Regulatory activity of distal and core RNA elements in Tombusvirus subgenomic mRNA2 transcription.

Positive-strand RNA viruses that encode multiple cistrons often mediate expression of 3'-encoded open reading frames via RNA-templated transcription of subgenomic (sg) mRNAs. Tomato bushy stunt virus (TBSV) is a positive-strand RNA virus that transcribes two such sg mRNAs during infections. We have previously identified a distal element (DE), located approximately 1100 nucleotides upstream from the initiation site of sg mRNA2 transcription, part of which must base pair with a portion of a core element (CE), located just 5' to the initiation site, for efficient transcription to occur (Zhang, G., Slowinski, V., and White, K. A. (1999) RNA 5, 550-561). Here we have analyzed further this long distance RNA-RNA interaction and have investigated the regulatory roles of other subelements within the DE and CE. Our results indicate that (i) the functional base-pairing interaction between these elements occurs in the positive strand and that the interaction likely acts to properly position other subelements, (ii) two previously undefined subelements within the DE and CE are important and essential, respectively, for efficient sg mRNA2 accumulation, and (iii) the production of (-)-strand sg mRNA2 can be uncoupled from the synthesis of its (+)-strand complement. These data provide important insight into the mechanism of sg mRNA2 transcription.

Many (ϩ)-strand (i.e. messenger-sensed) RNA viruses possess polycistronic coding organizations (1). However, this coding strategy poses a problem for translation of sequentially encoded open reading frames (ORFs) 1 within these genomes. The difficulty arises due to the 5Ј to 3Ј and linear nature of conventional ribosome scanning, which generally allows for only the first ORF encoded in a message to be translated efficiently (2). A coping strategy used commonly by a wide variety of (ϩ)-strand RNA viruses is to synthesize smaller viral mRNAs via RNA-templated transcription (3). These so-called subgenomic (sg) mRNAs represent 5Ј-truncated 3Ј-coterminal copies of the genome that permit efficient translation of their most 5Ј-proximal ORFs (3).
There is compelling evidence for two distinct mechanisms for the transcription of sg mRNAs. The first mechanism, which has been studied most extensively in Brome mosaic virus, involves the initiation of transcription at a localized internal promoter within the full-length (Ϫ)-strand of the genome (4,5). The second defined mechanism, which occurs in the arteriviruses, involves the discontinuous synthesis of (Ϫ)-strands that are then used as templates in the production of sg mRNAs, which include a portion of the 5Ј-untranslated region of the genome (6,7). A third mechanism that has been proposed is that of premature termination of (Ϫ)-strand synthesis during the copying of the genome and subsequent use of the 3Ј-truncated product as a template for the transcription of sg mRNAs (8 -10). Although there is no compelling experimental evidence validating this latter model, it is supported indirectly by the observation that (Ϫ)-strand sg mRNAs do accumulate in various (ϩ)-strand RNA viral infections (8,(11)(12)(13). However, the origin and function of these (Ϫ)strand sg mRNAs remain to be determined.
Tomato bushy stunt virus (TBSV) is the prototype member of the family Tombusviridae (14). Its (ϩ)-strand RNA genome is 4.8 kilobases in length and encodes five functional ORFs (see Fig. 1A). The 5Ј-terminally encoded p33 and its readthrough product p92 are the only viral proteins required for viral RNA synthesis, and both are translated directly from the viral genome (15). The three ORFs encoded more 3Ј in the genome are involved in viral assembly, movement, and suppression of host defense mechanisms and are translated from two sg mRNAs that are synthesized during infections (Fig. 1A) (16,17). RNA sequence elements within the genome that are required for transcription of the smaller sg mRNA2 have been identified previously (18,19). Efficient transcription of this message requires sequences both at the site of initiation, termed the core element (CE), as well as sequences some ϳ1100 nucleotides (nt) upstream from the start site, termed the distal element (DE) (Fig. 1A). The DE and CE can each be divided further into subelements A, B, and C based on their relative positions and/or structural properties. The sequence of the CE-C subelement is highly conserved within tombusvirus genomes and is located just 5Ј to the sg mRNA2 initiation site (Fig. 1C) (18). The DE-C element is not well conserved within the genus Tombusvirus and displays no significant complementarity to CE-C. In contrast, the DE-A and CE-A subelements and the DE-B and CE-B subelements are complementary to each other, and previous studies have provided compelling evidence that base pairing of the former pair is required for efficient sg mRNA2 synthesis (Fig. 1, B and C) (18). A functional role for the DE-A/CE-A base-pairing interaction was supported by comparative sequence analysis of tombusvirus genomes as well as deletion and compensatory-type mutational analyses (18). Interestingly, long distance base-pairing interactions have also been found to be important for sg mRNA transcription in the unrelated virus Potato virus X. (20,21). This suggests that long range RNA interactions involved in regulating sg mRNAs may be a common feature in diverse groups of (ϩ)-strand RNA viruses. However, the functions of these far-spanning interac-tions may be quite different within the unique contexts of these distinct viral genomes.
In a previous study we confirmed the functional requirement for the DE-A/CE-A base-pairing interaction (18); however, these analyses did not determine whether this interaction occurred in the (ϩ)-or (Ϫ)-strand of the genome nor did they investigate the specific role of the interaction in sg mRNA2 transcription. To address these questions and to determine the possible roles of other yet uncharacterized subelements, we have carried out additional analyses of the DE and CE. Our results indicate that the DE-A/CE-A base-pairing interaction is functional in the (ϩ)-strand and that the DE-C and CE-C subelements represent newly defined components that are important and essential, respectively, for efficient sg mRNA transcription. Additionally, by localizing both the DE and CE at different positions within the genome, we have deduced that the likely function of the DE-A/CE-A interaction is to position other subelements optimally. Furthermore, mutational analysis of the initiating nucleotide for sg mRNA2 transcription has provided clues to the origin of (Ϫ)-strand sg mRNAs and, together with other results, provided important new insight into the mechanism of sg mRNA2 transcription.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The plasmid T-100, containing a full-length cDNA copy of the wild type (WT) TBSV genome, has been described previously (14). Construction of mutant derivatives of T-100 containing modifications to the DE and/or CE (i.e. Psg20/26 and ⌬Psg1) have also been described (18). All mutant viral constructs used in this study were generated by standard recombinant DNA cloning techniques and polymerase chain reaction (PCR)-based oligonucleotide-mediated mutagenesis (22) and are derivatives of the previously described constructs pT-100, Psg20/26, or ⌬Psg1. Additionally, new mutant constructs were sequenced across all PCR-derived regions to confirm that only the desired modifications were present.
The mutants Psg35G, Psg34A, Psg34U, and Psg34C are derivatives of T-100 (18) in which the initiation nucleotide for sg mRNA2 transcription was either maintained (Psg35G) or substituted with an adenylate, uridylate, or cytidylate (Psg34A, Psg34U, and Psg34C, respectively.) Psg35G was generated with primer pair PB30 and PG35 (5Ј-GC-CCCCACCGGTCTGCAGAATGATGTTTCTCTAATTTAGTG) followed by digestion of the PCR product with AgeI and EcoRI and replacement of the corresponding fragment in T-100. Psg34A, Psg34U, and Psg34C were generated in an identical manner except that the PCR product used for replacement was generated using primer pair PB30 and the degenerate primer PG34 (5Ј-GCCCCCACCGGTCTGCAGAATGATGT-TTCTCTAATTTAGTGTGTCCTGCGAGGGGCCTCTT(A/T/C)TAACA-AGACCAGTTCATGG). All of these constructs are identical to T-100 except for the single nucleotide substitutions indicated and a 45-nucleotide deletion directly 5Ј to the AgeI site in T-100.
In Vitro Transcription and Protoplast Inoculation-DNA templates for in vitro transcription were prepared by linearizing plasmids containing viral cDNAs with SmaI. Viral RNA transcripts were synthesized using the AmpliScribe T7 Transcription kit (Epicentre Technologies) (23). Protoplasts were prepared from 6-to 8-day-old cucumber cotyledons, and purified protoplasts (ϳ3 ϫ 10 5 ) were inoculated with 5 g of each viral RNA transcript and incubated in a growth chamber under fluorescent lighting at 22°C for 24 h (23).
RNA Analysis-Total nucleic acids were isolated from protoplasts as described previously (23) except that preparations for the detection of (Ϫ)-strand viral RNAs were precipitated in ethanol with 0.1 M sodium acetate rather than 2 M ammonium acetate. For (ϩ)-strand viral RNA detection, aliquots of the total nucleic acid preparation (a tenth) were separated in 1.4% agarose gels and subjected to Northern blot analysis using a 32 P-5Ј-end-labeled oligonucleotide probe (P9) complementary to the 3Ј-terminal 23 nt of the TBSV genome (23).
(Ϫ)-Strand RNAs were detected by Northern blotting following electrophoretic separation of glyoxal-treated samples (22). Minus-sensed viral RNAs were detected with ␣-32 P-labeled in vitro RNA transcripts corresponding to the 3Ј-terminal 380 nt of the TBSV genome. Prehybridization and hybridization of the RNA probe were performed in UltraHyb solution (Ambion) as recommended by the manufacturer. For both (ϩ)-and (Ϫ)-strand detection, radioanalytical quantification of Northern blots was performed using an InstantImager (Packard Instrumental Company). Free energy values for predicted helices were calculated using MFOLD (24,25).

Preferential Destabilization of the DE-A/CE-A Interaction in the (ϩ)-or (Ϫ)-Strand-Previous
results indicated a requirement for a base-pairing interaction between the DE-A and CE-A for efficient sg mRNA2 transcription (18). However, since the majority (i.e. 10 of 12) of the base pairs involved are of the Watson-Crick variety, this interaction could potentially occur in either the (ϩ)-or (Ϫ)-strand (Fig. 1C). In the present study

Regulation of Tombusvirus Subgenomic mRNA2 Transcription
we sought to answer this question regarding polarity by introducing base substitutions into DE-A or CE-A that would progressively and preferentially destabilize the DE-A/CE-A basepairing interaction in either the (ϩ)-or (Ϫ)-strand ( Fig. 2A). These mutations were introduced into the previously described mutant genome Psg20/26 that is amenable to modification due to the introduction of convenient restriction enzyme sites near the DE-A and CE-A subelements (18). Psg20/26 lacks both the DE-B and CE-B subelements ( Fig. 2A) but maintains sg mRNA2 accumulation at ϳ80% that of the WT genome (18). Preferential (Ϫ)-strand destabilization was accomplished by introducing one, two, or three GU base pairs into DE-A/CE-A (via nucleotide substitutions) within the (ϩ)-strand of Psg20/ 26, thereby creating mutant derivatives Psg20/26m1, Psg20/ 26m2, and Psg20/26m3, respectively ( Fig. 2A). These non-Watson-Crick GU base pairs were predicted to be less destabilizing to the (ϩ)-strand interaction than to the corresponding (Ϫ)-strand interaction that would contain complementary CA mismatches ( Fig. 2A). Conversely, one, two, or three CA mismatches were introduced into DE-A/CE-A in the (ϩ)-strand of Psg20/26 to preferentially destabilize the (ϩ)strand interaction, thereby creating Psg20/26m4, Psg20/26m5, and Psg20/26m6, respectively ( Fig. 2A).
In vitro generated transcripts corresponding to the WT genome (Fig. 1A), T-100, and mutant genomes were synthesized in vitro and inoculated into cucumber protoplasts. Northern blot analysis was then performed on isolated total nucleic acids to determine whether the modifications introduced affected the level of sg mRNA2 accumulation (Fig. 2B). The relative levels of sg mRNA2 were determined by radioanalytical analysis of the blot, and the values generated represent ratios of sg mRNA2 levels to their corresponding genomic RNA levels, all normalized to that for Psg20/26 (Fig. 2B). Mutants Psg20/26m1 through Psg20/26m3, which were predicted to be preferentially destabilized in the (Ϫ)-strand DE-A/CE-A interaction, showed a maximum decrease in sg mRNA2 accumulation for Psg20/ 26m3 to a level ϳ60% that of the Psg20/26 control. In contrast, Psg20/26m4 through Psg20/26m6, which were predicted to be preferentially destabilized in the (ϩ)-strand interaction, showed a more pronounced progressive maximal decrease in sg mRNA2 accumulation to ϳ20% that of the Psg20/26 control. These data indicate that the (ϩ)-strand DE-A/CE-A interaction is more sensitive to disruption than its (Ϫ)-strand counterpart and thus favor a functional role for this interaction in the (ϩ)-strand.
Alternatively, the more prominent inhibitory effects observed for Psg20/26m4 through Psg20/26m6 could be related to the changes in nucleotide identities in DE-C (as opposed to their base-pairing potential). To address this possibility we systematically restored base-pairing competence for Psg20/ 26m4, Psg20/26m5, and Psg20/26m6 by introducing additional substitutions into DE-A that would regenerate base pairing while maintaining the original substitutions in CE-A, thus creating Psg20/26m1/4, Psg20/26m2/5, and Psg20/26m3/6, respectively ( Fig. 2A). Analysis of these mutants showed that sg mRNA2 levels were near Psg20/26 control levels in all cases (Fig. 2C). Since the original U to C substitutions were retained in these mutants, the dominant functional property is likely (ϩ)-strand base pairing rather than nucleotide identity.
Analysis of Viral Genomes Containing Localized DE/ CEs-In the TBSV genome the DE and CE are separated by ϳ1100 nt. The reason for the long distance positioning of these two elements is not clear but may be related to the mechanism(s) by which they operate. In an attempt to gain insight into the function of these elements, we created a series of mutants in which the two elements were positioned directly adjacent to one another in the primary sequence. In these mutant genomes, the ϳ1100-nt-long sequence normally separating the DE and CE was reduced to a 6-nt-long restriction enzyme site (XbaI, Fig. 3A). For construction of the first set of mutants (C-series, Fig. 3B), the previously generated mutant genome ⌬Psg1 was used as the base construct (18). In ⌬Psg1, sg mRNA1 synthesis is completely inactivated by five nucleotide substitutions in and near its site of initiation (18). Additionally, and more relevant to this work, the DE is deleted from its natural upstream position (Fig. 3A), resulting in only very low levels of sg mRNA2 accumulation (ϳ10% that of T-100, Fig.  3C). To determine whether sg mRNA2 synthesis could be restored by reintroducing the DE proximal to CE in ⌬Psg1, the entire DE sequence was inserted just 5Ј to the CE (Fig. 3B). The ⌬Psg1 derivative generated, Psg55C (Fig. 3, A and B), was then inoculated into protoplasts to assess the effect on sg mRNA2 accumulation. Northern blot analysis of accumulating viral RNAs revealed a considerable increase in sg mRNA2 levels to ϳ50% that of the T-100 control (Fig. 3C). Additional mutants were also constructed in which various subelements were deleted. In Psg51C and Psg50C, either the DE-B/CE-B subelements or both the DE-B/CE-B and DE-A/CE-A subelements, respectively, were deleted from their localized positions (Fig. 3A). These modifications resulted in relative sg mRNA levels of ϳ25 and ϳ40%, respectively (Fig. 3C). These results suggest that neither the DE-B/CE-B nor the DE-A/CE-A subelement interactions are required for the DE-mediated increase in sg mRNA2 accumulation observed and instead implicate DE-C as a possible key subelement for improved activity. To further test this latter concept, the DE-C subelement was deleted from the Psg55C and Psg51C contexts to generate Psg36 and Psg56, respectively (Fig. 3A). Both of these DE-C-lacking genomes exhibited reduced levels of sg mRNA2 accumulation similar to that of ⌬Psg1 (Fig. 3C), supporting an important role for DE-C within this CE-localized context.
Next the same group of localized DE/CE segments (Fig. 3A) was repositioned near the normal DE locale. The structure of the localized DE/CEs in this second set of mutants was identical to those in the first; however, the genomic context differed. Although these mutants were also derivatives of ⌬Psg1, their genomes were ϳ1100 nt shorter than ⌬Psg1 due to the absence of the normal intervening sequence between the DE and CE in these constructs (Fig. 3D). Consequently the genomes of these mutants exhibit increased electrophoretic mobility (Fig. 3E). This particular genomic structure was chosen over an alternative in which the CE would be deleted from its normal location and inserted just 3Ј to the DE (while maintaining the original ϳ1100 nt 5Ј to the CE) because the resulting sg mRNA transcribed from such a mutant would be ϳ1100 nt longer than the normal sg mRNA2. This highly modified sg mRNA would be predicted to have significantly altered properties that would invalidate any comparisons with WT sg mRNA2. The analysis of Psg55D, which contained the entire CE element positioned near the DE locale, revealed an average sg mRNA2 level ϳ70% that of T-100 (Fig. 3E) Total nucleic acids were isolated from infected cucumber protoplasts after a 24-h incubation and analyzed as described under "Experimental Procedures." Below, bar graphs show the relative accumulation levels of sg mRNA2 that represent ratios of sg mRNA2 levels to their corresponding genomic levels, all normalized to that for Psg20/26. Viral RNA accumulation was quantified by radioanalytical analysis of Northern blots using a Packard Instant Imager. The values shown represent means (with standard deviations) from three separate protoplast infections. respectively, resulted in rather modest decreases in sg mRNA2 accumulation levels to ϳ60 and ϳ40%, respectively (Fig. 3E). This lack of strong dependence on these subelements is similar to that witnessed for the corresponding mutants localized at the CE (Fig. 3C) and again suggests that the DE-B/CE-B and DE-A/CE-A subelements are dispensable for the basal activity observed (i.e. Ͼ3-fold over that of ⌬Psg1) for these localized structures. Interestingly, the deletion of the DE-C subelement in Psg36D and Psg56D did not lead to any major decreases in sg mRNA2 accumulation (Fig. 3E). These latter results are in contrast to those observed for Psg36C and Psg56C where deletion of DE-C resulted in levels of sg mRNA2 accumulation similar to that of ⌬Psg1 (Fig. 3C). This finding indicates that the requirement for DE-C is context-dependent.
Analysis of the CE-C and DE-C Subelements and Initiating Nucleotide-To explore the role of the newly defined and previously uncharacterized CE-C subelement, two mutant genomes were constructed within a Psg20/26 context. In the first mutant, Psg20/26⌬CEC, the entire CE-C subelement (5Ј-AGGGGCCUCUU) was deleted; however, the initiating guanylate located just 3Ј to CE-C (see Fig. 1C) was maintained (Fig.  4A). For the second mutant, Psg20/26mCEC, the sequence of CE-C was mutated to 5Ј-GCUCUCGUGAG to maintain the original nucleotide composition but not the primary sequence (Fig. 4A). Again the original initiating guanylate just 3Ј to CE-C was maintained. When tested for their ability to direct sg mRNA2 transcription, both showed very low levels of sg mRNA2 accumulation (Fig. 4B). This result defines CE-C as a new subelement that is essential for efficient sg mRNA2 accumulation.
The requirement for the DE-C subelement in the more natural Psg20/26 context was also examined by analyzing mutant Psg52 in which the DE-C was deleted (Fig. 4A). Deletion of this subelement led to a significant decrease in the relative accumulation of sg mRNA2 (to Ͻ50% that of Psg20/26, Fig. 4C). This notable reduction implicates DE-C as being important for optimal sg mRNA2 accumulation levels when it is positioned distally in a WT-like genomic context.
To investigate the role of the initiating nucleotide located just 3Ј to CE-C, the guanylate present in the control genome Psg35G was mutated to an A, U, or C, generating the Psg35Gderivatives Psg34A, Psg34U, and Psg34C, respectively (Fig.  5A). Each of the three substitutions led to dramatic decreases (Ͼ10-fold) in sg mRNA2 accumulation levels (Fig. 5B). In contrast, when (Ϫ)-strand viral RNAs were analyzed for the same infections there was no corresponding decrease in the relative (Ϫ)-strand sg mRNA2 accumulation levels, and, in two cases, relative levels instead increased (Fig. 5C). These observations indicate that the initiating nucleotide is specifically required for efficient accumulation of sg mRNA2 (ϩ)-strands and suggests that it can also influence, albeit to a lesser degree, the relative abundance of sg mRNA2 (Ϫ)-strands.

Role of the DE-A/CE-A Interaction-
In this study we have carried out a more detailed analysis of the RNA subelements implicated in the regulation of sg mRNA2 transcription in TBSV. Previous results from our laboratory have confirmed the importance of the DE-A/CE-A base-pairing interaction for efficient sg mRNA2 transcription (18). Our current results suggest that this interaction is functional primarily in the (ϩ)-strand (Fig. 2B). Strengthening of this interaction by the substitution of AU with GC base pairs did not dramatically alter sg mRNA2 accumulation, although slightly elevated levels were observed when three GC base pairs were introduced (Fig. 2C). This suggests that this interaction may already be at its minimally required stability for near-optimal activity. The (ϩ)-strand activity defined for the DE-A/CE-A interaction also likely extends to the DE-B/CE-B interaction that was proposed to play an auxiliary role in promoting and/or stabilizing the DE-A/CE-A interaction (18). Additional support that these two interactions occur in the (ϩ)-strand comes from previous comparative sequence analyses of corresponding elements in other species of the genus Tombusvirus (18). Sequence comparisons of the predicted secondary structures revealed a preponderance of GU, GA, and GG non-Watson-Crick base pairs within the analyzed helices. Such noncanonical base pairs are common in rRNA secondary structure and are predicted to be significantly more stable than their corresponding (Ϫ)-strand counterparts (26). Taken together, these data support the functional interaction of DE-A/CE-A, and by extension DE-B/CE-B, in the (ϩ)-strand of the viral genome.
What is the function of the DE-A/CE-A interaction in the (ϩ)-strand? A similar 8-base pair-long secondary structure is predicted to form in Red clover necrotic mosaic virus (RCNMV), however it occurs just 2 nt upstream from the initiation site of sg mRNA synthesis (10). Another difference in the RCNMV interaction is that it occurs in trans as it involves base pairing between the two RNA segments comprising the bipartite genome. For the RCNMV interaction, it was proposed that formation of this helix acts to stall the viral polymerase during (Ϫ)-strand synthesis and that this, in turn, causes it to terminate prematurely and generate (Ϫ)-strand sg mRNAs (10). These (Ϫ)-strand sg mRNAs would then be used as templates for (ϩ)-strand sg mRNA transcription. Our previous studies showed that the DE-A/CE-A interaction is essential for enhanced sg mRNA2 synthesis when the DE and CE are separated by ϳ1100 nt within the genome (18). However, our current results indicate that when the DE and CE are placed directly adjacent to each other, there is no strict requirement for the DE-A/CE-A interaction. Localized elements containing only the DE-C/CE-C subelements were still able to direct notable sg mRNA2 transcription (Fig. 3, C and E). This observation suggests that the primary role of the DE-A/CE-A interaction within the context of the WT genome may be in the appropriate relative positioning of other subelements, such as DE-C and/or CE-C, thereby facilitating their activities (see below). Nevertheless, additional roles for this interaction cannot be precluded currently.

Roles of the DE-C and CE-C Subelements-
The DE-C, unlike the DE-A and DE-B, is not predicted to have any significant complementarity to its counterpart in the CE, CE-C (Fig. 1C).
Our results indicate that the necessity of the DE-C subelement for sg mRNA2 accumulation is context-dependent. When the DE was repositioned near the normal CE location, deletion of DE-C led to a dramatic decrease in sg mRNA2 accumulation (Fig. 2B). However, when the CE was localized near the DE no such effect was observed upon deletion of CE-C (Fig. 2C). In the more natural context of Psg20/26, the DE-C was also found to be important for optimal sg mRNA2 accumulation (Fig. 4C). Based on its structural properties, a possible role for this element may simply be to not base pair with CE-C. This would in turn leave CE-C free to interact functionally with other sequences and/or proteins. This idea is consistent with the observation that there is no significant sequence identity among DE-C elements from different members of the genus Tombusvirus, yet a common feature is that they all exhibit little complementarity to their corresponding CE-C subelements (18). In contrast, the sequence comprising the CE-C is highly conserved for members of this genus and is located just 5Ј to the site of initiation of sg mRNA2 (18). This subelement was found to be essential for efficient sg mRNA2 accumulation (Fig. 4B). In RCNMV the sequence just 5Ј to the initiation site of its sg mRNA interacts with a complementary viral sequence to mediate sg mRNA transcription (10). Due to the relatedness of these viruses (i.e. they are both members of the family Tombusviridae), it is possible that sg mRNA2 transcription in TBSV also requires base pairing of its similarly positioned sequence, CE-C. The lack of complementarity between CE-C and DE-C would preclude the latter from acting in this capacity. However, possible base-pairing partners for the CE-C sequence have been identified within the TBSV genomic sequence and are currently being analyzed to determine whether they participate in regulating sg mRNA2 transcription.
Role of the Initiating Nucleotide and Insights into the Mechanism of sg mRNA2 Transcription-The initiating nucleotide for sg mRNA2 synthesis has been mapped by primer extension to a guanylate just 3Ј to the CE-C (27). Our results have shown that for efficient transcription of sg mRNA2 there is a strict requirement for the identity of the initiating nucleotide to be a guanylate (Fig. 5B). However, the same strict requirement was not observed for the production of (Ϫ)-strands corresponding to sg mRNA2 (Fig. 5C). These (Ϫ)-strands have been previously implicated as possible templates for the synthesis of (ϩ)-strand sg mRNA2 (18); however, exactly how they are formed remains unclear. For example, the sg mRNA2 (Ϫ)-strands could accumulate either as dead-end products synthesized from sg mRNAs transcribed from the full-length (Ϫ)-strand of the genome or, alternatively, could represent prematurely terminated products synthesized during (Ϫ)-strand synthesis of the (ϩ)-strand genome. In the latter case, the sg mRNA2 (Ϫ)strands could potentially serve as intermediates by acting as templates for the transcription of (ϩ)-strand sg mRNA2. Our finding that (Ϫ)-strand accumulation does not correlate with (ϩ)-strand accumulation argues against the (Ϫ)-strands originating from (ϩ)-strand sg mRNA2. This independence of (Ϫ)on (ϩ)-strand sg mRNA accumulation has also been observed for Flock house virus (13). For TBSV, in some cases, the relationship between the two oppositely sensed strands is actually inverse, with lower (ϩ)-strand accumulation corresponding to higher (Ϫ)-strand accumulation, as for Psg34A and Psg34U (Fig. 5, B and C). These results support the concept that the (Ϫ)-strand sg mRNAs observed are not derived from (ϩ)-strand sg mRNA2. Instead they could represent prematurely terminated products that arise during (Ϫ)-strand synthesis of the viral genome that, in turn, could serve as templates for sg mRNA2 synthesis. The inability of the (Ϫ)-strands generated from the mutants (Psg34A, Psg34C, and Psg34U) to direct transcription of (ϩ)-strands could be related to the strict requirement of tombusvirus polymerases to initiate efficiently only opposite to an appropriately positioned C within promoters in viral templates (28).
Taken together, our results are consistent with the concept that sg mRNA2 transcription could occur via a premature termination mechanism. However, other alternatives, such as the observed effects being the result of post-transcriptional processes, cannot be ruled out at the present time. Additional experiments are planned to further investigate the mode by which sg mRNA2 is expressed.