Diverse Gene Junctions of Respiratory Syncytial Virus Modulate the Efficiency of Transcription Termination and Respond Differently to M2-Mediated Antitermination

Abstract

The ability of the diverse gene junctions of respiratory syncytial (RS) virus to signal the termination of transcription was analyzed. Nine dicistronic subgenomic replicons of RS virus were constructed; each contained one of the RS virus gene junctions in its natural upstream and downstream sequence context. The RNA synthesis activities of these subgenomic replicons were analyzed in the absence and presence of the M2 protein, which we showed previously to function as a transcription antiterminator. Our data showed that the efficiency with which the polymerase terminated transcription was affected by the gene junction that it encountered. The M2 protein significantly decreased the efficiency of the termination of transcription, resulting in increased levels of readthrough transcription at all the gene junctions. The diverse gene junctions fell into three broad groups with respect to their ability to signal transcription termination. One group of gene junctions (NS1/NS2, NS2/N, M2/L, and L/trailer) showed inefficient termination in the absence or the presence of the M2 protein. A second group of gene junctions (N/P, P/M, M/SH, SH/G, and G/F) terminated transcription efficiently. The SH/G gene junction terminated transcription with the greatest efficiency and produced low levels of readthrough transcripts in the absence or the presence of the M2 protein, correlating with the absence of SH/G polycistronic transcripts in RS virus-infected cells. The F/M2 gene junction was particularly sensitive to the M2 protein: it efficiently signaled termination in the absence of the M2 protein but produced high levels of readthrough transcripts in the presence of the M2 protein. This result suggests that the M2 protein may regulate its own production by negative feedback. The data presented here show that the different gene junctions of RS virus do modulate RS virus transcription termination. The M2 protein reduced termination at all gene junctions. The magnitude of antitermination due to the M2 protein, however, varied at the different gene junctions. The data presented here indicate that the mechanism for the regulation of RS virus gene expression is more complex than was previously appreciated.

Human respiratory syncytial (RS) virus is the leading viral cause of lower respiratory tract disease in infants worldwide. It is the type species of the Pneumovirus genus of the family Paramyxoviridae. The single-stranded, negative-sense RNA genome is 15,222 nucleotides in length and tightly bound by the nucleocapsid (N) protein (16). During infection, 10 major subgenomic mRNAs are generated; each one encodes 1 of the 10 viral proteins detected in infected cells (7).

The viral components of the RNA-dependent RNA polymerase are the phosphoprotein (P) and the large polymerase (L) protein (24, 28). These proteins, in conjunction with the N protein-encapsidated genomic RNA, are sufficient to promote virus RNA replication and transcription in cells (14, 15, 28). A fourth viral protein, the M2 protein, has been reported to increase the processivity of the viral polymerase during transcription (6, 14). We recently demonstrated that the M2 protein of RS virus functions as a transcription antiterminator, causing the polymerase to fail to terminate at RS virus gene junctions, thus increasing readthrough transcription and the production of polycistronic mRNAs (15).

The RS virus genome has a single site at its 3′ end for entry of the polymerase during transcription (11). Transcription occurs in a sequential polar fashion from 3′ to 5′. The exact mechanism by which discrete mRNAs are generated is unknown. The favored model is that they are generated by a termination-reinitiation mechanism, whereby the polymerase releases the mRNA corresponding to the upstream gene and reinitiates transcription at the start of the downstream gene. In this model, a percentage of the polymerase molecules that terminate transcription of the upstream gene fail to reinitiate at the start of the downstream gene, resulting in a gradient of mRNAs corresponding inversely to the distance of the gene from the 3′ end of the genome (1).

The junction between RS virus genes consists of a semiconserved 12- or 13-nucleotide sequence at the end of the upstream gene and a highly conserved 9-nucleotide sequence at the start of the downstream gene. These sequences are separated by an intergenic region, variable in size and sequence, which is not represented in the mRNAs (5). From this point on, the term “gene junction” will be used to refer to the region encompassing the gene end, the intergenic region, and the gene start sequences. It has been demonstrated for the prototypical single-stranded, negative-sense RNA virus vesicular stomatitis virus that the sequences at the gene junctions play a critical role in the regulation of sequential transcription (2, 3, 17, 25). The gene end sequence and the base composition and size of the highly conserved intergenic region are critical for efficient termination of transcription. The conserved gene start sequence is required for efficient initiation of transcription (25).

Previously, each of the semiconserved gene end sequences of RS virus were analyzed for termination efficiency in constructs which isolated them from their natural surrounding sequences. These dicistronic constructs contained the virus gene end sequence at the end of the downstream gene immediately preceding the trailer region rather than their natural, variable intergenic region sequence. In this context, the gene end sequences were reported to signal the termination of transcription with equal efficiencies, with the exception of the NS1 and NS2 gene end sequences, which were less efficient terminators (18). Additionally, it should be noted that these studies were performed with helper RS virus to support RNA synthesis; thus, the effects of individual viral proteins on termination could not be assessed.

The nontranscribed intergenic regions of RS virus vary in both size and sequence from one gene junction to the next, ranging from 1 nucleotide (the N-P intergenic region) to 52 nucleotides (the G-F intergenic region [5]). In addition, the M2 and L genes overlap by 68 nucleotides, so that the L gene start sequence occurs upstream of the M2 gene end sequence. Analysis of transcription in RS virus-infected cells showed that mRNAs generated by readthrough transcription occurred for all of the gene junctions, except the SH/G junction (10). Also, data from RS virus in vitro transcription assays showed that different levels of transcriptional attenuation occurred at different gene junctions (1). These reports indicated that the variation in the gene junctions affected transcription. Another analysis of the RS virus diverse intergenic regions with a reconstituted system and cell cultures reported that the different sequences had no differential effects upon sequential transcription (19). However, that analysis examined the variant intergenic regions in the context of a single invariant gene end sequence (that of the N gene) rather than the natural, variable gene end sequences by which they are normally preceded in the RS virus genome.

The discrepancies among the data obtained in vitro, the observations in infected cells, and the data obtained with a reconstituted system in cell culture prompted us to examine the ability of the RS virus gene junctions to signal the termination of transcription in their natural local sequence context. This was done by constructing subgenomic replicons, each containing an intergenic region surrounded by its authentic gene end and gene start sequences as well as a minimum of 50 nucleotides of the authentic upstream and downstream gene sequences. The data presented in this paper describe the effect of each gene junction upon the termination of transcription in the natural local sequence context as well as the effects of the M2 protein on the efficiency of the termination of transcription at these diverse gene junctions. The roles of termination and readthrough transcription in the regulation of gene expression are discussed.

MATERIALS AND METHODS

cDNA constructs.

The construction of cDNAs for the expression of RS virus proteins and subgenomic replicon RNAs with a recombinant vaccinia virus T7 expression system was described previously (28). Sequences encoding the N, P, L, and M2 proteins (pN, pP, pL, and pORF1, respectively) were cloned behind a T7 RNA polymerase promoter in plasmid pGEM3 (15, 28). The generation of a cDNA encoding a subgenomic replicon of RS virus strain A-2, termed pWT, was previously described by Yu et al. (28). A single nucleotide change at position 4 of the leader sequence was introduced into pWT, as this change had been shown previously to increase RNA synthesis from subgenomic replicons (8). The resulting cDNA encodes an RNA analogous to the RS virus genome, the production of which is under the control of a T7 RNA polymerase promoter, and the 3′ end is generated by cleavage by a hepatitis delta virus ribozyme. This construct comprises the background into which each of the RS virus intergenic junctions was cloned.

Each intergenic junction was amplified by PCR from a cDNA containing the entire RS virus genome. In all cases (with the exceptions of the M/SH and F/M2 gene junctions), the 5′ (in the positive sense) PCR primer had an engineered terminal MunI site, and the 3′ primer had an engineered terminal BglII site. All PCR products were digested with MunI and BglII and ligated into MunI/BglII-cut pWT. For the clone containing the F/M2 gene junction, the 5′ PCR primer was designed with a 5′-terminal EcoRI site, as the F/M2 intergenic sequence contains a MunI site. In addition, two stop codons were engineered into the M2 sequence of this construct immediately following the first translation initiation codon of the M2 open reading frame. This was done to prevent M2 protein expression from the subgenomic replicon, which could complicate the interpretation of the effect of the M2 protein on transcription from this template. The construction of the clone containing the M/SH gene junction was described previously (15). In each case, the subgenomic replicon contained the leader sequence (44 nucleotides), the first 380 nucleotides of the NS1 gene, the intergenic junction and surrounding sequences described below, the final 67 nucleotides of the L gene, and the trailer sequence (154 nucleotides). The fragments of the genome cloned for each gene junction are as follows (nucleotide numbers are given 5′ to 3′ of the positive-sense replication product): pNS1/NS2, nucleotides 476 to 972; pNS2/N, nucleotides 1036 to 1664; pN/P, nucleotides 2177 to 2556; pP/M, nucleotides 2963 to 3649; pM/SH, nucleotides 3895 to 4497; pSH/G, nucleotides 4548 to 5226; pG/F, nucleotides 5367 to 6013; pF/M2, nucleotides 7437 to 8150; and pM2/L, nucleotides 8332 to 8812.

DNA transfections, radioactive labeling, and electrophoretic analysis of RNA.

RS virus-specific RNA synthesis from subgenomic replicons was assayed with a recombinant vaccinia virus T7 expression system. Transfections, labeling, and electrophoresis were performed as previously described (15). Briefly, HEp-2 cells were grown in minimum essential medium (GIBCO Laboratories) supplemented with 5% fetal bovine serum in 60-mm dishes. Cells infected with recombinant MVA vaccinia virus expressing T7 polymerase were transfected with cDNAs encoding subgenomic replicons and the trans-acting proteins required for RNA synthesis. In each case, cells were transfected with 6 μg of cDNA encoding a subgenomic replicon, 5 μg of pN, 2 μg of pP, 2 μg of pL and, where indicated, 0.3 μg of pORF1 (encoding the wild-type M2 protein). Transfection of a constant amount of pORF1 resulted in the synthesis of a constant amount of M2 protein. At 16 h posttransfection, RS virus-specific RNAs were labeled with [³H]uridine (33 μCi/ml) in the presence of actinomycin D (10 μg/ml) and cytosine arabinoside (50 μg/ml).

After a 5-h labeling period, cells were harvested, and cytoplasmic extracts were prepared as described previously (27). RNAs were purified by phenol extraction and recovered by ethanol precipitation.

RNase H analysis.

Digestion of RNAs in the presence of oligo(dT) was performed in order to remove poly(A) tails to optimize electrophoretic separation. Purified RNAs were resuspended in 20 μl of water and incubated with an equal volume of 2× RNase H buffer and 1 μg of oligo(dT) (18-mer) for 20 min at room temperature. Poly(A) tails were removed from RNAs by digestion with 1 U of RNase H and incubation at 37°C for 30 min. RNAs were precipitated by the addition of 150 μl of water, 5 μl of 4 M NaCl, 15 μg of yeast tRNA, and 500 μl of ethanol (3, 4). Precipitated RNAs were analyzed by electrophoresis in 1.75% agarose–urea gels and detected by fluorography (21, 26).

Quantitation of RNAs.

Fluorographs of RNA-containing gels were subjected to densitometric analysis with a Howtek Scanmaster 3 scanner and Pdi Quantity One software. Relative molar quantities of RNAs were calculated on the basis of uridine content. Levels of readthrough transcription at each gene junction were determined by calculating the quantity of discrete transcripts which were initiated at the first gene start signal and which failed to terminate at the first gene end signal as a percentage of the total quantity of discrete mRNAs which were initiated at the first gene start signal.

RESULTS

To examine the role of the diverse gene junctions in transcription, nine RS virus dicistronic subgenomic replicons were generated; each contained one of the RS virus gene junctions (Fig. 1) (see Materials and Methods for details). Each of the junctions was flanked by at least 50 nucleotides of the naturally occurring upstream and downstream gene sequences, allowing analysis of transcription across each junction in the local sequence context in which it occurs naturally in the virus genome (5).

FIG. 1.

Diagram of construction of cDNAs encoding subgenomic replicons containing the RS virus gene junctions. The nucleotides of the upstream and downstream genes and the sequence of each intergenic junction are shown. Each gene junction was cloned into a plasmid between the sequences of the 3′ end (leader [le] sequence and the first 380 nucleotides of the NS1 gene) and the 5′ end (last 67 nucleotides of the L gene and the trailer [tr] sequence) of the RS virus genome. The predicted products of RNA synthesis from a generic dicistronic subgenomic replicon are shown. Tφ, T7 RNA polymerase terminator; HδV, hepatitis delta virus ribozyme sequence; T7, T7 RNA polymerase promoter; ig, intergenic region; rep., replication product; r/t, readthrough transcript; A_n, poly(A) residues. Dots are inserted to maintain space alignment between conserved sequences.

The cDNAs encoding each of the dicistronic replicons were transfected into HEp-2 cells infected with recombinant MVA vaccinia virus expressing T7 RNA polymerase, and RNA replication and transcription were examined by cotransfection of cDNAs expressing the N, P, and L proteins as described previously (15, 28). With this system, the transcriptional activity of each of the replicons was analyzed in the absence and presence of the M2 protein, which was expressed from a plasmid containing ORF1 only of the M2 gene. RNAs were examined by direct metabolic labeling with [³H]uridine in the presence of actinomycin D and cytosine arabinoside and visualized by agarose-urea gel electrophoresis followed by fluorography (21, 26). The products of RNA synthesis from each subgenomic replicon in the absence and presence of the M2 protein are shown in Fig. 2. Each of the replicons was capable of directing the production of positive- and negative-sense replication products, a monocistronic upstream mRNA (mRNA1) and a monocistronic downstream mRNA (mRNA2). If termination failed to occur at the end of either or both genes, three readthrough transcripts could be produced: failure to terminate at the end of mRNA2 would lead to readthrough into the trailer sequence (r/t A), failure to terminate at the end of mRNA1 would lead to readthrough into mRNA2 (r/t B), and failure to terminate at the end of mRNA1 and mRNA2 would lead to the production of a transcript containing mRNA1, mRNA2, and the trailer sequence (r/t C) (Fig. 1). It should be noted that, due to the manner in which the gene junctions were cloned, the sizes of mRNA1 and mRNA2 varied for each of the subgenomic replicons, as described in Materials and Methods. The identity of each RNA species produced from each replicon was determined by RNase H analysis with oligonucleotides of known sequence which could anneal to either mRNA1 or mRNA2 (data not shown) (2, 4, 15).

FIG. 2.

RNA synthesis from dicistronic subgenomic replicons containing the different gene junctions of RS virus in the absence and presence of the M2 protein. HEp-2 cells infected with recombinant MVA vaccinia virus expressing T7 RNA polymerase were transfected with cDNAs encoding subgenomic replicons each containing one of the RS virus gene junctions, pN, pP, pL and, where indicated, pORF1, encoding the M2 protein. Cells were exposed to [³H]uridine in the presence of actinomycin D and cytosine arabinoside. Total RNA was digested with RNase H in the presence of oligo(dT) and analyzed by agarose-urea gel electrophoresis followed by fluorography. RNAs were synthesized from a subgenomic replicon containing the NS1/NS2 gene junction (A), a subgenomic replicon containing the NS2/N gene junction (B), a subgenomic replicon containing the N/P gene junction (C), a subgenomic replicon containing the P/M gene junction (D), a subgenomic replicon containing the M/SH gene junction (E), a subgenomic replicon containing the SH/G gene junction (F), a subgenomic replicon containing the G/F gene junction (G), a subgenomic replicon containing the F/M2 gene junction (H), and a subgenomic replicon containing the M2/L gene junction (I). rep., replication products; r/t, products of readthrough transcription. See the legend to Fig. 1 for details.

The RNAs were quantitated by densitometric analysis of fluorographs of the RNA gels, and the values were normalized for uridine content, as [³H]uridine was used to label the RNAs. To quantitate transcription termination at these junctions, the frequency of readthrough transcription was calculated by dividing the quantity of discrete transcripts initiating at the first gene start signal but failing to terminate at the first gene end signal (r/t B + r/t C) by the total quantity of transcripts initiating at the first gene start signal (mRNA1 + r/t B + r/t C) (Fig. 1). Thus, the percentage of readthrough is represented by 100 × [(r/t B + r/t C)/(mRNA1 + r/t B + r/t C)]. Experiments with each subgenomic replicon were repeated a minimum of three times; the averaged data are represented graphically in Fig. 3.

FIG. 3.

Effects of the RS virus gene junctions on the termination of transcription programmed by RS virus subgenomic replicons. The percent molar abundance of transcripts failing to terminate (% r/t [y axis]) at the specified RS virus gene junction (x axis) was calculated in the presence and absence of the M2 protein by densitometric analysis of fluorographs generated from at least three separate experiments and is graphically represented. The standard deviation calculated for each plotted quantity is represented by an error bar. tr, trailer.

Our results show that the gene junctions varied in their efficiency of transcription termination in both the absence and the presence of the M2 protein. The diverse gene junctions were found to fall into three broad groups with respect to transcription termination efficiency: those which signaled termination efficiently (N/P, P/M, M/SH, SH/G, and G/F), those which signaled termination inefficiently (NS1/NS2, NS2/N, M2/L, and L/tr), and that for which termination efficiency was highly dependent on the absence or presence of the M2 protein (F/M2). As shown in Fig. 3, the presence of the M2 protein decreased termination and increased the frequency of readthrough transcription at each gene junction, in agreement with our previous report of M2 protein activity at the M/SH gene junction (15).

The SH/G gene junction showed the most efficient termination and lowest readthrough frequency of all the gene junctions (Fig. 2F and Fig. 3). Readthrough at the SH/G junction was increased approximately fourfold in the presence of the M2 protein, but the readthrough frequency was still only 6% (Fig. 3). This result is consistent with the previous finding that a dicistronic transcript corresponding to SH and G mRNAs was never found in RS virus-infected cells, although polycistronic transcripts involving all the other gene junctions were found (7).

The N/P, P/M, M/SH, and G/F gene junctions all exhibited efficient termination in the absence of the M2 protein, as shown by the low frequency of readthrough transcription (<5%) (Fig. 2C, D, E, and G and Fig. 3). In the presence of the M2 protein, readthrough at each of these junctions was enhanced 6- to 10-fold.

Termination of transcription was inefficient at the NS1/NS2 and NS2/N gene junctions in the absence of the M2 protein, as shown by high levels of readthrough transcripts (27 and 10%, respectively) (Fig. 2A and B and Fig. 3). The efficiency of termination was further decreased and readthrough transcription was enhanced three- to sevenfold when the M2 protein was present. The propensity for readthrough at these gene junctions correlates with a previous report showing that the NS1 and NS2 gene end signals were less efficient transcription terminators than the other RS virus gene end signals (18).

The M2/L gene junction is unusual in that the start of the L gene precedes the end of the M2 gene so that the two transcription units overlap by 68 nucleotides (5). Previously, Collins et al. demonstrated that the production of full-length mRNAs corresponding to the L gene was lower than expected due to this overlap (9). This result was postulated to occur due to premature termination of the L mRNA as the polymerase recognized the M2 gene end signal, thus producing a 64-nucleotide polyadenylated mRNA. Data in Fig. 2 and 3 show that transcription terminated poorly at this junction in the absence and presence of the M2 protein, yielding high levels of readthrough transcripts in both cases. In the absence of the M2 protein, approximately 20% of the polymerase failed to terminate and read through the junction. When the M2 protein was present, readthrough occurred approximately 85% of the time.

The ability of the F/M2 gene junction to signal transcription termination was highly dependent upon the absence or presence of the M2 protein. In the absence of the M2 protein, readthrough transcription occurred at a low frequency equivalent to that seen for the N/P, P/M, M/SH, and G/F gene junctions (Fig. 3). However, the addition of the M2 protein led to a 15-fold increase in readthrough transcription, rather than the 6- to 10-fold increase seen for the aforementioned junctions. This result also differed from results obtained for the junctions which terminated poorly in the presence of the M2 protein (NS1/NS2, NS2/N, and M2/L junctions), as these junctions allowed a significant amount of readthrough in the absence of the M2 protein. The F/M2 junction terminated transcription efficiently in the absence of the M2 protein but not in the presence of the M2 protein and was particularly sensitive to the presence of the M2 protein.

The frequency of readthrough transcription at the L/trailer junction, present in all of the dicistronic replicons, was compared for all the constructs. Readthrough transcription frequency was calculated by dividing the quantity of transcripts which contained the trailer sequence by the total quantity of transcripts containing the mRNA2 sequence. Thus, the percentage of readthrough is represented by 100 × [(r/t A + r/t C)/(mRNA2 + r/t A + r/t B + r/t C)]. The L/trailer junction terminated transcription inefficiently, allowing readthrough transcription to occur 25% of the time, in the absence of the M2 protein (Fig. 3). In the presence of the M2 protein, termination efficiency decreased less than twofold, showing that this junction was less sensitive to the effect of the M2 protein than the others. This junction behaved the same way in all of the subgenomic replicons.

DISCUSSION

The gene junctions of nonsegmented negative-sense RNA viruses are comprised of a gene end sequence, required to signal the termination of transcription; an intergenic region not found in the monocistronic mRNAs; and a gene start sequence, which signals transcription initiation. Both the gene end and the intergenic sequences of RS virus vary from gene to gene (5).

We examined the transcription termination efficiency of each gene junction in its natural sequence context in the presence and absence of the M2 protein. The data presented here show (i) that the individual gene junctions of RS virus modulate the efficiency of transcription termination, (ii) that the M2 protein increases the frequency of readthrough transcription at all RS virus gene junctions, but (iii) that the junctions vary in their sensitivity to the presence of the M2 protein.

Previous work indicated that the M2 protein increases the processivity of the RS virus polymerase (6, 14). This finding is consistent with the action of other transcription antiterminators, which also increase the processivity of the polymerase in order to facilitate readthrough transcription (13). However, the readthrough transcription observed in the presence of the M2 protein is not simply a function of increased processivity allowing more polymerase molecules to reach the end of a gene. We found no correlation between the size of the upstream gene and the frequency of readthrough transcription in either the presence or the absence of the M2 protein, arguing against a simple nonspecific effect of processivity. In fact, it is apparent that the polymerase can transcribe the length of the genome in the absence of the M2 protein (as shown by significant levels of mRNA1 and mRNA2) but terminates less efficiently when the M2 protein is present (Fig. 2, particularly panels A, D, and G).

The SH/G gene junction was the most efficient at terminating transcription. This result is consistent with the observation that no polycistronic mRNAs containing SH and G mRNA sequences were detected in RS virus-infected cells (10). Assuming that the downstream cistrons of a polycistronic transcript would be less efficiently translated, the termination characteristics of the SH/G gene junction might suggest that the ratio of SH protein to G protein must be tightly controlled, perhaps maximizing the amount of G protein by preventing readthrough transcription.

Termination at the F/M2 gene junction was most sensitive to the absence or presence of the M2 protein. In the absence of the M2 protein, this junction was an efficient terminator; however, when the M2 protein was present, the highest (approximately 15-fold) increase in readthrough transcription occurred. These results may indicate that the M2 protein can regulate its own production. We hypothesize that when the levels of the M2 protein are low, the polymerase terminates at the end of the F protein gene, allowing the production of monocistronic M2 protein mRNAs, which can be efficiently translated. When M2 protein levels are high, the polymerase can read through the F/M2 junction, thus decreasing M2 protein production by transcribing the M2 protein sequence as the downstream cistron of a polycistronic mRNA and therefore reducing its ability to be translated (20). This model describes a negative feedback mode of regulation and suggests a higher order of gene regulation in RS virus than was previously observed.

Overall, the findings presented here on the efficiency of termination at the variant gene junctions of RS virus agree with previously published work in which Northern blot analysis was used to map the order of genes in the RS virus genome by the detection of readthrough transcripts (7, 10). In these studies, readthrough transcripts corresponding to NS1/NS2, NS2/N, and F/M2 were abundant, whereas a readthrough transcript containing SH and G mRNA sequences was undetectable.

The M2/L gene junction is unusual in its organization in that the downstream gene start precedes the upstream gene end; thus, the two transcription units overlap (5). This junction allowed a relatively high level of readthrough transcription (20%) in the absence of the M2 protein; the level was increased fourfold in the presence of the M2 protein. The production of mRNA encoding the polymerase protein in all nonsegmented negative-sense RNA viruses is strongly down-regulated. It was reported that the overlap at this junction caused premature termination of the L protein mRNA and, thus, led to attenuation of L protein production (9). The high levels of readthrough transcription that we observed at this junction might also contribute to the attenuation of polymerase production. It is unclear whether the overlap had an effect on readthrough transcription or whether the gene end signal and surrounding sequences caused the increase in readthrough transcription.

We demonstrated previously that the M2 protein reduced the efficiency of transcription termination, thus enhancing readthrough transcription and the production of polycistronic mRNAs (15). The results presented here show that the M2 protein enhanced readthrough transcription at all of the variable RS virus gene junctions but that it did so to different extents. Our hypothesis for why the M2 protein should do this was that readthrough transcription would, to some extent, override transcriptional attenuation, which could severely down-regulate the expression of promoter-distal genes in a virus such as RS virus, which has 10 genes and a single polymerase entry site. By increasing readthrough transcription at certain gene junctions, the polymerase could gain deeper access into the genome. Variation in susceptibility to readthrough for the different gene junctions implies that specific regulation of transcription termination occurs at each gene junction. The efficiency of translation of the second cistron in a polycistronic transcript should be greatly reduced; thus, readthrough transcription will lead to an overall decrease in the expression of the product of the downstream open reading frame of a polycistronic mRNA (20). These findings add another facet to our model for the function of the M2 protein. While readthrough transcription would allow more polymerase molecules to access 3′-distal genes, it would also down-regulate the expression of proteins from cistrons which are not 5′ most in polycistronic mRNAs. Therefore, the gene junctions appear to locally regulate the ratio of upstream to downstream gene products. From previous studies of RS virus and other nonsegmented negative-sense RNA viruses, it is known that the ratios of certain virus proteins are of critical importance for processes during virus replication (12, 15, 22, 23, 28). We therefore propose that the levels of M2 protein-mediated readthrough transcription allowed by the RS virus gene junctions are important for maintaining the correct ratios of the different viral proteins in order to facilitate efficient viral replication.

The fact that both the gene end signals and intergenic regions vary at each gene junction makes it difficult to deduce the signals involved in regulating termination efficiency. It is probable that certain gene end signals work in conjunction with certain intergenic sequences in order to terminate transcription with a specific efficiency. We noted that four of the five junctions with the greatest propensity for readthrough transcription (NS1/NS2, NS2/N, F/M2, and M2/L) possess only four uridines in the uridine tract of the gene end sequence, whereas the junctions at which termination occurred more efficiently have at least five uridines in the uridine tract. Barr et al. found that the efficiency of the termination of transcription for vesicular stomatitis virus was highly sensitive to the length of the uridine tract: removal of even one uridine from the highly conserved run of seven uridines at the gene ends led to a suppression of transcription termination (2).

In summary, the results presented here demonstrate that the intact natural RS virus gene junctions modulate the efficiency of the termination of transcription. The effects of different gene end signals on transcription termination were examined previously with a constant, heterologous sequence context, and the efficiencies of termination were reported to be the same for all gene ends, except those of NS1 and NS2, which terminated approximately 40% less efficiently (18). Diverse intergenic regions were also analyzed, again in a constant sequence context which was quite different from that in which these regions are found in the virus genome. Under these conditions, the diverse intergenic regions were found to have no effect on the termination of transcription (19). Our results differ from these previous studies and show that the sequence context of the individual elements of the natural RS virus gene junctions affect the manner in which they function. The effects of variable gene end sequences and interplay with variable intergenic sequences are currently under investigation. In addition, the M2 protein affects termination at different gene junctions to different extents. Our findings suggest that the mechanism for the regulation of RS virus gene expression is more complex than the traditional model of gene regulation in nonsegmented negative-sense RNA viruses, in which levels of gene expression are determined by the location of the gene on the genome with respect to the 3′ polymerase entry site.