Initiation of Genomic Plus-Strand RNA Synthesis from DNA and RNA Templates by a Viral RNA-Dependent RNA Polymerase

Abstract

In contrast to the synthesis of minus-strand genomic and plus-strand subgenomic RNAs, the requirements for brome mosaic virus (BMV) genomic plus-strand RNA synthesis in vitro have not been previously reported. Therefore, little is known about the biochemical requirements for directing genomic plus-strand synthesis. Using DNA templates to characterize the requirements for RNA-dependent RNA polymerase template recognition, we found that initiation from the 3′ end of a template requires one nucleotide 3′ of the initiation nucleotide. The addition of a nontemplated nucleotide at the 3′ end of minus-strand BMV RNAs led to initiation of genomic plus-strand RNA in vitro. Genomic plus-strand initiation was specific since cucumber mosaic virus minus-strand RNA templates were unable to direct efficient synthesis under the same conditions. In addition, mutational analysis of the minus-strand template revealed that the −1 nontemplated nucleotide, along with the +1 cytidylate and +2 adenylate, is important for RNA-dependent RNA polymerase interaction. Furthermore, genomic plus-strand RNA synthesis is affected by sequences 5′ of the initiation site.

Replication of the viral genome is an essential feature of viral pathogenesis. In RNA viruses whose genomes can be translated directly after entry into the cell, the genomic plus-strand RNA serves as the template for minus-strand synthesis. The minus-strand RNA then serves as the template for generating multiple copies of genomic and, where applicable, subgenomic plus-strand RNAs (4). Our laboratory studies viral RNA replication in brome mosaic virus (BMV) as a model system. BMV is a plant-infecting tripartite plus-strand RNA virus and is the type member of the bromovirus group of plant viruses of the alphavirus-like superfamily (8). BMV has three genomic RNAs, which are designated RNA1 (3.2 kb), RNA2 (2.8 kb), and RNA3 (2.1 kb). RNA1 and RNA2 are monocistronic and encode the 1a protein and 2a protein, which are essential for the replication of the BMV genome (2). RNA3 is dicistronic and encodes the 3a movement protein and the coat protein.

Replication of viral RNA is facilitated by the RNA-dependent RNA polymerase (RdRp), a complex composed of 1a, 2a, and unidentified cellular proteins. The BMV RdRp can accurately initiate minus-strand RNA synthesis from input plus-strand templates and can initiate the synthesis of subgenomic plus-strand RNA products from input minus-strand templates (1, 9, 12, 14, 15, 21, 22). In contrast, little is known about sequence and structural elements required for directing genomic plus-strand RNA synthesis, in large part because an in vitro assay for initiation of BMV genomic RNA synthesis was not previously reported. In this work, our studies of RNA synthesis from DNA templates led us to propose that BMV genomic plus-strand RNA synthesis require a nontemplated nucleotide. The addition of this nucleotide to the 3′ end of BMV minus-strand RNAs led to the initiation of genomic plus-strand RNAs in vitro. We also present detailed analyses of the sequence requirements necessary for accurate initiation of genomic plus-strand RNA syntheses.

MATERIALS AND METHODS

Synthesis of DNA and RNA templates for the RdRp assay.

Synthetic deoxyoligonucleotides used as template for RdRp assays were purchased from Operon Technology. All oligonucleotides were quantified by spectrophotometry, adjusted to the desired concentration, and visually inspected after being stained with toluidine blue after gel electrophoresis. RNA templates were made by PCR and used for in vitro transcription. One of the PCR primers contained a T7 promoter to allow transcription directly from the PCR products. Typically, each of 30 PCR cycles consisted of 30 s of denaturation at 94°C, annealing at 5°C before the lowest oligonucleotide melting temperature (T_m), and elongation at 72°C. PCR products were purified by standard methods (20) followed by in vitro transcription (Ampliscribe; Epicentre). Transcripts were purified by anion-exchange chromatography on Qiagen Tip-20 columns as specified by the manufacturer. The concentration of RdRp templates was determined by toluidine blue staining following denaturing polyacrylamide gel electrophoresis (PAGE) and by spectrophotometry.

RdRp activity assays.

BMV RdRp was prepared from infected barley as described by Sun et al. (26). Standard RdRp activity assay mixtures consisted of 40-μl solutions containing 20 mM sodium glutamate (pH 8.2), 4 mM MgCl₂, 12 mM dithiothreitol, 0.5% (vol/vol) Triton X-100, 2 mM MnCl₂, 200 μM ATP, 500 μM GTP, 200 μM UTP, 242 nM [α-³²P]CTP (400 Ci/mmol, 10 mCi/ml; Amersham), the desired amount of template, and 5-10 μl of RdRp. Following incubation for 90 min at 30°C, the reaction products were extracted with phenol-chloroform (1:1, vol/vol) and precipitated with 6 volumes of ethanol, 10 μg of glycogen, and 0.4 M (final concentration) ammonium acetate.

Products of the RdRp reactions were digested with 2.5 U of S1 nuclease for 10 min at 30°C in the buffer supplied by the manufacturer (Promega). Denaturing loading buffer (45% [vol/vol] deionized formamide, 1.5% [vol/vol] glycerol, 0.04% [wt/vol] bromophenol blue, 0.04% [vol/vol] xylene cyanol) was added to the S1-treated products. The samples were heated at 90°C for 3 min and then analyzed by PAGE on 10% acrylamide gels containing 7 M urea. All the gels were exposed to film at −80°C, and the amount of label incorporated into newly synthesized RNAs was determined with a PhosphorImager (Molecular Dynamics).

RESULTS

DNA templates for RNA synthesis.

While RNA is the preferred template for the BMV RdRp, accurate initiation of RNA synthesis can use DNA templates (23). Initiation of RNA synthesis from a DNA template can take place from either a penultimate cytidylate or an internal cytidylate in processes resembling the synthesis of genomic minus-strand and subgenomic RNAs. Furthermore, the interaction between RdRp and DNA, as measured by a template competition assay, is remarkably similar to the interaction between RdRp and RNA (23). The ease of manipulation of DNA templates through standard chemical synthesis makes DNA an attractive substitute for determining the requirements for RNA synthesis from RNA templates.

To examine synthesis from DNA templates, deoxyoligonucleotide d(−1/13) was used as the prototype. d(−1/13) contains the sequence complementary to the template for the BMV subgenomic RNA (nucleotides [nt] 1241 to 1252 of BMV RNA3). This template directs the synthesis of two products of 13 and 14 nt; the latter is the result of nontemplated nucleotide addition by the BMV RdRp (21). The −1 nt is a guanylate, which, along with the initiation cytidylate (+1), has previously been demonstrated to be necessary for RNA synthesis (23). Two guanylates were added at the 5′ end to allow transcription initiation by T7 polymerase and to direct the incorporation of radiolabeled cytidylates.

The requirements for the initiation of RNA synthesis from the penultimate nucleotide were examined first. Consistent with the report of Siegel et al. (23), the removal of the 3′-terminal guanylate reduced RNA synthesis to 10% of that for d(−1/13) (Fig. 1A, lane 1). However, the −1 guanylate can be replaced with a uridylate, cytidylate, or adenylate and RNA synthesis is then between 38 to 63% of that for d(−1/13) (lanes 3 to 5). Addition of a cytidylate at the −1 position should place a potential initiation nucleotide at both the 3′ terminus and the penultimate position. However, initiation still took place from the penultimate cytidylate as judged by the mobility of the resultant RNA. The addition of two nucleotides (3′ AG) 3′ of the initiation cytidylate decreased RNA synthesis to 33%. However, the resulting products were indistinguishable in size from those produced by d(−1/13), suggesting that initiation still took place from the authentic cytidylate. Addition of 3 nt (3′ AAG) 3′ of the initiation cytidylate resulted in synthesis at 14% of that from d(−1/13). These results indicate that the penultimate cytidylate is the preferred nucleotide for initiation and that additional sequence 3′ of the initiation nucleotide can affect the efficiency of RNA synthesis.

FIG. 1.

Synthesis of plus-strand RNA by using DNA templates. (A) Template d(−1/13) containing the sequence complementary to nt 1241 to 1252 of BMV RNA3 are shown with the initiation cytidylate indicated by an arrow. Changes of the 3′ ends of d(−1/13), the +2 adenylate, the +3 uridylate, and the +4 adenylate were used for RNA synthesis by BMV RdRp. The changes are indicated above the autoradiogram of the RdRp products. The positions of the 13- and 14-nt products are shown on the left. The reaction products were separated by denaturing PAGE (12% polyacrylamide) and visualized by autoradiography. (B) Summary of the effect of nucleotide changes focusing on the 3′ end of the initiation site. All results presented were from at least three independent trials. (C) Summary of the effect of nucleotide changes at positions +2, +3, and +4 in d(−1/13). (D) Effect of changes to a guanylate in the first six positions in d(−1/13) on plus-strand RNA synthesis.

Replacement of nucleotides at positions +2 to +7.

The sequence immediately 5′ of the initiation site is rich in A and U nucleotides. Marsh et al. (14) have proposed that this sequence contributes to efficient RNA synthesis. All possible changes of nucleotides from positions +2 to +4 were made to examine the requirements for RNA synthesis (Fig. 1A, lanes 8 to 16). Replacement of the +2 adenylate with uridylate resulted in synthesis at levels similar to that for d(−1/13). However, changing the +2 adenylate to a guanylate or a cytidylate reduced synthesis to less than 30% of that for d(−1/13) (Fig. 1C). Transition from uridylate to cytidylate at the +3 position was acceptable for RNA synthesis while transversion to purines decreased RNA synthesis to less than 20% of that for d(−1/13) (Fig. 1A, lanes 1 to 13). Changing adenylate to either uridylate or cytidylate at the +4 position resulted in levels of RNA synthesis similar to those for d(−1/13), while a change to a guanylate decreased RNA synthesis to 21% (Fig. 1A, lanes 14 to 16; Fig. 1C).

The presence of guanylates at positions +1 to +4 was detrimental for efficient RNA synthesis (Fig. 1D). To examine this correlation further, positions +5 and +6, normally uridylates, were individually changed to guanylate. A change of U to G at +5 yielded the same RNA synthesis as for d(−1/13), while a change of U to G at +6 resulted in 40% of the synthesis for d(−1/13). A change of A to G at +7 also resulted in the same level of synthesis as for d(−1/13) (data not shown). These results suggest that the presence of a guanylate at the first four positions of the template would decrease the efficiency of RNA synthesis but that this requirement may be relaxed starting at the +5 position.

Interaction between DNA template and RdRp.

Changes at various positions along the DNA template could affect either the ability of the template to interact stably with RdRp or the efficiency of nucleotide incorporation into the nascent RNA. We used a template competition assay (21) to distinguish between these two possibilities. The reaction mixture contains limiting amounts of RdRp, a reference RNA, r(−20/15), which directs the synthesis of a 15-nt product, and various concentrations of the competitor template. Many of the competitor templates chosen for this analysis were observed to decrease RNA synthesis (Fig. 2). For simplicity, the experiments described below used r(−20/15) as the reference template. Competitor d(−1/13) at fivefold molar excess reduced synthesis from r(−20/15) to 60%. A more severe reduction was observed when d(−1/13) was present at 10-fold molar excess of r(−20/15). Removal of the −1 nucleotide resulted in a template that could no longer compete with r(−20/15) for RNA synthesis even at 10-fold molar excess (Fig. 2). The ability to inhibit RNA synthesis was partially restored when the −1 nucleotide of the template was a cytidylate, consistent with its being a more effective template. When the initiation nucleotide was at the fourth position from the 3′ end due to the addition of 3′ AAG, the ability to inhibit synthesis was again reduced.

FIG. 2.

Effect of nucleotide changes in d(−1/13) on the ability of the resultant DNA template to compete for RdRp. An RNA template, r(−20/15), directing the synthesis of a 15-nt product, was used as a reference. The amounts of RNA synthesis generated from r(−20/15) in the presence of the different competitors are listed as percentages relative to synthesis in the absence of competitor. DNA competitors were used at 5- and 10-fold molar excess with respect to r(−20/15). All results were from at least three independent trials. ND, not determined.

To determine whether the initiating cytidylate is also involved in the stable interaction with RdRp, competitors were tested in molar excess of the reference template. At fivefold molar excess, a change of the cytidylate to a guanylate (+1C/G) reduced the ability of the template to stably interact with RdRp. A significant decrease in synthesis occurred when +1C/G was present at 10-fold molar excess, suggesting that the ability of the template to interact with RdRp is partially retained. Previous studies with the BMV subgenomic RNA promoter showed that a change of the initiation nucleotide was able to reduce, but not abolish, stable interaction with RdRp (21).

In contrast to the importance of the −1 and +1 positions, changes in upstream positions along the template had less dramatic effect on the stable interaction with RdRp. Although some nucleotide substitutions at the +2 position reduced the ability to direct RNA synthesis, their ability to compete for interaction with RdRp were similar to that of wild-type (wt) d(−1/13) (Fig. 1C) (24). Similarly, changes at position +3 or +6 did not significantly affect the ability of the template to interact with RdRp (Fig. 2).

A more quantitative analysis of the interaction between DNA templates and the BMV RdRp was performed. DNA competitor was added in increasing molar amounts relative to r(−20/15) to determine the concentration of the competitor DNA required to reduce synthesis by 50% (IC₅₀). Lower IC₅₀s indicate that the competitor can interact more stably with RdRp. The IC₅₀ of wt d(−1/13) was 175 nM. An oligonucleotide with deoxythymidines instead of deoxyuridine had a similar IC₅₀ value of 150 nM (Fig. 3). These results show that the C-5 methyl group that distinguishes thymines from uracils does not have a detectable effect on the ability of the DNA template to interact with RdRp. Consistent with the inhibition assay results shown in Fig. 2, removal of the 3′-terminal guanylate or changing of the initiation cytidylate to a guanylate raised the IC₅₀ to ca. 1 mM. These results indicate that the primary determinants for RdRp-DNA template interaction lie within nt −1 and +1. Additional sequence upstream of the initiation site must affect RNA synthesis at a level beyond stable interaction with the BMV RdRp.

FIG. 3.

Concentration of DNA templates needed to reduce RNA synthesis from r(−20/15) by 50%. The percentage of synthesis from r(−20/15) directing the synthesis of a 15-nt product was measured in the presence of increasing amounts of competitor DNA. The IC₅₀s are given within the boxed region. The results for d(−1/13)dT were not plotted, to reduce the complexity of the figure.

Synthesis of genomic plus-strand RNAs from minus-strand endscripts.

The results of the DNA template studies described above suggest that an additional nucleotide 3′ of the initiation cytidylate may be required. Siegel et al. (21) have demonstrated that BMV RdRp could add a nontemplated nucleotide at the 3′ end of an RNA molecule. Since the presence of this nontemplate nucleotide in the minus-strand RNA would not direct the incorporation of an extra nucleotide in the viral RNAs, it may have been missed in routine analyses of viral sequences.

We wanted to examine the requirements for the initiation of BMV genomic plus-strand RNA synthesis by using RNA templates. Short RNAs of 58, 46, and 51 nt, each with an extra 3′ guanylate, were generated to correspond to the minus strand of BMV RNA1, RNA2, and RNA3, respectively. These short transcripts were termed “endscripts,” since they represented the minus-strand 3′ ends of BMV RNAs (Fig. 4B). All three endscripts should contain the complete stem-loop structure observed to be required for plus-strand genomic RNA synthesis in vivo (18). By using a standard RdRp assay, we observed that endscripts B2(−)46G, B1(−)58G, and B3(−)51G were all able to direct plus-strand synthesis (Fig. 4C, lanes 1 and 2, 5 and 6, and 9 and 10, respectively). A high-resolution gel shows that these bands may be composed of fragments of two distinct sizes, possibly due to nontemplated terminal nucleotide addition by the BMV RdRp (24). As a control for the specificity of RNA synthesis, we tested endscripts of the minus strand of cucumber mosaic virus (CMV) RNA2 and RNA3, i.e., CMV2(−)G and CMV3(−)G, respectively. CMV2(−)G, of 64-nt, was able to direct synthesis at only 10% of B2(−)46G, and CMV3(−)G, of 54 nt, was unable to direct any synthesis of discrete RNA products by the BMV RdRp (Fig. 4C, lanes 13 to 16). The synthesis from BMV endscript is thus species specific. To determine whether initiation of BMV endscripts took place from the penultimate cytidylate, the +1 cytidylate was mutated to a guanylate in all three BMV endscripts. These mutants were greatly reduced in their ability to direct plus-strand synthesis (Fig. 4C, lanes 3 and 4, 7 and 8, and 11 and 12), indicating that genomic plus-strand RNA synthesis is initiated from the cytidylate presumed to be used in vivo (2). A smear, presumed to be misinitiated RNAs, is observed in lanes with mutant B3(−)51G endscripts (lanes 11 and 12). Initiation of genomic plus-strand RNA is approximately one-third of initiation of a 46-nt RNA from the subgenomic promoter (1). However, these results clearly indicate that in vitro BMV RdRp is able to specifically distinguish BMV promoters and initiate genomic plus-strand synthesis.

FIG. 4.

Initiation of genomic plus-strand RNAs directed by minus-strand endscripts. (A) Comparison of the 3′ sequences of BMV and CMV minus-strand RNAs. The nontemplated guanylate added to each template is shown in bold type. The initiation cytidylate is denoted by an arrow. (B) Predicted secondary structures of the 3′ ends of BMV RNA1, RNA2, and RNA3, i.e., B1(−)58G, B2(−)46G, and B3(−)51G, respectively. The structure predictions were generated by the MFOLD program (10). (C) Initiation of genomic plus-strand RNAs from minus-strand endscripts. RdRp reaction products were separated by denaturing PAGE (12% polyacrylamide) and visualized by autoradiography. The amounts of RNA synthesized from various templates relative to B2(−)46G (% Syn) are shown at the bottom of the autoradiogram. The results presented are an average from three independent trials. The sizes of the RNA products are indicated on the side of the autoradiogram. The symbol φ represents the products of a control reaction with no added template. Endscripts that are initiation competent are indicated by +, while initiation-incompetent endscripts are indicated by −. C2(−)G and C3(−)G are endscripts of CMV RNA2 and CMV RNA3, respectively. (D) Synthesis of a 200-nt genomic plus-strand RNA. The endscript with a guanylate replacing the +1 cytidylate is indicated by +1c/g. −G denotes a 200-nt endscript without the designed nontemplated guanylate; +G denotes a 200-nt endscript with a guanylate at the 3′ end of the RNA. RNA synthesis from B2(−)200+G, B2(−)200Δ-1, and B2(−)200+1C/G was 100, 27, and 0% respectively. The results presented are an average from three independent trials. M denotes a reaction designed to produce a molecular mass marker of 203 nt.

Next, we wanted to determine whether it was possible to generate longer fragments of genomic plus-strand RNAs by using minus-strand endscripts. Two endscripts were made: B2(−), of 200 nt, lacking a designed 3′-terminal nontemplated guanylate, and B2(−)200G, containing a −1 guanylate. B2(−)200G was able to direct efficient synthesis of a product of the expected length in comparison to the molecular mass marker (Fig. 4D, lanes 1 and 5). In four independent experiments, synthesis from B2(−)200 was at 30% of that from B2(−)200G (Fig. 4D) (24). The synthesis observed in the absence of a 3′ nontemplated nucleotide might be due to addition of an extra nucleotide(s) by the T7 polymerase during transcription (6, 17). To confirm that initiation of the 200-nt product took place from the penultimate cytidylate, the +1 cytidylate was mutated to a guanylate. This mutant endscript was unable to direct plus-strand synthesis (Fig. 4D, lane 3). In addition to the 200-nt product, a prominent 100-nt product was observed. The 100-nt RNA appears to be due to internal initiation of synthesis, since mutation of the +1 cytidylate did not affect its synthesis (Fig. 4D). Since shorter endscripts are more conducive to analysis of template requirements, all experiments described below were performed with the B2(−)46G as the prototype.

Effect of nucleotide changes near the initiation cytidylate.

In addition to the −1 position, we wanted to determine the effect of changes at +1, +2, and +3 on RNA synthesis. Consistent with previous observations, changing the +1 cytidylate to a guanylate was detrimental and virtually abolished RNA synthesis (Fig. 5A, lane 3). Changing the +2 adenylate to a guanylate reduced synthesis to 8% of that of the wt (lane 4). These results are comparable to observations made with DNA templates containing identical changes at the +1 and +2 positions (Fig. 1C). However, changing the +3 uridylate to an adenylate did not affect RNA synthesis from an RNA template, which is in contrast to what was observed when a DNA template with the identical nucleotide change was used as the template (compare Fig. 1A, lane 12, and Fig. 5A, lanes 1 and 5). The lack of a 2′ OH and/or the presence of a 5′-methyl group at specific positions may combine to alter the interaction between the template and RdRp.

FIG. 5.

The effects of nucleotide changes near the initiation cytidylate on RNA synthesis. (A) Effect of nucleotide changes near the initiation cytidylate. Changes from B2(−)46G (top sequence) are indicated in bold type. − indicates the absence of the −1 nucleotide. (B) Effects of the identity of the 3′ nontemplated nucleotide on genomic plus-strand initiation. The initiation cytidylate is indicated by an arrow. Substitutions of the 3′ nontemplated nucleotide is indicated in bold type. The synthesis directed by the different endscripts is given as a percentage relative to B2(−)46G. The results presented are from three independent trials.

Effect of the minus-strand 3′ end on RNA synthesis.

To further analyze the effect of the 3′ nontemplated nucleotide on RNA synthesis, we generated endscripts containing different 3′ nucleotides. The absence of a nontemplated nucleotide at the 3′ end significantly reduced RNA synthesis (Fig. 5A, lane 2). The −1 guanylate could be replaced by a uridylate, adenylate, or cytidylate with only moderate reduction of the levels of plus-strand synthesis (Fig. 5B, lanes 1 to 4). These results indicate that while the 3′ nontemplated nucleotide is necessary for efficient RNA synthesis, the identity of the nucleotide at this position is not crucial. However, the addition of two nucleotides (3′ AG) or three nucleotides (3′ AAG) 3′ of the initiating cytidylate reduced plus-strand RNA synthesis to less than 15% of wt (24), indicating that the BMV RdRp prefers the initiating cytidylate to be at the penultimate position in an RNA template.

The nucleotide changes could affect either the stability of interaction with RdRp and/or the ability to direct synthesis by RdRp. To distinguish between these two possibilities, we carried out competition assays with r(−20/15) as the reference template. As a nonspecific competitor, a 53-nt RNA unrelated to BMV (PCRII/53) was used. PCRII/53 caused a slight decrease in synthesis from the reference RNA r(−20/15), to 66%, when present at 250 nM, the highest concentration tested, suggesting that there is limited nonspecific RdRp-RNA interaction (Table 1). The B2(−)46G endscript was a more effective competitor, reducing synthesis from r(−20/15) with an IC₅₀ of 120 nM (Table 1). This concentration is similar to the 175 nM obtained with the DNA template, d(−1/13) (Fig. 3). Removal of the −1 nucleotide at the 3′ end (Δ−1) or changing the +1 cytidylate to a guanylate (+1C/G) decreased the competitiveness of these RNAs (Table 1); they were unable to inhibit RNA synthesis from the reference template to less than 50%, even at the highest concentration tested (24). Therefore, the IC₅₀s of these RNAs are listed as >250 nM. Changes at the +2 position (+2A/G) also resulted in a slight reduction in the ability of the template to interact with the RdRp complex, but not to the degree observed for changes at −1 or +1 positions (Table 1). Changes at the +3 position (+3U/A) yielded an IC₅₀ similar to that of wt B2(−)46G (Table 1). Competition assays using CMV2(−)G and CMV3(−)G endscripts showed that BMV RdRp does not bind to CMV minus strand templates (24). This is not surprising since the 3′ sequence (3′ GCA) in the CMV RNA was identical to the BMV 3′ sequence (Fig. 4A). However, the inability of BMV RdRp to direct efficient RNA synthesis by using CMV templates suggests that in addition to the 3′ nucleotides, upstream sequences are required for efficient RNA synthesis.

TABLE 1.

Mutations in endscripts can affect the ability to interact with RdRp

Construct	3′ Sequencea	IC50 (nM)
B2(−)46G	3′ GCAUUU… N46 5′	120
B2(−)Δ−1	3′ -CAUUU… N46 5′	>250
B2(−)Δ−1,+1,+2	3′ ---UUU… N44 5′	>250
B2(−)+1C/G	3′ GGAUUU… N46 5′	>250
B2(−)+2A/G	3′ GCGUUU… N46 5′	235
B2(−)+3U/A	3′ GCAAUU… N46 5′	140
Nonspecific RNA	3′ AUAGGU… N53 5′	>250

^aNucleotides in bold indicate a change from the wild-type sequence. 

To examine the spatial relationship between the 3′ initiation site and upstream sequences, endscripts containing one, two, or three initiation sites (C₊₁A₊₂U₊₃) were constructed and assayed (Fig. 6). Endscript containing two initiation sites generated products of 46 and 49 nt, corresponding to initiation from the authentic and the second initiation sites, respectively. In terms of abundance relative to B2(−)46G, the 49-nt RNA was at 89% while the 46-mer was at 25% (Fig. 6). When three initiation sites were present, products of 52, 49, and 46 nt corresponding to initiation from all three potential sites were observed at 69, 43, and 19%, respectively. Since the 3′-terminal initiation site always yields the largest amount of product, this result confirms our previous observation that RdRp has a preference for the penultimate cytidylate (Fig. 6). However, it is also interesting that in the presence of additional initiation sites, the authentic initiation site is still used. This observation suggests that an appropriate spacing between the 3′ initiation site and sequences or structure requiring the sequence 5′ of the initiation site may influence the efficiency of plus-strand RNA synthesis.

FIG. 6.

Effect of multiple initiation sites on RNA synthesis. The authentic initiation site is indicated by an arrow in the first construct marked 1. Additional initiation sites added to the 3′ end of B2(−)46G are indicated by arrows 2 and 3. The three initiation sites should generate products of 46, 49, and 52 nt. RNA synthesis directed by the different initiation sites from their respective templates are presented relative to initiation from cytidylate 1 in B2(−)46G. Products initiated from the three potential initiation sites are indicated on the right as 1, 2, and 3 respectively.

5′ requirements to initiate plus-strand RNA synthesis.

The B2(−)46G endscript is predicted to fold into a stable stem-loop structure whose subdomains we named L1, L2, A1, and A2 (Fig. 7A). To determine whether RNA structures or sequences 5′ of the initiation site were important for directing plus-strand synthesis, a set of deletions were constructed and tested. Deletion of nucleotides originally in positions 3 to 11, expected to perturb the A2 stem, reproducibly increased synthesis to 200% of that for wt B2(−)46G (compare Fig. 7B lanes 1 and 2 with lanes 5 and 6). Deletion of nucleotides originally in positions 17 to 26, expected to disrupt the L1 loop, resulted in an RNA that directed 70% of wt RNA synthesis (lanes 9 and 10). In both cases, initiation took place from the authentic +1 cytidylate, since a change to a guanylate resulted in RNAs that failed to direct synthesis (lanes 7 and 8 and lanes 11 and 12). The large amount of RNA synthesis from templates with the L1 deletion was probably due to replacement of the L1 loop region with 5′ sequence, as we demonstrate below.

FIG. 7.

Requirements for plus-strand RNA synthesis. (A) The predicted secondary structure of B2(−)46G with the stems (A1 and A2) and loops (L1 and L2) indicated by brackets. The initiation cytidylate is denoted by an arrow. (B) Analysis of the sequence in the predicted stem-loop region. The templates used in the specified reactions are indicated at the top of the autoradiogram. Endscripts that have a +1 are indicated by +, while endscripts with changes of the initiation cytidylate to a guanylate are indicated by −. The sizes of the RdRp products are denoted on the right. The amounts of plus-strand synthesis directed by B2(−)Δ3–11 and B2(−)Δ17–26 were 200 and 70%, respectively, compared to synthesis directed by B2(−)46G after correcting for the number of CMP units incorporated. The results presented are an average from three independent trials. (C) Additional analysis of sequences required for efficient RNA synthesis. The templates used and whether they can direct the initiation of RNA synthesis (+ or −) are indicated at the top of the autoradiogram. The sizes of the RNA products are denoted on the right of the autoradiogram. The amounts of synthesis, after adjusting to the number of radiolabelled CMP units incorporated, from B2(−)26G, B2(−)26TV, B2(−)22G, and B2(−)16G were 100, 17, 22, and 5% respectively. (D) Alignment of the sequences in B2(−)26G, B2(−)Δ3–11 (deletion of A2 stem region), B2(−)Δ17–26 (deletion of L1 loop region), and B2(−)26TV (transversion of nt 17 to 24). The two guanylates as well as the two adenylates putatively required for efficient synthesis are shown in bold type.

To determine further the effect of 5′ sequences, endscripts with 5′ truncations were made and tested. When the template sequence was reduced to 26 nt in B2(−)26G, molar amounts of plus-strand RNA synthesis remained comparable to those from B2(−)46G (Fig. 7C, lanes 4 and 5). However, a further deletion of 4 and 10 nt in endscripts B2(−)22G and B2(−)16G, respectively, reduced RNA synthesis to 22 and 5%, respectively (lanes 10 and 11 and lanes 13 and 14). These results suggest that nt 17 to 26 may be required for efficient synthesis. To confirm this without changing the length of the RNAs, transversion of nt 17 to 24 in the context of the 26-mer, B2(−)26TV, was tested and found to result in only 17% synthesis in comparison to B2(−)26G (lanes 7 and 8). In all cases, initiation took place from the authentic +1 cytidylate, since a mutation at the initiation cytidylate failed to direct RNA synthesis (lanes 3, 6, 9, 12, and 15). The results indicate that positions 17 to 26 contain an element(s) important for RNA synthesis.

The deletion of nt 17 to 26 in the context of B2(−)46G reduced synthesis to 70% of wt. This is unexpected because transversion of the same sequence in the context of B2(−)26G reduced synthesis to 17% of wt. The incongruity of these results prompted us to examine more closely the sequence in the 5′ portion of B2(−)Δ3–11 and B2(−)Δ17–26. Despite our changes, both endscripts contain sequences that are similar between nt 17 to 26. For example, both B2(−)Δ3–11 and B2(−)Δ17–26 contain the two guanylates that are naturally present in B2(−)46 and both contain two adenylates 4 or 5 nt 3′ of the guanylates (Fig. 7D). Furthermore, these two adenylates are absent in the nonfunctional template B2(−)26TV. Stawicki and Kao (25) have demonstrated that RdRp can have a certain amount of flexibility in adjusting to minor spatial perturbations that may allow the recognition of the two adenylates in B2(−)Δ17–26 despite the 1-nt difference in their spacing. Presently we do not have any experimental evidence to show that the two adenylates or the 5′ guanylates are directly interacting with RdRp. However, the similarity in the 5′ sequences in B2(−)Δ3–11 and B2(−)Δ17–26 may explain the observations that both efficiently directed RNA synthesis.

Since the 5′ end affects initiation of RNA synthesis, we asked whether sequence from positions 17 to 26 is sufficient for interaction with RdRp. To address this question, an RNA containing B2(−) nucleotides 11 to 26 and an RNA of the same length with the transversions at positions 17 to 26 were chemically synthesized for use in template competition assays. Addition of either oligonucleotide to 10-fold molar excess with respect to the reference RNA r(−20/15) failed to reduce RNA synthesis. As controls, assays carried out in parallel with competitor B2(−)26G or B2(−)26TV were able to compete efficiently (24). These results indicate that nucleotides in the region from positions 11 to 26 in the absence of the 3′ sequence are not sufficient for stable interaction with RdRp.

DISCUSSION

We have demonstrated for the first time accurate in vitro initiation of genomic plus-strand RNA synthesis by BMV RdRp. The use of DNA templates that mimicked the 3′ end of the BMV RNA minus strand allowed us to determine that the recognition of minus-strand templates by RdRp in vitro required a nontemplated nucleotide. Addition of this nontemplated nucleotide to the 3′ ends of the minus strands of all three BMV RNAs allowed accurate initiation of genomic plus-strand RNAs in a species-specific manner. Finally, efficient RNA synthesis is affected by the sequence 5′ of the initiation site.

Comparison of synthesis from RNA and DNA templates.

RNA synthesis directed by DNA templates is about 8% as efficient as RNA synthesis with RNA templates (23). However, the requirements for template recognition by RdRp were similar for both DNA and RNA templates. We found that the +1 initiation cytidylate, the +2 adenylate, and the nontemplated −1 nucleotide were important for RNA synthesis and recognition of the RNA by RdRp. However, at the +3 position, a U-to-A change is tolerated in an RNA template but not in a DNA template. This result may indicate that the 2′ hydroxyls in the initiation site contribute to stable RdRp-template interaction and that its absence in DNA templates forces RdRp to require additional upstream contact sites.

Despite slight differences in the use of DNA and RNA templates by BMV RdRp, DNA templates can be useful in establishing conditions for studies of RdRp-RNA interaction. Working with DNA has many technical advantages. For example, DNAs longer than 100 nt can be chemically synthesized with precise ends whereas chemical synthesis of RNA is presently inefficient for stretches longer than 40 nt. Furthermore, RNA templates generated by T7 RNA polymerase often contain additional nucleotides at the 3′ termini due to the propensity of T7 RNA polymerase to add nontemplated nucleotides (6). Lastly, a wider range of modified nucleotides is available as deoxynucleotides, allowing a more in-depth probing of the nucleotide moieties required for the RdRp-RNA interaction.

RdRp-template interaction.

RNA synthesis from the end of a template requires 1 nt 3′ of the initiation nucleotide. We observed this requirement for the initiation of plus-strand RNA synthesis from both DNA and RNA templates (Fig. 1 and 5A). There is precedence that a nontemplated nucleotide may be a general requirement for RNA synthesis. This requirement was observed with BMV minus-strand RNA synthesis from the tRNA-like promoter (26). A nontemplated nucleotide at the 3′ end of the minus strand has been found in CMV and Semliki Forest virus (7, 29). A nontemplated guanylate is required for the production of CMV-associated satellite RNA (CARNA5) (30). In CMV, Semliki Forest virus, and CARNA5, the extra nucleotide at the 3′ end was a guanylate, and the results of the in vitro experiments with BMV endscripts suggest that there is a preference for a guanylate for initiation of plus-strand RNA synthesis.

The nontemplated nucleotide may be added by the terminal transferase activity in the RdRp complex. Such an activity appears to be present in many polymerases, including those from BMV (21), poliovirus (16), vaccinia virus (3), and bovine viral diarrhea virus NS5B (31). The hepatitis C virus NS5B may be an exception in that the terminal nucleotide transferase activity observed in some preparations of the hepatitis C virus RdRp may be due to a cellular protein (13).

The extranucleotide 3′ of the initiating cytidylate provides stability in the RdRp-RNA interaction. Templates lacking the nontemplate nucleotide were reduced in both RNA synthesis and the ability to compete for interaction with RdRp (Fig. 2 and Table 1). While the nontemplated nucleotide may bond with RdRp, the bonds are not base specific, since nucleotides other than guanylates were capable of directing efficient RNA initiation. However, there do appear to be constraints in the number of residues 3′ of the initiation cytidylate, since more than one nontemplated nucleotide reduced RNA synthesis and interaction with RdRp. We note that the recombinant RdRp of bovine viral diarrhea virus does not appear to require any nucleotide 3′ of the initiation nucleotide for efficient RNA synthesis (11), suggesting that the steric constraints of catalytic pockets of different RdRps may vary.

Role of 5′ sequences.

The 3′ ends of BMV and CMV RNAs have identical sequences (Fig. 4A). However, the CMV RNAs are poor templates for RNA synthesis, indicating that additional sequence 5′ of the initiation site are important for efficient synthesis, and may confer specificity in viral RNA synthesis. The 5′ sequences may affect synthesis by three possible mechanisms: (i) modulation of abortive synthesis, (ii) antitermination of RNA synthesis (5, 19), or (iii) direct initiation at the 3′ end (but not RdRp binding). It is possible that BMV RdRp will abort synthesis when using non-BMV templates. However, since abortive synthesis usually terminates before the first 10 phosphodiester bonds are formed (27, 28), we speculate that the BMV sequence between nt 17 and 26 may not be directly involved in abortive synthesis but could be inducing elongation or affecting initiation.

Pogue and Hall (18), have suggested that the putative plus-strand stem-loop structure and the ICR2-like sequences present within the loop region are important for genomic plus-strand synthesis. When a putative plus-strand stem-loop structure was disrupted, RNA replication was greatly reduced (18), whereas disruption of the putative minus-strand stem-loop structure was not correlated with effects on viral replication (18). These results led Pogue and Hall to propose that the plus-strand RNA is involved in additional rounds of plus-strand RNA synthesis. We have demonstrated that accurate initiation of plus-strand RNA synthesis could take place from minus-strand templates in the absence of plus-strand RNAs (and the ICR-like sequences). In addition, disrupting the predicted A2 stem structure does not affect the ability of the template to direct RNA synthesis. While our results do not rule out possible contributions of the plus-strand RNA in vivo, they raise the possibility that the results of Pogue and Hall (18) reflect some requirements other than those required for initiation of genomic plus-strand RNA synthesis. Our data further suggests that the sequence complementary to ICR2, nt 17 to 24, is involved in RNA synthesis. Transversion of the sequence complementary to ICR2, in the context of a 26-nt minus-strand RNA, or the deletion of this sequence greatly reduced the ability of the template to direct synthesis (Fig. 7C). Taken together, these observations suggest that the minimal promoter sequence required for directing plus-strand initiation should be present within the first 26 nt of the minus-strand RNA.

The initiation of genomic plus-strand RNA in vitro completes the last class of RNA promoters used by the BMV RdRp. Synthesis of genomic plus-strand RNA from endscripts was about one-third as productive as synthesis directed by a subgenomic plus-strand promoter. We would rank the amount of RNA synthesized from the three BMV promoters from most to least efficient as follows: subgenomic RNA > genomic plus-strand RNA > minus-strand RNA. This relative abundance suggests that different viral RNA promoters have inherently different abilities to interact with RdRp and/or direct nucleotide polymerization. More work is needed to understand the mechanism of asymmetric RNA synthesis in a viral infection.