INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The arterivirus Equine arteritis virus (EAV) is a plus-stranded RNA virus belonging to the order Nidovirales, which includes coronaviruses and arteriviruses (4, 11). The family Arteriviridae consists of EAV, Porcine reproductive and respiratory syndrome virus (PRRSV), murine Lactate dehydrogenase-elevating virus (LDV), and Simian hemorrhagic fever virus (SHFV) (see reference 34 for a recent review).

The EAV genome is a single 12.7-kb RNA molecule (11). Two large replicase polyproteins, the ORF1a and ORF1ab proteins, are encoded in the 5' three-fourths of the genome. Translation of the ORF1b-encoded part of the ORF1ab protein involves a ribosomal frameshift (7). These polypeptides are proteolytically cleaved to yield 12 nonstructural proteins (36, 37, 44, 45, 48). In addition to genome replication, these proteins function in the production of a 3'-coterminal nested set of subgenomic (sg) RNAs, from which the viral structural proteins are translated (Fig. 1A).

View larger version (29K): [in this window] [in a new window]   FIG. 1.   (A) Schematic diagram of the EAV genome organization and expression. The regions of the genome specifying the leader (L) sequence, the replicase gene (ORF1a and -1b), and the structural protein genes are indicated. The nested set of EAV mRNAs (genome and sg mRNA2 to -7) is depicted below. Black boxes, positions of leader and major body TRSs. (B) Close-up of the 3'-terminal one-third of the EAV genome. Numbers, previously identified body TRSs (see text); solid vertical lines, UCAAC sequences; dashed vertical lines, previously identified noncanonical leader-body junction sites.

EAV contains six or seven structural proteins (9, 10, 16, 35): the nucleocapsid protein N, the nonglycosylated, triple-spanning membrane protein M, the small envelope protein E, and four glycoproteins, GP_2b, GP₃, GP₄, and GP₅. It is unclear whether GP₃ is a structural component of the arterivirus particle in view of the conflicting data reported for PRRSV GP₃ (see reference 34 for a review). The genes encoding the EAV structural proteins (ORF2a to -7; Fig. 1) are located in the 3'-proximal 3 kb of the genome. From 5' to 3', the genes encode proteins in the following order: E-GP_2b-GP₃-GP₄-GP₅-M-N. mRNA2 is presumed to be bicistronic and is thought to encode both E and GP_2b. The other five genes (encoding GP₃ to N) are expressed from sg mRNA3 to -7.

The EAV sg RNAs are synthesized in different but constant molar ratios in the infected cell. The smallest sg RNA contains only the 3'-terminal open reading frame (ORF) of the genome (the N gene), while each next-larger sg RNA, with the exception of mRNA2, contains one additional ORF. Thus, all sg RNAs, except for the smallest RNA, are structurally polycistronic. However, with the exception of mRNA2, only the 5'-proximal ORF of each sg mRNA is assumed to be translated. Consequently, the expression of the EAV structural proteins can be primarily regulated at the level of sg mRNA synthesis.

The most characteristic feature of nidovirus sg mRNA transcription is that the sequences of the sg mRNAs are noncontiguous in the genome. At their 5' ends, they include a common leader sequence (156 to 221 nucleotides [nt] for arteriviruses and 65 to 98 nt for coronaviruses), which is identical to the 5' end of the genomic RNA (see references 19 and 34 for recent reviews). Thus, all nidovirus-specific RNAs, including the genome, are both 5' and 3' coterminal (Fig. 1A). The leader and body parts of the sg RNAs are connected by a short, conserved sequence element termed the transcription-regulating sequence (TRS), or intergenic sequence for coronaviruses. This sequence is present at the 3' end of the leader (leader TRS) as well as upstream of (almost) every structural protein gene (body TRSs). The body TRS has been shown to be necessary for sg RNA synthesis (17, 22, 42, 46). A number of models have been put forward to explain the discontinuous step in nidovirus sg RNA synthesis (2, 31, 33, 38). Our recent data (46) are most consistent with the model of Sawicki and Sawicki (32), in which the discontinuous step occurs during minus-strand synthesis, producing minus-strand sg-sized templates for the synthesis of sg mRNAs in a process that resembles similarity-assisted RNA recombination (3, 27, 46).

The consensus sequence of the EAV TRS is 5' UCAAC 3' (6, 8). Recently, by site-directed mutagenesis of TRSs in the infectious cDNA clone of EAV, we have shown that base pairing between the complement of the RNA7 body TRS in the minus strand and the leader TRS in the plus strand is absolutely required for sg mRNA7 synthesis (46). However, at least for EAV, the mere presence of a UCAAC sequence in the genome is not sufficient for a leader-body junction event to occur (6, 7). The EAV genome contains multiple UCAAC sequences that appear to be transcriptionally silent. Upstream of ORF4, -5, and -7, two UCAAC boxes are present, whereas ORF2a and -2b are preceded by three UCAAC sequences (Fig. 1B) (11). Furthermore, the replicase gene contains a number of internal 5' UCAAC 3' sequences. In our previous study (6), RNA3 was shown to use three alternative leader-body junction sites, TRSs 3.1, 3.2, and 3.3, of which TRSs 3.1 and 3.3 are noncanonical junction sites. However, for the other EAV sg mRNAs mentioned above only a single active TRS was identified. The situation is different in PRRSV, where the use of two alternative junction sites has been documented for sg RNA4, -5, and -7 (isolate VR-2332 [28]) and RNA3 (isolate ISU-79 and some other isolates [24]).

Here, during the extension of our body TRS mutagenesis studies, two additional TRSs were discovered: one upstream of ORF4 and one upstream of ORF5. We show that, despite the fact that their primary sequences are almost identical, these alternative TRSs differ a lot in transcriptional activity. By mutagenizing the alternative TRSs of RNA3, -4, and -5, a strict correlation between individual RNA transcription levels and production of infectious progeny virus was established. Moreover, our results suggest that the glycoproteins that are expressed from these mRNAs are limiting factors during the viral life cycle and that the discontinuous step in sg mRNA synthesis is a crucial point in the regulation of structural protein expression.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and virus. Baby hamster kidney (BHK-21; ATCC CCL10) cells were maintained and used for RNA transfection experiments as described previously (40, 43). They were also used for infections with wild-type (wt) EAV (Bucurus strain) (12) and mutant viruses.

Mutagenesis of the EAV full-length cDNA clone. Translationally silent TRS mutations were introduced into a full-length cDNA clone of EAV (43) with the help of PCR mutagenesis (20). The oligonucleotides used for mutagenesis (purchased from Eurogentec, Seraing, Belgium) are listed in Table 1. Using standard recombinant DNA techniques (30), PCR fragments containing the mutations were cloned into shuttle vectors containing EAV sequences and were sequenced completely. PCR-based sequencing was carried out using the GeneAmp PCR system 2400 (Perkin-Elmer) and the ABI PRISM kit (Perkin-Elmer). Sequence reactions were analyzed using the ABI PRISM 310 genetic analyzer (Perkin-Elmer). To facilitate cloning, novel restriction sites were engineered in the EAV sequence by the introduction of translationally silent mutations. Subsequently, mutations were put into the full-length clone.

In vitro RNA transcription, transfection, wt EAV infection, RNA isolation, and Northern blots. Methods for in vitro RNA transcription from XhoI-linearized wt or mutant full-length EAV cDNA clones, as well as the procedures for transfection of BHK-21 cells, have been described previously (43). Infections with wt EAV were carried out as described previously (9) with a multiplicity of infection of 10 to 20 PFU/cell. For first-cycle RNA analysis, cells were kept at 39.5°C for 12 to 14 h and RNA was isolated using acidic phenol as described previously (47) and separated in denaturing agarose-formaldehyde gels. Gels were hybridized with ³²P-labeled oligonucleotide E154, which is complementary to the 3' end of the genome and which recognizes all plus-strand EAV RNAs (8) (Table 1). After hybridization, gels were analyzed using a Personal FX molecular imager and Quantity One software (both from Bio-Rad).

RT-PCR. For reverse transcription-PCR (RT-PCR) analysis, poly(A)-containing RNA was isolated directly from the cell lysates using oligo(dT)-carrying Dynabeads (Dynal, Oslo, Norway) according to the manufacturer's instructions. cDNA synthesis was carried out at 42°C for 1.5 h using Moloney murine leukemia virus reverse transcriptase (Life Technologies Inc.) and antisense primers E372, E359, and E122, which map to the regions downstream of the RNA3, -4, and -5 body TRSs, respectively. Subsequently, the cDNA was used as the template for an sg RNA-specific PCR (leader-body PCR). Oligonucleotides E371, E380, and E370, which are located downstream of the RNA3, -4, and -5 body TRSs, respectively, were used as antisense primers, and oligonucleotide E157, located in the EAV leader sequence, was used as the sense primer (Table 1). The PCR consisted of 35 cycles, each comprising 45 s of denaturation at 95°C, 45 s of annealing at a temperature calculated to be 5°C below the lowest melting point of the primers involved, and 2.5 min of extension at 72°C. The 35 cycles were followed by a 5-min incubation at 72°C. Products were analyzed in 2% agarose gels and photographed using a GelDoc apparatus (Bio-Rad) and Quantity One software. For sequence analysis, PCR products were isolated from gel using a GeneClean kit (BIO 101 Inc.) and used directly as a template for the sequencing reaction.

Primer extension analysis. Poly(A)-containing RNA was incubated with 10 pmol of ³²P-labeled primer (E371 and E372 for the detection of RNA3-specific extension products, EU153 and E359 for RNA4, and E370 and E122 for RNA5) in the presence of 40 mM methylmercury hydroxide in a total volume of 10 µl for 10 min at room temperature. After the incubation, methylmercury hydroxide was neutralized by adding 4 µl of a 1.4 M -mercaptoethanol solution. Primer extension reactions were performed in a total volume of 30 µl containing 5 µl of 6× first-strand buffer (Life Technologies Inc.), 1 mM deoxynucleoside triphosphates, 10 mM dithiothreitol, 40 U of RNase Out (Life Technologies Inc.), and 200 U of Superscript reverse transcriptase (Life Technologies Inc.) for 1.5 h at 42°C. Subsequently, reaction products were applied to an 8 M urea-6% acrylamide gel. A ³³P-labeled sequence reaction performed with an M13mp18 single-stranded DNA and M13 forward primer using the T7 sequencing kit (Pharmacia), as well as a ³²P-labeled 100-bp ladder (Life Technologies Inc.), were used as markers. Gels were analyzed and individual bands were quantitated as described above. All calculations were based on two or three experiments with two different primers.

IFAs. Transfected cells were seeded on coverslips and fixed with 3% paraformaldehyde in phosphate-buffered saline (PBS) at 12, 24, or 36 h posttransfection. Immunofluorescence assays (IFAs) were carried out as described previously (41). Rabbit antisera directed against nonstructural protein 2 (nsp2) (37) and nsp3 (29) were used to monitor virus replication. Rabbit antisera recognizing the E (35), GP_2b (9), and M (9) proteins and mouse monoclonal antibodies 6D10 (anti-GP₅ [1]) and 3E2 (anti-N [21]) were used as primary antibodies to detect EAV structural proteins. As secondary antibodies, a Cy3-conjugated donkey anti-rabbit immunoglobulin G (IgG) antibody and a fluorescein isothiocyanate-conjugated donkey anti-mouse IgG antibody (both from Jackson ImmunoResearch Laboratories) were used. The samples were examined with an Olympus fluorescence microscope equipped with a digital camera and Qfluoro software (Leica).

Infectious-center assays and plaque assays. For infectious-center assays, transfected cells were mixed, in different proportions, with untransfected BHK-21 cells directly after transfection. Cells were allowed to attach for 4 h, and then the medium was replaced with an overlay of medium containing 1.5% agarose and 2% fetal calf serum. For plaque assays, medium was collected from transfected cells at 24 h posttransfection and used to infect fresh BHK-21 cells. After a 1-h infection, cells were washed, and an agarose overlay was applied. At 3 to 6 days postinfection (p.i.), cells were fixed with 10% formaldehyde in PBS and stained with crystal violet.

Direct RT-PCR analysis of plaque-derived RNA. Plaques were picked and resuspended in PBS, incubated for 1 h at 37°C with 0.5 mg of proteinase K/ml-0.5% sodium dodecyl sulfate, extracted twice with phenol-chloroform-isoamyl alcohol, ethanol precipitated in the presence of 10 µg of yeast RNA as the carrier, and resuspended in 8 µl of water to which 10 pmol of the appropriate RT primer was added. cDNA synthesis, followed by PCR, was carried out as described above, except that oligonucleotides E373, E374, and E375 were used as the sense PCR primers to amplify the RNA3, -4, and -5 TRS regions in the genome, respectively (Table 1). PCR products were purified from gel and sequenced as described above, using oligonucleotides E400, E401, and E402 as the sequencing primers.

RNA structure predictions. The RNA structure of the 3'-terminal region of the EAV genome (GenBank accession no. Y07862) was predicted by using a previously described genetic algorithm implemented in the program STAR (15, 39).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effects of body TRS mutagenesis on the transcription of the corresponding sg RNA and production of infectious progeny virus. Previously, the absolute requirement of a body TRS for EAV sg RNA synthesis was shown for RNA7 only (46). To extend this study to the other sg RNA species, mutations were introduced in the EAV full-length cDNA clone in all canonical-body TRSs (UCAAC boxes) (6, 8), which were previously shown to serve as leader-body junction sites for the transcription of sg RNA2 to -7 (Table 2, mutants M2 to M7). Because all EAV structural protein genes are overlapping, every body TRS is also a part of the coding sequence for an upstream structural protein. In order to affect only sg RNA transcription, not the sequences of structural proteins encoded by upstream ORFs, the mutations introduced into TRSs were designed to be translationally silent.

View larger version (90K): [in this window] [in a new window]   FIG. 2.   Northern blot analysis of virus-specific RNA isolated from cells transfected with the wt EAV cDNA construct and with the constructs in which major TRSs used for synthesis of sg mRNA2 to -7 were disrupted by mutations (see Table 2).

View larger version (64K): [in this window] [in a new window]   FIG. 3.   Immunofluorescence analysis of BHK-21 cells transfected with TRS mutants M3.2, M4, M5, and M7. At 12 and 24 h p.t., cells were fixed and stained for nsp3 (EAV replication marker) to test for spread of the EAV-specific signal.

EAV uses alternative body TRSs for transcription of sg RNA3, -4, and -5. The observations described above prompted us to analyze the TRS mutants for RNA3, -4, and -5 in more detail. To discriminate between leaky transcription from mutated TRSs and the use of alternative TRSs, a sg mRNA-specific RT-PCR approach was designed, similar to that used by den Boon et al. (6) and van Marle et al. (47) (Fig. 4A). First, cDNA was synthesized using an RT primer located downstream of a body TRS region in an sg mRNA. Due to the nested structure of EAV mRNAs, such an RT primer also binds to the corresponding positions in all larger viral mRNAs, including the genome. Next, a PCR specific for the leader-body junction region of the desired mRNA was performed, using an internal antisense primer located just upstream of the RT primer in the body of the mRNA and a sense primer located in the leader sequence. This approach allowed us to obtain and separate by size RT-PCR products corresponding to the different mRNA species originating from all the potential TRSs upstream of the binding site of the antisense PCR primer. To make sure that such alternative mRNAs with different leader-body junctions would be detected with this method, we selected antisense primers that were located within the coding sequence of the corresponding ORF. After separation in agarose gel, PCR products were purified, and leader-body junctions were sequenced directly using an internal primer located in the leader.

View larger version (56K): [in this window] [in a new window]   FIG. 4.   (A) Scheme of the mRNA-specific RT-PCR strategy. For clarity, a hypothetical gene with two upstream TRSs (A and B) (triangles) is depicted. Black and white arrows, RT and PCR (sense and antisense) primers, respectively. The genomic RNA (g RNA) and two sg RNAs originating from TRSs A and B are shown. Double lines, PCR products corresponding to sg RNAs A and B. (B) RT-PCR specific for mRNA3 (lanes M3.2 and WT3), mRNA4 (lanes M4 and WT4), and mRNA5 (lanes M5 and WT5). WT lanes, reactions performed on RNA from the wt EAV constructs; lanes M3.2, M4, and M5, RNA from M3.2, M4, and M5 constructs, respectively. , sg RNA2; , sg RNA3.1; , sg RNA3.2; , sg RNA3.3; , sg RNA4.1; , sg RNA4.2; , sg RNA5.1; , sg RNA5.2. A 100-bp ladder was used as a size marker.

Relative abundances of RNA3 and -4 subspecies. To measure the relative use of alternative TRSs, we developed primer extension assays specific for EAV sg RNAs similar to those used by de Vries et al. (8). Because the primer was always used in a large molar excess compared to the template RNA, the relative abundances of individual products in the same reaction could be assumed to be directly proportional to the relative abundances of the corresponding RNA subspecies. A radioactively labeled oligonucleotide was hybridized to poly(A)-containing RNA isolated at 12 h p.i. from wt EAV-infected cells. This step was followed by cDNA synthesis, and the reaction products were separated in a denaturing polyacrylamide gel (Fig. 5A). For each TRS region, a pair of antisense primers located in the corresponding ORF was used, one proximal to the TRSs and one more distal, so that sg mRNAs containing each of the potential leader-body junctions would be detected.

View larger version (70K): [in this window] [in a new window]   FIG. 5.   Quantitation of the relative amounts of RNA3, -4, and -5 subspecies. (A) Primer extension analysis specific for RNA3 (lanes 1 and 2), RNA4 (lanes 3 and 4), and RNA5 (lanes 5 and 6) subspecies. For each mRNA-specific reaction pair, RNA from the cells infected with wt EAV was hybridized with an antisense primer, which was either close to (uneven lanes) or more downstream of (even lanes) the TRSs. , sg RNA2; , sg RNA3.1; , sg RNA3.2; , sg RNA3.3; , sg RNA4.1; , sg RNA4.2; , sg RNA5.1. A 100-bp ladder as well as the products of a sequencing reaction were used as size markers. (B) Semiquantitative RT-PCR analysis aimed at estimating the relative amounts of the RNA5.1 and -5.2 subspecies. Eight fivefold serial dilutions of RNA from M5.1 (top) or M5.2 (bottom) were subjected to RNA5-specific RT-PCR analysis. Indicated are the positions of PCR products corresponding to RNA species 4 and 5.2 (top) or 4 and 5.1 (bottom). (C) The same approach was used to estimate the relative amounts of RNA3.1 and -3.3 subspecies. RNA is from M3+3 (top) or M3.2 (bottom). Indicated are the positions of PCR products corresponding to RNA species 2 and 3.3 (top) or 2 and 3.1 (bottom). A 100-bp ladder was used as a size marker for both panels B and C.

Effects of single and double knockouts of alternative RNA3, -4, and -5 TRSs on mRNA synthesis. To study the importance of the use of alternative body TRSs for viral mRNA synthesis and production of infectious progeny virus, additional mutant constructs were designed (Table 2, M3.1 to M5+5) to inactivate TRSs 3.1, 4.1, and 5.2. For each RNA, we also constructed double mutants (TRS 3.1 plus 3.2 in M3+3; TRS 4.1 plus 4.2 in M4+4; TRS 5.1 plus 5.2 in M5+5). These novel mutants were transfected and analyzed as described above for the initial set of TRS mutants.

View larger version (49K): [in this window] [in a new window]   FIG. 6.   RT-PCR specific for sg RNA3 (A), RNA4 (B), and RNA5 (C). For each set of RNA-specific reactions, RT-PCR was performed on poly(A)-containing RNA from single as well as double TRS mutants and on RNA from a wt EAV construct. A 100-bp ladder was used as a size marker.

7

3

6

7

Effects of single and double knockouts of alternative RNA3, -4, and -5 TRSs on production of infectious progeny virus. Cells transfected with the six additional constructs described in the previous paragraph were also used for IFAs at 12 and 24 h p.t. using an anti-nsp3 antiserum. The phenotype of M3.1, as revealed by this assay, was similar to the phenotype of M3.2 described above. Both mutant viruses were infectious, and the infection was found to spread through the monolayer rapidly. Surprisingly, this was also true for M3+3 (data not shown), which was assumed to rely on ORF3 expression from a single, minor RNA3 subspecies, RNA3.3.

Virus titers and plaque phenotypes of single- and double-knockout TRS 3, 4, and 5 mutants. To characterize the TRS mutants in more detail, we first determined the virus titers produced upon transfection of these constructs. Cells were transfected with wt or mutant RNAs, and infectious-center assays and/or IFAs were done in order to determine the transfection efficiencies (data not shown). At 24 h p.t., the medium of transfected cultures was harvested and used for triplicate plaque assays. Cells were fixed after 3, 4, 5, or 6 days. Table 2 shows the titers of the mutant viruses and the wt control at 24 h p.t. normalized to their individual transfection efficiencies, as well as the average plaque diameters at 3 days p.i. For all TRS mutants analyzed, there was a clear difference in virus production between mutants using major RNA subspecies (3.1, 3.2, 4.2, or 5.1) for the production of corresponding structural proteins and mutants using minor RNA subspecies (3.3, 4.1, or 5.2). Furthermore, the differences in titers between M3.1 and M3.2 (1:1.1), M3.1 and M3+3 (15:1), and M4.1 and M4.2 (30:1) corresponded very well to the differences in molar amounts between the RNA subspecies produced from the active TRSs in these mutants (see above).

Sequence analysis of the progeny of TRS mutants. At 3 or 4 days p.i., plaques of TRS mutants were picked, and virion RNA was isolated as described in Materials and Methods. The relevant body TRS regions in the viral genome were amplified by RT-PCR, and PCR products were sequenced directly (Table 2).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Heterogeneity of arterivirus sg mRNAs. To express their structural proteins, arteriviruses generate a 3'-coterminal nested set of at least six sg mRNAs (reviewed in references 11 and 34). Generally, only the most 5'-proximal ORF is assumed to be expressed from each of these sg mRNAs. However, there are some notable exceptions: mRNA2 of EAV and probably the corresponding mRNA of all other arteriviruses (35) have been postulated to be functionally bicistronic. Moreover, SHFV generates two extra sg mRNA species, one of which is again presumed to be functionally bicistronic, in connection with the postulated duplication of three structural protein genes (14).

The use of alternative body TRSs for EAV mRNA3, -4, and -5. In this report, we have described novel alternative body TRSs that are used for the generation of alternative subspecies of EAV mRNA4 and -5. In both cases, as well as for the previously described alternative body TRSs for RNA3, all TRSs are located upstream of the translation initiation codons of the respective ORFs. Thus, sg mRNAs produced from these TRSs can be used to express the corresponding structural proteins (Fig. 7A). While there are no alternative AUG codons downstream of the most-upstream body TRSs for RNA3 and -4 (TRS 3.1 and 4.1), there is an AUG codon between TRSs 5.1 and 5.2 that is followed by a potential ORF of 59 codons, which is in the 1 frame relative to the ORF that encodes the GP₅ protein. However, this start codon is in an unfavorable context, and in this study, the GP₅ protein has been shown to be translated from RNA5.1. Thus, all subspecies of EAV mRNA3, -4, and -5 most likely express the same glycoprotein, i.e., GP₃, GP₄, or GP₅, respectively.

View larger version (18K): [in this window] [in a new window]   FIG. 7.   (A) RNA3, -4, and -5 body TRS regions. The sequences are aligned with respect to the translation initiation for the corresponding structural protein. (B) Putative secondary structure of the RNA3, -4, and -5 body TRS regions. The predicted structures are shown only for the regions immediately adjacent to TRSs; other regions are indicated by dots. Additional 3-bp segments at positions 10193 to 10195 and 10230 to 10232 (not shown), possible in the loop containing TRS 3.1 and TRS 3.2, provide a gain in free energy of only 1.3 kcal/mol and therefore are estimated to be rather labile. TRSs are shaded.

RNA structure of the plus-strand template may determine body TRS activity during discontinuous minus-strand transcription. It has been proposed that nidovirus discontinuous transcription resembles similarity-assisted RNA recombination (3, 27, 46). During this process, a nascent minus strand is translocated from a donor (body) template to the acceptor (leader) template, a process that may be guided by sequence similarity between the leader and body TRSs and by the secondary structure of the TRS regions. The leader TRS in the genome was predicted to be located in a loop "presented" by a striking hairpin structure (46), a mechanism that would resemble a number of antisense RNA-regulated control mechanisms that are based on interactions between single-stranded tails and hairpin loops (reference 46 and references therein). The 3' part of the stem of the leader TRS hairpin is located downstream of the TRS, and therefore it is only present in the genomic plus strand. Consequently, the formation of similar, alternative leader TRS hairpins in one or multiple subgenomic mRNAs is considered unlikely. Therefore, if the leader TRS hairpin indeed directs the reinitiation of transcription, the genomic leader in the (plus) sense would be the sole acceptor molecule during discontinuous minus-strand synthesis. The frequency with which the nascent strand is released from the body TRS, a factor that is likely to be a major determinant of the transcriptional activity of a TRS, may be determined by the local secondary structure of the plus-strand template, from which the minus-strand body is transcribed. To investigate this issue, we generated RNA structure predictions for all EAV body TRS regions, including those described in this report. An intriguing relationship was revealed: body TRSs that are used to produce major sg RNA species tend to be located in single-stranded regions of the predicted structure of the genomic template, while less-active and inactive potential TRSs (5' UCAAC 3' boxes) are fully or partially base paired. This suggests that the RNA structure of the donor template may regulate the discontinuous step in minus-strand RNA synthesis. A single-stranded body TRS may be more "open" for interactions that can promote attenuation of minus-strand synthesis and release of the nascent strand. Clearly, the possibility that the structure of the nascent minus strand itself may also play a role in this process cannot be excluded. Experiments to test this hypothesis in the EAV system are currently under way.

Role of alternative body TRSs in the viral replication cycle. In this paper, we have functionally tested the significance of the use of alternative body TRSs for the transcription of EAV RNA3, -4, and -5 using both biochemical and virological assays. We have been able to show a strict quantitative correlation between the levels of these individual sg mRNAs and production of infectious progeny virus. Recently, we have reported that the expression of all EAV proteins encoded by ORF2a to -7 is required for production of infectious progeny virus (26). The present study not only confirmed this for at least three of these proteins, GP₃, GP₄, and GP₅, but also revealed that they are produced by the virus in optimized quantities. For example, for mutants M3.1 and M3.2, a twofold reduction in the amount of mRNA3 (and that of the corresponding structural protein) resulted in a corresponding reduction of the titer of infectious progeny virus (Table 2, M3.1 and M3.2). This suggests that the expression of EAV structural proteins is tightly regulated at the level of sg RNA synthesis. By determining the rate of minus-strand release at a body TRS, the RNA structure of the plus-strand template may be part of such a regulatory mechanism.