INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Vesicular stomatitis virus (VSV) is a nonsegmented negative-strand RNA virus from the Rhabdoviridae family. Although VSV has served as a useful model to study virus entry, replication, and assembly, the mechanism by which the VSV polymerase transcribes and processes its transcripts has been a topic of debate for many years. Moreover, progress in this area has been hampered by the inability to introduce site-specific mutations into the VSV genome. The recent development of systems that allow directed genetic manipulation (reverse genetics) of VSV and the ability to recover full-length virus (22, 39) or minivirus particles (26, 35) entirely from cDNAs have now provided additional tools to dissect and understand the events that occur during VSV transcription.

The replication cycle of VSV begins when an endocytosed virion fuses with the endosomal membrane and the ribonucleocapsid core particle (RNP) is uncoated and released into the host cell cytoplasm. Uncoating occurs either during or immediately following fusion of the viral envelope with the endosomal membrane and results in the release of the matrix (M) protein from the RNP. The uncoated RNP, which consists of the genomic RNA tightly encapsidated by ~1,200 molecules of the nucleocapsid (N) protein, serves as a template for both transcription of mRNAs and replication of the genome by the viral polymerase, which is also brought into the cell by the incoming virion.

One of the characteristic features of VSV transcription, as well as that of other nonsegmented, negative-strand RNA viruses, is that mRNA synthesis is both sequential and polar (19, 20). For VSV, the polymerase first transcribes a small 47-nucleotide RNA called the leader from the extreme 3' end of the genome, and then each of the five mRNAs encoding the VSV proteins are synthesized in the order in which they appear from the 3' end of the genome. For the gene junctions that have been studied, the downstream gene is transcribed approximately 30% less than that of the upstream gene (19), and as a consequence, the abundance of the five mRNAs also follows the order of genes on the genome (e.g., N>P>M>G>L). In addition to transcription and replication, the VSV polymerase is responsible for capping and methylating (17, 25, 27, 28) as well as polyadenylating (32) VSV transcripts.

The recent demonstration that a sequence of 23 conserved nucleotides, which is found at the 3' and 5' junction of all VSV genes, can direct the expression of foreign genes within recombinant VSV genomes provided the first evidence that this sequence contains all the cis-acting signals necessary to polyadenylate and terminate transcripts from the upstream gene and then to reinitiate transcription at the adjacent downstream gene (31). The conserved sequence 3'-AUAC(U)₇-5' which is found at the end of each gene is critical for both polyadenylation and termination of VSV transcripts (3, 18). When this sequence is encountered, the polymerase reiteratively copies, or stutters, over the seven U's to produce a poly(A) tail approximately 150 nucleotides in length (32). Immediately following the polyadenylation signal, there are two nontranscribed intergenic nucleotides, which are usually 3'-GA-5' (29). Recently, it was suggested that this dinucleotide is important for efficient termination of the upstream transcript because certain nucleotide substitutions resulted in higher levels of readthrough transcripts at the mutated gene junction (4, 36). Substitutions at the first nucleotide were found to have the greatest effect, while single substitutions at the second position resulted in only slightly higher levels of readthrough transcripts (4). Based on those results, it was proposed that the minimum polyadenylation and termination signal is 3'-AUAC(U)₇G-5' (3). However, it appears that the second nucleotide does play a small role in upstream mRNA termination, since nucleotide changes at both positions resulted in higher levels of readthrough transcription than single nucleotide changes at the first position (36). Still others question the interpretation that the dinucleotide is important for efficient termination, because only minimal effects on the upstream transcript levels were observed when this sequence was deleted (18). Following the intergenic dinucleotide is the sequence 3'-UUGUCnnUAC-5' (with n being any nucleotide). The first three nucleotides of this sequence are most important for efficient gene expression because mutations at these positions severely reduced the amount of mRNA produced from the mutated gene (36). Presumably, this sequence is important for reinitiation following polyadenylation and release of the upstream mRNA, although it is possible that the reduced mRNA levels from the mutated gene may be the result of mRNA instability due to the lack of a 5' cap or prevention of some other postinitiation step by the polymerase.

Although the precise mechanisms responsible for transcript initiation and termination have not been defined, several different models have been proposed over the last two decades to explain certain features of VSV transcription (2). There is some evidence that the viral polymerase may initiate at internal sites on the genome (7, 37), and it has been suggested that the polarity observed during VSV transcription is the result of elongation of internally initiated polymerase complexes being dependent on transcription of the upstream gene (37). However, the most widely accepted model is that the viral polymerase initiates transcription de novo from the extreme 3' terminus of the genomic RNA (12) and then genes are sequentially transcribed by a start-stop mechanism (12, 20). Based on our analysis of transcription start site mutants, we proposed an additional feature to the start-stop model which suggested that after polyadenylation and termination of the upstream message, the polymerase scans the intergenic junction and initiates transcription at the first U it encounters, even if the U is not in the context of an optimal start sequence (36). Recently, an alternative model which suggests that the 5' end of a transcript is generated following an RNA cleavage event was revived and modified (33). The chemistry of an RNA cleavage event would explain the atypical 5' cap found on VSV mRNAs in which the  and  phosphates of the cap are provided by the cap donor (GDP) and only the  phosphate of the 5' RNA acceptor is used in the 5'-5' triphosphate bridge (1). In this current model, the last two nucleotides of the poly(A) tail of the upstream message serve as the primer for transcript initiation at the adjacent downstream gene. The priming event occurs by a forward slippage of the nascent poly(A) tail over the intergenic dinucleotide and subsequent base pair formation between the last two nucleotides of the poly(A) tail and the first two nucleotides (UUGUC...) of the downstream start sequence. Addition of the 5' cap results from a GDP-dependent, polymerase-mediated cleavage event.

In this study, we used a transcriptionally active, dicistronic VSV minivirus (G and M gene minigenome [GMMG]) to test several aspects of the cleavage-capping model (33) and the U-dependent start model (36) for initiation of VSV transcription. Through insertional mutagenesis at the VSV intergenic dinucleotide within GMMG, we determined that the polymerase can efficiently perform all of the transcriptional events at gene junctions containing at least a 6-nucleotide intergenic region. However, not all sequences were tolerated by the VSV polymerase, suggesting that, in contrast to the proposed cleavage-capping model, the polymerase appears to scan the region and is affected in a sequence-specific manner. We also show that the first U following the polyadenylation signal of the upstream gene is not the sole determinant for transcript initiation. These data support a model for VSV transcription which includes a scanning polymerase that samples sequences downstream of the polyadenylation signal until an appropriate start site is encountered.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Expression plasmids and minigenome constructs. Plasmids encoding the wild-type (wt) GMMG as well as the VSV Indiana N, P, G, and L proteins have been described previously (14, 35). To generate minigenome mutants with insertions in the G and M intergenic region, we performed region-specific mutagenesis by using a PCR-based strategy. Genomic-sense, degenerate oligonucleotides spanning the M-G gene junction [5'-CATAGTGACGCGTAAACAGATCGATCTCTGTT(N/V₄)AGTTTTTTTCATAGGG-3'; N = G, A, T, or C; V = G, A, or C] were used with an antigenomic-sense oligonucleotide (MW-28, 5'-TATAGGGCCCTCGCGAAGACAAACAAACCATTATTATC-3') complementary to the leader region (in boldface type) to generate PCR products from a wt pBS-GMMG template. The primers used to generate the intergenic mutants were synthesized by delivering equimolar amounts of the indicated phosphoramidites at each target position. For the VSV New Jersey G-L intergenic mutants, primers that included the sequences shown in Fig. 1 were used in the PCR. Following PCR amplification using standard conditions (14), the PCR products were digested with MluI and StuI. The resulting 210- to 229-bp fragments were used with a StuI-to-BglII fragment (391 bp) from the M gene in a three-way ligation to replace the corresponding regions in the wild-type pBS-GMMG plasmid. Direct sequence analysis (thermal cycle dideoxy sequencing method; New England Biolabs) was used to identify plasmids containing the site-specific mutations and to confirm that deleterious mutations were not introduced during PCR amplification.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Sequences of GMMG intergenic insertion mutants. A diagram of pBS-GMMG is shown. Expression from the T7 promoter results in the synthesis of a genome-sense RNA which becomes encapsidated by the N protein and serves as a template for the synthesis of the M and G mRNAs by the VSV polymerase (35). The sequence of the M-G gene junction is enlarged, and the inserted sequences are shown below it. All insertions were between the wt nontranscribed intergenic dinucleotide (3'-GA-5') and the G gene start sequence. The VSV polymerase is represented by the shaded rectangles and ovals. HDV, hepatitis delta virus ribozyme sequence; T, the T7 terminator. N corresponds to G, A, T, or C; V corresponds to G, A, or C. Dashes indicate nucleotides that are identical to the nucleotides found in the NJ-wt sequence.

Minigenome recovery and passaging. Recoveries were performed essentially as described previously (36). Baby hamster kidney (BHK-21) cells in 35-mm-diameter dishes were infected with vTF7-3 (15) and then transfected with 10 µg of plasmid DNA encoding either wt or mutant VSV minigenomes, together with 5, 4, 3, and 1 µg of plasmids encoding the N, P, G, and L proteins, respectively, by using TransfectACE (30). Culture supernatants were harvested 18 to 24 h posttransfection.

RNA analysis (primer extension and Northern blot assays). All RNA analyses were performed exactly as described before (36). For primer extensions, we used an oligonucleotide (3188; 5'-GTCATTATGCCAATTTAAATC-3') complementary to a sequence approximately 180 nucleotides from the 5' end of the G mRNA. The sequencing ladder was generated by using the same end-labeled primer and wt pBS-GMMG plasmid, unless otherwise indicated. For Northern blot and primer extension analysis, total RNA was isolated at 5 to 6 h postinfection after infection with third- or fourth-passaged (P3 or P4) supernatants. The RNAs used for both Northern blot analysis and primer extension assays were isolated from the same minigenome infections except for the 3'-GAGGCA-5' and 3'-GAGGUU-5' mutants in Fig. 4, in which the RNA for each assay was isolated from separate minigenome infections. The templates used to generate the G mRNA-specific and the M mRNA-specific probes were described previously (35).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Previous data from saturation mutagenesis of the nontranscribed intergenic dinucleotide and the G gene start sequence suggested that the intergenic dinucleotide is important for efficient termination of the upstream gene mRNA whereas the first 3 nucleotides of the start sequence are most critical for efficient expression of the downstream gene. From these data, we suggested that the polymerase scans the intergenic region and attempts to initiate transcription at the first U encountered (36). Recently, Shuman proposed a novel cleavage-capping mechanism to explain the unique capping reaction chemistry and mRNA synthesis that occur during VSV transcription (33). In order to (i) test our U-dependent initiation hypothesis, (ii) test one aspect of the cleavage-capping model, and (iii) further examine the role that the conserved intergenic sequences play during polyadenylation and termination of the upstream mRNA and reinitiation at the next downstream gene, we utilized insertional mutagenesis at the M-G gene junction in the context of the dicistronic minigenome, GMMG. Either 1 or 4 nucleotides was inserted after the wild-type nontranscribed intergenic dinucleotide and before the G start sequence (Fig. 1). We also constructed a GMMG variant containing the 21-nucleotide intergenic region found between the G and L genes of VSV New Jersey (Ogden strain). Nucleotide substitutions were then made within this intergenic sequence either to create a minimal start sequence (5'-UUGUC...3') (Fig. 1, NJ-2 and NJ-3) or to disrupt the potential base-pairing between the margins of this intergenic region (Fig. 1, NJ-1 and NJ-2). Base-pairing between the intergenic margins was suggested in the cleavage-capping model to be necessary to loop out the extra intergenic sequence (33). A total of 22 GMMG mutants were constructed and analyzed for M and G mRNA expression.

Differential effects of intergenic insertions on transcript initiation and accumulation. To examine the effects of 1- or 4-nucleotide insertions on M and G mRNA expression, total RNA from cells replicating either wild-type or mutant minigenomes were analyzed by Northern blot analysis. G and M antisense probes were used to detect G mRNA, M mRNA, and two types of M-G RNA species, which were the positive-sense replicative intermediate and an M-G dicistronic mRNA. The replicative intermediate and M-G mRNA from GMMG-infected cells are approximately the same size because the latter is polyadenylated.

View larger version (38K): [in this window] [in a new window]   FIG. 2.   Northern blot analysis of single-nucleotide insertion mutants. Approximately 4 µg of total RNA was fractionated on a 1% agarose-formaldehyde gel, transferred to a nylon membrane, and hybridized with a mixture of two antisense RNA probes specific for either G or M mRNA sequence. The inserted nucleotide for each mutant is underlined.

View larger version (61K): [in this window] [in a new window]   FIG. 3.   Analysis of "random" 4-nucleotide insertion mutants. (A) Northern blot analysis. Approximately 4 µg of total RNA was probed essentially as described in the legend to Fig. 2. Lane 1, wt VSV; lane 2, wt GMMG; lanes 3 to 9, 4-nt intergenic insertion mutants. The percentage of G mRNA expression for each mutant compared to that of wt GMMG is shown below each lane and was determined by densitometry of digitized fluorograms. Inserted nucleotides are underlined. (B) Primer extension. The total RNA was hybridized to an end-labeled oligonucleotide complementary to sequences in the G mRNA. Following extension, the products were resolved on a denaturing 6% polyacrylamide gel. A sequencing ladder generated from pBS-GMMG and the end-labeled oligonucleotide is also shown. The conserved intergenic dinucleotide is boxed, and the position of the wt initiating nucleotide is marked by an asterisk. The amount of total RNA used for mutant GMMG extension reactions ranged between 2 and 100 times more than that used for wt GMMG. The nucleotides where aberrant initiation occurred are indicated (lanes 4 to 8, outline font). A sixfold-longer exposure of lanes 5 to 8 is shown in the panel on the right.

The VSV polymerase does not attempt to initiate transcription at the first U encountered. To determine if the polymerase attempts to initiate transcription at the first U encountered following polyadenylation of the upstream mRNA, we substituted one or two U's at nucleotides 3 through 6 in the 3'-GAGGCA-5' mutant (nucleotides 3 to 6 are underlined), which expressed wild-type levels of M and G mRNAs, and then examined the level of G mRNA expressed from each mutant minigenome. Northern blot analysis (Fig. 4A) revealed that a U at position 3, 4, or 5 did not drastically alter downstream G expression relative to that seen for the 3'-GAGGCA-5 mutant (lanes 4 to 6 and 3, respectively). In addition, U residues at both positions 3 and 4 did not drastically alter G mRNA expression (lane 8). However, a U at position 6 or two U's at positions 5 and 6 almost completely eliminated G mRNA expression (lanes 7 and 9). These data indicate that the presence of a U directly preceding the downstream start sequence is deleterious for gene expression.

View larger version (46K): [in this window] [in a new window]   FIG. 4.   Analysis of U-dependent transcription initiation. (A) Northern blot analysis. Approximately 4 µg of total RNA was probed essentially as described in the legend to Fig. 2. Inserted nucleotides are underlined. The percentage of G mRNA expression for each mutant that contained a U ranged from 83 to 106% relative to that of the parental 3'-GAGGCA-5' mutant and was determined by densitometry of digitized fluorograms. Inserted nucleotides are underlined. RI, replicative intermediate. (B) Primer extension analysis. The initiating nucleotide from the G mRNA was determined essentially as described in the legend to Fig. 3. One hundred times more RNA was used for the mutants shown in lanes 7 and 8. Since different amounts of RNA were used for this assay, the differences in signal intensity for the mutants compared to that of wt GMMG are not quantitative.

Analysis of an intergenic variant. Whereas all intergenic junctions of VSV Indiana are 2 nucleotides (usually 3'-GA-5'), the G-L intergenic junction from VSV New Jersey contains an extra 19 nucleotides between the wt (3'-GA-5') dinucleotide and the L gene start sequence. To examine this G-L gene junction, we generated a GMMG mutant (Fig. 1, NJ-wt) which contained the intergenic sequence from the Ogden strain of VSV New Jersey. Northern blot analysis (Fig. 5A) showed that the G mRNA level was reduced to approximately 25% of wt (VSV Indiana) levels (Fig. 5A, compare lanes 2 and 3), and primer extension analysis showed that the NJ-wt G gene transcripts were initiated at two sites (Fig. 5B). The majority of the transcripts initiated at the consensus start sequence located 21 nucleotides downstream from the M gene poly(A) signal. A less abundant transcript which was 18 nucleotides larger was also detected (Fig. 5B). This transcript would have initiated at the fourth nucleotide of the intergenic sequence. Interestingly, no transcripts were initiated at any of the U's within the sequence, further confirming that the polymerase does not attempt to initiate transcription at the first U following the upstream polyadenylation sequence.

View larger version (39K): [in this window] [in a new window]   FIG. 5.   Mutagenesis of the 21-nucleotide G-L intergenic region from VSV New Jersey, Ogden strain. (A) Northern blot analysis. Approximately 4 µg of total RNA was probed essentially as described in the legend to Fig. 2. Lane 1, wt VSV Indiana; lane 2, wt GMMG; lane 3, NJ-wt G-L intergenic variant; lane 4, mutant NJ-1 that eliminates the potential base pairing at margins; lane 5, mutant NJ-2 that eliminates potential base pairing at margins and that contains an internal minimal start sequence; lane 6, mutant NJ-3 that contains an internal minimal start sequence; lane 7, VSV New Jersey intergenic region that contains two optimal start sequences (2X Start). RI, replicative intermediate. (B) Primer extension analysis. The initiating nucleotide of the G mRNA was determined essentially as described in the legend to Fig. 3. To obtain equivalent band intensities, four times more total RNA was used for the primer extension reactions of mutant GMMGs that had reduced G mRNA levels. Sequences for the various intergenic mutants with the corresponding start sites are shown. The Indiana wt sequence is shown on the left; the sequences of the NJ-wt and the 2xStart mutant are shown on the right. Intergenic sequences are boxed. Nucleotides that are in parentheses and/or are underlined indicate changes from the NJ-wt sequence. Alternative start sites are indicated by , , and .

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In this study, we used insertional mutagenesis to test several aspects of two models that have been proposed for initiation of VSV transcription (33, 36). One model proposed that the 5' end of transcripts are generated following a polymerase-mediated cleavage-capping event in which the 3' end of the upstream polyadenylated mRNA is used as a primer while the intergenic region is simultaneously looped out (33). Another model suggested that after polyadenylation and termination of the upstream message, the polymerase scans the intergenic junction and initiates transcription de novo at the first U it encounters, even if the U is not in the context of an optimal start sequence (36). The data presented here demonstrate unequivocally that this later model is not completely correct and that other factors are most certainly involved in determining at which nucleotide the VSV polymerase initiates transcription. However, the data also show that the polymerase does sample the sequence of the intergenic region, presumably by a nontranscriptive scanning mechanism, and can respond by initiating transcripts at suboptimal, nonconsensus start sites within the intergenic region. Therefore, longer intergenic regions are not simply bypassed or looped out, as suggested in the cleavage-capping model.

No sequence patterns emerged to indicate why some of the inserted sequences were bypassed (e.g., 3'-GAGGCA-5' and 3'-GACCAG-5') while others were not (e.g., 3'-GACGGG-5') nor why some of the sequences influenced the polymerase to initiate at a suboptimal start sequence. Perhaps the sequence itself or the events resulting from initiation at a suboptimal start sequence cause a portion of the polymerase population to disengage from the template. Alternatively, since suboptimal start sequences can be utilized and the resulting transcripts do not accumulate to high levels, the conserved sequence found at the beginning of each VSV gene may not be needed only for transcript initiation. Perhaps this sequence is also crucial for a later step, such as capping, methylation, and/or elongation of the nascent mRNA, and therefore is critical for transcript stability. We are currently screening a larger population of these insertion mutants to ascertain if there is a predictive rule regarding which sequences will be tolerated by the polymerase. One rule did emerge in that all initiations at suboptimal start sequences which resulted in detectable amounts of mRNA occurred at a C. This observation supports our previous conclusion (36) that there is a strong bias for the VSV polymerase to initiate transcripts with a purine.

Several nonsegmented negative-strand RNA viruses contain nontranscribed intergenic regions with a consensus length and sequence (3'-GA-5' for VSV and 3'-GAA-5' for Sendai virus, parainfluenza virus type 3, and measles virus [6, 16, 29, 34]), while others contain intergenic regions with both heterogeneous length and sequence (respiratory syncytial virus, rabies virus, Ebola virus, and Marburg virus [8, 13, 38]). Recently, it was shown, by using a similar type of minigenome system for respiratory syncytial virus, that the heterogeneous intergenic regions of respiratory syncytial virus do not appear to increase or decrease the attenuation that occurs at each gene junction (21). Therefore, while our results suggest that the VSV polymerase responds to the nontranscribed intergenic region in a sequence-specific manner, it appears that not all of the polymerases from the order Mononegavirales are affected by the intergenic region in the same manner.

From previous data (4, 36) and the data presented here, it is clear that transcript levels are severely decreased when a U directly precedes the start sequence. Primer extension analysis showed that a U at this position resulted in the accumulation of very low levels of transcripts which were initiated at this U. Presumably, the decreased levels of transcript are a result of altering the context of the start sequence. If the mechanism of transcription proceeds through some variation of the cleavage-capping model, then perhaps when the start sequence contains three U's (e.g., 3'-UUUGUC-5'), the poly(A) sequence of the upstream transcript forms a more stable hybrid through base pairing with three U's instead of the normal two. The increased stability of this hybrid may alter the strained complex proposed to be required for the polymerase-mediated attack by GDP across the ApA dinucleotide and inhibit the cleavage event. However, our previous observations (36) that nucleotide substitutions at position 3 of the transcription start site (3'-UUGUC-5') can also severely reduce the amount of transcript produced and the fact that the polymerase can utilize suboptimal start sequences indicate that there are additional sequence-specific requirements for transcript initiation that cannot be explained by the cleavage-capping model.

One of the early models for VSV transcript initiation proposed that some fraction of polymerase molecules can initiate transcription at internal start sequences (37). More recent studies with the polR mutant of VSV have provided compelling evidence that internal initiation can occur at the leader-N gene junction in vitro (7); however, the authors cautioned that internal initiation may be specific for the leader-N gene junction and suggested that sequential transcription likely occurs once the N gene is transcribed. Although we did not examine the leader junction, our results with a GMMG mutant which contained two optimal start sequences after the M-G gene junction support this conclusion because we detected only transcripts that were initiated at the first consensus start sequence. If the polymerase had been able to initiate internally, we would have expected to see transcripts which initiated at both sites. The data obtained with this mutant also eliminated a potential mechanism for attenuation, namely, that some fraction of polymerase molecules bypass the start sequence and then disengage from the template at some point downstream. Had this been the mechanism, we again would have expected to see transcripts which had initiated at both sites, leading to an increase in the levels of the G gene transcript compared to wt levels.

By using a GMMG variant that contained the intergenic region from the G-L gene junction of the New Jersey serotype (Ogden strain), we found that expression from the downstream G gene was reduced 75% relative to that of wt (VSV Indiana) GMMG. Although it has not been determined if there is a similar increase in attenuation at the G-L gene junction of VSV New Jersey, our data suggest that this extra sequence may have been maintained to down-regulate polymerase expression. It is known that some negative-strand RNA viruses utilize a variety of strategies to down-regulate expression of not only their polymerase (9, 10, 24) but of other genes as well (5). We also found that transcripts were initiated from two sites. Approximately 22% initiated at the fourth nucleotide of the intergenic region, while the majority initiated at the consensus start site. These results contrast those of Luk et al., who used primer extension to map the start site of the L protein transcript in VSV New Jersey-infected cells as well as following in vitro transcription (23). Analysis of in vivo-generated transcripts indicated that only the consensus start sequence was used, whereas it appeared that in vitro-generated transcripts were initiated at two distinct sites, one following the intergenic dinucleotide and one at the consensus start sequence. The differences between the initiation sites that we observed and those previously reported (23) could be due to inherent differences between the VSV Indiana and VSV New Jersey polymerases and their ability to navigate this intergenic sequence. However, we cannot exclude the possibility that the differences are due to other factors such as the ratios of the N, P, and L proteins following transfection compared to that found in VSV New Jersey-infected cells.

Collectively, our data suggest that the nontranscribed intergenic region is not simply bypassed or looped out during transcription. While other aspects of the cleavage-capping model were not tested (i.e., that initiation of transcription occurs via a concomitant cleavage and capping event), this model is worthy of further investigation because of its potential to explain the unusual chemistry of the 5' cap present on VSV mRNAs and the possible identification of a primer for transcript initiation. To further test this model, we are currently sequencing the 5' ends of mRNAs generated from mutants in which the first two nucleotides of the start sequence have been altered. Should the poly(A) tail of the upstream message serve as the primer for the downstream transcript, one might expect to see A's as the initiating nucleotides as opposed to the corresponding nucleotide change that was made in the template. Defining the mechanism responsible for VSV transcript initiation and capping will likely provide important clues as to how other viruses in the order Mononegavirales transcribe their genes. Although the fundamental mechanisms may be similar, sequence differences at the transcription start and stop polyadenylation sites and the potential utilization of different cellular factors important for polymerase activity may provide avenues to develop specific antiviral agents capable of selectively inhibiting the replication of members of this group of medically important pathogens.