Increased Readthrough Transcription across the Simian Virus 5 M-F Gene Junction Leads to Growth Defects and a Global Inhibition of Viral mRNA Synthesis

Abstract

Recombinant simian virus 5 (rSV5) mutants containing substitutions in the M-F intergenic region were generated to determine the effect of increased readthrough transcription on the paramyxovirus growth cycle. We have previously shown, using an SV5 dicistronic minigenome, that replacement of the 22-base M-F intergenic region with a foreign sequence results in a template (Rep22) that directs very high levels of M-F readthrough transcription. An rSV5 containing the Rep22 substitution grew slower and to final titers that were 50- to 80-fold lower than those of wild-type (WT) rSV5. Cells infected with the Rep22 virus produced very low levels of monocistronic M and F mRNA, consistent with the M-F readthrough phenotype. Surprisingly, Rep22 virus-infected cells also displayed a global decrease in the accumulation of viral mRNA from genes located upstream and downstream of the M-F junction, and overall viral protein synthesis was reduced. Second-site revertants of the Rep22 virus that had regained WT transcription and growth properties contained a single base substitution that increased the M gene end U tract from four to eight residues, suggesting that the growth defects originated from higher-than-normal M-F readthrough transcription. Thus, the primary growth defect for the Rep22 virus appears to be in viral RNA synthesis and not in morphogenesis. A second rSV5 virus (G14), which contained a different foreign M-F intergenic sequence, grew to similar or slightly higher titers than WT rSV5 in some cell types and produced ∼1.5- to 2-fold more mRNA and viral protein. The data support the hypothesis that inhibition of Rep22 virus growth is due to increased access by the polymerase to the 5′ end of the genome and to the resulting overexpression of L protein. We propose that the elevated naturally occurring M-F readthrough which is characteristic of many paramyxoviruses serves as a mechanism to fine-tune the level of polymerase that is optimal for virus growth.

For the nonsegmented negative-sense RNA viruses, the viral polymerase is thought to gain access to the viral genes through a single 3′ promoter (9). The junction between the tandemly linked viral genes contains important cis-acting signals that direct the viral polymerase to carry out sequential and polar stop-start transcription. At the end of a viral gene, the polymerase can terminate transcription and release a poly(A)⁺ mRNA before reinitiating mRNA synthesis at a downstream transcription start site (reviewed in references 1 and 24). Alternatively, attenuation of transcription can occur such that the polymerase terminates transcription but does not reinitiate at the next gene (19), and this results in a gradient of mRNA transcription. In addition, the viral polymerase can ignore the gene junction termination signals to produce a dicistronic readthrough mRNA. In the work described here, we have analyzed the effect on virus growth of a gene junction mutation that leads to high readthrough transcription across the paramyxovirus M-F gene junction.

The nonsegmented negative-sense RNA viruses fall into two groups based on the structure of their intergenic regions. For some viruses, such as vesicular stomatitis virus (VSV) and Sendai virus (SeV), the sequence at each gene junction in the genome is highly conserved and it would be expected that polymerase activities and transcriptional attenuation would be similar at each junction. In the second group of viruses, such as respiratory syncytial virus (RSV) and simian virus 5 (SV5), there is a high degree of variability at each gene junction (see references 7, 8, and 21 for sequences). Recent evidence suggests that for these viruses, each gene junction may act differentially to control polymerase functions and attenuation (11, 12, 14, 33).

For the paramyxo- and rhabdoviruses, the junction between each of the genes is composed of a gene end, a nontranscribed intergenic region, and a gene start site. In the case of the conserved VSV gene end regions, the invariant 3′-AUACU₇-5′ gene end (35) contains a stretch of seven uridyl (U) residues directing the viral polymerase to poly(A) nascent mRNAs through a stuttering mechanism (37) and additional signals that promote termination of transcription (2). Reverse genetics experiments have shown that polyadenylation-termination by the VSV polymerase requires an intact U7 tract. The removal of even a single U residue results in high levels of readthrough products, in which the polymerase ignores the gene junction signals to produce a dicistronic mRNA (2, 18).

Polyadenylation-termination by the SV5 polymerase differs significantly from that of VSV in the stringency of the length of a functional gene end U tract. The SV5 gene end sequences are diverse and contain U tracts which vary in length from four to seven residues (Fig. 1A). Our previous work has shown that the variable U tracts at the SV5 gene end regions are functionally equivalent in templating the addition of a poly(A) tail to viral RNAs (34). The variability in SV5 gene end U tracts can be an important structural feature of the nucleocapsid template, since we have shown that the SV5 U tract serves a previously unrecognized role as a spacer to maintain a distance of at least six bases between a gene end and gene start site (34).

FIG. 1.

SV5 gene junction sequences and structure of rSV5 M-F intergenic mutants. (A) Sequence of SV5 gene junctions. The SV5 gene end, intergenic, and gene start sequences are listed as genomic RNA (3′ to 5′). (B) Structure of rSV5 M-F intergenic mutants. The SV5 genome is depicted as a rectangle, with vertical bars denoting the intergenic regions. The 3′ leader (le) and 5′ trailer (tr) are shown as black boxes. The M gene end, intergenic, and F gene start sequences corresponding to the WT rSV5 or the mutant Rep22 and G14 templates are listed 3′ to 5′. The M gene end U tract is underlined. The transcription phenotypes of dicistronic M-F minigenomes harboring these intergenic regions are listed, as determined by Rassa and Parks (33).

The intergenic region separating a gene end from the gene start site is normally not transcribed by the viral polymerase, and the role of these sequences in gene expression is not completely understood. The intergenic regions are highly conserved for many nonsegmented negative-strand viruses. For example, each of the SeV intergenic regions is composed of the trinucleotide 3′-GAA-5′ (genome sense), the only exception being the 3′-GGG trinucleotide at the HN-L junction (sequences compiled in reference 21). Likewise, the VSV intergenic regions are usually composed of the dinucleotide 3′-GA-5′ (35). In a VSV minigenome system, alterations of the conserved intergenic GA dinucleotide can result in templates that direct high levels of readthrough transcripts and in some cases the alteration can affect reinitiation of transcription at a downstream gene start (3, 40, 41).

SV5 differs significantly from both VSV and SeV by having individual intergenic regions that are highly variable, differing both in sequence and overall length (Fig. 1A). This variability at intergenic regions is also a property of other paramyxoviruses (7, 8, 21, 22), including RSV, human parainfluenza virus type 2 (HPIV-2), mumps virus, and simian virus 41 (SV41). In a minigenome system, the structurally diverse RSV intergenic regions can have a major effect on modulating viral transcription (12). SV5 is similar to RSV in this regard, because minigenomes with sequence substitutions in the 22-base M-F intergenic region can be defective in transcription termination or in reinitiation (33).

A striking feature of viruses with variable gene junctions is the elevated readthrough transcription which occurs at the M-F junction (see reference 32 for examples). The synthesis of high levels of M-F dicistronic mRNA would be expected to reduce expression of the F protein, and we have hypothesized that this may be a mechanism to control F protein translation. An alternative hypothesis is that M-F readthrough transcription is a mechanism to control attenuation at this junction. A direct consequence of readthrough transcription is that more polymerase molecules should have access to genes encoded downstream of the M-F. Thus, M-F readthrough transcription may control expression of the L polymerase gene. To test these hypotheses we have determined the effect of increased M-F readthrough transcription on SV5 growth. A recombinant SV5 (rSV5) harboring a mutation in the M-F intergenic sequence which leads to elevated M-F readthrough transcription was found to have a global defect in viral transcription. Based on the finding that viral transcription and not morphogenesis is the major defect in the growth of this mutant, we propose that M-F readthrough serves as a mechanism to fine-tune the level of polymerase that is optimal for virus growth.

MATERIALS AND METHODS

Cells, viruses, and plaque assays.

Monolayer cultures of cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS). Vaccinia virus MVA expressing T7 RNA polymerase was the kind gift of B. Moss and was grown and titers were determined on BHK cells (5, 45). SV5 plaque assays were performed on CV-1 cells by using an overlay containing 1% agar with DMEM, 2% FCS, and 10 mM HEPES (pH 7.2). Cells were fixed at 4 days postinfection (p.i.) with 3.7% formaldehyde, and plaques were visualized after being stained with 0.1% crystal violet.

Construction of SV5 plasmids containing altered M-F junctions and recovery of rSV5 from cDNA.

SV5 minigenomes containing the Rep22- and G14-altered M-F intergenic regions have been described previously (33), and the mutant sequences are shown in Fig. 1B. The full-length infectious cDNA clone pBH276 (13) was kindly provided by B. He and R. A. Lamb (Northwestern University). To introduce these sequences into the SV5 cDNA genome, the 1,370-bp Asp718-SphI fragment encoding the Rep22 or G14 M-F gene junction was excised from pMF2-Rep22 or pMF2-G14 (33) and inserted into the corresponding sites in the M-F region of plasmid pBH276 to generate pRSV5-Rep22 and pRSV5-G14, respectively.

rSV5 was recovered by using pRSV5-Rep22 or -G14 as described previously (13), with minor modifications (36). Briefly, 3.5-cm-diameter dishes of A549 cells were infected at a multiplicity of infection (MOI) of ∼3 with vaccinia virus MVA (45) and then transfected with plasmids encoding the NP (1.2 μg), P (0.2 μg), and L (1.5 μg) genes along with the plasmid encoding the full-length viral antigenome (∼5 μg) as described previously (26). After overnight incubation, the medium in each well was replaced with 2 ml of DMEM containing 10% FCS and ∼10⁵ uninfected CV-1 cells. Medium was harvested 2 days later, clarified by low-speed centrifugation, and used undiluted to infect a fresh monolayer of MDBK cells. After 3 days, the medium from this infection was harvested and used to isolate plaques on CV-1 monolayers. Virus stocks were generated from single plaques by growth in MDBK cells. The presence of the nonviral intergenic region was confirmed by cloning and sequencing reverse transcription-PCR (RT-PCR) products derived from viral RNA. Wild-type (WT) rSV5 was similarly isolated from pBH276 (13).

One-step growth analysis and determination of virus yields.

To determine the kinetics of virus release in a one-step growth cycle, ∼10⁶ cells in 3.5-cm-diameter dishes were infected at an MOI of ∼5 for 1 h. The cells were washed twice and covered with 1 ml of DMEM containing 2% FCS. Medium was removed at 0, 4, 8, 12, and 24 h p.i., clarified by centrifugation (2 min, 14,000 × g), adjusted to 0.75% bovine serum albumin (BSA), and stored frozen for determination of titer by plaque assay.

Analysis of viral transcription products.

The accumulation of SV5 RNA in infected cells was analyzed as described previously (33). Briefly, cells were mock infected or infected at an MOI of ∼5, which results in >75% of the cells being infected, as assayed by indirect immunofluorescence. Total intracellular RNA was isolated with Trizol reagent (Life Technologies), and poly(A)⁺ RNA was isolated with oligo(dT) cellulose (New England Biolabs). RNA samples were analyzed by Northern blotting with genome-sense ³²P-labeled riboprobes corresponding to the following SV5 gene sequences: NP (bases 1 to 180 [28]), P (504 to 642 [42]), M (1079 to 1265 [38]), F (160 to 407 [30]), SH (1 to 283 [16]), and HN (302 to 559 [15]). Quantitation of RNA transcription products was performed by the PhosphorImager instrument and software (Molecular Dynamics). The amount of SV5 RNA in each sample was normalized to the amount of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA analyzed in parallel by using an antisense riboprobe corresponding to bases 1 to 335 of the rat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene.

Isotopic labeling of polypeptides and immunoprecipitation analysis.

Rabbit polyclonal antibodies specific for the SV5 NP, P, M, F, and L proteins were prepared (Covance Inc.) using the following antigens. NP antiserum was raised to a synthetic peptide corresponding to the C-terminal 11 amino acids (28). P, M, and F antisera were raised to fusion proteins consisting of the bacterial maltose-binding protein (MBP) linked to SV5 protein fragments from the P-specific portion of the P/V gene (42), the entire M protein (38), or F protein residues 230 to 438 (30). The MBP-P, MBP-M, and MBP-F proteins were expressed in Escherichia coli after induction with isopropyl-β-d-thiogalactopyranoside (IPTG), and purified by chromatography on amylose-Sepharose columns as described by the manufacturer (New England Biolabs) before being used as immunogens. Anti-L sera was raised to a His-tagged polypeptide (L-His) corresponding to SV5 L protein residues 1 to 350 (28). L-His was expressed in E. coli after induction with IPTG and purified by chromatography on Ni-NTA columns as described by the manufacturer (Qiagen) before being used as an immunogen. Serum specific for sodium dodecyl sulfate (SDS)-denatured HN has been described previously (27).

Cells infected with the individual rSV5s were radiolabeled for 15 min at 12 h p.i. using 100 μCi/ml Tran[³⁵S]label, washed in phosphate-buffered saline, and lysed in 1% SDS. Samples were immunoprecipitated with an excess of polyclonal rabbit antiserum as described previously (10, 27) before being analyzed on 10% polyacrylamide gels (23).

To prepare radiolabeled virions, 6-cm-diameter dishes of mock-infected or rSV5-infected MDBK cells were radiolabeled from 12 to 24 h p.i. with 100 μCi of Tran[³⁵S]label/ml in medium consisting of 20% DMEM plus 80% Cys⁻Met⁻ DMEM. After the medium was clarified by low-speed centrifugation, extracellular virions were centrifuged (5 h at 37,000 rpm and 18°C, in an SW41 rotor) through a 20% glycerol cushion. Pellets were resuspended, and equal aliquots from each sample were analyzed directly on 10% polyacrylamide gels.

RESULTS

Growth properties of rSV5 containing foreign sequences at the M-F junction.

Our previous work has established that substitution of the 22-base M-F intergenic region with the Replace 22 (Rep22) foreign sequence shown in Fig. 1B resulted in a minigenome template that directed very high levels of an M-F readthrough transcription product (∼80% of total F mRNA [33]). To determine the effect of increasing M-F readthrough transcription on SV5 growth, the 22-base M-F intergenic region in the full-length SV5 cDNA (13) was replaced with the Rep22 mutation, and recombinant virus was isolated from cDNA. A second full-length SV5 cDNA was made to contain an altered M-F intergenic region such that a G residue which normally follows the M gene end U tract was included as the first intergenic residue. In addition, 14 new nonviral sequences were added downstream of the G residue (G14; Fig. 1B). A minigenome template containing this G14 M-F substitution directed transcription to levels matching those of the WT M-F minigenome (33). Based on the results from the minigenome system, it was anticipated that the Rep22 and G14 substitutions would yield viruses with increased M-F readthrough and WT transcription properties, respectively. rSV5 was recovered from both of these plasmids as described previously (13), and the resulting viruses were designated Rep22 and G14. A sequence analysis of RT-PCR products derived from Rep22 and G14 viral RNA confirmed the presence of the engineered M-F intergenic mutations.

The Rep22 and G14 viruses displayed plaque morphologies that were distinct from those of WT rSV5. Compared to WT virus, G14 virus plaques were consistently clearer, with well-defined edges (Fig. 2A). Contrary to our expectations for this virus, it will be shown below that the G14 virus produces elevated levels of viral RNA and protein, which is consistent with this virus being more efficient than WT virus in promoting the cell-cell fusion that is measured by plaque assays in CV-1 cells. Plaques from the Rep22 virus were cloudy and not well resolved above the monolayer background. In addition, occasional large, clear plaques were detected. Evidence will be presented below that these are revertants that have restored WT growth phenotypes.

FIG. 2.

Plaque morphology and growth rates for rSV5s with nonviral sequences at the M-F junction. (A) Plaque morphology of M-F mutants. CV-1 monolayers were mock infected or infected with WT rSV5, G14, or Rep22 virus, and plaques were detected by crystal violet staining. The G14 mutant virus produces clearer plaques than WT SV5, while the Rep22 mutant produces cloudy plaques. CV-1 (panel B) or MDBK (panel C) cells were infected at an MOI of ∼5, and medium was harvested at the indicated times p.i. Extracellular virus titers were determined by plaque assay on CV-1 monolayers. Values represent the mean of two independent experiments.

To determine if the M-F intergenic substitutions affected the rate of virus growth, cells were infected at a high MOI and a single-step growth analysis was carried out. As shown in Fig. 2B and C, WT and G14 viruses showed similar growth rates as well as final virus titers when measured in CV-1 or MDBK cells. In contrast, Rep22 virus had a much slower growth rate and the accumulation of extracellular virus at 12 h p.i. was ∼50- to 80-fold lower than that seen for WT rSV5. The final yield of the rSV5 M-F intergenic mutants was not significantly affected by incubation at lower or higher temperatures, indicating that they were not temperature sensitive for growth (G. D. Parks, J. C. Rassa, and K. R. Ward, unpublished results). In the experiments presented below, MDBK cells were chosen for further analysis of the M-F mutant virus growth properties. Together, these data indicate that different sequence substitutions in the SV5 M-F intergenic region can have a differential effect on plaque morphology, virus growth rates, and final titers of virus.

Changes in the M-F intergenic region can lead to viruses with increased or decreased viral mRNA transcription.

To determine if the M-F mutant viruses were altered in transcription, the accumulation of viral mRNA was measured in MDBK cells that were infected with WT, Rep22, or G14 viruses. At 12 h p.i., poly(A)⁺ RNA was isolated and analyzed by Northern blotting with riboprobes specific for the P, M, F, and HN genes. As shown in Fig. 3A, the level of monocistronic M and F mRNA in cells infected with the Rep22 mutant virus was much lower than that found for WT rSV5. This outcome was anticipated based on our previous results showing that ∼80% of the F mRNA from a Rep22 M-F minigenome was detected as a dicistronic M-F readthrough mRNA transcript (33). However, for unknown reasons, transcription in the Rep22 virus-infected cells differed from that of the minigenome system in that the dramatic decrease in monocistronic M and F mRNA (Fig. 3A) was not accompanied by a corresponding increase in dicistronic M-F readthrough products. Since our Northern blot assay measures the accumulation of viral mRNA, this difference in the two systems may reflect differential turnover of the dicistronic M-F transcript.

FIG. 3.

Differential mRNA synthesis for rSV5 mutants with nonviral M-F intergenic sequences. MDBK cells were mock infected (M) or were infected with WT rSV5, Rep22 virus (Rep, panel A), or G14 virus (G14, panel C) at an MOI of ∼5. Poly(A)⁺ RNA was isolated at 12 h p.i. and analyzed by Northern blotting with ³²P-riboprobes specific for cellular GAPDH or the indicated SV5-specific mRNAs. The positions of the monocistronic and dicistronic readthrough mRNAs are indicated. (B) Quantitation of transcription in Rep22 virus-infected MDBK cells. For the indicated genes, the amount of monocistronic mRNA accumulated at 12 h p.i. was normalized to the amount of GAPDH mRNA and is expressed as a percentage of WT rSV5 analyzed in parallel. Lines above the bars indicate the standard deviation from the mean of three independent experiments.

Surprisingly, cells infected with Rep22 virus also accumulated less mRNA from genes located 3′ (NP and P genes) and 5′ (SH and HN genes) of the M-F junction. After normalizing to the level of cellular GAPDH mRNA in each sample, the accumulation of M and F mRNA in Rep22 virus-infected cells was ∼12 and ∼3% of WT levels, respectively. Relative to WT virus-infected cells, the levels of transcripts from genes upstream of the Rep22 M-F junction were ∼26% for NP and ∼40% for P, while those of genes downstream of the Rep22 M-F junction were ∼28% for SH and ∼30% for HN (Fig. 3B). Thus, the Rep22 virus is altered in two aspects of viral transcription: a severe decrease in the accumulation of M and F monocistronic mRNA and, to a lesser extent, a global decrease in transcripts from the other viral genes.

Analysis of viral mRNA from cells infected with the G14 virus showed an overall increase in the level of P, M, F, and HN transcripts (Fig. 3C). After normalizing to the level of cellular GAPDH mRNA in each sample, MDBK and CV-1 cells infected with G14 virus were found to accumulate ∼five- to twofold more viral mRNA, depending on the individual gene analyzed. This result was not anticipated, since in the minigenome system, the G14 mutation has no detectable effect on transcription. The slight increase in G14 mRNA was not due to differences in MOI, since immunofluorescence assays showed equal numbers of infected cells (not shown). However, as expected, the fraction of viral transcripts present as a dicistronic readthrough mRNA (e.g., NP-P, M-F, and SH-HN) was not significantly different from that seen for WT rSV5.

The above results were obtained with mRNA isolated from cells at 12 h p.i. To determine if the altered transcription phenotypes were also evident at an early time after infection, poly(A)⁺ RNA was isolated at 4 h p.i. from MDBK cells that were infected with the M-F mutant viruses. Although the overall level of viral mRNA was much lower at 5 h p.i. than at 12 h p.i., the same altered transcription phenotypes were evident: cells infected with Rep22 and G-14 viruses synthesized much less and slightly more mRNA than WT rSV5, respectively (data not shown). These results suggest that the altered transcription phenotypes of the M-F mutants are characteristics of their nucleocapsid templates and are not indirect consequences of the accumulation of viral proteins late in the infectious cycle. Taken together, the data support our hypothesis that M-F readthrough acts to regulate transcription and not the level of F mRNA and protein, since the defect in Rep22 virus growth is accompanied by a global decrease in viral transcription.

The M-F intergenic mutants produce altered levels of viral proteins and virions.

The rate of protein synthesis was analyzed in cells infected with the M-F intergenic mutants to determine if the changes in virus transcription were accompanied by changes in the rate of translation. Mock-infected or rSV5-infected MDBK cells were pulse-labeled with ³⁵S-amino acids at 12 h p.i. Viral proteins were immunoprecipitated from cell lysates with polyclonal sera specific for the individual SV5 proteins before analysis by SDS-polyacrylamide gel electrophoresis (PAGE). As shown in Fig. 4A, the rate of viral protein synthesis in cells infected with Rep22 virus was much less than that in WT virus-infected cells and this was found for proteins encoded by genes located both upstream (e.g., P) and downstream (e.g., HN) of the M-F junction. Similar results were obtained with Western blotting to measure accumulation of viral proteins (data not shown). The defect in translation of Rep22 proteins appeared to be more severe for genes located farther from the 3′ end of the genome. For example, Rep22 virus-infected cells synthesized ∼five- to eightfold less NP and P than WT virus-infected cells, but the relative rate of synthesis of Rep22 virus M, F, HN, and L proteins was much lower. Cells infected with G14 virus synthesized approximately the same level of protein as that found for WT rSV5 (Fig. 4A), although in some cell types there was a 1.5- to 2-fold increase in translation rate (see below). Thus, the transcription phenotypes seen for the M-F intergenic mutants are similarly reflected in the rate of viral protein synthesis.

FIG. 4.

Viral protein synthesis in cells infected with M-F intergenic mutants. (A) Protein synthesis. MDBK cells were mock infected or were infected with WT rSV5, Rep22 virus (Rep), or G14 virus (G14). Cells were radiolabeled with Tran[³⁵S]label for 15 min at 12 h p.i. Samples from cell lysates were immunoprecipitated with the indicated polyclonal antiserum before analysis by SDS-PAGE. In the HN panel, the faster-migrating species is the NP protein that immunoprecipitates nonspecifically. (B) Proteins contained in extracellular virions. MDBK cells infected with the indicated viruses were radiolabeled from 12 to 24 h p.i. with Tran[³⁵S]label as described in Materials and Methods. Extracellular virions were isolated by centrifugation through a 20% glycerol cushion, and the proteins contained in the resulting pellets were analyzed directly by SDS-PAGE.

Altered protein synthesis in virus-infected cells could affect the ratio of proteins in progeny virions. To examine virion-associated proteins, MDBK cells were mock infected or infected with Rep22, G14, or WT rSV5. After being radiolabeled with ³⁵S-amino acids from 12 to 24 h p.i., extracellular virus was isolated by centrifugation as described in Materials and Methods and the viral proteins were examined by SDS-PAGE. A typical result is shown in Fig. 4B. Cells infected with G14 virus released as much and in some cases slightly more than WT levels of viral protein (Fig. 4B, lane G14). When the amount of virion-associated proteins was normalized to the level of NP, the ratio of proteins in the G14 virions did not differ significantly from that in WT SV5. In contrast, the overall amount of virion protein released from cells infected with Rep22 virus was much less than that of WT (Fig. 4B, lane Rep), with the level of NP and M decreased by ∼8- to 10-fold. In the case of Rep22 virions, the relative level of L and to a lesser extent P was slightly higher than that of WT SV5 virions, while the relative level of M was unchanged. These results indicate that the level of virions released from cells infected with the M-F mutants is a reflection of overall levels of mRNA transcripts and intracellular protein. In the case of the Rep22 virions, the level of polymerase-associated proteins is altered slightly.

In a variety of cell types, the yield of virus from cells infected with Rep22 was always lower than that of WT rSV5 (Fig. 5A). The Rep22 virus growth defect is not due to an inability to counteract the host interferon system, since Rep22 grew to lower titers than WT virus in Vero and BHK cells, two cell types which do not respond to alpha/beta interferon. The growth defect for Rep22 virus was always accompanied by a lower rate of viral protein synthesis than that seen for cells infected with WT rSV5, as shown in Fig. 5B for SV5 M protein. In contrast to the results for Rep22 virus, the final virus yields for G14 were very similar to those for the WT, with the exception of TK-143 cells for which G14 virus yield was ∼20-fold higher. In some cell types, the rate of viral protein synthesis was slightly higher in cells infected with G14 virus than with WT rSV5 (e.g., CV-1 and TK-143), but this enhanced protein synthesis was not strictly correlated with increased virus yield. This result suggests that some other factor determines the efficiency of virus production.

FIG. 5.

Protein synthesis and growth of M-F intergenic mutants in various cell types. (A) Yield of rSV5 M-F intergenic mutants in various cell types. Cells were infected with WT rSV5, Rep22, or G14 at an MOI of ∼5, and virus titer in the medium at 24 h p.i. was determined by plaque assay on CV-1 monolayers. Values represent the mean of two independent experiments, with vertical bars indicating the standard deviation from the mean. (B) Expression of M protein in four cell types. The indicated cell types were infected with WT rSV5, Rep22, or G14 virus. At 12 h p.i., cells were radiolabeled and M protein was immunoprecipitated as described in the legend to Fig. 4. For each virus-cell combination, duplicate immunoprecipitations were carried out with a sample using half as much cell lysate (right lane of each pair).

Second-site revertants of the Rep22 phenotype have restored WT mRNA transcription and virus growth properties.

Plaque assays of the growth-defective Rep22 virus showed occasional large clear plaques among the cloudy plaques that are typical of this M-F intergenic mutant virus (∼1 in 50 plaques; Fig. 2A). The growth properties of two plaque isolates were analyzed in single-step growth assays to determine if these clear plaques represented revertants of the Rep22 growth defect. As shown in Fig. 6A, revertants 1 and 2 showed growth rates and final virus yields in MDBK cells which were much greater than those of the rSV5 Rep22 parental virus and which closely matched that of rSV5 WT. Northern blot analysis of poly(A)⁺ RNA from MDBK cells infected with revertants 1 and 2 showed that the accumulation of all viral mRNAs was much higher than that of Rep22 virus-infected cells. As shown in Fig. 6B for the M and F mRNAs, cells infected with the revertant viruses accumulated ∼1.5 times higher levels of M mRNA than rSV5 WT, but the relative levels of F mRNA were very similar. Thus, the Rep22 revertant viruses have mRNA transcription profiles that are similar to but not identical to that of rSV5 WT.

FIG. 6.

Second-site revertants of the rSV5 Rep22 mutant phenotype have restored WT virus growth and gene expression properties. MDBK cells that were mock infected (M) or infected with WT rSV5, Rep22, Rep22-revertant virus 1 (R1) or -revertant virus 2 (R2) were analyzed in a single-step growth assay (A) and for mRNA synthesis (B), as described in the legends to Fig. 2 and 3, respectively. Sequence analysis of RT-PCR products from R1- and R2-infected cells revealed a single C to U point mutation (arrow, panel C) which results in an extension of the M gene end U tract from four to eight residues.

To determine if the WT growth properties of revertants 1 and 2 correlated with changes in the M-F intergenic region, cDNAs were generated by RT-PCR, using total RNA from virus-infected MDBK cells and primers which flank the M-F junction. Sequence analysis of eight cloned PCR products detected only a single substitution within ∼100 bases flanking the revertant 1 and 2 M-F junction. Surprisingly, this substitution was not in the nonviral sequences that compose the altered M-F intergenic region. As shown in Fig. 6C, Rep22-revertant viruses contained a single substitution in the C residue directly flanking the M gene end U tract. Thus, the WT phenotype of both revertant viruses correlates with a second-site substitution which extends the M gene end U tract from four to eight residues.

DISCUSSION

For the nonsegmented negative-sense RNA viruses, the distance of a viral gene from the 3′ genomic promoter is the major factor dictating the level of transcription (reference 44 and reviewed in reference 1). However, recent evidence suggests that cis-acting signals at the gene junctions may also contribute to controlling the relative level of mRNA synthesis from a particular viral gene (11, 12, 14, 33). Our mutational analysis has focused on the M-F intergenic region, since it has been proposed that this junction may represent a critical site for transcription control (20, 32, 33). In this study, we have tested the hypothesis that M-F readthrough transcription serves as a mechanism to control viral transcription and growth.

Readthrough transcription at most paramyxovirus gene junctions is relatively low, but the M-F junction is an exception to this rule for a number of viruses, including HPIV-1, -2, and -3, measles virus, SV41, and SV5 (4, 6, 29, 39, 43). In cells infected with WT SV5, ∼40% of the F mRNA is detected by Northern blotting in an M-F dicistronic form (32). M-F readthrough transcription has been reported to be even higher in the case HPIV-2 (∼60%; Parks et al., unpublished), HPIV-1 (∼80% [4]), and SV41 (100% [43]). It is likely that each of these viruses has evolved a characteristic level of M-F readthrough transcription for optimal growth, but the effect on virus growth of altering this polymerase function has not been previously examined.

Our results support the proposal that the Rep22 growth defect is due to increasing M-F readthrough transcription to levels which are much higher than that which occurs naturally (i.e., from ∼40 to ∼80%). In the context of an infected cell, the Rep22 virus synthesizes much less monocistronic M and F mRNA, consistent with the defect originally described for the Rep22 minigenome (33). However, an unexpected result is that Rep22 virus-infected cells also display a global defect in transcription and translation of genes both upstream and downstream of the M-F junction. The global defect in Rep22 transcription is evident even at early times after infection, suggesting this is an inherent property of the nucleocapsid template or of the input virions. Second-site revertants of Rep22 contain a single base substitution in the M gene end U tract, and this change correlated with the restoration of WT levels of M and F mRNAs and with WT growth kinetics. Taken together, these data support the hypothesis that the Rep22 substitution has a negative effect on virus growth as a consequence of directing very high levels of M-F readthrough transcription. We have previously identified SV5 M gene end mutations which decrease M-F readthrough transcription from ∼40 to <10% (32). Work is in progress to determine if rSV5 that synthesizes less M-F readthrough mRNA has a converse phenotype, one in which growth is enhanced.

We have proposed two functions for naturally occurring M-F readthrough transcription. First, the level of F protein synthesis should be affected by the level of M-F readthrough transcription, since the F protein coding sequences would be locked into a dicistronic M-F mRNA. Thus, readthrough transcription could be a mechanism to control potentially toxic effects of F protein or the level of F needed for virion assembly. Second, increasing M-F readthrough transcription should lead to increased synthesis of viral mRNA from genes downstream of F (i.e., SH, HN, and L protein) due to lower attenuation of transcription. Thus, a second hypothesis is that M-F readthrough transcription controls the synthesis of polymerase protein at levels needed for optimal growth. Since the Rep22 growth defect appears to be in viral RNA synthesis and not morphogenesis, the data support the hypothesis that the inhibition of growth is due to increased access by the polymerase to the 5′ end of the genome and to the resulting overexpression of L protein relative to NP or P protein.

The ratio of polymerase-associated proteins can be an important factor in optimal RNA synthesis for the nonsegmented negative-strand viruses. Support for this comes from early complementation studies (25) as well as from reverse genetics systems (e.g., 17, 26, 31, 46). More recently, recombinant viruses have been recovered with altered levels of L transcription. For rabies virus and SeV, increased synthesis of L mRNA correlated with enhanced virus growth (11, 20). However, increased synthesis of L protein does not always lead to an increase in virus growth properties. For example, He and Lamb (14) have generated a series of rSV5 viruses in which the HN-L junction was individually replaced with each of the other SV5 junctions. Recombinant viruses that accumulated a higher L to HN mRNA ratio had slower growth kinetics and formed smaller plaques. We propose that the Rep22 virus is an extreme example of this phenotype, such that very early after infection the large increase in M-F readthrough transcription results in disproportionate levels of newly synthesized L protein relative to P or NP, and this leads to a global reduction in accumulation of viral mRNA and to defective growth. An alternative hypothesis is that the Rep22 virus has disproportionate levels of virion-associated polymerase and this leads to a defect in primary transcription.

Surprisingly, the substitution of foreign sequences at the M-F junction did not always have a negative effect on SV5 growth, as evidenced by the G14 mutant virus which in some cells gave slightly higher than WT levels of mRNA. This contrasts with the results from the SV5 M-F minigenome system, where transcription from a template containing the G14 mutation was indistinguishable from the WT minigenome (33). A factor that could account for these differences with the G14 M-F junction is that the level of polymerase-associated proteins changes significantly during the course of an infection, while the level of plasmid-derived proteins is relatively constant in the minigenome system. Thus, small differences in transcription that were not evident in the minigenome system would be amplified in the course of a virus infection. The global increase in transcription with rSV5 G14 is similar to that reported recently for a recombinant SeV which was altered such that reinitiation at the F gene start site is more efficient (20). The resulting SeV displayed a small global increase in viral transcription that was attributed to decreased attenuation at the M-F junction. Based on this result, we propose that the normal SV5 M-F intergenic region contains a weak inhibitory sequence which normally acts to modulate attenuation across this junction and the G14 substitution removes this inhibitory sequence to allow an increase in L protein expression. In our model, the key difference between the Rep22 growth defect and the enhanced G14 growth properties is the extent to which there is an increase in L transcription and/or translation. Readthrough transcription would allow very high levels of expression at a critical time early after infection, while the subtle G14 mutation would allow a slow accumulation that is evident only late in infection.

The Rep22 revertants contain a point mutation outside of the foreign intergenic sequence that results in an extension of the M gene end U tract from four to eight residues. The link between the reversion to WT transcription and growth properties and this second-site substitution suggests that extending the M gene end U tract overcomes an inhibitory effect of the downstream Rep22 intergenic sequence. In the minigenome system, efficient M gene transcription termination can be restored to the Rep22 template in one of two ways: extending the M gene end U tract to more than four residues or by adding back the first G residue in the intergenic region such that a U4-G gene end-intergenic combination is created (33). Thus, the results from the revertant viruses are consistent with the results from our mutational analysis with the M-F minigenome. We propose that the diverse SV5 gene end and intergenic regions are not separate cis-acting regions but rather they can act together to create signals for modulating polymerase activities (33).

The C to U mutation in the Rep22 revertants creates a new M gene end region (3′-UAAGU₈; Fig. 6C) that has little similarity to the SV5 gene end consensus sequence 3′-A/UUUCU_4 to 7 (32). Based on our previous results (33), it is possible that the combination of a new gene end and a foreign intergenic region (Rep22) creates a signal on the nucleocapsid template or in the nascent mRNA chain which is efficiently recognized for termination by the SV5 polymerase. Alternatively, one of the polymerase components (L, P, or NP) may have accumulated a mutation(s) which expands the already diverse sequences recognized as termination signals by the SV5 polymerase. These two hypotheses can be distinguished with our minigenome system by determining if a dicistronic minigenome template containing the Rep22 revertant M-F junction can direct efficient M gene end termination using plasmid-derived WT NP, P, and L proteins.