INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Rotaviruses, members of the family Reoviridae, are the major cause of acute dehydrating diarrhea in infants and young children (16). The rotavirus genome consists of 11 segments of double-stranded (ds) RNA and is contained in the viral core along with the minor structural proteins, VP1 and VP3 (10). These two proteins function as the RNA-dependent RNA polymerase (7, 41) and guanylyltransferase (31), respectively, of the virus. The shell of the core is made up of 60 dimers of VP2 arranged as a T=1 icosahedral lattice (20). VP2 has nonspecific RNA-binding activity (18) and directly interacts with dsRNA lining the VP2 shell (33). Cores which are purified from virions and then disrupted by incubation under hypotonic conditions (open cores) have been shown to possess replicase activity which catalyzes the synthesis of dsRNA from rotavirus mRNA (5). Analysis of recombinant proteins in cell-free replication systems has demonstrated that VP1 and VP2, when combined, have replicase activity and are able to catalyze the synthesis of dsRNA (30, 46).

Rotavirus mRNAs are largely monocistronic, possessing 5' cap structures but lacking 3' poly(A) tails (8). Cores surrounded by a protein matrix consisting of 260 trimers of VP6 are referred to as double-shelled particles (32) and have an associated transcriptase activity which directs the synthesis of viral mRNA (3, 6). The fact that transcriptase activity is associated with double-shelled particles, and not with cores, indicates that VP6 may play an important role in mRNA synthesis. Infectious virions (triple-shelled particles) consist of double-shelled particles which are covered by a shell of protein composed of 780 molecules of the glycoprotein VP7 and 120 molecules of the spike protein VP4 (38).

Among the 11 segments of dsRNA that make up the viral genome, conserved sequences are present only at the 5' and 3' termini. The consensus sequence at the 5' terminus is 5'-GGC-poly(A/U), and that at the 3' terminus is 5'-aUgugaCC-3,' with fully conserved residues shown in uppercase letters (8). Sequence analyses of different strains of rotavirus have shown that the entire 5' and 3' untranslated regions (UTRs) of homologous genome segments are highly conserved, sometimes more so than the nucleotide sequence of the open reading frame of the segments (8). The functions of the conserved termini of the viral RNAs are not entirely clear, but these regions may contain cis-acting signals which are important for replication, transcription, and packaging. Indeed, recent work with cell-free replication systems has shown that within the conserved sequence at the 3' end of viral mRNA is a cis-acting signal which is essential for promoting minus-strand synthesis, i.e., the minimal essential promoter (29, 43, 44). Analysis of mutant viral RNAs in vitro has also revealed that rotavirus mRNAs contain two other signals that, although not essential, enhance minus-strand synthesis (29, 44). One of these enhancement signals is located immediately upstream of the minimal essential promoter at the 3' end of the mRNA, while the other is located at the 5' end of the mRNA. Because the enhancement signals are located in regions of the viral mRNA that are not highly conserved, it is possible that secondary structures common to all 11 species of viral RNA, as opposed to the primary nucleotide sequence of the regions, are responsible for the enhancement of minus-strand synthesis.

From the location of the cis-acting signals that promote minus-strand synthesis and from the predicted secondary structure of the SA11 gene 8 mRNA, complementary regions within the conserved sequences at the ends of the RNA are proposed to interact to form a modified panhandle structure that functions as the promoter for minus-strand synthesis (Fig. 1). Two features of the predicted secondary structure of the gene 8 mRNA are that the 3'-terminal 12 nucleotides are not base paired and that the 5' UTR forms a stem-loop motif. We report here that mutations in the gene 8 mRNA which increase the extent of complementarity between the 5' and 3' ends significantly decrease the efficiency of minus-strand synthesis. While mutant RNAs with complete or near-complete terminal complementarity replicated poorly, they were able to compete for VP1. These data indicate that the sequence making up the minimal essential promoter must be single stranded for the mRNA to function as an efficient template for minus-strand synthesis but that the strandedness of the minimal essential promoter does not affect the ability of the polymerase to recognize the viral RNA.

View larger version (38K): [in this window] [in a new window]   FIG. 1.   The optimal secondary structures of wild-type and mutant SA11 gene 8 RNAs and wild-type OSU gene 9 RNA were predicted based on free-energy minimization with the mfold program. Shown are the portions of the secondary structures illustrating the computed interaction between the 5' and 3' ends of the RNA. Mutations in the sequences are presented in red, lowercase letters. The conserved residues that are part of the 3' essential cis-acting replication signal are shown in blue, and the location of deletions is shown in green. kc/m, kcal/mol.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Preparation of open cores. Rotavirus SA11-4F virions were propagated in MA104 cells and purified by CsCl centrifugation (29). The method of Chen et al. (5) was used to prepare open cores from purified virus. The concentration of protein in open core preparations was determined with a Bio-Rad protein assay kit.

Removal of genomic dsRNA from open cores. Open cores (100 µg) were treated for 60 min at 30°C with 340 U of micrococcal nuclease (Worthington Biochemical Co., Lakewood, N.J.) per ml in a reaction mixture (500 µl) containing 10 mM Tris-HCl (pH 8.0), 10 mM NaCl, and 1 mM CaCl₂. After complete digestion of dsRNA in the open core preparation was confirmed by electrophoresis on a 1% agarose gel, the nuclease was inactivated by the addition of EGTA to 3 mM.

Construction of templates for T7 transcription. PCR was used as follows to introduce mutations in the gene 8 cDNA and to directly link the gene 8 cDNA to a promoter for T7 RNA polymerase. The plasmid SP65g8R, containing a full-length gene 8 cDNA of simian rotavirus SA11, was digested with BamHI and HindIII to release the gene 8 cDNA insert (35). The insert was gel purified with Qiaex II reagents (Qiagen, Valencia, Calif.) and used as the template in amplification reaction mixtures containing 25 U of Pfu DNA polymerase (Stratagene, La Jolla, Calif.) per ml under the following conditions: 94°C at 1 min, 48°C at 1 min, and 72°C at 2 min (30 cycles). The amplified DNA produced by Pfu polymerase is blunt ended and does not contain 3'-overhang A residues. To generate the G8-5'-B1 mutant DNA under the control of a promoter for T7 RNA polymerase, the plus-sense primer T7G8/5'-B1 (5'-TAATACGACTCACTATAGGtcacatAAGCGTCTCAGTCGCCGTTTG-3') and the minus-sense primer G8/3'-wt (5'-GGTCACATAAGCGCTTTCTATTCT-3') were used for PCR. To produce the G8-5'-B2 and G8-5'-B4 DNAs, the plus-sense primers T7G8/5'-B2 (5'-TAATACGACTCACTATAGGCTacatAAGCGTCTCAGTCGCCGTTTG-3') and T7G8/5'-B4 (5'-TAATACGACTCACTATAGAAGCGTCTCAGTCGCCGTTTG-3'), respectively, and the minus-sense primer G8/3'-wt were used instead. The G8-5'/3'-B1 and G8-3'-B1 DNAs were generated by including the plus-sense primers T7G8/5'-B1 and T7G8/5'-wt (5'-TAATACGACTCACTATAGGCTTTTAAAGCGTCTCAGTCG-3'), respectively, in amplification reactions along with the minus-sense primer G8/3'-B1 (5'-GgctgtgcAA GCGCTTTCTATTCTTGC-3'). Sequences within primers that define the promoter for T7 RNA polymerase are underlined, and sequences in the primers that represent mutations of the wild-type gene 8 sequence are shown in lowercase letters. The amplified DNAs were purified by phenol-chloroform extraction and then used in T7 transcription reactions to produce gene 8 RNAs. Unless otherwise noted, the 5' and 3' ends of the transcripts made from the amplified DNAs are identical in sequence to those of authentic (wild-type) gene 8 mRNA.

Synthesis of viral RNAs by run-off transcription. Viral RNAs were synthesized with Ambion MEGAscript T7 transcription kits according to the protocol supplied by the manufacturer (29). To prepare ³²P-labeled RNA, GTP was reduced to one-fifth of the recommended amount and 40 µCi of [-³²P]GTP (800 Ci/mmol) was added to an otherwise standard 40-µl reaction mixture. After incubation, unincorporated nucleotides were removed from transcription products by passage through Sephadex G-25 spin columns. The RNA quality was assessed electrophoretically (5), and the RNA concentration was determined spectrophotometrically.

Cell-free replication assay. Replication assays were carried out as described by Chen et al. (5) with some modification. Specifically, the standard 20-µl reaction mixture contained 50 mM Tris-HCl (pH 7.2), 5 mM magnesium acetate, 5 mM dithiothreitol, 200 µM (each) ATP, CTP, and GTP, 20 µM UTP, 10 µCi [-³²P]UTP (800 Ci/mmol), 0.1 µg of micrococcal nuclease-treated open cores, and 0.1 µg of template RNA. Unless otherwise noted, reaction mixtures of competition assays contained 0.1 µg each of reporter RNA and competitor RNA. Reaction mixtures were incubated for 2 h at 32°C. Replication products were resolved by electrophoresis on 12.5% polyacrylamide gels containing sodium dodecyl sulfate (19) and detected by autoradiography. Bands of ³²P-labeled dsRNA were quantified with a Molecular Dynamics PhosphorImager (model 445SI).

Oligonucleotide inhibition assay. Three deoxyoligonucleotides were tested for the ability to inhibit the synthesis of dsRNA in replication assays, G8-WT-5' (5'-GGCTTTTAAAGCGTCTCA-3'), which corresponds in sequence to the first 18 nucleotides of the gene 8 mRNA; G8-B1-5' (5'-GGtcacatAAGCGTCTCA-3', mutant residues shown in lowercase letters), which corresponds in sequence to the first 18 nucleotides of the G8-5'-B1 RNA and is fully complementary to the 3'-terminal 14 nucleotides of the gene 8 mRNA; and G8-507 (5'-TCCTTCTGCAGTTATTGTAGTTTC-3'), which is complementary in sequence to nucleotides 484 to 507 of the gene 8 mRNA. In inhibition assays, oligonucleotides were added to standard reaction mixtures lacking open cores. After incubation for 1 h at 32°C, open cores were added and the reaction mixtures were incubated an additional 2 h at 32°C. The dsRNA products were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and quantified with a PhosphorImager.

Electrophoretic mobility shift assay. To generate the RNA probe for the mobility shift assays, SP72g8R40 was first cleaved with SacII and blunt ended with T4 DNA polymerase (5). The ³²P-labeled RNA probe, SP72-v3'-40, was then synthesized from the linearized plasmid with the Ambion MEGAshort transcription system according to the protocol of the manufacturer, except that the concentration of UTP was reduced to one-fourth and 50 µCi of [-³²P]UTP (800 Ci/mmol; Amersham) was included per 20 µl of reaction mixture. The RNA product was 73 nucleotides in length, with the sequence of the last 40 residues identical to that found at the 3' terminus of the gene 8 mRNA. The remaining 33 nucleotides of the probe represent SP72-specific sequences. The RNA probe was gel purified as described elsewhere (27).

Data interpretation. All experiments described in this study were repeated a minimum of three times, and the results obtained from the repetitions were consistent with one another. Unless otherwise noted, the values provided in the figures are the nonaveraged results of a single experiment and are representative of the values obtained for the other repetitions of the same experiment.

Secondary structure of rotavirus RNAs. The secondary structures of rotavirus RNAs were computed with the mfold program, version 2.3, developed by Zuker and Turner (37, 42, 48, 49) and made available on the home page of Michael Zuker of Washington University, St. Louis, Mo. (http://www.ibc.wustl.edu/~zuker/).

Nucleotide sequence accession numbers. The GenBank accession numbers of the NSP2 genes analyzed with the mfold program are LO4529 (DS1), X57944 (Hu-SG2), X94562 (IS2), LO4530 (NCDV), X06722 (OSU), AB009625 (PO-13), Z21640 (RF), LO4530 (SA11), LO4533 (Ty-1), J02420 (UK), and LO4534 (Wa). The accession number for OSU gene 9 is X04613.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Predicted secondary structure of the gene 8 mRNA. Previous studies with a number of viruses, including picornaviruses (24), tobamoviruses (12, 21), human immunodeficiency virus type 1 (13, 23), and Q (2), have shown that secondary structures in viral RNAs can play important roles in genome replication, assembly, and translation. As an initial step toward identifying secondary structures in rotavirus plus-strand RNAs that might be involved in the synthesis of minus-strand RNAs, the RNA encoding NSP2 of simian (SA11, gene 8), avian (PO-13 and Ty-1), bovine (NCDV, RF, and UK), human (DS1, Hu-SG2, IS2, and Wa), and porcine (OSU) strains of rotavirus was analyzed with the mfold program (version 2.3) made available by M. Zucker. The mfold program predicts the secondary structures of RNAs based on their minimal computed free energies according to the rules established by Turner and others (37, 42). All of the 20 most-stable secondary structures calculated for the SA11 gene 8 mRNA predicted that the 5' and 3' ends of the RNA would interact in the manner illustrated in Fig. 1A. The major features of the structure produced by the interaction of the ends of the RNA are that (i) the 3'-terminal 12 residues are non-base paired, including the minimal essential promoter; (ii) the 5' UTR (residues 1 to 45) forms a stem-loop motif; (iii) the initiation codon (residues 46 to 48) is not base paired and is located immediately downstream of the stem-loop motif of the 5' UTR; and (iv) residues downstream of the initiation codon (residues 49 to 80) and upstream of the non-base-paired 3'-terminal stretch (residues 1015 to 1045) interact to form a panhandle that juxtaposes the 5' and 3' ends of the RNA.

Structural similarity of SA11 gene 8 and OSU gene 9 RNAs. Analysis of mutated SA11 gene 8 RNAs in a cell-free replication system has shown that both the 5' and 3' ends of the RNA contain cis-acting signals that are important for minus-strand synthesis (29). Folding of the gene 8 RNA in a way that allows interaction of the 5' and 3' termini would provide a mechanism by which the cis-acting signal at the 5' end could be brought close to those at the 3' end to form a structure that serves as the promoter for minus-strand synthesis. Besides the gene 8 RNA of SA11, cell-free replication assays have shown that the 5' and 3' end of the VP7 RNA (gene 9) of OSU also contain cis-acting signals that enhance minus-strand synthesis (43, 44). To address the possibility that the OSU gene 9 RNA, like the SA11 gene 8 RNA, might fold such that its 5'- and 3'-terminal replication signals were close to each other, the mfold program was used to predict the folded structure of the gene 9 RNA. Of the 22 most-stable structures predicted, all showed that the 5' and 3' ends of the OSU gene 9 RNA would interact in a manner nearly identical to that exhibited by the SA11 gene 8 RNA (Fig. 1B). These results support the hypothesis that interactions between the termini of these RNAs could lead to the formation of a structure that promotes minus-strand synthesis.

Base pairing of the 3' terminus of the RNA template inhibits minus-strand synthesis. A key feature of the computed secondary structure for the wild-type gene 8 RNA is that the last 12 bases of the RNA are non-base paired (Fig. 1A). To determine whether synthesis of minus-strand RNA on the gene 8 template RNA requires that this region be single-stranded, the 5'-terminal sequence was mutated such that the 5' and 3' ends of the gene 8 mRNA were either fully (G8-5'-B1) or partially (G8-5'-B2) complementary (Table 1). The proposed secondary structure of the mutant RNA, G8-5'-B1, predicts that the first and last 14 bases of the RNA will base pair and that the stem-loop of the 5' UTR of the wild-type RNA will be lost, with a new stem-loop generated beginning with residue 15 (Fig. 1D). The structure of G8-5'-B2 is predicted to be like that of G8-5'-B1, except that the third residues from the 5' and 3' ends of the RNA cannot base pair (Fig. 1E). The ability of G8-5'-B1 and G8-5'-B2 to function as templates for dsRNA synthesis was evaluated in a cell-free replication system containing SA11 open cores. The results showed that compared with assays containing wild-type gene 8 RNA, assays containing G8-5'-B1 and G8-5'-B2 synthesized approximately 100- and 10-fold-less dsRNA, respectively (Fig. 2 and Table 2).

View larger version (45K): [in this window] [in a new window]   FIG. 2.   Synthesis of dsRNAs from wild-type and mutant template RNAs by the cell-free replication system. Reaction mixtures contained micrococcal nuclease-treated open cores, [32P]UTP, and either 0.1 µg of the indicated template RNA or no RNA (Without). The 32P-labeled dsRNA products were resolved by electrophoresis on a 12.5% polyacrylamide gel and detected by autoradiography.

View larger version (16K): [in this window] [in a new window]   FIG. 3.   Effect of mutagenesis at the 5' end of the gene 8 mRNA on synthesis of dsRNA in vitro. Wild-type RNA (Wt) and the mutant gene 8 mRNAs G8-5'-B1, -B2, -B4, and -B6 were used as templates in the cell-free replication system. The dsRNA products of the reaction mixtures were resolved by gel electrophoresis and quantified with a PhosphorImager. The relative values of dsRNA are shown with the value for the wild type normalized to 100%.

View larger version (53K): [in this window] [in a new window]   FIG. 4.   Partial rescue of the defective template activity of the G8-5'-B1 RNA by introduction of compensatory mutations in the 3' terminus. The 32P-labeled dsRNA products made in reaction mixtures containing wild-type or mutant gene 8 template RNA or no template RNA (Without) were resolved on a 12.5% polyacrylamide gel and detected by autoradiography. Band intensities were determined by phosphorimaging.

Inhibition of RNA replication by complementary oligonucleotides. The change of six nucleotides at the 5' end of wild-type gene 8 RNA hypothetically allows the ends of G8-5'-B1 to fully anneal and thus form a "closed" panhandle structure (Fig. 1D). The fact that G8-5'-B1 replicates poorly suggests that the panhandle structure of the wild-type RNA must include a single-stranded 3' terminus in order for the viral RNA polymerase to efficiently replicate the RNA. If this is the case, annealing of a complementary oligodeoxynucleotide to the 3' terminus of the RNA would be expected to inhibit replication. To test this possibility, the Wt-5', B1-5', and G8-507 oligonucleotides were synthesized. The sequence of the Wt-5' oligonucleotide is identical to the first 18 residues of the wild-type gene 8 RNA, while that of the B1-5' oligonucleotide is identical to the first 18 residues of the mutant G8-5'-B1 RNA and therefore has the potential to form a complete duplex with the 3' terminus of the wild-type RNA. The control oligonucleotide G8-507 is complementary to residues 484 to 507 of the gene 8 mRNA. The oligonucleotides were added in varying concentrations to replication assays containing wild-type gene 8 RNA, and the dsRNA products were detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and quantified with a PhosphorImager (Fig. 5). At concentrations of 20 nM, the B1-5' oligonucleotide inhibited replication by more than fivefold, whereas the Wt-5' and G8-507 oligonucleotides inhibited replication only slightly (<twofold). At 500 nM, the presence of the B1-5' oligonucleotide reduced replication by more than 100-fold, while the level of inhibition caused by the Wt-5' and G8-507 oligonucleotides was less than threefold (Fig. 5). These results support the conclusion that base pairing of the predicted 3' single-stranded terminus of gene 8 wild-type mRNA interferes with its ability to function as a template for replication.

View larger version (38K): [in this window] [in a new window]   FIG. 5.   Effect of complementary oligonucleotides on the replication of the gene 8 RNA in vitro. Wild-type gene 8 template RNA (0.1 µg) was replicated either alone (lane 3) or in the presence of the indicated concentrations of the oligonucleotides Wt-5', B1-5', or G8-507 (lanes 4 to 12). The 32P-labeled dsRNA products were resolved on a 12.5% polyacrylamide gel and quantified by phosphorimaging. The amount of dsRNA synthesized in the absence of oligonucleotide (lane 3) was considered to be 100%. Lane 1, 32P-labeled genomic dsRNA of SA11-4F; lane 2, no template RNA was added to the reaction mixture.

G8-5'-B1 and G8-5'-B2 RNAs interfere with replication of wild-type gene 8 mRNA. The G8d45-543 RNA retains the complete sequence of the 5' and 3' UTRs of the wild-type gene 8 mRNA but lacks residues 45 to 543. The mfold program predicts that the G8d45-543 RNA contains the 5' stem-loop of the wild-type RNA but that this structure is not positioned opposite the 3' UTR of the RNA and therefore that G8d45-543 does not contain a 5'-to-3' panhandle (Fig. 6A). Like the G8-5'-B1 and G8-5'-B2 RNAs, we observed that the G8d45-543 RNA replicates efficiently in replication assays when the mutant RNA is the only template that is present in the reaction mixture (Fig. 2). However, in assays containing equal amounts of G8d45-543 RNA and wild-type gene 8 mRNA, replication of the mutant template was reduced by approximately 10-fold and therefore is inefficient in comparison to replication of the wild-type template (Fig. 6B). This same phenotype has been described for other gene 8 RNAs that contain deletions at or near their 5' ends (29, 44). The three most-likely possibilities to explain the inhibitory effect that wild-type RNA has on the replication of the G8d45-543 RNA are (i) the viral RNA polymerase, i.e., VP1, binds the wild-type RNA more efficiently than G8d45-543 RNA; (ii) the viral RNA polymerase initiates minus-strand synthesis more efficiently on the wild-type RNA than the G8d45-543 RNA; or (iii) a combination of (i) and (ii). To determine whether the G8-5'-B1 and G8-5'-B2 RNAs were like wild-type RNA and were able to competitively interfere with the replication of the G8d45-543 RNA, replication assays were performed with equal amounts of G8d45-543 RNA and either G8-5'-B1 or G8-5'-B2 RNA. The analysis showed that despite their replication-defective phenotypes, the G8-5'-B1 and G8-5'-B2 RNAs inhibited the replication of G8d45-543 to at least the same extent (>10-fold) as wild-type gene 8 RNA (Fig. 6B). Thus, the mechanism by which wild-type RNA and the mutant RNAs, G8-5'-B1 and G8-5'-B2, interfere with replication of G8d45-543 is probably the same and does not correlate with the level of replication of the competitor RNA in the assay. The inhibitory effects are rotavirus specific given that the replication of G8d45-543 was not affected by the addition of 10-fold excess of yeast RNA in reaction mixtures (Fig. 6B).

View larger version (20K): [in this window] [in a new window]   FIG. 6.   Inhibition of the replication of the G8d45-543 RNA by wild-type and mutant RNAs. (A) The optimal secondary structure for the 5' and 3' ends of the G8d45-543 RNA is shown. Note that the 5' end of the RNA forms the same stem-loop structure as predicted for SA11 wild-type gene 8 mRNA but that the 5' and 3' ends of the RNA do not interact due to the deletion of residues 45 to 543 (Fig. 1A). (B) Replicase assays contained 0.1 µg of the internal deletion mutant RNA, G8d45-543 and no competitor RNA (Without), 0.1 µg of competitor RNA (G8-Wt, G8-5'-B1, G8-5'-B2, or Yeast 1X), or 1.0 µg of competitor RNA (Yeast 10X). The 32P-labeled dsRNA products were resolved on a 12.5% polyacrylamide gel and quantified with a PhosphorImager. The amount of G8d45-543 dsRNA synthesized in the absence of competitor RNA was considered to be 100%, and the relative amount of G8d45-543 dsRNA synthesized in the presence of the competitor RNAs is given (percentage of replication of reporter RNA).

View larger version (40K): [in this window] [in a new window]   FIG. 7.   Inhibition of gene 8 RNA replication by competitor RNAs. 32P-labeled wild-type gene 8 RNA (0.1 µg) was replicated alone (lane 2) or in the presence of the indicated amount (in micrograms) of cold competitor RNA. The 32P-labeled dsRNA products were resolved by gel electrophoresis and detected by autoradiography. The intensities of the 32P-labeled bands were determined with a PhosphorImager. The amount of gene 8 dsRNA made in the absence of competitor RNA was considered to be 100%. Lane 1, 32P-labeled genomic dsRNA of SA11-4F.

G8-5'-B1 efficiently binds RNA polymerase. We have shown that the G8-5'-B1 mutant, although poorly replicated, is able to inhibit the replication of both wild-type and G8d45-543 RNAs. The simplest explanation for this observation is that the mutations in G8-5'-B1 RNA do not affect the ability of the RNA to bind the polymerase, and therefore the RNA is able to specifically interfere with the replication of other template RNAs. In a previous study, Patton (27) used short gene 8-specific RNA probes in electrophoretic mobility shift assays to show that the viral RNA polymerase, VP1, specifically recognizes and binds to the 3' end of rotavirus mRNA. Because of the length of the RNA, it was not possible to use G8-5'-B1 as the probe in electrophoretic mobility shift assays and thereby directly explore the question of whether the mutant RNA efficiently binds VP1. To overcome this limitation, competition assays were performed instead to evaluate the ability of the G8-5'-B1 RNA, gene 8 wild-type RNA, and yeast RNA to competitively interfere with the binding of VP1 to the ³²P-labeled probe, SP72-v3'-40. This probe is 73 nucleotides in length, and the sequence of its last 40 nucleotides is identical to that present at the 3' end of the gene 8 RNA. As shown by assay with baculovirus-expressed recombinant VP1, ³²P-labeled SP72-v3'-40 is able to interact with VP1 to form a complex that can be detected in the electrophoretic mobility shift assay (Fig. 8A). The results of the competition assays showed that for all concentrations of competitor RNA examined (0.2, 1.0, and 4.0 µg), wild-type RNA and G8-5'-B1 interfered with the formation of the VP1-SP72-v3'-40 complex to a similar extent (Fig. 8B). This interference was specific given that yeast RNA had no effect on the interaction of VP1 and the probe. These data indicate that although G8-5'-B1 is replicated 100-fold less efficiently than wild-type gene 8 RNA, it is able to bind VP1 as efficiently as the wild-type RNA. Hence, events other than merely the binding of the RNA polymerase to the RNA template are important for efficient synthesis of minus-strand RNA.

View larger version (34K): [in this window] [in a new window]   FIG. 8.   The replication-defective RNA, G8-5'-B1, competitively interferes with the formation of VP1-probe complexes. (A) The sequence of the 73-nucleotide SP72-v3'-40 probe is given, and the portion of the probe that corresponds to the last 40 residues of the gene 8 RNA is underlined. 32P-labeled SP72-v3'-40 probe (10 pmol) and 2.5 µg of rabbit liver tRNA were incubated alone, with open cores or with purified recombinant VP1 (60 ng) or VP2 (400 ng) (30). (B) 32P-labeled SP72-v3'-40 probe (27 ng; 1.2 pmol) was incubated alone (lane 1) or with open cores and no competitor RNA (lane 11), or with 0.2 µg (0.6 pmol), 1.0 µg (2.8 pmol), or 4.0 µg (11.4 pmol) of G8-wt, G8-5'-B1 RNAs, or yeast RNA. The molar ratios of competitor RNA to probe in assays containing 0.2, 1.0, and 4.0 µg of G8-wt and G8-5'-B1 RNA were 0.5, 2.3, and 9.5, respectively. VP1-probe complexes were detected in reaction mixtures by electrophoresis on nondenaturing 8% polyacrylamide gels and by autoradiography.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Sequence analysis has shown that the 5' and 3' ends of rotavirus RNAs, like those of other RNA viruses with segmented genomes, e.g., reovirus (25), orbivirus (34), wound tumor virus (1, 45), and influenza virus (9, 39), are highly conserved. This evolutionary conservation suggests that the terminal sequences contain signals essential for replication and translation of rotavirus RNA. Indeed, recent studies with cell-free replication systems have shown that the conserved sequence at the 3' end of rotavirus mRNA forms the minimal cis-acting signal required for minus-strand synthesis (29, 44). These studies also revealed that the 5' end of at least two types of rotavirus mRNAs contains a signal required for maximal levels of minus-strand synthesis. Based on the location of the cis-acting replication signals and on secondary-structure predictions, we and others have proposed that rotavirus mRNAs fold in a manner such that interaction of the 5' and 3' termini is allowed, thereby producing a panhandle structure that serves as the promoter for minus-strand synthesis (15, 26, 28, 29). The structure of the rotavirus mRNA is suggested therefore to be similar to the panhandles formed by the RNAs of influenza virus, which have been shown to promote RNA synthesis (11, 14, 39). Besides rotavirus mRNAs, analysis of the secondary structure of other Reoviridae mRNAs, including those of wound tumor virus (1), reovirus (4), and rice dwarf virus (40), has indicated that the mRNAs of these viruses may also fold to produce panhandle-shaped molecules.

In this study, we have used current RNA-folding algorithms to predict the secondary structures of all 11 rotavirus NSP2 genes available in GenBank. The computed secondary structure of the SA11 NSP2 mRNA predicts that the last 12 nucleotides of the RNA are single-stranded and that the 5' UTR folds to form a stem-loop structure. The 5' stem-loop and the 3' single-stranded tail are proposed to be held adjacent to each other by base pairing between the region of the RNA immediately downstream of the 5' UTR and upstream from the 3' single-stranded tail. The mfold program predicted that the most-stable secondary structures for the NSP2 mRNA of six of the other eight mammalian strains of rotavirus (DS-1, Hu-SG2, NCDV, OSU, RF, and Wa) formed a panhandle structure like that of SA11 NSP2 mRNA. The two exceptions were for the UK and IS2 strains of virus. In the case of UK, while the most-stable structures did not show the SA11-type panhandle structure, the two next-most-stable structures did, and these two differed only minimally in free energy (<2 kcal/mol) from the most-stable structure. For the IS2 strain, no dominant structure emerged for the NSP2 mRNA. Instead, of the 10 stable structures computed for the mRNA, four distinct classes of structures were predicted, with one class like that of the SA11 panhandle. Analysis of the predicted secondary structures of other rotavirus genes (e.g., OSU VP7 mRNA, porcine CN86, and CC86 gene 11 mRNAs) suggests that ends of other mRNAs may interact to form a panhandle and a 3' single-stranded tail. Of particular note, analysis of the reovirus T1L S2 gene by the mfold program indicated that the ends of the S2 mRNA interact to form a 5' and 3' panhandle and a 3' single-stranded tail that are remarkably similar to those predicted for the SA11 NSP2 and the OSU VP7 mRNAs.

The results of experiments (present and previous) have demonstrated that deletion or mutation of residues making up the 5' UTR reduces the efficiency at which the rotavirus NSP2 mRNA replicates, particularly in competition assays that also contain wild-type RNA (Fig. 3) (29, 44). This suggests that the 5' UTR and its secondary structure play a key but an as-yet-undefined role in minus-strand synthesis. However, studies on the internal deletion mutant G8d45-543 indicate that the mere presence of the 5' UTR and its secondary structure is not sufficient to allow the mRNA to undergo maximum levels of replication, even when the 3' half of the mRNA is present. Instead, analysis of G8d45-543 indicates that along with sequences in the 5' and 3' UTRs, an internal sequence of the RNA is required for maximal RNA replication, perhaps because it promotes folding of the RNA so that the 5' stem-loop and 3' tail are correctly juxtaposed. From these results, it is possible to infer that two types of cis-acting replications exist within the viral template mRNA, those that contain recognition signals that stimulate the binding of viral proteins necessary for RNA synthesis (e.g., VP1) and those that are necessary for the folding of the RNA so that the promoter for minus-strand synthesis is created.

The last 12 nucleotides of the rotavirus gene 8 mRNA contain a cis-acting signal that is essential for minus-strand synthesis and is predicted to project as a single-stranded tail from the panhandle region of the RNA. In this study, we used two approaches to show that the predicted single-stranded nature of the 3' tail is important to the function of the essential replication signal. Firstly, the introduction of mutations into the 5' end of the gene 8 mRNA that caused complete (G8-5'-B1) or near-complete (G8-5'-B2) complementarity between the 5' end and the 3' tail were found to significantly inhibit minus-strand synthesis (>10-fold). Furthermore, compensatory mutations that were introduced into the 3' end of the replication-defective G8-5'-B1 and G8-5'-B2 RNAs that decreased the extent of complementarity between the 5' end and 3' tail were found to stimulate minus-strand synthesis. The fact that deletion of residues 2 to 8 of the gene 8 mRNA (G8-5'-B4) produced a template which could be replicated efficiently and, except for the presence of a 3' single-stranded tail, was structurally identical to the replication-defective G8-5'-B1 RNA provides compelling evidence that the 3' end of the mRNA must be single-stranded for efficient minus-strand synthesis. Secondly, the presence of an oligonucleotide (B1-5') in the cell-free replication that was fully complementary to the 3' tail of the gene 8 mRNA inhibited the synthesis of minus-strand RNA by more than 100-fold (500 nM). In contrast, the same concentration of an oligonucleotide that was not complementary to the 3' tail and that corresponded in sequence to the first 18 nucleotides of the gene 8 mRNA had only a minimal effect on minus-strand synthesis (<threefold). Taken together, these data indicate that the 3' end of the template mRNA must be single stranded for the 3' cis-acting replication signal to function and for dsRNA synthesis to occur. The fact that the 3' cis-acting replication signal may not function unless it is single stranded may explain why the viral replicase in open-core preparations is unable to use the endogenous dsRNA genome as a template for the synthesis of minus-strand RNA.

Although replication defective, the mutant G8-5'-B1 RNA was shown to be able to competitively interfere with the replication of other rotavirus mRNAs in vitro and to do so as efficiently as wild-type mRNA. Hence, the G8-5'-B1 RNA is phenotypically wild type with respect to its ability to recruit at least one of the trans-acting protein factors in the cell-free system that is required for minus-strand synthesis. A competitive electrophoretic mobility shift assay was used to show that the G8-5'-B1 RNA binds the RNA polymerase VP1 as efficiently as wild-type RNA. While the recognition signal for VP1 in the mRNA has not been precisely mapped, these data indicate that whether the 3' essential replication signal is single or double stranded has no effect on the ability of the polymerase to bind to the RNA. Furthermore, the fact that gene 8 mutant RNAs containing deletions of residues 1 to 10 (g8Rd1 to -10 [29]), 4 to 50 (g8Rd4 to -50 [29]), and 45 to 453 (G8Rd45 to -543) retain the ability to replicate to a level within threefold of the wild-type mRNA suggests that the 5' portion of the mRNA does not contain the recognition signal for the RNA polymerase. Instead, as was indicated in previous studies with electrophoretic mobility shift assays (27), the polymerase recognition signal probably resides in the 3'-terminal region of the mRNA.

The finding that the G8-5'-B1 RNA can efficiently bind VP1 but is replication defective supports the conclusion that the association of the polymerase with the template RNA is not the sole factor which determines whether an RNA undergoes replication. Indeed, it is known that VP2 is also essential for RNA replication (22, 46) and that, when combined, VP1 and VP2 possess replicase activity that stimulates the synthesis of dsRNA from wild-type mRNA in vitro (30). VP2 is the most abundant protein in the open-core replication system and has been shown to have RNA-binding activity that is nonspecific and that recognizes both single-stranded RNA and dsRNA (18). As a result, both VP1 and VP2 can be expected to associate with the mutant template G8-5'-B1 in the cell-free replication system but this interaction does not lead to efficient levels of dsRNA synthesis. Thus, the replication-defective phenotype of G8-5'-B1 is probably related not to a defect in the ability to recruit the minimum two proteins required for RNA replication but, instead, to a defect in the ability of these proteins to initiate minus-strand synthesis on the template. More precisely, the data provide evidence that minus-strand initiation, but not VP1 and VP2 binding, requires that the 3' cis-acting replication exist in a single-stranded form, i.e., an open form.

In summary, the rotavirus gene 8 mRNA is predicted to fold to form a panhandle with a single-stranded 3' tail. Data were obtained with a cell-free replication system showing that the 3' tail must exist in a single-stranded form if the mRNA is to function as an efficient template for minus-strand synthesis. While RNA-folding programs do not take into account the effect that proteins may have on the secondary structure of the RNA, the rotavirus core proteins VP1, VP2, and VP3 may serve to promote the formation of the panhandle. In particular, since it is thought that the polymerase VP1 binds the 3' end of mRNA (27, 31) and the capping enzyme VP3 binds the 5' end of mRNA (31), and both these proteins bind to adjacent sites at the NH₂ end of VP2 (47), the VP1-VP2-VP3 complex may catalyze the formation of the panhandle by bringing the 5' and 3' ends of the RNA close to each other. Given that the 3' end of the template RNA for replication must be single stranded for minus-strand synthesis to occur, it remains unclear how the RNA polymerase is able to initiate mRNA synthesis on the double-stranded genome segments. Perhaps a mechanism exists by which the 5' end of the dsRNA segments is partially melted, thus providing a single-stranded cis-acting signal at the 3' end of the minus strands that allows initiation of plus-strand synthesis.