Transcription and replication initiate at separate sites on the vesicular stomatitis virus genome

Abstract

The RNA-dependent RNA polymerase of the nonsegmented negative-strand RNA viruses carries out two distinct RNA synthetic processes: transcription of monocistronic, capped, and polyadenylated subgenomic messenger RNAs, and replication by means of the synthesis of a full-length positive-sense copy of the genome. The template for both processes is the negative-sense genomic RNA tightly encapsidated by the viral nucleocapsid protein. By applying UV transcriptional mapping to engineered variants of vesicular stomatitis virus, we discovered that, in infected cells, transcription and replication are controlled by initiation at different positions on the viral genome.

The genome of vesicular stomatitis virus (VSV), the prototypic rhabdovirus, comprises 11,161 nucleotides of negative-sense RNA. This RNA, tightly encapsidated by the viral nucleocapsid (N) protein, forms the template for the RNA-dependent RNA polymerase (RdRP), the viral components being a 241-kDa large subunit (L) and a phosphoprotein (P) (1, 2). The RdRP uses the genome as a template for two reactions: (i) transcription of a short leader RNA (Le+) and 5 mRNAs that encode N, P, matrix (M) protein, glycoprotein (G), and L; and (ii) replication to yield full-length antigenomic and then genomic strands.

The VSV gene order was determined by transcriptional mapping using ultraviolet (UV) radiation. Low-dose UV radiation induces formation of covalent dimers between adjacent thymine (T) or uracil (U) residues in nucleic acids (3–5). These block progression of polymerase during transcription and thus show the distance polymerase traverses to transcribe a gene. Such experiments determined the VSV gene order as 3′-N-P-M-G-L-5′ and showed the mRNAs were synthesized in an obligatory sequential manner after polymerase entry at a single 3′ promoter (6–8). The abundance of viral mRNAs was shown to decrease with distance from the promoter such that N > P > M > G > L (9). This relative abundance reflected a localized transcriptional attenuation, whereby 30% of polymerase molecules reaching each gene junction failed to transcribe the downstream gene (10).

The accepted model for VSV transcription is the stop-start model, which posits that polymerase enters the genome at a single site and that transcription of a downstream gene depends on termination of transcription of the upstream gene (11). The N gene is preceded by a 50-nt leader region, which is transcribed into a 47-nt RNA (Le+) (12–14). This leader region is too small to contribute to the UV sensitivity of the 1,336-nt N mRNA. Thus UV mapping studies could not determine whether transcription of N required prior transcription of Le+.

In vitro reconstitution studies suggested a single polymerase entry site. In these experiments, polymerase was added to template in the presence of limiting nucleoside triphosphates (NTPs), and the products of de novo synthesis were analyzed (15). In the presence of ATP and CTP, the major product was 5′-pppApC-3′, representing the first 2 nucleotides of Le+. The tetranucleotide 5′-pppApApCpA-3′, corresponding to the first 4 nucleotides of the N mRNA, was found only in a two-step reaction. In this reaction, all NTPs were present, the reaction was stopped, NTPs were removed, and the products of synthesis in the presence of ATP and CTP were analyzed. These experiments indicated that the RdRP could access the template only at the extreme 3′ end, and that transcription of N required prior transcription of Le+.

Studies on a VSV mutant, polR1, with a single amino acid change in the template-associated N protein, provided evidence that conflicts with the single-initiation site model (16). PolR1 virus synthesized N mRNA in excess over Le+, which suggested a two-entry site model for RNA synthesis: 3′ entry for synthesis of Le+ and the full-length antigenome, and direct entry at the first gene-start site for synthesis of mRNAs (17).

Reverse genetic systems developed for the nonsegmented negative-sense (NNS) RNA viruses (18, 19) led to identification of signals that regulate gene expression. Analysis of the VSV intergenic junction defined signals for mRNA termination and initiation and provided support for the stop-start model (20–27). Analysis of the VSV genomic termini identified sequences essential for transcription and replication (28–33). However, these studies were unable to address the stop-start model of sequential transcription at the Le-N gene junction.

Determination of the polymerase initiation site is pivotal to understand global regulation of RNA synthesis in VSV, and by extrapolation other NNS viruses. If transcription and replication initiate at nucleotide 1, then during synthesis of Le+ a regulatory event must occur to determine whether polymerase transcribes or replicates the genome. This regulatory event was postulated to be encapsidation of nascent Le+ with N protein (34, 35). Conversely, if transcription and replication initiate at separate positions, regulation must occur before synthesis.

We examined whether polymerase initiates transcription at the 3′ end of the genome, or directly at the first gene start. Recombinant viruses were generated to test this question by using the technique of UV-mapping. These viruses contained an additional small gene (I), either 60 or 108 nt, inserted just downstream of the leader region. In addition, the sequence of the leader region was altered to increase or decrease the number of adjacent U residues, such that the UV sensitivity of Le+ would be altered in a predictable manner. The effect of these changes on the UV sensitivity of Le+ and the I mRNA was examined in vitro and in infected cells. These experiments confirmed that transcription of the first gene required prior transcription through the leader region in vitro. In infected cells, however, we demonstrate that polymerase initiates transcription directly at the first gene start sequence.

Materials and Methods

Recombinant Viruses.

An infectious cDNA clone of VSV, pVSV1(+) (36), was modified by standard methods (37). The I gene sequences were 3′-UUGUCAUUAGUCUUAAGAGCUCUACCGUAUGACAUUGGUAUUCCGGUAUACUUUUUUUGA-5′ (I-60), and 3′-UUGUCAUUAGUCUUAAGAGCUCGUGUUGGUUGGUCUUGCACUUUUUCGCAGGACGCACAUCGCUUGACGCUACCCGUAUGACAUUGGUAUUCCGGUAUCUUUUUUUGA-5′ (I-108). Potential sites for UV-induced U dimers are underlined. Viruses were recovered, amplified, and purified as described (36). Sequences were confirmed by reverse transcription–PCR on genomic RNA by using a primer that annealed to the first 15 nt of the genome and a second primer designed to anneal to the complement of nucleotides 145–166 in the N gene of the wild-type (WT) genome.

UV Irradiation.

Purified virus was diluted in PBS to 2 × 10⁹ plaque-forming units (pfu)/ml. Aliquots (0.25 ml) were dispensed into 6 wells of a 24-well plate on ice, exposed to UV radiation at 31 erg⋅mm⁻²⋅s⁻¹ for 0–8 min at a distance of 54 cm from a 30-W UV bulb. Irradiations were performed in three groups: (i) rV1+60 WT, rV1+60incU, and rV1+60decU for in vitro analysis; (ii) rV1+60 WT, rV1+60incU, and rV1+60decU for in vivo analysis; and (iii) rV1+60 and rV1+108 for in vivo analysis.

Analysis of RNA Synthesis in Infected Cells.

Before infection, baby hamster kidney (BHK-21) cells were treated with 10 μg/ml cycloheximide (CHX) and 10 μg/ml actinomycin-D (act-D) for 15 min. Cells were infected with VSV at a preirradiation multiplicity of infection of 100, in duplicate. Virus was adsorbed for 45 min and cells were exposed to [³H]uridine in the presence of act-D and CHX from 0.25 to 5.25 h after infection. Cells were harvested, cytoplasmic extracts were prepared, and RNAs were analyzed on agarose urea gels (18, 31).

VSV RNA Synthesis in Vitro.

No quantitative difference in the UV sensitivity of the I mRNA was obtained when VSV was detergent disrupted or intact before UV irradiation (not shown). In vitro transcriptions used detergent-activated virions (38). Purified virus activated with 0.05% Triton N-101/0.4 M NaCl (5 min, 25°C) was diluted with 10 mM Tris⋅HCl, pH 8.1, before irradiation. Reaction mixtures contained 1 × 10⁸ pfu of virus in 100 μl that contained 30 μl of nuclease-treated rabbit reticulocyte lysate (Promega), 10 μl of 10× buffer (0.3 M Tris⋅HCl, pH 8.1/0.33 M NH₄Cl/0.07 M KCl/0.045 M Mg(OAc)₂/0.01 M DTT/0.002 M spermidine/10 mM ATP/5 mM each GTP, CTP, and UTP), and 5 μl of actinomycin-D (1 mg/ml). Duplicate reactions were incubated at 30°C for 5 h.

Primer Extension Analysis.

The I mRNAs and Le+ RNAs were detected by primer extension, using specific oligonucleotide primers extended by Superscript II reverse transcriptase (Invitrogen) at 50°C. The I mRNA was detected by primer (5′-TTTTTTTTTTTTCATA-3′) designed to anneal to all VSV mRNAs. Leader RNAs were detected by using primers designed to anneal to nucleotides 47–33 of WT Le+ (5′-GTTTCTCCTGAGCC-3′), Inc-U Le+ (5′-TTTTTTCATAGGCC-3′), and Dec-U Le+ (5′-GTCTCTCCTGAGCC-3′). Primers were end-labeled by using [γ-³³P]ATP and T4 polynucleotide kinase (Invitrogen). Labeled primers were purified from unincorporated [γ-³³P]ATP by using a QIAquick nucleotide removal kit (Qiagen). Labeled primer (0.2 pmol) was annealed with 1/25 of the total RNA from a 60-mm dish or an in vitro transcription reaction. Products were analyzed by electrophoresis on denaturing 6% polyacrylamide gels and quantitated by densitometric scanning (39).

Results

VSV Engineered to Contain an Additional Gene Inserted at the Leader–N Gene Junction and Modifications to the Sequence of the Leader Region.

To examine whether polymerase initiates transcription directly at the first gene-start sequence or must first transcribe Le+, we generated recombinant viruses that allowed UV mapping to discriminate these two possibilities. We used a recombinant VSV (rV1+60) that contained an additional 60-nt gene (I) inserted between the leader region and the N gene (39). This additional gene comprised the gene-start (3′-UUGUCNNUAG-5′), 37 nt of stuffer sequence, and the gene-end and intergenic dinucleotide (3′-AUACUUUUUUUGA-5′). In addition, two other viruses were generated that altered the number of UV target sites in the leader region (Fig. 1A and Table 1). The rationale was that if transcription initiated at the 3′ end of the genome, these alterations would affect the UV sensitivity of the I-60 mRNA. Conversely, if the RdRP bypassed the leader region and instead initiated transcription directly at the first gene-start, these changes would not affect the UV sensitivity of I-60. Previously we identified sequences in the leader region essential for transcription and replication (33). Guided by these analyses, we generated two variants of rV1+60: one in which the number of adjacent U residues in leader was increased (rV1+60incU) and a second in which they were decreased (rV1+60decU) (Fig. 1A). The sequence of the leader region of these viruses was confirmed by reverse transcription–PCR (data not shown). All viruses grew to similar titers (>10⁹ pfu/ml) and had similar plaque sizes, indicating that viral growth was not measurably affected (data not shown).

Figure 1.

Effect of UV radiation on VSV mRNA synthesis in infected cells. (A) Schematic of VSV with an additional gene between leader and the N gene and alterations to the leader sequence. The negative-sense genome is shown 3′-5′. The sequences of the leader regions are shown. le, leader; tr, trailer; Inc-U, increased-U leader; Dec-U, decreased-U leader. Note positions of nucleotide identity are indicated by a -. (B) Effect of UV radiation on mRNA synthesis. Purified viruses were exposed to UV radiation before infection of BHK cells. RNAs were labeled, extracted, treated with RNase H before electrophoresis, and visualized by fluorography. The mobility of the individual mRNA species is indicated (L, G, N, P, M). Note: (i) the P and M mRNAs comigrate; and (ii) the small size and low U content of the I mRNA make its detection difficult by direct metabolic labeling.

Table 1.

Number of U dimers that can form on UV irradiation

Virus	Leader	Maximum no. of U dimers
		I-60		I-108
		3′	Internal	3′	Internal
Dec-U	3	11	8
WT	6	14	8	20	14
Inc-U	8	16	8

The maximum number of covalent U dimers that could form on UV irradiation are shown. For I-60 and I-108 the values are shown for initiation of synthesis at the 3′ end (3′) or for initiation of synthesis directly at the first gene start site (Internal). Sequences are shown in Fig. 1A and Materials and Methods. To calculate the number of U dimers we used the following assumptions: U3 can only form 1, U4 and U5 can form a maximum of 2, and U6 and U7 can form a maximum of 3. 

The effect of these alterations on viral mRNA synthesis and on the UV sensitivity of the N, P, M, G, and L mRNAs was examined. Viruses were resuspended in PBS at a titer of 2 × 10⁹ pfu/ml and exposed to UV radiation at a dose of 31 erg⋅mm⁻²⋅s⁻¹ for 0–2 min. Duplicate confluent monolayers of BHK cells were infected with VSV at a preirradiation titer of 1 × 10⁸ pfu. Cells were pretreated with actinomycin-D to inhibit DNA-dependent RNA synthesis, and cycloheximide to prevent protein synthesis and thus prevent replication of undamaged genomes. This treatment ensured that the only viral RNA synthesis that occurred was primary transcription. Viral RNAs were labeled, extracted, annealed with oligo(dT) before exposure to RNase H to remove polyadenylate tails, analyzed by electrophoresis, and visualized by fluorography (Fig. 1B). The sensitivity of the viral mRNAs to UV radiation were similar for rV1+60, rV1+60decU, and rV1+60incU in that L > G > M > P > N. This result demonstrated that the alterations to the viral genome had not altered transcription, and confirmed the sequential nature of transcription described for WT VSV (6–8). Because of their size, the contribution of the leader region and the 60-nt gene to the UV sensitivity of the N, P, M, G, and L mRNAs was undetectable. In addition, the low U content of the I-60 mRNA made it difficult to detect this species by metabolic labeling.

Effect of Altering the U Content of the Leader Region on the UV Sensitivity of Leader RNA.

To examine the effect of alterations to the U content of the leader region on the UV sensitivity of Le+, VSV transcription was analyzed in vitro. Detergent-disrupted virus was exposed to UV radiation at a dose of 31 erg⋅mm⁻²⋅s⁻¹ for 0–8 min. In vitro transcriptions were performed in duplicate and RNAs were recovered. The abundance of each Le+ was determined by primer extension (Fig. 2A). For each virus, the 47-nt product of primer extension on the Le+ was detected, and its abundance diminished as the dose of UV radiation increased. Reactions were performed under conditions of primer excess to ensure quantitative detection. Leader RNA abundance was determined by densitometric scanning of the autoradiographs. The amount of Le+ remaining after varied doses of UV radiation was expressed as a percentage of the average value obtained from the unirradiated control, and was plotted (Fig. 2B). The rate of decrease of synthesis for each Le+ was linear as demonstrated by the lines of best fit. This result indicated that inhibition of synthesis followed single-hit kinetics. According to the Poisson distribution, each member of the population receives, on average, a single UV hit when the abundance of the RNA is reduced to 37%. This 37% survival dose was determined as 39,060, 20,460, and 12,090 erg⋅mm⁻² for the Le+ synthesized by Dec-U, WT, and Inc-U, respectively. Thus the UV sensitivity of Le+ depended on the number of adjacent U residues in the leader region. The maximum number of U dimers that could form in each template was 3, 6, and 8 for Dec-U, WT, and Inc-U, respectively (Table 1). These numbers predict relative survival doses of 1.00, 0.50, and 0.375 for Dec-U, WT, and Inc-U, in good agreement with the measured values of 1.0, 0.5, and 0.3. These data thus show that altering the number of adjacent U residues in the leader region affected the UV sensitivity of Le+ in a predictable manner.

Figure 2.

Effect of alterations to the leader region on the UV sensitivity of the leader RNA synthesized in vitro. (A) Primer extension analysis of the leader RNAs. Reverse transcriptions were performed with primers designed to anneal to positions 33–47 of the individual leader RNAs. Products were analyzed on denaturing 6% polyacrylamide gels, and the individual leader RNAs are shown (Le). (B) Quantitation of the leader RNAs. The autoradiographs in A were scanned and analyzed. For each Le+ the amount remaining was plotted as the percentage of synthesis directed by the unirradiated control. ■, WT; ●, Inc-U; ▴, Dec-U.

Effect of UV Radiation on mRNA Synthesis in Vitro.

To determine whether transcription of the first gene required transcription through the leader region, we examined the UV sensitivity of the I-60 mRNA synthesized in vitro. The I-60 mRNA was detected by primer extension using oligonucleotide primer 5′-TTTTTTTTTTTTCATA-3′ (Fig. 3A). This primer annealed to each VSV mRNA by virtue of their common 3′-terminal sequence 5′ … UAUG(A)_n. The products of extension on the viral N, P, M, G, and L mRNAs would be large (>800 nt), and are therefore not visualized. For each virus, the product of primer extension on the I-60 mRNA was detected, and its abundance diminished with increasing doses of UV radiation (Fig. 3A). The I-60 mRNA abundance was determined and plotted (Fig. 3B). The rate of decrease of synthesis of I-60 was linear as demonstrated by the lines of best fit, confirming single-hit kinetics. The 37% survival dose for I-60 synthesized by Dec-U, WT, and Inc-U was determined as 11,720, 9,110, and 6,880 erg⋅mm⁻², respectively. Thus, altering the number of adjacent U residues in the leader region affected the UV sensitivity of I-60. The maximum numbers of U dimers that can form in each template are 11, 14, and 16 for Dec-U, WT, and Inc-U, respectively (Table 1). Thus the predicted relative sensitivity of I-60 was 1.00, 0.79, and 0.69 for Dec-U, WT, and Inc-U, respectively. This result agrees with the ratio of experimentally derived values of 11,720:9,110:6,880 or 1.0:0.8:0.6. These data showed that altering the number of adjacent U residues in the leader region affected the UV sensitivity of the I-60 in a predictable manner. This observation is consistent with 3′ entry and obligatorily sequential transcription of Le+ before synthesis of I-60, and not with direct initiation of transcription at the first gene-start. Thus we confirmed previous work demonstrating that transcription initiates at the 3′ end of the genome in vitro (15). These data also extended the utility of UV-mapping experiments by demonstrating that the U content of an RNA genome can be manipulated to affect the UV sensitivity of a downstream transcript in a predictable manner.

Figure 3.

Effect of alterations to the leader region on the UV sensitivity of the I mRNA synthesized in vitro. (A) Primer extension analysis of the I mRNA. Reverse transcriptions were performed and the products were analyzed. (B) Quantitation of the I-60 mRNAs. The autoradiographs in A were scanned and analyzed. ■, WT I-60; ●, Inc-U I-60; ▴, Dec-U I-60.

Effect of UV Radiation on I mRNA Synthesis in Infected Cells.

The experiments described above showed that transcription initiated at nucleotide 1 with synthesis of Le+ before transcription of the first gene in vitro. To examine this issue in vivo we determined the UV sensitivity of I-60 synthesized by each virus in infected BHK cells. The I-60 mRNA was detected by primer extension analysis (Fig. 4). For rV1+60incU, the primer detected leader RNA in addition to the I-60 mRNA, because the sequence at the end of the leader region of this virus was identical to that found at the end of each VSV gene (3′-AUACUUUUUUUGA-5′). The abundance of I-60 and the Inc-U Le+ was determined (Fig. 4B). For each RNA, the rate of decrease of synthesis was linear as shown by the respective lines of best fit; thus inhibition of synthesis followed single-hit kinetics. In contrast to the in vitro analysis described above, each of the lines of best fit was tightly clustered. Linear regression analysis was used to determine the 37% survival doses (Table 2). For 3′ entry and transcription of Le+ before I-60, the maximum numbers of U dimers that could form in each template were 8, 11, 14, and 16 for Inc-U (Le+), Dec-U (I-60), WT (I-60), and Inc-U (I-60), respectively (Table 1). This finding yields a ratio of predicted survival doses of 1.00:0.73:0.57:0.50. In contrast, the measured values 5,000, 5,250, 5,130, and 4,190 erg⋅mm⁻² (Table 2) yielded relative sensitivities of 1.00:1.05:1.03:0.84. These data are inconsistent with 3′ entry of polymerase and sequential transcription of Le+ before I-60. Rather they indicate that transcription initiated directly at the first gene-start sequence, which predicts a ratio of relative sensitivities of 1.0:1.0:1.0:1.0. A separate slopes linear regression analysis (40) comparing all lines was performed. This analysis indicated that the slopes for Inc-U (Le+), Dec-U (I-60), and WT (I-60) were not significantly different from one another (P > 0.3). The slope for Inc-U (I-60) was slightly less than the other three slopes (P < 0.001). However, this difference was not compatible with obligatorily sequential transcription of leader RNA before synthesis of the I-60 mRNA. For these data to support obligatorily sequential transcription of Le+ before I-60, the slope of the Inc-U (I-60) would need to differ 2-fold from the existing Inc-U (Le+). This difference would result in the hypothetical dashed line for Inc-U (I-60) instead of the experimentally determined solid line (Fig. 4B). Consequently these data show that polymerase initiates transcription directly at the first gene-start site in infected cells. These findings were confirmed by multiple independent experiments that compared the relative sensitivity of the Inc-U (I-60) and Dec-U (I-60) by using a different primer (data not shown), and by comparing the sensitivity of I-60 with a larger mRNA (I-108) as described below.

Figure 4.

Effect of alterations to the leader region on the UV sensitivity of the I mRNA synthesized in infected cells. (A) Primer extension analysis of the I mRNA. Reverse transcriptions were performed and the products were analyzed. Note that because of sequence identity at the end of the leader RNA, the primer anneals to the leader synthesized from rV1+60incU. This product is shown (Le). (B) Quantitation of the I mRNAs. The autoradiographs in A were scanned and analyzed. ■, WT I-60; ●, Inc-U I-60; ▴, Dec-U I-60; ♦, Inc-U Le+. Two values for each time point were plotted to generate the graphs shown. Because of the tight distribution of the data points, many points are superimposed on one another. Statistical analysis of the data is presented in Table 2. Dashed line represents the predicted UV sensitivity for Inc-U (I-60) if transcription were obligatorily sequential.

Table 2.

Fitted regression analysis for UV sensitivities of RNA synthesized in infected cells

RNA	Slope	R2, %	37% survival dose, erg⋅mm−2
Inc-U (Le+)	−0.371  ± 0.028	99.3	5,000
Dec-U (I-60)	−0.352  ± 0.030	98.7	5,250
WT (I-60)	−0.360  ± 0.028	99.0	5,130
Inc-U (I-60)	−0.441  ± 0.022	99.6	4,190

Regression analysis was performed on the data generated from densitometric scanning (Fig. 4A). These values were expressed as percentages (Fig. 4B). The fitted regression lines are of the form ln(y) = intercept + slope(exposure time). The R² value is a measure of how tightly the data values fit about the regression line with maximum possible value of 100%. Values for the slopes are given within 95% confidence limits. 

UV Sensitivity of the First Gene Is Altered by Its Base Composition.

The above experiments indicated that in infected cells, polymerase bypassed the leader region and instead initiated transcription directly at the first gene-start. To confirm this, we generated an additional recombinant VSV (rV1+108) that contained a 108-nt gene (Table 1) inserted between the WT leader region and the N gene and compared the relative UV sensitivity of I-60 and I-108. Purified viruses were exposed to UV radiation for 0–5 min before infection of cells. RNAs were extracted, and the I mRNA was detected by primer extension. Synthesis of I mRNA decreased as the extent of UV irradiation increased (Fig. 5A). The abundance of the I mRNAs was determined and plotted to obtain the graph shown in Fig. 5B. The 37% survival doses were 3,000 erg⋅mm⁻² for I-108 and 5,400 erg⋅mm⁻² for I-60, a 1.8-fold difference in relative sensitivity. The predicted relative sensitivity of the I-60 and I-108 mRNA are 1:1.43 (14:20) for polymerase that initiates transcription at the 3′ end of the genome, or 1:1.75 (8:14) for polymerase that initiates directly at the first gene-start. Consequently, the measured 1.8-fold difference is compatible with polymerase entry at the first gene-start sequence rather than at the extreme 3′ end of the genome.

Figure 5.

Effect of UV radiation on the synthesis of the I mRNA in infected cells. (A) Primer extension analysis of the I mRNA. Reverse transcriptions were performed and the products were analyzed. The I mRNAs are indicated I-60 and I-108. (B) Quantitation of the I-60 and I-108 mRNAs. The autoradiograph in A was scanned and analyzed. ■, I-60; ▴, I-108.

Discussion

We used UV transcriptional mapping to examine whether the VSV RdRP initiated transcription at the 3′ end of the genome, or directly at the first gene-start site. We inserted a 60-nt gene (I) immediately downstream of the leader region, and altered the number of potential UV target sites in the leader region by changing the number of adjacent U residues. Analysis of RNAs synthesized in vitro showed that the UV sensitivity of Le+ was proportional to the maximum number of U dimers that can form on exposure to UV radiation. These in vitro experiments also showed that the UV sensitivity of the I mRNA was affected by these alterations to the leader region, confirming earlier in vitro studies that showed transcription initiates at the 3′ end of the genome with obligatorily sequential synthesis of Le+ before transcription of the first gene.

In contrast, analysis in infected cells provided evidence that was incompatible with 3′ entry and obligatorily sequential synthesis of Le+ before transcription of the first gene. Rather, the in vivo experiments showed that the RdRP initiated transcription directly at the first gene-start sequence. Three lines of evidence supported these conclusions: (i) the UV sensitivity of I-60 mRNA synthesized in vivo, (ii) the demonstration that the I-60 mRNAs and the Inc-U Le+ RNA had similar UV sensitivities, and (iii) the relative sensitivities of the 60- and 108-nt genes. These data cannot eliminate the possibility that some RdRP initiates transcription at the 3′ end of the genome in infected cells; however, these data show that the RdRP predominantly initiates transcription directly at the first gene-start site in infected cells as there was no major contribution of the leader gene or its varied U composition to the UV sensitivity of the I-60 mRNA in cells (Fig. 4). Consequently, these results have considerable implications for the mechanism of regulation of gene expression in VSV and, by extrapolation, other NNS RNA viruses. In addition, these experiments extend the utility of UV-mapping studies by demonstrating that the sequence of a genome can be specifically altered to affect the target sensitivity of a downstream transcript.

Transcription.

The stop-start model of sequential transcription is widely accepted. This model states that the RdRP enters the genome at a single site and initiates transcription at position 1, synthesizing Le+. Termination of Le+ is essential to allow polymerase to transcribe the N mRNA. In turn, termination of the N mRNA is essential for transcription of P, and so on. The experiments described here show that polymerase initiates transcription directly at the first gene-start in infected cells, a finding that is incompatible with the stop-start model applying at the leader–N gene junction. Our results do not suggest that polymerase enters internally at the other gene-starts on the genome as was initially proposed in a multiple entry site model. Such a model is incompatible with the original UV-mapping experiments (6–8) and the data shown in Fig. 1. Rather, we propose a modification to the stop-start model of sequential transcription (Fig. 6). We suggest that during transcription, polymerase is recruited to the template through recognition of specific sequence elements that include the previously defined RdRP-binding site, and probably other sequence elements identified as essential for mRNA synthesis (29, 33). Transcription initiates directly at the N gene-start, possibly by direct positioning of the polymerase active site. During replication, polymerase is recruited to the 3′ end of the genome through recognition of specific sequence elements that may overlap those defined for transcription. However, in this case the active site is positioned so synthesis initiates at nucleotide 1. Alternatively, all polymerase could enter the genome at position 1, and during transcription scan through the leader region to initiate transcription at the first gene-start.

Figure 6.

Model for RNA synthesis: Schematic of the VSV genome depicting the leader region (Le) and the N and P genes. During transcription the RdRP, a complex of L (large oval) and a trimer of P (small oval), binds to specific sequences and initiates synthesis at the N gene start. The products of this reaction are the VSV mRNAs, of which the N and P mRNAs are shown. The 5′-terminal cap is depicted by a black diamond, and the 3′ polyadenylate tail by A(n). During replication, the RdRP initiates at the 3′ end of the genome. Initiation at the 3′ end provides leader RNA (not shown) and the full-length antigenome. Replication requires protein synthesis to supply N protein for encapsidation of nascent RNA. N protein (hatched oval) is kept soluble by interaction with P, in a 2:1 complex. Positions of initiation are indicated by the black triangles.

Separate Initiation Sites for Transcription and Replication.

A central unresolved aspect of the molecular biology of NNS viruses is how transcription and replication are coordinately regulated. Immediately on infection of a cell, viral genomes are transcribed by the input polymerase to yield the mRNAs. Viral protein synthesis is required for replication to supply a source of N protein for encapsidation of the nascent strand. These observations led to a model proposing RdRP activity was switched during Le+ synthesis from transcription to replication by the availability of N protein (34, 35). While a requirement for N protein to encapsidate the nascent RNA strand is indisputable (41), there is little evidence that N protein switches RdRP activity from transcription to replication.

The finding that transcription and replication initiate at separate positions on the genome is incompatible with the possibility that N switches RdRP activity from transcription to replication during synthesis of Le+. Rather, it suggests that transcription and replication are regulated through the use of separate initiation sites. Initiation site choice could be regulated by a modification to the RdRP, the template, or both. Polymerase initiating at the 3′ end would synthesize leader RNA or the antigenome, a process that requires N protein for encapsidation (41, 42). The suggestion of two forms of polymerase is attractive, as specific mutations in polymerase can separately affect transcription and replication (43–45).

The N RNA template could also regulate the site of polymerase initiation. The polR1 mutant of VSV has a single amino acid change in N (R146H), which perturbs the ratio of products that initiate at the 3′ end or at the N gene-start in vitro (17). This change may alter the template structure, the ability of polymerase to bind to template, or both. Cryoelectron microscopy identified four forms of the Sendai virus template in infected cells (46). The form of the template may influence the RdRP entry site, and polR1 may adopt a conformation that favors internal entry.

Previously we showed that the ability of a VSV subgenomic replicon to act as template for transcription or replication was influenced by the extent of complementarity between the genomic termini. Increased complementarity favored replication at the expense of transcription (31, 33). In view of the distinct entry sites for transcription and replication, it seems possible that polymerase entry is influenced by the interaction of the termini, such that when the termini interact the RdRP initiates replication at the extreme 3′ end of the genome, and when the termini do not interact, the RdRP initiates transcription at the first gene-start site.

Transcription in Infected Cells Versus in Vitro.

Previous work demonstrated that transcription initiated at the genomic 3′ end in vitro (15), and we confirmed this for each recombinant virus in vitro (Figs. 2 and 3). However, in vivo, transcription was shown to initiate internally at the first gene-start for each virus (Fig. 4). These findings show that the mutations engineered into the genome did not alter the mechanism of template recognition by polymerase differentially between the viruses. Rather, they show that polymerase initiated at a separate site in vitro compared with in vivo. Precisely why transcription initiates at different positions in vivo and in vitro was not determined, but possible explanations include the host-cell environment, the low-pH transition that occurs during infection of cells, and the abundance of M protein. Several host-cell factors may be required for RNA synthesis (47–50), and their abundance likely differs in infected cells versus in vitro. The pH-dependent transition during internalization of virions may alter RdRP template interaction. Altering the pH of in vitro transcription reactions alters the abundance of products that initiate at the 3′ end of the genome and read through the leader–N gene junction (17). Finally, the role of M in transcription is poorly understood; however, it affects the ratio of RNA oligomers that correspond to Le+ and N mRNA (51), and transcriptional attenuation (52). Determining which (if any) of these contribute to the difference in vitro versus infected cells may reveal how RdRP initiation site choice is regulated.

In conclusion, we provided evidence that in infected cells, the polymerase of VSV initiates transcription directly at the first gene-start sequence, and not through the prior transcription of a leader RNA. These experiments thus define separate initiation points for transcription and replication on the VSV genome, and thereby suggest a previously unrecognized means of regulation of the two processes.

Abbreviations

VSV

vesicular stomatitis virus

RdRP

RNA-dependent RNA polymerase

N

nucleocapsid

L

large subunit

P

phosphoprotein

M

matrix

G

glycoprotein

Le+

leader RNA

NNS

nonsegmented negative-sense

WT

wild type

pfu

plaque-forming units

Inc-U

increased-U leader

Dec-U

decreased-U leader