INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Respiratory syncytial virus (RSV) is a nonsegmented negative-strand RNA virus in the Pneumovirus genus of the Paramyxoviridae family (6, 18). RSV is a major cause of serious respiratory disease in infants and adults and is a major target for vaccine and antiviral drug development. As is typical for the nonsegmented negative-strand RNA viruses, RSV genomic RNA is associated tightly with the nucleocapsid N protein to form an RNase-resistant helical nucleocapsid. This encapsidated genomic RNA is the template used by the viral polymerase to synthesize the positive-sense RNAs, namely, the 10 subgenomic viral mRNAs and the antigenomic RNA. The antigenome is a complete positive-sense copy of the genome. It is an intermediate in RNA replication and, like the genome, is encapsidated with N protein. In addition to the N protein, the nucleocapsid-associated proteins include the large L protein and the phosphoprotein P. The L protein contains conserved polymerase motifs and likely directs RNA synthetic functions as well as capping and methylation. The P protein appears to serve both as a polymerase cofactor and as a chaperone that keeps free N protein soluble and available for assembly with nascent genomic or antigenomic RNA (16). In contrast to other nonsegmented negative-strand viruses, RSV mRNA synthesis involves an additional viral protein, the M2-1 protein, which confers transcriptional processivity and increases the frequency of polymerase readthrough across intergenic junctions (7, 8, 10, 14).

The polymerase can engage in transcription, producing subgenomic mRNAs, or in RNA replication, a two-step process producing in turn antigenomic RNA and progeny genomic RNA (6, 18). To initiate either of these processes, the polymerase is presumed to bind at a genomic promoter at the 3' end of the genomic template. Transcription involves a stop-start mechanism guided by conserved signals at the gene boundaries. Specifically, each gene begins with a 10-nucleotide gene start (GS) signal, which directs transcriptional initiation, and ends with a 12- to 13-nucleotide gene end (GE) signal, which directs polyadenylation and release of the completed transcript (16, 17). The polymerase proceeds down the genome, transcribing each gene in turn. The same template, and ostensibly the same promoter, is used for the synthesis of the antigenome, the first step in RNA replication, but the GS and GE signals are ignored. In addition, it is thought that nascent replication products are encapsidated cosynthetically. The factors which determine whether the genomic template engages in transcription versus replication are not well understood. One proposal for nonsegmented negative-strand RNA viruses in general is that there is a balance between replication and transcription which is mediated by encapsidation of the nascent antigenomic RNA (18). However, at least in the case of RSV, the proposed switch to replication at the expense of transcription could not be reproduced in a minigenome system by overexpression of the N and P proteins (11).

There is a high degree of sequence identity between the 3' ends of genomic RNA and antigenomic RNA: specifically, 10 of the first 11 nucleotides (nt) and 21 of the first 26 nt are identical, after which the amount of sequence identity is insignificant (7, 20). It seems likely that this high degree of sequence identity is due to a conserved 3' promoter structure. It might also reflect the presence of a conserved encapsidation signal at the 5' end of the encoded antigenomic or genomic RNA. As a consequence of these conserved sequences, genomic and antigenomic RNA each has complementarity between its 3' and 5' ends, and thus each has the potential to form a panhandle structure. It has been suggested that base pairing between the 3' and 5' ends of the genome of the rhabdovirus vesicular stomatitis virus (VSV) is important in RNA replication (27), although other studies with the paramyxovirus Sendai virus appear to rule out a role for terminal complementarity (26).

In the present study, we used an RSV minigenome system to study the roles of the conserved terminal sequences in encapsidation, RNA replication, and transcription. We focused on the seven terminal nucleotides of the 5' trailer region, which are complementary with the leader except at position 4. This sequence might be part of a possible encapsidation signal in the genomic RNA, and its complement at the 3' end of the antigenome is likely to be part of the antigenomic promoter. We examined mutations at each of the seven positions to determine their importance in encapsidation and RNA synthesis. We also examined deletions of part or all of the trailer region. None of the point mutations affected encapsidation of the genome produced by T7 polymerase or its function as a template for producing antigenome and mRNA, but mutations at four of these positions blocked the accumulation of progeny genome, while mutations at three other positions had an intermediate effect or no inhibitory effect. The trailer deletions did not affect encapsidation or the template function of genome produced by T7 polymerase, providing evidence that neither transcription nor replication involves terminal complementarity. We also describe the importance of carefully controlling for effects on the magnitude of T7-mediated transcription of the genome plasmid.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and transfection. HEp-2 cells were maintained in Opti-MEM (Life Technologies, Inc., Gaithersburg, Md.) containing 2% fetal bovine serum (FBS) and tested every 2 months to confirm the absence of mycoplasma. Transfections were performed in 35-mm wells of HEp-2 cells with a mixture of pTM1-N (0.4 µg), pTM1-P (0.2 µg), pTM1-L (0.1 µg), and minigenome plasmid (0.2 µg), as well as 12 µl of LipofectACE (Life Technologies, Inc.), unless otherwise stated (13). This mix (0.2 ml) was incubated in the absence of serum for 45 min at room temperature before the addition of 0.8 ml of Opti-MEM-2% FBS that included vTF7-3, a recombinant vaccinia virus that expresses T7 polymerase (12). Cells were transfected overnight in the absence of antibiotics, and medium was replaced the next morning with Opti-MEM-2% FBS. Transfected cells were harvested at 44 h and RNA extracted with Trizol (Life Technologies, Inc.). To analyze encapsidated RNA, cells were resuspended in 10 mM Tris (pH 7.5)-10 mM NaCl-1 mM CaCl₂-1% Triton X-100-0.5% sodium deoxycholate containing 1 µg of micrococcal nuclease (S7 nuclease; Boehringer Mannheim GmbH, Mannheim, Germany) per ml and gently vortexed, incubated at 30°C for 30 min (1, 21), and extracted with Trizol.

Northern blotting. RNA was electrophoresed on 1.5% agarose gels containing 0.44 M formaldehyde, transferred to Nytran Plus by using a Turbo Blotter (Schleicher & Schuell, Keene, N.H.) and 8 mM NaOH as buffer, and fixed by UV cross-linking (Stratalinker; Stratagene). ³²P-labeled RNA probes were synthesized in vitro by T7 RNA polymerase (Boehringer Mannheim) in the presence of [³²P]CTP (Amersham Life Science, Inc., Arlington Heights, Ill.) and incubated with the blots, as previously described (12). Positive-sense probe was synthesized from antigenome C4 plasmid which was linearized at an NcoI site located in the downstream region of the chloramphenicol acetyltransferase (CAT) gene, resulting in a transcript of approximately 600 nt which included the upstream (leader) end of the antigenome and most of the CAT open reading frame (ORF) but lacked the downstream (trailer) end (12). Negative-sense probe was synthesized from minigenome C2 plasmid which was linearized at the XbaI site marking the upstream end of the CAT gene, resulting in a transcript of approximately 852 nt which included the complete CAT ORF and the downstream (trailer) end of the genome (12). The blots were exposed to film (BioMax Film; Kodak, Rochester, N.Y.) with an enhancing screen at 70°C and to phosphor screens for quantitation in a STORM phosphorimager (Molecular Dynamics, Sunnyvale, Calif.).

Plasmid mutagenesis and isolation. Genome trailer mutants 3A, 4G, 4C, 5A, and 7C were prepared by PCR amplification of a PstI-HindIII fragment which begins at the end of the CAT gene in the minigenome C2 cDNA and spans the trailer region and T7 promoter. Amplification was done with a positive-sense primer which hybridized within the CAT gene and a mutagenic negative-sense primer which hybridized to the T7 promoter and the end of the trailer region and contained the desired mutation. The PCR product was cloned into the PstI-HindIII window of minigenome plasmid C2. Mutants 1C, 2G, and 6C were prepared by the method of Byrappa et al. (3). Briefly, two primers were used, one of which contained the desired mutation, abutting each other but facing in the opposite direction on the genome plasmid, C2. These primers were used to PCR amplify the plasmid with Vent polymerase (New England Biolabs, Inc.) for 18 cycles. The tube was immediately removed from the thermocycler, and EDTA was added to stop the reaction. The product was isolated by agarose gel electrophoresis, ligated, and used to transform competent Escherichia coli DH10B cells. All mutations were confirmed by nucleotide sequence analysis.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effect of trailer mutations on T7 transcription. We examined the effects of nucleotide substitutions in the last seven positions of the 5'-terminal trailer region of RSV. This was done with a reconstituted transcription-replication system based on the negative-sense RSV-CAT minigenome C2 (13), as outlined in Fig. 1. The cDNA is bordered on the 5' end relative to the encoded genome by the core promoter for T7 RNA polymerase. There are three nonviral G residues between the trailer and the core promoter in order to increase promoter efficiency. The cDNA is bordered on the 3' end by a hammerhead ribozyme and a T7 terminator. Ribozyme-mediated cleavage would generate the 3' end of the genome. T7 RNA polymerase is provided by infection with the recombinant vaccinia virus, vTF7-3 (12).

View larger version (37K): [in this window] [in a new window]   FIG. 1.   Steps involved in reconstituted encapsidation, RNA replication, and transcription of the wt RSV-CAT C2 minigenome and derivatives containing nucleotide substitutions. The cDNA-encoded genome is 934 nt long and contains, in 3'-to-5' order, the 44-nt RSV leader region, the 10-nt NS1 GS signal, the upstream 32 nt of the nontranslated region of the NS1 gene, a 669-nt negative-sense copy of the CAT ORF, the last 12 nucleotides of the nontranslated region of the L gene, the 12-nucleotide L GE signal, and the 155-nucleotide trailer region. (Step 1) Naked minigenomic RNA is synthesized from transfected plasmid by T7 RNA polymerase supplied by the coinfecting vaccinia virus recombinant, vTF7-3. Self-cleavage by the ribozyme is indicated with an open triangle. Other cotransfected plasmids bearing individual RSV ORFs (not shown) are transcribed by T7 polymerase and then translated into support proteins. (Step 2) Soluble N and P support proteins (open circles and filled squares, respectively) encapsidate the naked RNA, rendering it resistant to digestion with micrococcal nuclease. This step is very inefficient, probably because it is not associated with the RSV polymerase (see Discussion). (Step 3) This step illustrates the 7C mutation in the genome trailer, in which position 7 relative to the 5' end of the genome is changed from A to C (negative sense). The three nonviral G residues present at the 5' end of the T7 transcript are not shown and, in any event, do not interfere with replication either of minigenomes or of complete recombinant virus (7, 24). (Step 4) The encapsidated plasmid-supplied genome (parental or mutant) serves as template for the synthesis of antigenome (RNA replication) and CAT mRNA (transcription) by the reconstituted RSV polymerase. The antigenome is encapsidated efficiently, probably concurrent with synthesis. (Step 5) The 7C trailer mutation at the 5' end of the trailer, from step 3 above, results in a complementary substitution in the 3' end of the encoded antigenome. (Step 6) The second step of RNA replication, the synthesis of genome, is blocked in the 7C mutant (Results). (Step 7) This illustrates a second point mutation, at leader position 7, which was introduced into the 7C trailer mutant in order to restore terminal complementarity at position 7. (Step 8) The 7G leader mutation results in a complementary mutation in the 5' end of the encoded antigenome.

View larger version (8K): [in this window] [in a new window]   FIG. 2.   Mutations introduced into the 5' trailer region of the wt C2 minigenome and effect on the intracellular accumulation of T7 transcript. The 5'-terminal 7 nt of the C2 genome are shown as a negative-sense sequence, 3' to 5'. The various single point mutations which were introduced are shown. Those listed above the wt sequence are ones which either did not affect or increased the level of intracellular T7 transcript. Mutations listed below are ones which reduced the level of T7 transcript.

View larger version (30K): [in this window] [in a new window]   FIG. 3.   Effects of individual point mutations in the minigenome trailer region on T7 transcription, genome encapsidation, synthesis of positive-sense mRNA and antigenome, and synthesis of negative-sense progeny genome. Duplicate wells of HEp-2 cells were infected with vTF7-3 to provide T7 RNA polymerase and transfected (13) with wt minigenome plasmid C2 (lanes 1 to 3) or mutant genome plasmid 1C (lanes 4 to 6), 2G (lanes 7 to 9), 3A (lanes 10 to 12), 4G (lanes 13 to 15), 4C (lanes 16 to 18), 5A (lanes 19 to 21), 6C (lanes 22 to 24), or 7C (lanes 25 to 27). Lanes marked "0" received no additional plasmids (lanes 1, 4, 7, 10, 13, 16, 19, 22, and 25). Lanes marked "NPL" received N, P, and L support plasmids, the latter encoding a nonfunctional L protein (lanes 2, 5, 8, 11, 14, 17, 20, 23, and 26). Lanes marked "NPL" received N, P, and functional L plasmids (lanes 3, 6, 9, 12, 15, 18, 21, 24, and 27). RNAs were harvested at 44 h, and lysates were prepared. One lysate of each pair was processed directly for RNA purification, and the resulting total intracellular RNA was subjected to electrophoresis on formaldehyde gels and analyzed by Northern blot hybridization with positive-sense (A) or negative-sense (C) riboprobe. The other lysate of each pair was subjected to digestion with micrococcal nuclease (N'ase), and the remaining RNA was purified and analyzed by Northern blot hybridization with positive-sense riboprobe (B). Film exposure for panels B and C was 12 h, and for panel A it was 1.5 h. Quantitation of the hybridization patterns shown in this figure are presented in Table 1.

Effect of trailer mutations on the encapsidation of the T7 genome transcript. We then investigated whether any of the nucleotide substitutions in the plasmid-derived mutant genomes affected encapsidation, namely, their assembly into nucleocapsids (Fig. 1, step 2). This was done in the experiment shown in Fig. 3 by measuring the amount of micrococcal nuclease-resistant genome synthesized in cells receiving genome plasmid alone or receiving in addition the N, P, and L support plasmids, the latter encoding a nonfunctional L protein. This mix supplied the proteins (N and P) necessary for encapsidation without reconstituting the viral polymerase. When no support plasmids were provided, genomic RNA would not be encapsidated and should be sensitive to degradation by micrococcal nuclease (Fig. 3, lanes "0" in panel B versus panel A, and note that the exposure time for panel B was eight times that of panel A; also, see Table 1), although as noted below there sometimes was a background of residual RNA due to incomplete nuclease digestion. When the N and P support plasmids were provided, a band of protected genome was evident for the wt genome and for each mutant (Fig. 3B, lanes marked NPL). This species was quite abundant for each genome except for the 4C mutant (Fig. 3B, lane 17). The reduced accumulation of protected genome in that case reflects the reduced efficiency of T7-expressed, plasmid-derived genome (Fig. 3A, lane 16), as described above. Thus, each genome retained the capacity to be assembled into a nucleocapsid by N and P proteins supplied in trans (Fig. 3B, lanes NPL). The proportion of T7 RNA polymerase-produced genome that was encapsidated relative to the amount of T7 transcript containing the ribozyme varied from 10 to 41% (Table 1, column NPL). It should be noted that these values do not exactly reflect the efficiency of encapsidation of the plasmid-derived RNA. This is because the T7 transcript containing uncleaved ribozyme represents only a fraction of the total plasmid-derived material, much of which was likely degraded. Hence, it serves as a surrogate marker. In addition, the apparent differences in encapsidation efficiency between certain minigenomes were not consistent between experiments and thus represent experimental variability rather than significant differences.

Effect of trailer mutations on the production of progeny antigenome. If plasmid-derived genomic RNA is encapsidated properly, the genomic promoter at its 3' end (Fig. 1, step 3) should be fully functional, since this end was not subjected to mutagenesis. However, there is the possibility that RNA synthesis is influenced by interaction between the genomic termini, as suggested for VSV (27). To investigate the ability of each genome to form a functional template for the synthesis of positive-sense RNA (Fig. 1, step 4), the total intracellular RNA from cells infected and transfected as described above was analyzed by Northern blot by using a negative-sense riboprobe (Fig. 3C, note that the length of film exposure for panel C was the same as for panel B and eight times longer than for panel A). As would be expected, positive-sense RNAs were not synthesized either in the absence of support plasmids or when N and P were provided together with nonfunctional L (for example, Fig. 3C, lanes 1 and 2). When N, P, and functional L were provided, the genomic template was copied into a discrete band of positive-sense antigenomic RNA and a smear of subgenomic mRNA (Fig. 3C, lane 3). The mRNA appears as a smear because it contains a high proportion of prematurely terminated species due to the absence of the M2-1 antitermination factor (8, 10). The M2-1 protein was specifically omitted here because otherwise the intense band of full-length mRNA obscured the antigenomic RNA. Analysis of positive-sense RNAs from lysates treated with micrococcal nuclease confirmed that the antigenome was completely protected, whereas most of the mRNA was degraded (not shown).

Effect of the trailer mutations on the function of the antigenomic promoter. Once antigenomic RNA is synthesized and encapsidated, it should be able to direct the production of progeny genome by the RSV polymerase (Fig. 1, step 6). This is clearly evident for the wt genome by examination of total negative-strand RNA (Fig. 3A, compare lane 2 to lane 3). The correctly sized genome migrates more rapidly than the T7 transcript genome-ribozyme, and an increase in its amount in the presence of the complete polymerase is indicative of RNA replication. The amount of genome-sized RNA produced by T7 polymerase (Fig. 3A, lane 2) was barely detectable, while the amount of genome that accumulated in the presence of the L protein was as prominent as the T7 transcript (lane 3). A similar pattern of amplification was observed with mutants 4G and 4C (lanes 15 and 18), though in the case of 4C the much-lower level of amplified genome reflected the much-lower level of T7 transcript, as described above. Mutants 1C and 5A exhibited some genome amplification, though the amount produced was less than the T7 transcript (lanes 6 and 21), indicating a decrease in the efficiency of replication.

Comparison of genome template activities. We estimated the relative efficiency of each genome template for the synthesis of positive-sense RNA. This was done by dividing the total amount of positive-sense RNA product for each genome (Fig. 3C, lanes 3, 6, 9, 12, 15, 18, 21, 24, and 27; Table 1, column "total positive-sense RNA") by the corresponding amount of encapsidated genome available as template (Fig. 3B, lanes 3, 6, 9, 12, 15, 18, 21, 24, and 27; Table 1, column "NPL"). The values for positive- and negative-sense RNA came from different blots and cannot be compared directly, but the quotient obtained for each genome provided an estimate in arbitrary units of relative template activity that can be compared between genomes (Table 1, column "Relative genome template activity"). The values were very similar, varying by 2.2-fold or less, and thus the wt and various mutant genomes were very similar with regard to the ability to serve as template for the synthesis of positive-sense RNA.

Effects of double and triple trailer mutants. We also prepared mutant genomes that combined the 3A, 5A, and 7C mutations in pairs or altogether (Fig. 4). The various mutant genomic RNAs produced by T7 polymerase were encapsidated (Fig. 4B, compare lanes 4 and 5, 7 and 8, 10 and 11, 13 and 14, and 16 and 17; note that the exposure time was identical for all panels). The encapsidated mutant genomes functioned as template for the synthesis of both antigenome and mRNA (Fig. 4C, lanes 6, 9, 12, 15, and 18), and the antigenome was encapsidated (Fig. 4D, lanes 6, 9, 12, 15, and 18). In addition, none of the mutant genomes was amplified detectably in the presence of the replicase, as judged by comparing the amount of encapsidated genome present in the presence of N and P versus N, P, and L (Fig. 4B, compare lanes 5 and 6, 8 and 9, 11 and 12, 14 and 15, and 17 and 18). In the particular experiment shown in Fig. 4, the amount of encapsidated genome which accumulated in the presence of N, P and L for the double mutant 3A 5A (panel B, lane 15) was marginally greater than the amount which accumulated in the presence of N and P (lane 14), but this was not observed in repeat experiments and thus represents experimental variability, which is not unexpected in experiments involving multiple samples and nuclease treatments. In summary, the double and triple trailer mutants behaved like the individual mutants. In particular, the presence of the various combinations of two or three trailer mutations did not affect the encapsidation and template function of genomic RNA.

View larger version (50K): [in this window] [in a new window]   FIG. 4.   Effects of multiple point mutations in the minigenome trailer region on T7 transcription, genome encapsidation, synthesis of positive-sense mRNA and antigenome, and synthesis of negative-sense progeny genome. The trailer mutations 3A, 5A, and 7C were inserted individually (3A and 7C), in pairs (3A, 7C and 3A, 5A), and as a triplet (3A, 5A, 7C) into the C2 minigenome. The experiment was performed basically as described in the legend for Fig. 3 except that the nonfunctional L plasmid was not included in the N and P plasmid condition (lanes 2, 5, 8, 11, and 14). Results from the individual mutations and from the combined mutant are presented.

Effect of restoring terminal complementarity on RNA replication. The point mutations at positions 1 to 3 and 5 to 7 reduce the extent of terminal complementarity of both genomic and antigenomic RNA. Previously, terminal complementarity was shown to enhance the replication of VSV genomes (27). Thus, disruption of complementarity by the point mutants described here might be a factor in the inhibition of genome synthesis. Since all of the genomes synthesized antigenomic RNA, and in the same proportion to mRNA as wt (Fig. 3C), the disruption of terminal complementarity apparently did not affect the ability of the genome to function as template. However, it was possible that complementarity is more important for the function of the antigenome template.

View larger version (57K): [in this window] [in a new window]   FIG. 5.   Restoration of complementarity between the genome ends at the position 7 point mutation. Trailer mutant 7C (lanes 4 to 6) is compared with a mutant containing both the trailer 7C mutation and a compensatory mutation in the leader, 7G (lanes 7 to 9). This second mutation restores the complementarity of the ends to the wt level. A third mutant with only the leader 7G mutation was also tested (lanes 10 to 12). Each of these mutants and the wt minigenome, C41, contains a delta ribozyme instead of the hammerhead ribozyme used in genomes described in the previous experiments. Transfections were performed in HEp-2 cells as described in Fig. 3. Blots in panels A, B, and C were exposed to film for 2 h. Panel D is the same blot as panel C, but it was exposed to film for 6 h to confirm the absence of antigenome in lanes 9 and 12.

Effect of large trailer deletions on genome encapsidation and template function. As described above, single substitutions in the last 7 nt of genomic RNA, or double or triple mutations at positions 3, 5 and 7, had no discernible effect on its encapsidation or template function. We therefore examined the effect of two large terminal deletions. In mutant 24, the last 24 nt of the 155-nt trailer region were deleted, whereas mutant 325 sustained deletion of the complete trailer region and the adjacent GE signal of the CAT transcription cassette. The sequence (negative sense) at the 5' terminus of each deletion genome is 3'-UGCUCUAUAA-5' for 24 and 3'-AAAGUGGUAC-5' for 325 compared to 3'-AAAAAGAGCA-5' for the wt genome (mutant nucleotide assignments which are identical to wt are underlined). Both of these genomes were cleaved with the hepatitis delta virus ribozyme. The two mutants were examined in parallel with wt genome C41 and a version of trailer mutant 2G, both of which also employ the hepatitis delta virus ribozyme.

View larger version (69K): [in this window] [in a new window]   FIG. 6.   Effects of deletion at the 5' genome trailer region on T7 transcription, genome encapsidation, synthesis of positive-sense mRNA and antigenome, and synthesis of negative-sense progeny genome. HEp-2 cells were infected with vTF7-3 and transfected with wt minigenome plasmid C41 (wt, lanes 1 to 4) or mutant minigenome plasmids 24 (lanes 5 to 8), 325 (lanes 9 to 12), or trailer 2G (lanes 13 to 16). Lanes marked "0" received no additional plasmids (lanes 1, 5, 9, and 13). Lanes marked "NP" received N and P support plasmids (lanes 2, 6, 10, and 14). Lanes marked "NPL" received N, P, and L plasmids (lanes 3, 7, 11, and 15). Lanes marked "NPLM2-1" received N, P, L, and M2-1 plasmids (lanes 4, 8, 12, and 16). RNAs were harvested and processed with (B and D) or without (A and C) micrococcal nuclease treatment, as indicated, and were analyzed by Northern blot hybridization with a negative (C and D)- or positive (A and B)-sense riboprobe, as indicated.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The trailer region is thought to have two roles during RNA replication: it encodes the antigenomic promoter, and it might contain a sequence that initiates genome encapsidation. We tested the effects of substitution at each of the seven terminal trailer nucleotide positions. None of the mutations affected encapsidation or template activity of the T7-transcribed genome. However, most of the substitutions prevented the encoded antigenome from serving as a template to synthesize progeny genome. As a result, RNA synthesis by the reconstituted RSV polymerase was restricted to that of positive-sense molecules, namely, the synthesis of mRNA and the synthesis and encapsidation of antigenomic RNA. The genome template was not amplified and instead was restricted to that supplied from the plasmid. Thus, the synthesis of mRNA, and the synthesis and encapsidation of antigenomic RNA, can be analyzed under conditions where the abundance of the template is controlled.

An important control in this work was that we examined the abundance of T7-transcribed genome, the pool for nucleocapsid formation. We also directly monitored the abundance of initial encapsidated genome and of genome after amplification by replication. These controls typically have not been part of published studies, but it is clear that they are critical in any situation where mutations can affect the synthesis or stability of the T7-transcribed replicon. It was well known that the sequence immediately downstream of the T7 core promoter can affect its activity and that purines yield higher activity than pyrimidines and G more than A. However, the length of sensitive sequence was not appreciated. For example, Milligan et al. (19) found strong effects in vitro only for the first two positions. We had attempted to preclude such effects by including three nonviral G residues as the first nucleotides of the transcript, which mimic the first three transcribed nucleotides of the authentic promoter. Indeed, this was sufficient to eliminate effects in vitro (unpublished data). Surprisingly, however, we found that substitutions involving the first 5 nt of the trailer region were associated with strong effects on transcription intracellularly. Together with the three nonviral G residues, this indicates that the first eight transcribed nucleotides influence the efficiency of T7 transcription, and the first seven strongly so, varying over a range of 10-fold or more. Since these experiments were performed in transfected cells rather than in the test tube, we cannot rule out effects on transcript stability due to the point mutations. However, the simpler and more likely explanation is that these substitutions affect promoter activity. Furthermore, identical transfection experiments were performed in 293 cells, in which naked RNA is considerably more stable. These experiments showed the same pattern of T7 transcript accumulation (data not shown), supporting the idea that the differences observed here were not due to effects on transcript stability. Consistent with the higher activities observed here with purine residues, the five strongest promoters in bacteriophage T7 initiate transcription with six purines, GGGAGA, whereas all but one of the 10 weaker promoters have pyrimidine substitutions within these 6 nt (9). Furthermore, the footprint of T7 polymerase has been shown to extend 5 nt beyond the core promoter (5). We suspect that the effect of these initial nucleotides was not evident in in vitro transcription reactions because the large excess of added purified polymerase compensated for reductions in the efficiency of initiation.

Not surprisingly, the amount of genome which became encapsidated was influenced by the level of T7 transcript, and in turn the level of RSV-mediated RNA synthesis was strongly influenced by the level of encapsidated template. If we had not controlled for this variable, it would have led to misinterpretation of the effects of certain mutations on replication. For example, the amount of antigenome and mRNA which accumulated in cells in the presence of N, P, and L with the 4C trailer mutant was substantially lower than for the wt, 2G, and 4G mutants and was similar to that of the 5A and 6C genomes (Fig. 3C and Table 1). Taken by itself, that observation would have suggested that the 4C genome is defective in RNA synthesis, but such a conclusion would have been incorrect. On the contrary, examination of the controls indicated that 4C was one of only two mutants (the other being 4G) in which RSV-directed RNA synthesis appeared to be relatively unaffected. Instead, the low level of positive-sense RNA synthesis reflected the low accumulation of initial T7 transcript. Conversely, the 2G mutant resembled the wt genome in the total amount of positive-sense RNA synthesized intracellularly. However, the obvious interpretation that this mutation did not greatly affect RNA synthesis would have been incorrect since, on the contrary, the genome was completely defective in amplification, and the high level of encapsidated genome observed was due solely to the enhanced synthesis of initial T7 transcripts.

None of the trailer mutations affected encapsidation of T7-supplied genome. For both wt and mutant genomes, this process was inefficient, as has been noted previously in a VSV minigenome system (22), although values for the efficiency of encapsidation could not be determined because much of the plasmid-derived transcript appeared to be degraded. It may be that the RSV genome is not preferentially encapsidated under these conditions. Indeed, in some of our experiments a small fraction of mRNA also was encapsidated, as has been described in a Sendai virus minigenome system (4). In contrast, essentially all of the antigenomic RNA which accumulated due to the reconstituted RSV polymerase was encapsidated, and thus this process appeared to be much more efficient, although we cannot exclude the possibility that additional unencapsidated antigenome was made and quickly degraded. Also, in the case of genomes which were competent for genome amplification, such as the wt and the 4G and 4C mutants, essentially all of the amplified genome also was encapsidated. Thus, replicon RNA which was synthesized by T7 polymerase was encapsidated inefficiently, whereas that produced by the reconstituted RSV polymerase was encapsidated efficiently.

It is generally thought that encapsidation of the genomic and antigenomic RNAs of nonsegmented negative-strand RNA viruses initiates at a cis-acting signal near the 5' end of each RNA and occurs concurrent with chain elongation (18). There is evidence for such a signal in the case of VSV (2, 21, 25) and of rabies virus (29). If there is such a signal for RSV, it was not evident in the highly inefficient encapsidation of genome which was synthesized by T7 RNA polymerase. In these experiments, the synthesis of replicon RNA by T7 RNA polymerase or by the RSV polymerase occurred at the same time and in the same transfected cells, and thus ostensibly under the same conditions except for the difference in template (plasmid versus nucleocapsid) and polymerase (T7 versus RSV). In particular, the requirement that the nascent transcript be available for cosynthetic encapsidation should have been fulfilled in each case. However, there may be some unknown aspect of T7-mediated RNA synthesis which specifically interferes with cosynthetic encapsidation. In these experiments, none of the trailer point mutations, either alone or in combination, and neither of the two trailer deletions affected the efficiency of encapsidation of the T7-transcribed genome. Thus, from these experiments, there is no evidence of a cis-acting signal for RSV. This will be investigated further by examining smaller RNAs that contain just the 5' end of the genome or antigenome versus heterologous negative controls. An alternative explanation is that a specific RNA initiation signal does not exist for RSV and that the replicase complex (N, P, and L) directs encapsidation, whereas the transcriptase (N, P, L, and M2-1) does not. We are in the process of mapping leader positions important in the synthesis of encapsidated antigenome, and it may be that this will shed further light on whether there is an RNA encapsidation signal.

While none of the mutations affected encapsidation of the T7 produced genome, nucleotides 2, 3, 6, and 7 appeared to be absolutely critical for the function of the antigenomic template, and positions 1 and 5 also appeared to be important. We have not yet tested the effect in this system of substituting other nucleotides at these positions or of making substitutions at additional sequence positions. It is interesting that of the seven positions tested, only position 4 was flexible, in that A, C, and G are all capable of functioning. This nucleotide position is also the only one of the first 11 nt of the antigenomic promoter that is not conserved in the genomic promoter. The leader region (genomic promoter) of wt RSV strain A2 contains G at this position (negative sense) (20). Certain attenuated strains contain C at this position, although this substitution did not appear to be part of the attenuation phenotype (28). In other work, we found that this 4C leader substitution results in a substantial enhancement of genome RNA synthesis, which we have now determined to be an effect on enhancing the production of encapsidated antigenome (unpublished data). In contrast, in the trailer region, the substitution of the wt A assignment at trailer position 4 with C or G did not have much effect on RNA synthesis. This indicates a difference between the genomic and antigenomic promoters.

The point mutations which blocked the synthesis of progeny genome, namely, at trailer positions 2, 3, 6, and 7, might directly inactivate the antigenomic promoter. This is the explanation we favor. Alternatively, they might act by preventing encapsidation of the nascent genomic RNA. This explanation seems less likely since these mutations did not prevent encapsidation of genomic RNA synthesized from plasmid. In addition, the possibility existed that the inhibitory effect of the trailer mutations was due to reduced terminal complementarity, as has been described for VSV (27). However, restoration of terminal complementarity by the further introduction of a compensatory mutation in the genome leader did not restore RNA replication. Furthermore, genomes which had sustained large deletions at the end of trailer, and thus lacked terminal complementarity altogether, retained the capacity to function as templates for the synthesis of antigenomic RNA. This indicated that terminal complementarity of the RSV genome is not important for its function as template. Thus, we conclude that the mutations in the trailer region which inhibited replication did so by direct effects on the primary structure of a cis-acting element, probably the antigenomic promoter.

It is interesting that the one leader mutation which was examined here, the 7G mutation, inhibited accumulation of antigenomic RNA, a replication product, but did not inhibit mRNA synthesis. One possibility is that this substitution specifically affected promoter function for replication but not transcription. Alternatively, perhaps antigenomic RNA was produced but not encapsidated and therefore was degraded. These possibilities will be further investigated.