vitro RNA synthesis assay and manipulating the Le sequence in the minigenome assay, it was possible to explore the mechanism by which this RNA was initiated and terminated. We showed that initiation at Le +3 could occur independently of initiation at position +1, indicating that the GS-like signal at nt 3 to 12 of Le can be accessed directly by the RSV RdRp. Substitution of one or two nucleotides of the Le to make nt 3 to 12 identical to the L GS sequence resulted in an increased elongation of RNA to the first GE signal, showing that the initiation signal and/or the sequence at the 5′ end of the RNA transcript can directly or indirectly influence RdRp processivity. Together, these data provide new information regarding the relationship between the RSV RdRp and cis-acting sequences in the viral genome and suggest a model for RSV transcription initiation, as discussed below.
The finding that the RSV RdRp initiated transcription of an RNA from position +3 of the Le region is the first demonstration of an alternative initiation site within the Le promoter being used during NNS RNA virus infection and challenges the dogma that the only initiation site within the Le region is opposite the first nucleotide at position +1. Recent analysis of the RSV TrC promoter has also detected initiation at position +3 during RSV infection, which is not surprising given the similarity of the 3′ terminal sequences of the Le and TrC promoter regions (33). This finding raises the question of how the RSV RdRp engages with the promoter to accomplish initiation from two closely spaced sites. The Le region contains a sequence at nt 3 to 12 that strongly resembles the L GS signal (Fig. 3A). Thus, it seems likely that it is this signal sequence that engages the RdRp and directs initiation of RNA synthesis from the +3 position. The results obtained from the in vitro RNA synthesis assay revealed that initiation from position +3 could occur independently of initiation from position +1 (Fig. 3D). This finding is consistent with results from a previous minigenome study, which showed that deletion or substitution of nt 1 and 2 of the Le did not inhibit RNA synthesis from the +3 position (29). To date, we have been unable to detect initiation from position +1 of the Le region in the in vitro RNA synthesis assay, suggesting that either initiation at +1 is much less efficient or that additional sequence and/or different conditions are required. Our previous studies have indicated that replication initiation at position +1 can occur by an unusual nontemplated initiation mechanism, and we have suggested that the RdRp might utilize a primer to accomplish this (29, 32). A recent paper showing that the dengue virus RdRp can self-generate a primer in a template-independent manner to initiate from the +1 position on a template lends credence to this hypothesis (35). Therefore, it is possible that nt 3 to 12 of the promoters are responsible for recruiting the RdRp and directing initiation of RNA synthesis from +1 and +3 but that the RdRp needs to become loaded with an A-C dinucleotide primer for initiation at +1.
A combination of primer extension and Northern blot analysis of RNA purified from RSV-infected cells showed that whereas the RNA initiated from the +1 position could be extended to generate antigenome RNA, most of the RNA initiated at +3 was truncated, yielding short transcripts of ~25 nt in length (Fig. 1 and and2).2). It should be noted that the primer extension analysis performed would not distinguish if some RNA initiated at +1 was also terminated prematurely, and so the ~25-nt RNA transcripts might include RNA initiated from +1 as well as RNA initiated at +3. Transcripts of equivalent size are also generated from the TrC promoter (33). Although the Le and TrC regions are similar in sequence, nt 23 to 27 of these promoters, where the majority of +3 initiated transcripts would be terminated, are not identical and bear no obvious similarity to a GE signal. These observations, combined with the size heterogeneity of the transcripts, suggests that termination of their synthesis occurs as a consequence of the RdRp reaching a checkpoint in RNA synthesis, rather than recognizing a specific stop signal within the promoters. A checkpoint that is known to occur during RSV RNA virus synthesis is during mRNA transcription, if the mRNA transcript is not capped. However, in this case, the RNA is truncated at 45 to 50 nt (13; L. R. Deflubé and R. Fearns, unpublished data), suggesting that the checkpoint that comes into play within the Le and TrC promoters is distinct from that which aborts mRNA synthesis following capping failure. Based on our current understanding of NNS RNA virus RNA replication, it seems likely that the checkpoint within the Le and TrC promoters is related to encapsidation and that the RdRp aborts RNA synthesis from the wt promoter if the RNA is not becoming bound with N protein as the RNA is synthesized. Consistent with this hypothesis, the experiment presented in Fig. 2D showed that some of the ~25-nt RNA was degraded by nuclease treatment, indicating that it was not encapsidated. The fact that the remainder was nuclease resistant is difficult to interpret. While this RNA might have become encapsidated concurrently with RNA synthesis, it is possible that it became encapsidated after being released by the RdRp. It is also possible that it was protected from nuclease digestion due to binding to another viral or cellular protein. It should be noted that this result should be interpreted with caution. If the ~25-nt RNA were not capped or protein bound, it could be highly labile. This could bias the results of this experiment toward detection of RNA that was protein bound, which potentially could be a small fraction of the total population of ~25-nt RNA transcripts that were generated.
Analysis of RNA generated in the minigenome system suggested that the promoter checkpoint causing the transcripts initiated at +3 to be truncated could be overcome, at least in part, by a single nucleotide substitution in the Le promoter sequence, which increased the similarity of nt 3 to 12 to a GS signal. A 4G-to-C substitution was able to allow elongation of RNA initiated at +3 and resulted in the accumulation of LeC-mRNA1 transcripts (Fig. 4). A plausible explanation for this is that this single nucleotide change could have facilitated capping of some of the RNA initiated at +3. In the case of VSV, it has been shown that initiation of RNA synthesis and capping of the nascent transcript are both directed by the GS signal (11, 12). In addition, it has been shown that capping of VSV mRNA occurs when the transcript reaches 31 nt in length (36). It has been proposed that this is because the RdRp must have moved this far from the mRNA initiation site to allow the 5′ end of the nascent transcript to reach the capping domain on the L protein (36). Therefore, it is possible that the single nucleotide change in the Le promoter facilitated capping of transcripts that had been extended far enough for their 5′ termini to reach the capping site on the RSV L protein and that the addition of the cap allowed the RdRp to then elongate the RNA. Alternatively, it is possible that the substitutions had no effect on capping per se but affected RdRp processivity by another mechanism. We obtained evidence indicating that a significant proportion of RNA initiated from the +3 site of the 4C minigenome contained a cap (data not shown), consistent with the hypothesis that the RNA was elongated because it was capped. However, these experiments were performed in cells infected with MVA-T7 to supply the T7 RNA polymerase, and so it is possible that the cap was added posttranscriptionally by the vaccinia virus capping enzyme, rather than cotranscriptionally by the RSV RdRp (unfortunately, we were unable to obtain sufficient levels of RNA for analysis when BSR-T7 cells were used instead). It would also have been informative if we could have determined what effect the 4C and 4C7G substitutions had on the relative amounts of ~25-nt RNA versus LeC-CAT 1 mRNA. Unfortunately, in these experiments it was not possible to detect the ~25-nt transcripts by Northern blotting due to very high background in the minigenome system (data not shown). Therefore, it is not known to what extent the 4C and 4C7G substitutions diminished the levels of these transcripts, but presumably if the change in elongation was entirely due to cap addition, this could impact only the transcripts extended far enough for the cap to be added but would not affect the ability of shorter transcripts to become elongated. Thus, it remains unclear exactly how the 4C and 4C7G substitutions affected RdRp activity within the promoter. Nonetheless, the data presented here clearly show that the RNA synthesis initiation sequence, or its complement in the nascent RNA, can have an impact on the subsequent behavior of the RSV RdRp.
Experiments with the minigenome system also indicated that the RdRp that initiated RNA synthesis within the Le promoter and extended the RNA to the end of the first gene was able to continue in a stop-start transcription mode (Fig. 5). This finding indicates that in the case of RSV, transcription-competent RdRp is not constrained to initiate at the NS1 GS signal but can enter at the 3′ end of the template. The fact that the RdRp could initiate from a GS-like sequence at nt 3 to 12 of the Le, coupled with the fact that it frequently terminated this transcript within the Le region, suggests a modified model for initiation of sequential mRNA transcription in RSV, in which the primary initiation site for this process is at +3 of the Le (Fig. 6). According to this model, the RSV RdRp is recruited to the GS-like sequence at nt 3 to 12 of the Le promoter and initiates transcription at position +3. Initiation of RNA synthesis allows the RdRp to break promoter contacts and move along the template. However, the RNA initiated at +3 is not capped or encapsidated, causing the RdRp to recognize a checkpoint and release the RNA after synthesis of ~25 nt. Similarly to events occurring at a gene junction, at least some of the RdRp is able to remain attached to the template and is able to scan the template to access the U-stretch at the end of Le and the NS1 GS sequence located at position +45. Recognition of the NS1 GS signal would allow the RdRp to reinitiate RNA synthesis accurately at the start of the NS1 gene. A similar sequence of events could occur if the RdRp initiates at position +1 and the RNA fails to become encapsidated. On the other hand, if the RNA does become encapsidated, the RdRp is able to bypass the checkpoint and complete antigenome synthesis. Strong support for the hypothesis that RSV transcription begins with RNA synthesis initiation from position +3 comes from previously published experiments that showed that introduction of a single nucleotide substitution at Le nt 3, 5, 8, 9, 10, or 11, the nucleotides that are identical in sequence to the L GS signal (Fig. 3A), inhibited all transcription from a minigenome template (24). This model would also explain why addition of six or more nucleotides to the 3′ end of the RSV Le inhibited mRNA transcription in previous experiments (23): presumably, it is important for the GS-like initiation sequence to be near the 3′ end of the nucleocapsid template to be accessed. The model is also consistent with the finding that elongation of the transcript initiated at position +3, by increasing the similarity of Le nt 3 to 12 to a GS signal, resulted in reduction of initiation at +45 (Fig. 4). If the model presented in Fig. 6 were correct, initiation at +3 would be the dominant initiation event, occurring with greater frequency than initiation at +45. Unfortunately, we do not have a suitable assay to measure levels of initiation at +1, +3, and +45. Performing primer extension analysis using infected cell extracts, as shown in Fig. 1, can measure only steady-state levels of RNA initiated at the different positions, and as stated above, uncapped and unencapsidated RNA may be very labile. Nonetheless, even in cell extracts, the level of RNA initiated at +3 of the Le region was very significant compared to initiation at +1 (Fig. 1B, panel i), indicating that initiation of RNA synthesis at +3 is a relatively frequent event, consistent with it playing a role in initiation of sequential mRNA transcription.
One possible argument against the model presented here is that even when there was an intact GS signal at nt 3 to 12 of the Le region, a significant proportion of mRNA was initiated from position +45 (Fig. 4B). However, there are three possible sources of RdRp that could account for this: (i) RdRp that initiated at +1 but truncated the transcript within the Le region, (ii) RdRp that initiated at +3 but released the transcript before capping could occur, and (iii) RdRp that initiated RNA synthesis at position +3 and elongated the RNA beyond the ~25-nt checkpoint but terminated prematurely within the CAT 1 gene and scanned backwards to locate the GS at +45. With regards to the third possibility above, it was noticeable that even with an intact GS sequence at nt 3 to 12, there was still considerable premature termination of the Le-CAT1 mRNA transcript, yielding RNA transcripts longer than ~25 nt but only ~200 nt long. This is not typically observed during transcription and could suggest that there is a fundamental difference in the processivity of the RdRp when it initiates within the Le compared to when it initiates at +45. Alternatively, it is possible that the model presented above is incomplete and that both positions +3 and +45 are used as transcription initiation sites, as suggested previously (19). It is possible that initiation from position +3 is the primary transcription initiation event, used when the RdRp first engages with the nucleocapsid, and that once the N:RNA structure is opened by the RdRp engaging with the Le promoter, all subsequent initiations take place directly at position +45. While the data presented here and in previous studies do not distinguish between these two possible models, they do offer a new perspective on how transcription is initiated in RSV.
Finally, it is important to consider the consequence to the virus of producing large amounts of small RNA. If the ~25-nt transcripts are uncapped and unencapsidated, they have the potential to be recognized by RIG-I and to alert the cell to the presence of the virus. However, it has been shown that cellular La autoantigen binds to RSV LeC-containing RNA to subvert activation of RIG-I (37). It is also possible that these transcripts are not merely by-products of transcription and replication but play a functional role in infection. Consistent with this hypothesis is the finding that the Tr RNA of Sendai virus can bind TIAR (and also possibly TIA-1) and subvert the cellular stress granule response (38), and a study of the RSV Tr suggests it might play a similar role (39). In addition, small RNAs generated from the influenza virus promoter have been shown to bind the viral RdRp complex and influence transcription and replication (40, 41). An intriguing feature of the small RNAs generated from the RSV Le (and TrC) promoters is their size. At ~25 nt, they are similar in size to a fully processed microRNA; therefore, in future studies it will be of significant interest to determine if they are able to function in the cellular microRNA pathway.
We thank David McGivern for generating the minigenomes used in the assays illustrated in Fig. 5, as well as for making the initial observation that the 4C minigenome makes a Le-CAT1 readthrough RNA, Bernard Moss for providing MVA-T7, Peter Collins for providing the plasmids required for minigenome expression, and Sean Whelan and members of his group for helpful discussion.
This work was funded by NIH grant AI074903 to R.F. and NIH T32 grant AI07642 to the Department of Microbiology at Boston University School of Medicine to support C.Z.T.
Published ahead of print 2 January 2013