Role of the M2-1 Transcription Antitermination Protein of Respiratory Syncytial Virus in Sequential Transcription

Abstract

M2-1 protein of human respiratory syncytial virus (RSV) is a transcription antitermination factor that is important for the efficient synthesis of full-length mRNAs as well as for the synthesis of polycistronic readthrough mRNAs, which are characteristic of nonsegmented negative-strand RNA viruses. The contributions of these effects to RSV sequential transcription were investigated with minigenomes which contained one to five genes which were either foreign marker genes or authentic RSV genes. When evaluated on a promoter-proximal gene, the effect of M2-1 on the synthesis of full-length mRNA was much greater for a long (1,212- or 1,780-nucleotide) gene (up to a 615-fold increase) than for a short (274-nucleotide) gene (less than a 2-fold increase). This was independent of whether the gene contained non-RSV or RSV-specific sequence. Once the polymerase had terminated prematurely, it was unable to reinitiate at a downstream gene. These studies also confirmed that M2-1 enhances the synthesis of polycistronic mRNAs and that the magnitude of this effect varied greatly among different naturally occurring gene junctions. The synthesis of polycistronic mRNAs, which presumably involves antitermination at the gene-end signal, required a higher level of M2-1 than did the synthesis of the corresponding monocistronic mRNAs. M2-1 did not have a comparable antitermination effect at the junction between the leader region and the first gene. In a minigenome containing the NS1 and NS2 genes in their authentic sequence context, synthesis of full-length NS1 and NS2 mRNAs in the absence of M2-1 was remarkably high (36 and 57%, respectively, of the maximum levels observed in the presence of M2-1). In contrast, synthesis of mRNA from additional downstream genes was highly dependent on M2-1. Thus, RSV has the potential for two transcription programs: one in the absence of M2-1, in which only the NS1 and NS2 genes are transcribed, and one in the presence of M2-1, in which sequential transcription of the complete genome occurs. The dependence on M2-1 for transcription was greater for a gene in the fifth position from the promoter than for one in the third position. This indicates that under conditions where M2-1 is limiting, its concentration affects the gradient of transcription. Although M2-1 was found to have profound effects on transcription, it had no effect on replication of any minigenome tested, suggesting that it is not an active participant in RNA replication or regulation of RNA replication. Finally, since a permissive RSV infection is marked by a gradual increase in the intracellular accumulation of viral proteins including M2-1, we examined the relative abundances of various mRNAs during RSV infection for evidence of temporal regulation of transcription. None was found, implying that the availability of M2-1 during a permissive infection is sufficient at all times such that its concentration does not mediate temporal regulation of gene transcription.

Human respiratory syncytial virus (RSV) is the most important viral agent of serious pediatric respiratory tract disease worldwide (11). It is a member of the family Paramyxoviridae of order Mononegavirales, the nonsegmented negative-strand RNA viruses (28). The genome of RSV (strain A2) is 15,222 nucleotides (nt) in length and encodes 11 proteins. Three are associated with the nucleocapsid: the major RNA-binding nucleocapsid N protein, the P phosphoprotein, and the major polymerase subunit L. Three are transmembrane surface proteins: the fusion F glycoprotein, attachment G glycoprotein, and small hydrophobic SH protein. One is the internal virion matrix M protein. Two are nonstructural proteins: NS1 and NS2. Two are encoded by separate translational open reading frames (ORFs) of the M2 gene: the M2-1 and M2-2 proteins. The gene order of the genome is: 3′-NS1-NS2-N-P-M-SH-G-F-(M2-1/M2-2)-L-5′ (6, 9, 13).

Most aspects of RSV transcription and replication conform to the models based on the prototype members of the Mononegavirales: Sendai virus, a paramyxovirus, and vesicular stomatitis virus (VSV), a rhabdovirus (reviewed in references 26 and 33). The genome is tightly bound by N protein to form the nucleocapsid, which is the template for the viral polymerase. At the ends of the genome are a short noncoding leader and trailer which precede and follow, respectively, the above-mentioned genes, and which for RSV contain all the cis-acting signals required for RNA replication (reference 24 and unpublished observations). Genome transcription is initiated at a single promoter site located at the 3′ (leader) end and involves a sequential stop-start mechanism in which the polymerase is guided by short, conserved cis-acting signals present at the ends of each gene to produce a series of subgenomic mRNAs (1, 3, 13, 14). In RSV, each gene begins with a 10-nt gene-start (GS) signal, at which mRNA synthesis begins, and ends with a semiconserved 12- to 13-nt gene-end (GE) signal, which directs polyadenylation and release of the mRNA (24). The polymerase then apparently remains template bound and crosses the intergenic region without transcribing to resume synthesis at the next GS signal. There is a gradient of decreasing mRNA abundance (4, 20, 32) due to transcription attenuation, which for VSV was shown to occur primarily at the gene junctions (22). During RNA replication, the polymerase disregards the cis-acting transcription signals and synthesizes a complete positive-sense intermediate RNA, the antigenome. How the polymerase shifts between transcription and replication is currently not known.

The RSV N, P and L proteins, together with the RNA genome, are the virus-specific components required for RNA replication (17, 34). We previously showed that these components also direct transcription but that the efficient synthesis of full-length mRNAs requires the M2-1 protein (12). This indicated that M2-1 might be an elongation factor, preventing pausing or stalling, or an antitermination factor, preventing cessation of chain elongation and release of the nascent RNA. We favor the antitermination model (see Discussion) and use the term “antitermination” throughout this paper. Subsequent studies by Hardy et al. (18, 19) also indicated that M2-1 has antitermination activity but provided a somewhat different view. In their studies, the polymerase did not appear to be highly dependent on M2-1 for the synthesis of complete mRNA. Rather, the major effect of M2-1 was to inhibit termination at the GE signal to produce polycistronic readthrough mRNAs (GE-antitermination). These two M2-1-associated phenomena (which for the purpose of discussion we term here “intragenic antitermination” and “GE antitermination,” respectively) are not incompatible and most probably are different manifestations of the same activity. However, the two phenomena would not be identical in their effects on sequential transcription of the 15,222-nt genome. Since these previous studies had involved engineered genes which usually were short and in some cases contained foreign sequence, the effects of M2-1 during authentic RSV transcription remained unclear. This was analyzed here on minigenomes containing as many as five separate genes and including up to 3,432 nt of authentic sequence from the 3′ end of the RSV genome. In addition, we addressed related issues such as the fate of prematurely terminated polymerases and the effect of M2-1 on events at the leader-NS1 junction. Finally, in previous studies, we had observed that M2-1 had no effect on genome synthesis (12). Here we extended that observation and showed that M2-1 has no effect on antigenome synthesis, indicating that it is not actively involved in RNA replication.

MATERIALS AND METHODS

Protein expression plasmids.

pTM1 plasmids containing the ORFs of the N, P, M2-1, or L protein under the transcriptional control of the promoter for T7 RNA polymerase and the translational control of the internal ribosome entry site of encephalomyocarditis virus were constructed in previous work (12, 17).

Minigenome plasmids.

Plasmid C2 encodes the negative-sense minigenome C2 (see Fig. 2A), which contains, in 3′ to 5′ order, the 44-nt RSV leader region, the NS1 GS signal, the upstream 29 nt of the nontranslated region of the NS1 gene, a negative-sense copy of the chloramphenicol acetyltransferase (CAT) translational ORF, the last 12 nt of the nontranslated region of the L gene and the L GE signal, and the 155-nt trailer region (17). The cDNA is bordered at the 5′ end relative to the encoded minigenome by three G residues and the T7 RNA polymerase promoter (the G residues improve the efficiency of initiation by the T7 RNA polymerase) and at the 3′ end with a self-cleaving hammerhead ribozyme (17). C41 (see Fig. 3A) is identical to C2 except that it is followed instead by the hepatitis delta virus ribozyme (27). Plasmid containing minigenome MP-30 (see Fig. 1A and 3A) was generated by insertion of an oligonucleotide duplex encoding the RSV N-P gene junction (consisting of the N GE signal, the 1-nt N-P intergenic region, and the P GS signal) into the unique BspEI site within the CAT ORF of plasmid C41. The oligonucleotide duplex contained the sequence 5′-TCCGGgAGTTAATAAAAAATGGGGCAAATAGGATCcCCGGA, which is shown in positive sense with the GE and GS motifs underlined, the BspEI sites italicized, and single nucleotide substitutions which destroyed each site in lowercase. To generate plasmid containing minigenome RF-9 (see Fig. 1A), a fragment of the F gene of RSV was inserted into the XbaI site which lies within the first gene of MP-30 (see Fig. 1A). The XbaI site lies at the junction between the nontranslated region of NS1 and the start of the CAT ORF. The 938-nt fragment of the F gene was generated by PCR with the primers 5′-ccggTCTAGACAATCAACATGCAGTGCAGTTAGCAAAGGC and 5′-ccggTCTAGAGCATTGTCACAGTACCATCCTCTGTCAG, where the XbaI restriction site is shown in italics and nonspecific flanking sequence is shown in lowercase. Plasmid containing minigenome C2-F (see Fig. 2A) was generated by replacing the 660-nt CAT ORF of plasmid C2 with a 601-nt fragment of the F gene of RSV. The F-gene fragment was generated by PCR with the primers 5′-acaacaacaacaTCTAGATATAGAAACTGTGATAGAGTT and 5′-acaacaacaacaCTGCAGGCATGACACAATGGCTCCTAG, where XbaI and PstI sites are italicized and nonspecific flanking sequence is shown in lowercase, digested with XbaI and PstI, and ligated into the XbaI-PstI window of plasmid C2. To construct plasmid containing minigenome NS1-NS2-CAT (see Fig. 4A), a fragment of the genome of infectious recombinant RSV containing nt 1 to 1125 was generated by PCR with the plasmid encoding recombinant RSV (10) as the template and the primers 5′-CTGCGTTAGCAATTTAACTGTG, which hybridizes to the plasmid backbone upstream of the leader, and 5′-ttatttgccccatttttttggATCTTCTATCTTATATCTCTC, which hybridizes within the NS2-N intergenic region and has nonspecific flanking sequence (lowercase) and a BstXI site (italics). This fragment was digested with BstXI, which cuts the PCR product twice, once near the end of the leader region and once within the downstream primer. The 1,091-nt fragment that was generated was inserted into the unique BstXI site of plasmid C2, which lies within the leader region. The orientation of the insert was verified by restriction digest analysis. Thus, minigenome NS1-NS2-CAT is a chimera of the first 1,125 nt of infectious recombinant RSV (10), containing the leader, the NS1 and NS2 genes fused to the last 10 nt of the leader, followed by the remainder of the C2 minigenome including the CAT gene with its transcription signals and trailer region. Plasmids encoding minigenomes NS1-NS2-N/CAT and NS1-NS2-N-P-M/CAT (see Fig. 5A) were constructed by inserting the BstXI-AvrII or the BstXI-SpeI fragments, respectively, of the plasmid encoding recombinant RSV into the BstXI-XbaI window of plasmid C2. Thus, minigenome NS1-NS2-N/CAT contains the first 2130 nt and NS1-NS2-N-P-M/CAT contains the first 3432 nt of infectious recombinant RSV fused to the start of the CAT ORF of minigenome C2.

FIG. 2.

M2-1 is required for fully processive transcription of authentic RSV sequence as well as foreign sequence. (A) Structures of minigenomes C2, in which the gene is composed mainly of foreign CAT sequence, and C2-F, in which the gene is composed solely of RSV-specific sequence. The negative-sense oligonucleotide, 6682, hybridizes to the nontranslated region of NS1, which forms the 5′ end of each mRNA. (B and C) Northern blot analysis of positive-sense RNAs synthesized in HEp-2 cells which were transfected as described for Fig. 1 with plasmid encoding minigenome C2 (B) or C2-F (C) together with plasmids N and P (lane 1), N, P, and L (lane 2), or N, P, or L, and the indicated amounts of M2-1 (lanes 3 to 6). RNA was analyzed by Northern blot hybridization to oligonucleotide 6682.

FIG. 3.

Lack of effect of M2-1 on events at the leader gene junction. (A) Structures of minigenomes MP-30 and C41. Minigenome MP-30 is identical to C41 except that it contains the N-P gene junction inserted within the CAT gene. Oligonucleotide 5880 is specific against the positive-sense leader transcript, and oligonucleotide 6682 hybridizes to the nontranslated region of NS1. (B) Northern blot analysis of positive-sense RNAs from HEp-2 cells which were transfected as described for Fig. 1 with plasmid MP-30 or C41 and the support plasmids N and P, together with the indicated combinations of L and M2-1. RNA was analyzed by Northern blot hybridization with the indicated 5′-end-labelled oligonucleotide.

FIG. 1.

In the absence of M2-1, the RSV polymerase efficiently synthesizes short, but not long, mRNAs and does not reinitiate following intragenic termination. (A) Structures of the MP-30 and RF-9 minigenomes. Minigenome MP-30 is a dicistronic minigenome in which the CAT coding sequence has been broken into two genes of 274 and 495 nt separated by the N-P gene junction of RSV (see Materials and Methods). Minigenome RF-9 was constructed by inserting the RSV F sequence (hatched) into the first gene of MP-30 (dotted lines). GS and GE transcription signals are shown as small open and solid boxes, respectively; negative-sense oligonucleotide probes 6682, 5629, 5878, and 3756 are shown as short thick lines. These conventions are used in each figure. (B and C) Northern blot analyses of positive-sense RNAs synthesized in HEp-2 cells which were infected with vaccinia virus recombinant vTF7-3 and simultaneously transfected with plasmid MP-30 (B) or RF-9 (C), and the support plasmids N and P, together with the indicated combinations of L and M2-1. RNA was purified 48 h later, subjected to electrophoresis on formaldehyde gels, transferred to nitrocellulose, and analyzed by hybridization with the indicated 5′-end-labelled oligonucleotide.

FIG. 4.

Expression of positive-sense RNAs from a tricistronic minigenome, NS1-NS2-CAT, which contains the 3′-terminal 1,125 nt of the RSV genome, including the NS1 and NS2 genes, followed by the CAT gene. (A) Diagram of minigenome NS1-NS2-CAT and locations of the negative-sense oligonucleotide probes. Oligonucleotide 6682 (∗) hybridizes to the nontranslated region of NS1, which is represented twice in the NS1-NS2-CAT minigenome and thus detects the NS1 and CAT mRNAs. (B to D) Northern blot analyses of positive-sense RNAs synthesized in HEp-2 cells transfected with NS1-NS2-CAT plasmid together with plasmids N and P (lane 3), N, P, and L (lane 4), or N, P, L, and the indicated amounts of M2-1 plasmid (lanes 5 to 10). Lane 1 contains total RNA isolated from RSV-infected HEp-2 cells at 24 h postinfection, and lane 2 contains RNA isolated from uninfected cells. Hybridization was performed with the indicated oligonucleotide probe. (E to G) Quantitation of the mRNAs detected with oligonucleotides 6684, 6686, and 3756, respectively, which each hybridize to the downstream end of one of the three mRNAs. The RNA bands in each lane were normalized so that the mini-antigenome equalled 1,000 units.

FIG. 5.

M2-1 alters the transcription gradient. (A) Structures of minigenomes NS1-NS2-N/CAT and NS1-NS2-N-P-M/CAT. The first two genes and gene junctions of NS1-NS2-N/CAT are identical to those of recombinant RSV, and the third gene is a chimera which contains 1,005 nt of the N gene of RSV fused to CAT. Likewise, NS1-NS2-N-P-M/CAT contains the first four genes and gene junctions of RSV, and the fifth gene consists of 180 nt of the RSV M gene fused to CAT. (B to E) Northern blots of positive-sense RNAs synthesized in HEp-2 cells transfected with plasmid encoding either minigenome NS1-NS2-N/CAT (B and C) or NS1-NS2-N-P-M/CAT (D and E) and the support plasmids N and P (lane 10), N, P, and L (lane 2), or N, P, L, and the indicated amounts of M2-1 plasmid (lanes 3 to 9). Lane 1 contains total RNA from RSV-infected cells (panels B and D) or from cells which received plasmids encoding minigenome C2, N, P, L, and 100 ng of M2-1 (panels C and E).

Transfections.

Monolayers of HEp-2 cells in six-well dishes were transfected with the following mixture of plasmids per single well of a six-well dish: 200 ng of minigenome DNA, 400 ng of pTM1 N, 200 ng of pTM1 P, 100 ng of pTM1 L, and 100 ng of pTM1 M2-1; the transfections were done in reactions in which M2-1 was included, except for the transfections in Fig. 2, 4 and 5, in which the cells received the indicated amount of pTM1 M2-1. In the transfections in Fig. 2, 4 and 5, pTM1 vector with no insert was added at the appropriate level to maintain a consistent amount of input plasmid. The cells were simultaneously transfected with the above-mentioned plasmids and infected at 10 PFU per cell with vaccinia virus vTF7-3 (provided by Thomas Fuerst and Bernard Moss), which expresses the T7 RNA polymerase (16), as follows. A 0.1-ml volume of OptiMem (Life Technologies) containing the plasmids was mixed with 0.1 ml of OptiMem containing 12 μl of LipofectACE (Life Technologies), incubated at room temperature for 15 min, and mixed with 0.8 ml of OptiMem containing 2% fetal bovine serum and the vaccinia virus inoculum. After 22 to 24 h, the transfection-infection mixture was replaced with OptiMem containing 2% fetal bovine serum and actinomycin D at 2 μg/ml to inhibit further plasmid-based RNA synthesis. The actinomycin D-containing medium was removed after 2 h, fresh OptiMem containing 2% fetal bovine serum was added, and the cells were incubated for a further 24 h.

RSV infection time course.

Monolayers of HEp-2 cells in 25-cm² flasks were infected with RSV (A2) at a multiplicity of infection of 4 PFU/cell. Following a 1-h adsorption, the virus inoculum was removed, the cell monolayer was washed, and OptiMem containing 2% fetal bovine serum was added. Cells for the 0-h time point were harvested at this time. The remaining flasks were incubated at 37°C, and cells were harvested for protein and RNA samples at 3-h intervals.

RNA isolation and Northern blot hybridization.

Total intracellular RNA was extracted from cell pellets by using Trizol reagent (Life Technologies) as specified by the supplier, except that the RNAs were extracted with phenol-chloroform and ethanol precipitated following the isopropanol precipitation. Approximately 4 μg of each RNA sample was analyzed by electrophoresis in a 1.5% agarose gel containing 0.44 M formaldehyde, transferred to nitrocellulose (Schleicher & Schuell), and fixed by UV cross-linking (Stratagene). The blots in Fig. 1, 2, 3, 4, and 5 (panels B and D) were hybridized with 5′-end-labelled negative-sense oligonucleotide probes in 6× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–5× Denhardt’s solution–0.1% sodium dodecyl sulfate (SDS)–0.05% sodium pyrophosphate at 52°C for 12 h. The blots were washed in 6× SSC for 30 min. The probe specificities are indicated in the figures, and their sequences are shown in Table 1. The blots in Fig. 5C and E were hybridized with a negative-sense ³²P-labelled CAT-specific riboprobe, and the blots in Fig. 6B and C were hybridized with double-stranded, randomly primed ³²P-labelled DNA probes against NS1 and M2 as indicated in the figure legend. The hybridization conditions for these probes were 6× SSC–5× Denhardt’s solution–0.5% SDS–200 μg of sheared DNA per ml at 65°C for 12 h. The blots were washed in 2× SSC–0.1% SDS at room temperature for 30 min and then at 65°C for 2 h. PhosphorImager analysis was carried out with a PhosphorImager 445 SI (Molecular Dynamics), and densitometry analysis was carried out with a Personal Densitometer SI (Molecular Dynamics).

TABLE 1.

Sequences and specificities of the oligonucleotide probes used for Northern blot analysis

Oligonucleotide	Specificity	Sequence (5′-3′)a
5880	Leader transcript	TTTTTGGTTTATGCAAGTTTGTTGTACGCATTTTTTCCC
6682	NS1 NT region	AAGTGGTACTTATCAAATTCTT
6683	NS1 mRNA 5′ end	CTCAATGAATTGCTGCCCATCTCTAACCAAGGGAGT
6684	NS1 mRNA 3′ end	GCTAGTTGATATTAATTATAATTTATGGATTAAGATC
6686	NS2 mRNA 3′ end	CAGGCTCCATCTGGACTATGGAGTATAGTTATGCATAG
5629	CAT ORF 5′ end	CGGTCTGGTTATAGGTACATTGAGCAACTGACTGAAATGCCTCAAAATGTT
5878	CAT ORF internal	CATATTCTCAATAAACCCTTTAGGGAAATAGGCCAG
3756	CAT ORF 3′ end	ACTGTTGTAATTCATTAAGCATTCTGCCGA

^aEach oligonucleotide is negative sense and is written in the 5′-to-3′ orientation. 

FIG. 6.

Lack of temporal regulation of RSV transcription. HEp-2 cells were infected with RSV (A2) at a multiplicity of infection of 4, and samples were harvested for RNA and protein at 3-h intervals. (A) Western blot of protein samples isolated from 0 to 24 h postinfection (lanes 2 to 10). Lane 1 contained total protein from uninfected cells. The RSV proteins were detected with an antiserum raised against gradient purified RSV virions. (B and C) Northern blots of RNA samples isolated simultaneously with the protein samples and arranged in the same lane order. The blots in panels B and C were hybridized with double-stranded DNA probes labelled by random priming and specific to NS1 and M2, respectively. (D) For each time point, the amount of each mRNA was normalized relative to genome-antigenome as an internal standard and expressed relative to the 24-h time point as 100.

Western blot analysis of RSV proteins.

Cells were lysed in 2% SDS–50 mM Tris (pH 7.0)–0.63 M β-mercaptoethanol and clarified by passage through a QIAshredder column (Qiagen). Lysate from approximately 4 × 10⁴ cells was subjected to electrophoresis though a 12% polyacrylamide gel, and the separated polypeptides were transferred to nitrocellulose by conventional electrophoretic techniques. RSV-specific proteins were detected by incubation with rabbit antiserum raised against gradient-purified RSV virions followed by incubation with anti-rabbit immunoglobulin G conjugated to alkaline phosphatase (Vector Laboratories) followed by an alkaline phosphatase reaction with a 5-bromo-4-chloro-3-indolylphosphate/nitroblue toluidine (BCIP/NBT) color development system (Promega).

RESULTS

Effect of gene length on sequential transcription.

The effects of the M2-1 protein on RSV transcription were examined in a minigenome system in which transcription and replication are reconstituted from plasmid-supplied RNA and protein components (12, 17). First, we investigated the effect of gene length on mRNA synthesis in the absence of M2-1 by using two templates, MP-30 and RF-9. MP-30 was constructed from an existing CAT reporter minigenome (C41) by inserting the RSV N-P gene junction into the CAT gene to create a dicistronic minigenome which encodes two mRNAs of 274 nt (first gene) and 495 nt (second gene) (Fig. 1A). Minigenome RF-9 was constructed from MP-30 by inserting RSV F sequence into gene 1 of MP-30 to create a minigenome that has a relatively long first gene (1,212 nt) but in which the gene junction and the second gene are unaltered. Compared to the authentic RSV genes, the 1,212-nt gene 1 of RF-9 is almost equivalent in length to the N gene (1,203 nt) and is much shorter than the F gene (1,903 nt) and L gene (6,578 nt). The 495-nt gene 2 is comparable in length to the three shortest RSV genes, namely SH (410 nt), NS1 (532 nt), and NS2 (503 nt). The 274-nt gene 1 of MP-30 is shorter than any RSV gene. The positive-sense RNAs expressed from MP-30 or RF-9 RNA when complemented by N, P, and L, in the presence or absence of M2-1, were analyzed by Northern blotting with 5′-end-labelled, negative-sense synthetic oligonucleotides specific for the 5′ or 3′ ends of each mRNA (Fig. 1A).

When hybridized to the RNA synthesized from MP-30 in the absence of M2-1, oligonucleotides 6682 and 5629, specific to the two ends of gene 1, detected mini-antigenome and mRNA 1 (Fig. 1B, lanes 1 and 4). The gene 1 mRNA that was generated formed a distinct band on the Northern blot and was detected equally relative to mini-antigenome with both oligonucleotide probes, indicating that most of this mRNA was complete. When M2-1 was included in the reaction mixture, there was a slight increase (1.12-fold) in the level of mRNA 1 and no change in the expression of mini-antigenome, and an additional minor RNA was synthesized which was the correct molecular weight to be gene 1-gene 2 readthrough mRNA (Fig. 1B, lanes 2 and 5). When the transcription products of RF-9 were examined with the same oligonucleotide probes, a different pattern was observed. In the absence of M2-1, both oligonucleotides detected mini-antigenome but the upstream oligonucleotide 6682 detected a diffuse smear of mostly incomplete mRNA and the downstream oligonucleotide 5629 detected a trace amount of complete mRNA, indicative of extensive premature termination (Fig. 1C, lanes 1 and 4). When M2-1 was included in the reaction mixture, full-length mRNA 1 was increased 38-fold. However, even at high levels of M2-1 (lane 2), a small amount of incomplete mRNA was detected by oligonucleotide 6682 which was above the level of background in the control without L (lane 3). As with MP-30, inclusion of M2-1 did not affect the level of mini-antigenome RNA and resulted in the synthesis of a small amount of gene 1-gene 2 readthrough mRNA which comigrated with a background band in Fig. 1C, lane 2, but was readily discernible in Fig. 1C, lanes 5, 8, and 11. We also examined the transcription of a minigenome containing a 1,780-nt luciferase transcription unit (results not shown). When luciferase enzyme activity was used as a marker of full-length luciferase mRNA transcripts, barely any full-length mRNA was synthesized in the absence of M2-1, but inclusion of M2-1 increased the level of full-length luciferase mRNA 365- to 615-fold (data not shown). Taken together, these results show that if a gene is short, such as gene 1 of MP-30, most polymerase molecules which initiate are able to complete transcription of the gene independently of M2-1. However, if the gene is long, such as gene 1 of RF-9 or luciferase, the majority of polymerase molecules require M2-1 to avoid premature termination. These results also confirmed the observation of Hardy and Wertz (18) that M2-1 induces readthrough at the gene junction, although the amount of readthrough product at this particular junction (N-P) was only a small fraction of the total mRNA population.

To examine the expression of the downstream gene in these reactions, replicate blots were hybridized with oligonucleotides 5878 and 3756, which are specific to the two ends of the downstream mRNA (Fig. 1A). For MP-30, the downstream gene was transcribed in both the presence (Fig. 1B, lanes 8 and 11) and absence (lanes 7 and 10) of M2-1, although in the latter case much of mRNA 2 was terminated prematurely. In contrast, the downstream gene of minigenome RF-9 was transcribed efficiently only in reactions in which M2-1 was included and the upstream, long gene was transcribed (Fig. 1C, compare lanes 7 and 10 to lanes 8 and 11). This indicates that transcription of gene 2 is dependent on transcription of gene 1; polymerase which terminates within gene 1 cannot reinitiate transcription at gene 2. Thus, intragenic antitermination has two effects: synthesis of complete mRNA and delivery of polymerase molecules to the next downstream gene.

Foreign sequence does not significantly affect intragenic antitermination.

The minigenome templates used in Fig. 1 and in our previous study of M2-1 (12) contained foreign CAT or luciferase sequence whose heterologous nature might have affected the processivity of the polymerase. For example, the U content of the RSV genome is 1.46 times that of the negative-sense strand of the CAT gene whereas the C content of the negative-sense strand of the CAT gene is 1.48 times that of the RSV genome. Transcription of two minigenomes, i.e., C2, which is a CAT-containing minigenome (see Materials and Methods), and C2-F, which is similar to C2 except that the CAT sequence has been replaced with a similar length of negative-sense RSV F sequence, was compared (Fig. 2A). Positive-sense RNAs synthesized from C2 or C2-F complemented with N, P, and L and with different amounts of M2-1 were analyzed by Northern blot hybridization. The probe was a negative-sense oligonucleotide specific against the nontranslated region of the NS1 gene, which is represented at the upstream end of the mRNA encoded by each minigenome (Fig. 2A). As shown in Fig. 2B and C, the patterns of transcription from C2 and C2-F were very similar and the synthesis of CAT-containing or F-containing mRNA was equally sensitive to M2-1. Since the C2-F minigenome contains only RSV sequence, this shows that the dependence on the intragenic antitermination activity of M2-1 was the same for RSV-specific and heterologous sequences.

M2-1 does not affect transcription at the leader-NS1 junction.

The finding that M2-1 has GE antitermination activity, noted above and originally described by Hardy and Wertz (18), raised the possibility that it also affects events at the leader-NS1 junction. We examined the positive-sense RNAs generated from minigenomes C41 (a CAT-containing monocistronic minigenome which is identical to C2 except that it is cleaved by a different ribozyme) and MP-30 by Northern blot hybridization with negative-sense oligonucleotides specific to transcripts of the leader and to the 5′ end of the encoded mRNA 1 of MP-30 and CAT mRNA of C41 (Fig. 3A). Hybridization with oligonucleotide 6682 showed that the short gene 1 of MP-30 was transcribed efficiently in the absence of M2-1, with only a 1.9-fold increase conferred by M2-1 (Fig. 3B, lanes 8 and 9), comparable to the results shown in Fig. 1. As would be expected because of its greater gene length, in the absence of M2-1 C41 generated mostly truncated mRNA which migrated as a diffuse smear whereas inclusion of M2-1 yielded full-length mRNA (Fig. 3B, lanes 11 and 12). Mini-antigenome RNA was detected for both C41 and MP-30, and its abundance was unaffected by M2-1 expression.

A replicate blot was then probed with leader-specific oligonucleotide 5880 to determine how much mRNA 1 was present as a readthrough with the short leader RNA. Based on previous observations (8, 24), we expected a fraction (approximately 10 to 15%) of the promoter-proximal mRNA to contain attached leader RNA. The question was whether the abundance of this readthrough RNA was affected by the presence or absence of M2-1. When C41 or MP-30 RNA synthesized in the presence of M2-1 plasmid was analyzed with oligonucleotide 5880, the probe hybridized to the antigenome, to mRNA 1 in the case of MP-30 (Fig. 3B, lane 3) and to CAT mRNA in the case of C41 (lane 6). When RNA synthesized in the absence of M2-1 was analyzed, it was found that MP-30 yielded a significant level of leader containing mRNA (lane 2). Similar to the results obtained with oligonucleotide 6682, there was a 1.7-fold increase in the amount of leader-containing mRNA in the presence of M2-1. The finding that some MP-30 mRNA 1 was attached to the leader in the absence of M2-1 and that this amount was not augmented relative to monocistronic mRNA by M2-1 indicates that M2-1 does not affect readthrough at this junction. The leader-specific oligonucleotide did not hybridize to a specific mRNA band synthesized from minigenome C41 in the absence of M2-1 (Fig. 3B, lane 5), but this was expected since most of the mRNA synthesized under these conditions would be heterogeneous in size and therefore dispersed on the Northern blot. The absence of a small, mRNA 1-sized band in the C41 RNA confirmed that the small species in the MP-30 pattern was leader-mRNA 1 readthrough mRNA rather than free leader, which would not be expected to be found by this method.

Effect of M2-1 on transcription of a tricistronic minigenome.

It was of interest to examine the effect of M2-1 on transcription of the 3′-terminal region of the authentic RSV genome. This was done with a negative-sense minigenome, NS1-NS2-CAT (Fig. 4A) in which the 3′-terminal 1,125 nt, including the leader region and the complete NS1 and NS2 genes, was identical to the 3′-terminal sequence of the genome of infectious recombinant RSV and was followed by CAT as the third gene. We chose CAT rather than N, the third gene in the RSV gene order, so that its transcript could be distinguished from N mRNA expressed from the N support plasmid.

The NS1-NS2-CAT minigenome was expressed in cells in the presence of N, P, and L and in the absence or presence of increasing amounts of M2-1. RNA was analyzed by Northern blot hybridization with the negative-sense oligonucleotides in Fig. 4A, whose sequences are shown in Table 1. For simplicity, Fig. 4 shows only a subset of the Northern blots, specifically those which were hybridized with oligonucleotide 6683, specific to the upstream end of the NS1 mRNA (Fig. 4B), oligonucleotide 6684, specific to the downstream end of the same mRNA (Fig. 4C), and oligonucleotide 6682, specific to a sequence in the upstream end of the NS1 mRNA which is also present in the chimeric CAT gene and thus enables direct comparison of NS1 and CAT mRNAs (Fig. 4D). Quantitation of blots, all from the same experiment and representing probes 6684, 6686, and 3756, each of which hybridized to the downstream end of one of the three monocistronic mRNAs, is summarized in Fig. 4E, F, and G, respectively. This analysis showed that the synthesis of complete NS1 mRNA was not highly dependent on M2-1 and that, surprisingly, the same was true of NS2 mRNA (e.g., Fig. 4B and C, lanes 4, and Fig. 4F). A considerable amount of monocistronic NS1 and NS2 mRNA was synthesized in the absence of M2-1, and the inclusion of M2-1 increased the synthesis of either mRNA by only up to two- to threefold (Fig. 4E and F). In contrast, CAT mRNA, representing the third gene, was barely detectable in the absence of M2-1 (e.g., Fig. 4D, lane 4) and increased up to 27-fold when M2-1 was included (Fig. 4G). The much greater dependence of transcription of CAT on M2-1 is probably due to its position in the minigenome and its length, and not its foreign nature, based on the results in Fig. 2. Since CAT is considerably shorter than N, the authentic third gene, it is reasonable to assume that N would be transcribed even less efficiently in the absence of M2-1.

In contrast to monocistronic NS1 and NS2 mRNAs, the synthesis of readthrough mRNAs was strongly dependent on M2-1 (Fig. 4B to G), as described previously by Hardy et al. (18, 19). For example, the NS1-NS2 mRNA was barely detectable in the absence of M2-1, but in the presence of M2-1 its level increased to a maximum of 25% of the level of NS1 gene transcripts (Fig. 4E). Similarly, NS1-NS2 mRNA was 30% of the level of NS1 gene transcripts in RSV-infected cells (Fig. 4B to D, lanes 1), confirming that the GE antitermination effect had been faithfully reconstituted in the plasmid-based rescue system. Interestingly, transcription of NS2-CAT and NS1-NS2-CAT mRNAs was more dependent on M2-1 than was transcription of CAT mRNA, even though the polymerase must traverse the same length of template to generate each of these transcripts (Fig. 4D). This indicates that more M2-1 is required to synthesize readthrough mRNAs than the corresponding monocistronic mRNAs.

Effect of M2-1 on the transcription gradient of RSV.

We were interested in distinguishing between two possible models of M2-1–polymerase interaction. The first envisions a transient association between the polymerase and M2-1, with M2-1 cycling on and off the polymerase during transcription. Higher concentrations of M2-1 would favor reassociation of M2-1 with the transcribing polymerase and promote processivity and hence would alter the gradient of transcription. The second model envisions a stable conversion of the polymerase to a processive form by M2-1. According to the second model, the polymerase would still be subject to the attenuating effect of cis-acting elements but the gradient of transcription would be largely independent of the concentration of M2-1. This was investigated by comparing the effect of M2-1 on a tricistronic minigenome (NS1-NS2-N/CAT) and a pentacistronic minigenome (NS1-NS2-N-P-M/CAT) (Fig. 5A). By examining expression of CAT mRNA from each of these minigenomes, it was possible to determine if transcription of the fifth gene is more dependent on the level of M2-1 expression than is transcription of the third gene. The positive-sense RNAs expressed by the two minigenomes in the presence of increasing amounts of M2-1 were examined by Northern blotting with an oligonucleotide specific to the NS1 mRNA 3′ end (Fig. 5B and D) and a CAT-specific riboprobe (Fig. 5C and E) and quantitated with a PhosphorImager. The two minigenomes directed the synthesis of similar amounts of NS1 mRNA. This parity was confirmed when the two blots were normalized relative to a control lane of RNA from RSV-infected cells contained in each blot as a standard. In response to increasing amounts of M2-1, the synthesis of monocistronic NS1 mRNA was increased two- to threefold for either minigenome and the synthesis of polycistronic mRNAs was detectable only in reaction mixtures containing M2-1. This recapitulates the results shown in Fig. 4. The RNAs were then hybridized with a negative-sense CAT-specific probe to compare the accumulation of the N/CAT mRNA, the third gene of the NS1-NS2-N/CAT minigenome, with that of the M/CAT mRNA, the fifth gene of the NS1-NS2-N-P-M/CAT minigenome. In each case, a negligible amount of CAT-containing mRNA was synthesized in the absence of M2-1, and synthesis was greatly stimulated by the addition of M2-1. We analyzed each minigenome separately to normalize the amount of N/CAT or M/CAT synthesized at each concentration of M2-1 relative to its expression at the highest level of M2-1. This showed that at input amounts of M2-1 of 1, 2, 4, 8, 16, and 32 ng, the level of expression of N/CAT was <1, 9, 33, 38, 51, and 66% of its maximum, respectively, whereas that of M/CAT was 3, 6, 12, 27, 32, and 52% of its maximum, respectively. Thus, the expression of M/CAT, the fifth gene, was more dependent on M2-1 than was the expression of N/CAT, the third gene. While the difference was not great, it was reproducible. Presumably, genes which were separated by more than a single intervening gene would exhibit an even greater difference in dependence on M2-1.

Analysis of ratios of RSV mRNAs during infection.

It was of interest to determine if there was any alteration in the transcription pattern of RSV during an infection time course, correlating with changes in the level of M2-1. RSV-infected HEp-2 cells were harvested for RNA and protein analysis at 3-h intervals. RSV proteins were detected by Western blotting with antiserum raised against gradient-purified RSV virions (Fig. 6A). At 0 h (following a 1-h adsorption), trace amounts of N, P, and G proteins could be detected in the cell lysates due to either the input virus inoculum or de novo-synthesized protein or both. As time progressed, other RSV proteins, i.e., M, F1, SH, and M2-1, became detectable and increased in abundance; these proteins included two isoforms of M2-1, which migrated slightly ahead of the M protein. This shows that the intracellular concentration of M2-1 protein increases during infection, although the relative ratio of all proteins appeared constant, at least during the phase of infection during which all were detectable. Total intracellular RNA samples from the same infection time course were analyzed by Northern blotting with double-stranded DNA probes against NS1 and M2 (Fig. 6B and C, respectively). Genome-antigenome RNA could be detected at the earliest time point, probably at least in part reflecting the input inoculum, and its level increased during the infection. NS1 and M2 monocistronic mRNA and polycistronic mRNAs could be detected from 3 h postinfection, and their levels increased over time. By quantitating the levels of NS1- and M2-containing mRNAs relative to the genome-antigenome at the different times of infection, it was found that the ratio between NS1 and M2 monocistronic mRNA and polycistronic mRNAs did not alter during the time course (Fig. 6D). This showed that the pattern of RSV mRNA expression was consistent during infection of a permissive cell line. This observation also is consistent with studies in which RSV RNA synthesis was monitored by metabolic labelling at various times during infection, which indicated that the pattern of RSV mRNAs was essentially invariant (4).

DISCUSSION

We previously showed that a major activity of the M2-1 protein is to promote the synthesis of complete mRNAs, i.e., “intragenic antitermination” (12). A subsequent study by Hardy and Wertz (18) showed that M2-1 also promotes antitermination at the GE signal, i.e., “GE antitermination.” The two effects have different outcomes on the synthesis of monocistronic versus polycistronic mRNAs, and it was important to evaluate their roles in the expression of the RSV genome. In the present study, we analyzed the effect of M2-1 on transcription of a variety of minigenomes including ones which contained several authentic RSV genes. This confirmed the occurrence of both effects and provided an analysis of their relative contributions to the transcription pattern of RSV.

When analyzed on comparatively short genes of up to approximately 500 nt, the intragenic antitermination activity of M2-1 increased the synthesis of full-length mRNA by a factor of only two- to threefold (e.g., NS1 expression in Fig. 4 and 5). In contrast, longer genes were highly dependent on M2-1 for complete transcription. For example, transcription of full-length luciferase mRNA (whose gene length is 1,780 nt) was increased at least 365-fold in the presence of M2-1 (data not shown). A control study indicated that the polymerase, with or without M2-1, behaved similarly on templates containing RSV-specific or heterologous sequence. Thus template length is a major factor in the dependence on M2-1 for synthesis of full-length mRNA. Accordingly, the intragenic antitermination effect of M2-1 is less apparent when evaluated on shorter minigenomes but is a major factor on longer minigenomes and certainly would be a critical factor in transcription of the complete 15,222-nt RSV genome.

The M2-1 protein might facilitate the synthesis of complete mRNAs by inhibiting the polymerase from pausing and forming a stalled complex. In this case, M2-1 would be a processivity factor. Alternatively, it might inhibit the polymerase from bona fide termination and release of the nascent transcript, in which case it would be an antitermination factor. While this question is difficult to resolve in an intracellular system, there is indirect evidence for the latter idea. Specifically, 50% or more of the incomplete mRNA synthesized in the absence of M2-1 was shown to be polyadenylated based on binding to oligo(dT) (12, 17). We presume that this polyadenylation is mediated by poly(A) polymerase encoded by the coinfecting vaccinia virus recombinant, and indeed approximately the same fraction of unencapsidated minireplicon RNA becomes polyadenylated (17). The availability of the 3′ end of the incomplete mRNA for polyadenylation implies that it is free and hence had been released by termination. Thus, our working model is that the M2-1 protein is an antitermination factor and that the synthesis of complete mRNAs and readthrough at GE signals are two aspects of an antitermination activity. Admittedly, the latter clearly involves an override of a cis-acting element while the former might only involve overcoming nonspecific destabilization of the polymerase-template-transcript complex. Since authentic RSV transcription is sequential and involves a very long template, the possibility existed that polymerase which terminated prematurely within a gene would remain template bound and continue to migrate down the genome, much as the polymerase is thought to do on intergenic regions. Such a polymerase would be available to reinitiate at the next downstream GS signal. However, when a long gene was placed in the promoter-proximal position, no expression was observed for a short downstream gene in the absence of M2-1 (Fig. 1C). This showed that reinitiation did not occur. It also showed that transcription of the downstream gene is completely dependent on sequential transcription, a point which we have noted previously (15, 24). Thus, the intragenic antitermination activity of M2-1 plays a crucial role in moving the polymerase down the genome.

It was noticeable that even in the presence of high levels of M2-1 there was a considerable amount of prematurely terminated mRNA in the transfected cells, detected as a diffuse smear of RNA by probes against the 5′ end of the mRNA but not by probes against the 3′ end (e.g., Fig. 1C, lanes 2 and 5; Fig. 4B and C, lanes 4 to 10). This pattern was not seen in RSV-infected cells (Fig. 4B to D, lanes 1). It is formally possible that the reconstituted system is somehow slightly defective for transcription. This would not be due to a faulty support protein, since each of these support cDNAs was used to construct a recombinant RSV parent which has a wild-type phenotype (10) and in particular produces the same RNA pattern as biologically derived RSV (31). A second possibility is that prematurely terminated transcripts are generated in RSV-infected cells but are rapidly degraded because they are not polyadenylated (23) whereas incomplete transcripts generated by the reconstituted RSV polymerase are stabilized by the poly(A) polymerase activity encoded by the coinfecting vaccinia virus, as we previously demonstrated (17). If this is the case, intragenic termination would account for some of the transcription attenuation of RSV, in addition to attenuation at the gene junction regions as described for VSV (22). Gel electrophoresis analysis of prematurely terminated transcripts generated in both the presence and absence of M2-1 revealed a broad band of heterogeneously sized molecules, suggesting that intragenic termination generally occurs in a nonspecific manner. However, bands of somewhat greater intensity could often be observed within the dispersed RNA, indicating that there might be some sites within genes at which the polymerase is particularly prone to termination, in both the presence and absence of M2-1.

Hardy et al. (18) showed that an additional manifestation of the activity of M2-1 is to promote antitermination at GE signals, resulting in readthrough across gene junctions. More recently, these workers observed that the level of readthrough differs among the various naturally occurring gene junctions (19), which also was noted previously from examination of readthrough mRNAs isolated from RSV-infected cells (6, 7). In RSV-infected cells, readthrough was greatest across the NS1-NS2 and NS2-N gene junctions (6), which probably is explained in large part by our previous finding that the GE signals of the NS1 and NS2 genes are less efficient than those of the other genes (25). In the present study, we also observed higher levels of readthrough in minigenomes containing the NS1-NS2 junction (Fig. 4 and 5) compared to the N-P junction (Fig. 1). Indeed, we found with minigenome NS1-NS2-CAT (Fig. 4) that the level of readthrough at the NS1-NS2 and NS2-CAT junctions (the NS2-CAT junction consists of the NS2-N intergenic region fused to the last 10 nt of the leader) was sufficient to reduce the accumulation of monocistronic NS2 mRNA to below the level of monocistronic CAT mRNA (Fig. 4F and G). However, one major difference between our findings and those of Hardy et al. is in the magnitude of readthrough. Those workers observed very high frequencies of readthrough, for example greater than 55, 60, and 75% across the F-M2, NS1-NS2, and NS2-N junctions, respectively (19), such that the readthrough mRNAs were in considerable excess of the monocistronic mRNAs. These remarkable patterns are very different from those observed in authentic RSV infections in cell culture, where polycistronic mRNAs consistently are much less abundant than the monocistronic mRNAs. The low abundance of readthrough mRNAs in RSV infection is true whether the RNAs are detected by Northern blot analysis (6, 7; see above) or by metabolic labeling (4, 6, 7, 21) and thus does not appear to represent a technical difference. In RSV-infected cells, the most abundant readthrough mRNAs, the NS1-NS2 and NS2-N mRNAs, account for up to 30% of the transcription product of their respective genes whereas most other readthrough mRNAs account for less than 10% of transcription of their respective genes (present work and unpublished observations). One possibility is that the very high level observed by Hardy et al. is an artifact of the plasmid-based system, perhaps due to disproportionately high synthesis of the M2-1 protein. Alternatively, if the very high level of readthrough observed by Hardy et al. is authentic, one would have to postulate that degradation or processing of the polycistronic mRNAs occurs in RSV-infected cells to account for the observed proportions of polycistronic and monocistronic mRNAs.

In the present study, the conditions of reconstituted transcription produced a level of readthrough which was essentially indistinguishable from that of RSV-infected cells and thus can be used to assess the relative contributions of intragenic antitermination and GE antitermination to sequential transcription. Based on the level of readthrough mRNA in RSV-infected cells, GE antitermination would deliver up to 30%, and more typically 10% or less, of the polymerase across a gene junction, thus sparing this fraction from the attenuation thought to occur at that site (22). In comparison, the intragenic antitermination effect of M2-1 on transcription of a 1.78-kb template increased the delivery of polymerase 365-fold and thus has a much greater effect. This suggests that during transcription beyond the NS1 and NS2 genes, the intragenic antitermination effect of M2-1 plays the major role in delivering polymerase to downstream genes and that the GE antitermination effect makes a significant but quantitatively more minor contribution.

Whereas M2-1 has antitermination activity at GE signals, it did not appear to direct readthrough at the leader-NS1 junction based on measurement of the amount of mRNA containing attached leader sequence. This was somewhat unexpected since it is generally thought that the leader region contains some kind of termination signal at its junction with NS1. This finding suggests that events at the leader-NS1 junction are different from those at other RSV gene junctions, and it provided evidence against the possibility that M2-1 is involved in regulating the synthesis of mRNA (transcription) versus antigenome (replication). Consistent with this, the level of antigenome accumulation was unaffected by the presence or absence of M2-1 or by changes in its concentration (such as in Fig. 1 through 5).

It was unexpected that significant levels of both the NS1 and NS2 mRNAs would be made in the absence of M2-1. These two genes, together with their intergenic sequence, constitute 1,054 nt, and it might have been expected that there would be a greater dependence on M2-1, at least for expression of the NS2 gene. Perhaps there is some template feature which facilitates elongation and remains to be identified. It would not be due to the relatively low efficiency of the NS1 and NS2 GE signals, since, as mentioned above, these decrease rather than increase the synthesis of monocistronic NS2 mRNA. The finding that NS1 and NS2 alone can be efficiently expressed in the absence of M2-1 represents a potential second transcription program. This program might be an idling state in which NS1 and NS2, but none of the virion structural proteins, including the M2-1 protein, are expressed. Whether this occurs in vivo, for example in a persistent infection, is unknown. The possible consequences of expression of the NS1 and NS2 genes alone are unclear. Each of these genes can be deleted from recombinant RSV alone and in combination without ablating its ability to grow in cultured cells, although the resulting virus is attenuated for growth in vitro (5, 31, 31a). The NS1 protein has been implicated as a negative regulatory factor (2), but neither protein appears to play a role in virion morphogenesis or passage (30).

We examined whether variation in M2-1 expression causes an alteration of the RSV transcription gradient downstream of the NS2 gene. This indeed appeared to be the case, since comparison of the effect of M2-1 on a tricistronic and a pentacistronic minigenome indicated that more M2-1 is required to transcribe the fifth gene than to transcribe the third gene. It has been shown previously that the transcription gradient of measles virus can differ in different cell lines (29). This is a prior example of how the transcription gradient could be affected by a trans-acting factor(s), presumably by a host cell factor(s) in measles virus. An M2-1-dependent effect on the RSV transcription gradient is consistent with a model in which M2-1 oscillates on and off the polymerase during sequential transcription. At subsaturating concentrations of M2-1, the polymerase would tend to dissociate from the nucleocapsid and the downstream genes would be affected disproportionately. This seemed to discount the alternative possibility that M2-1 confers a stable processive state to the polymerase, since under that condition the expression of upstream and downstream genes should occur in the same relative proportion regardless of the level of M2-1.

The data shown in Fig. 4 also addressed the question whether intragenic antitermination and GE antitermination differed in their relative requirements for M2-1. For example, the synthesis of NS2-CAT mRNA from minigenome NS1-NS2-CAT was more dependent on M2-1 than was the synthesis of CAT mRNA (Fig. 4D). Since the polymerase must traverse exactly the same length of template in each case, the greater dependence of the readthrough mRNA presumably indicates that GE antitermination requires a higher level of M2-1. The model described above, in which M2-1 oscillates on and off the polymerase, could explain this. When M2-1 dissociates from the polymerase, the polymerase might be able to continue transcription for some distance, as is seen in the situation where M2-1 is absent, before either terminating transcription or reassociating with M2-1. In that time, if the polymerase encountered a GE signal, it would almost certainly terminate transcription to generate monocistronic mRNA, since readthrough mRNAs were not observed in the absence of M2-1. Thus, a low level of M2-1 might be sufficient to allow the polymerase to traverse a significant length of template generating monocistronic mRNAs but not permitting synthesis of readthrough mRNA. A higher level of M2-1 would augment the probability that the polymerase would be bound to M2-1 when it encountered the GE signal, facilitating readthrough mRNA synthesis.

There was no change in the gradient of gene expression or of the relative ratios of polycistronic mRNAs during RSV infection of HEp-2 cells, even though M2-1 protein levels increased significantly with time. This could be either because only a very low intracellular concentration of M2-1 is required for it to be saturating or because the ratio of M2-1 to other viral components (e.g., nucleocapsid) is the important factor and this ratio remains constant throughout infection. Thus, it appears that although both the intragenic antitermination and GE antitermination activities of M2-1 dramatically affect RSV gene expression, there is no temporal regulation due to M2-1 or any other factor, at least in a permissive cell line.