Effects of Defined Mutations in the 5′ Nontranslated Region of Rubella Virus Genomic RNA on Virus Viability and Macromolecule Synthesis

Abstract

The 5′ end of the genomic RNA of rubella virus (RUB) contains a 14-nucleotide (nt) single-stranded leader (ss-leader) followed by a stem-and-loop structure [5′(+)SL] (nt 15 to 65), the complement of which at the 3′ end of the minus-strand RNA [3′(−)SL] has been proposed to function as a promoter for synthesis of genomic plus strands. A second intriguing feature of the 5′ end of the RUB genomic RNA is the presence of a short (17 codons) open reading frame (ORF) located between nt 3 and 54; the ORF encoding the viral nonstructural proteins (NSPs) initiates at nt 41 in an alternate translational frame. To address the functional significance of these features, we compared the 5′-terminal sequences of six different strains of RUB, with the result that the short ORF is preserved (although the coding sequence is not conserved) as is the stem part of both the 5′(+)SL and 3′(−)SL, while the upper loop part of both structures varies. Next, using Robo302, an infectious cDNA clone of RUB, we introduced 31 different mutations into the 5′-terminal noncoding region, and their effects on virus replication and macromolecular synthesis were examined. This mutagenesis revealed that the short ORF is not essential for virus replication. The AA dinucleotide at nt 2 and 3 is of critical importance since point mutations and deletions that altered or removed both of these nucleotides were lethal. None of the other mutations within either the ss-leader or the 5′(+)SL [and accordingly within the 3′(−)SL], including deletions of up to 15 nt from the 5′(+)SL and three different multiple-point mutations that lead to destabilization of the 5′(+)SL, were lethal. Some of the mutations within both ss-leader and the 5′(+)SL resulted in viruses that grew to lower titers than the wild-type virus and formed opaque and/or small plaques; in general mutations within the stem had a more profound effect on viral phenotype than did mutations in either the ss-leader or upper loop. Mutations in the 5′(+)SL, but not in the ss-leader, resulted in a significant reduction in NSP synthesis, indicating that this structure is important for efficient translation of the NSP ORF. In contrast, viral plus-strand RNA synthesis was unaffected by the 5′(+)SL mutations as well as the ss-leader mutations, which argues against the proposed function of the 3′(−)SL as a promoter for initiation of the genomic plus-strand RNA.

Rubella virus (RUB) is the sole member of the Rubivirus genus of the Togaviridae family of animal viruses. This family also includes the Alphavirus genus, whose type species is Sindbis virus (SIN). In many regards, RUB is a typical representative of the togaviruses, although important differences between RUB and the alphaviruses have been discovered (reviewed in reference 7; the molecular biology of alphaviruses is reviewed in reference 31). As with the alphaviruses, the RUB virion contains an icosahedral nucleocapsid consisting of the single-stranded, plus-polarity genomic RNA of approximately 10,000 nucleotides (nt) and multiple copies of a single capsid protein C. The genomic RNA is capped and polyadenylated. The nucleocapsid is surrounded by a lipid envelope containing two viral glycoproteins, E1 and E2. In the cytoplasm of infected cells, the genomic RNA serves as an mRNA for translation of a 240-kDa polyprotein precursor that is posttranslationally cleaved by a viral papain-like cysteine protease into P150 and P90, two nonstructural proteins (NSPs) thought to function in virus RNA replication (1, 5, 15). The NSPs are encoded by an open reading frame (ORF) that covers the 5′ terminal two-thirds of the genome. It is in the processing of the NSP precursor that RUB differs most from alphaviruses. The processing of the alphavirus precursor involves a sophisticated cascade of temporally regulated cleavages resulting in production of four NSPs as well as a number of specific products of incomplete proteolysis. This process is believed to orchestrate the course of virus RNA synthesis in which the different polypeptides perform different specific functions (reviewed in reference 31). In contrast, the single cis cleavage mediated by the RUB NS protease would not allow such an elaborate regulation.

The RUB genomic RNA next functions as the template for the synthesis of a complementary genome-length, minus-polarity RNA which serves in turn as the template for production of new genomic plus strands as well as a subgenomic RNA. The subgenomic RNA contains sequences from the 3′-terminal one-third of the genome and is the mRNA for translation of the structural proteins (SPs; C, E2, and E1) encoded by the SP ORF. The individual SPs are generated by cotranslational cleavage of the polyprotein precursor by cellular signalase. Synthesis of the subgenomic RNA is initiated on the minus-strand template at a site located between the two ORFs.

One of the most intriguing similarities between alphaviruses and RUB is the conservation of three regions of homologous nucleotide sequence and/or structure which are thought to be regulatory signals for viral replication (3). One of these is a stem-and-loop structure at the 5′ end of the genomic RNA [5′(+)SL]. The RUB 5′(+)SL can stimulate translation of reporter genes both in vitro and in vivo and was shown to bind a number of cellular proteins, one of which is the La autoantigen (18, 24, 25). The functional significance of the cell protein binding is unknown.

In togaviruses, the complementary equivalent of the 5′(+)SL present at the 3′ end of viral minus-strand RNA [3′(−)SL] is thought to serve as a promoter for initiation of genomic plus-strand RNA by the viral replicase (31). The 3′(−)SL of both SIN and RUB have been shown to bind cellular proteins (18, 20, 21); in the case of SIN, one of these was identified as mosquito analog of the La autoantigen (22). In SIN, the 5′(+)SL and 3′(−)SL occur at the exact ends of the plus- and minus-strand RNAs. Site-directed mutagenesis of these structures conducted with an infectious clone of SIN revealed that only deletions at the immediate end of the genomic RNA (nt 5 or nt 2 to 4), which are located at the bottom of the stem, were lethal (19). Deletions downstream from this site of 1 to 15 nt resulted in viable viruses, some of which grew less well than the parental virus. Although the effect of these mutations on the viral macromolecular synthesis was not examined, these observations are currently considered as evidence of the importance of the 3′(−)SL for initiation of genomic RNA synthesis. In terms of this model, RUB is a convenient counterpart of SIN in that the 5′(+)SL is located 14 nt downstream from the exact 5′ end. Therefore, it should be possible to perform similar mutagenic analysis of RUB to distinguish whether the exact 5′ nucleotides, the stem-and-loop structure, or both are crucial for virus viability.

Another intriguing feature at the 5′ end of the RUB genome is the presence of a short ORF between nt 3 and 54 that could encode a 17-amino-acid peptide. This ORF overlaps the start of the NSP ORF at nt 41 in an alternate translational frame and thus could also potentially downregulate translation of the NSP ORF. In this study, we determined the 5′-terminal sequences of additional strains of RUB to ascertain that both the secondary structure and the short ORF were conserved. Then we used Robo302, an infectious clone of RUB (27), to mutagenize the genomic 5′ sequences to study the potential functional significance of these features.

MATERIALS AND METHODS

Cells and viruses.

Vero cells obtained from the American Type Culture Collection were maintained in Dulbecco’s modified Eagle medium (Gibco/BRL) containing 5% fetal bovine serum and gentamicin (10 μg/ml) at 35°C under 5% CO₂. Propagation and titration of RUB by plaque assay was done as previously described (5, 9, 34).

Construction of mutants.

Standard recombinant DNA techniques (13) were used in these protocols, with minor modifications. Restriction enzymes and T4 DNA ligase were obtained from New England BioLabs or Boehringer Mannheim Biochemicals. Sequencing of DNA was done by using dideoxy-chain termination sequencing kits from United States Biochemicals (Cleveland, Ohio).

Mutations were created by PCR using pairs of primers containing the desired mutations, ExTaq polymerase, which has a 3′-5′ exonuclease proofreading activity (PanVera Corp., Madison, Wis.), and EcoRI-linearized Robo302 plasmid as a template under conditions optimized for amplification of the high G+C content RUB sequences (27) with the exception that shorter polymerization times (30 s to 1 min) were used in the PCR cycles. To introduce mutations S3, S5, S10, D5, D_SL2, D_SL4, D_SL5, and I_SL1 (see Fig. 3) into Robo302, a HindIII-NcoI [39] (numbers in brackets indicate positions)-restricted PCR fragment containing the SP6 promoter and 5′ end of the RUB genome through the NcoI site at nt 39 containing the desired mutation was ligated overnight with roughly equimolar amounts of NcoI[39]-PinAI[816] and PinAI[816]-HindIII restriction fragments from Robo302 (the first fragment extends from the NcoI site at nt 39 through the PinAI site at nt 816 of the genome, while the second fragment contains the remainder of the RUB genome and pCL1921 vector). Mutations D_SL3, S_SL2, S_SL3, and S_SL4 were introduced by the same protocol, except that the HindIII-NcoI[39] mutation-containing fragment was produced by annealing two partially overlapping complementary oligonucleotides, filling in the single-stranded regions with Klenow enzyme, and restricting with HindIII and NcoI. To create mutations S1, S2, S4, S6, S7, S8, S9, D1, D2, D3, D4, D6, D7, D8, D9, D10, D11, S_SL1, and D_SL1, a HindIII-PinAI[816]-restricted PCR fragment (containing the SP6 RNA polymerase promoter and 5′ end of the RUB genome through the PinAI site at nt 816) containing the desired mutation was ligated with the PinAI[816]-HindIII restriction fragment of Robo302.

FIG. 3.

Mutations introduced into the 5′ terminus of Robo302. The wt (Robo302) sequence is boxed. Mutations in the ss-leader are shown above the wt sequence, and mutations in the 5′(+)SL are shown below the wt sequence. Mutations that were lethal are shaded, while viable mutations are indicated by + in the “Viability” column. Deviations from the wt plaque morphology exhibited by some of the mutants are indicated in the “Plaque phenotype” column (Robo302 forms large, clear plaques). In the “Method” column, the procedure of generation of each of the mutants is indicated: in E (express method), the ligation mixture was used directly for in vitro transcription followed by transfection of Vero cells, while in P (plasmid method), plasmids containing the mutation were generated prior to transcription and transfection. With both methods, mutations were verified by 5′ sequencing of mutant virion RNAs. The mutants marked by asterisks were found to contain second site mutations (see Table 1 for sequences).

With many of the mutants (Fig. 3), following ligation the restriction mixture was digested with EcoRI to linearize ligation products for runoff transcription, extracted with phenol-chloroform, and used directly in an in vitro transcription reaction. Alternatively, the ligation reaction was used to transform competent Escherichia coli MC1061 cells, and plasmids from several colonies were isolated and sequenced to confirm the presence of the desired mutation. Two to three clones of each of the mutants were linearized with EcoRI prior to in vitro transcription and transfection.

In vitro transcription, transfection of cells, and confirmation of mutations.

In vitro transcriptions using SP6 RNA polymerase in the presence of m⁷G(5′)ppp(5′)G cap analog and Lipofectin-mediated transfections of Vero cells with the resulting transcripts were done as described elsewhere (27, 29). Successful transcription was confirmed by agarose gel electrophoresis of an aliquot of the transcription reaction in the presence of ethidium bromide. The efficiency of in vitro transcription of the mutant templates was comparable to that of Robo302, with the exception of the S3 mutant, which produced approximately 10 times fewer transcripts. Generally, duplicate plates of cells were transfected with each transcription reaction; one plate was overlaid with agar, and the cells in the other plate were maintained in growth medium. Five to six days later, characteristic plaques (if present) were picked, and virus was eluted and amplified once in Vero cells to produce a stock for sequencing to confirm the mutation and for subsequent analysis of the mutant phenotype. If no plaques were observed, the growth medium from the duplicate transfected culture was harvested and used as the mutant virus stock. Sequencing to determine the genomic 5′ ends of different strains of RUB and to confirm the presence of mutations in the majority of mutants was done as described previously (27). Briefly, a 120-nt primer extension product (PEP) complementary to the 5′ end was synthesized on virion RNA template, poly(A) tailed with terminal deoxynucleotidyltransferase, and PCR amplified. The resulting PCR product was cloned into pGEM1 or pGEM2 plasmid vector. To confirm the D_SL3, S_SL2, S_SL3, and S_SL4 mutations, the protocol was modified such that oligonucleotide 73 (5′-CAAGGATCCAGAACCTCATCTAGGAG; the BamHI site used for cloning into pGEM1 vector is underlined) was used in the PCR amplification step. This oligonucleotide, complementary to nt 50 to 66, is closer to the 5′ end than oligonucleotides 36 and 292, used previously, and its use was found to increase the specificity of PCR amplification of the poly(A)-tailed PEP. To confirm the D_SL2, D_SL4, D_SL5, S_SL1, I_SL1, and D_SL1 mutations, the 5′ PEP was not poly(A) tailed and was instead directly PCR amplified by using oligonucleotide 74, containing nt 1 to 18 of the genome (the first five of these mutations were downstream from nt 18 and thus A tailing was not necessary; this method was also useful for D_SL1 due to the 3′-5′ exonuclease activity of Deep Vent polymerase used). For each mutation, several plasmid clones containing the complete 5′ terminus were identified and sequenced.

Analysis of virus macromolecular synthesis.

RUB-specific RNA and protein production was examined in cultures infected simultaneously. Vero cells grown in 35- and 60-mm-diameter plates (Corning Glass Works, Corning, N.Y.) and eight-chamber slides (Nunc, Naperville, Ill.) (for immunofluorescence assay [IFA]) were infected with Robo302 or mutant viruses at a multiplicity of infection (MOI) of 0.5 PFU/cell. At indicated times postinfection, total intracellular RNA was extracted from the infected monolayers in the 35-mm-diameter plates with TRI reagent (Molecular Research Center, Cincinnati, Ohio) according to the manufacturer’s protocol, redissolved in 40 μl of H₂O, and stored at −70°C. Aliquots of 4 μl were used for Northern blotting, done as described previously (2). [³²P]dCTP nick-translated Robo102 plasmid (34) was used as a probe for total RUB-specific RNA, and [³²P]-UTP-labeled RNA transcripts synthesized from template plasmid pRUB-SP-ORF, which contains the 3′-terminal 3,292 nt of the RUB genome (14), were used as strand-specific probes. pRUB-SP-ORF was linearized with HindIII and transcribed with T7 RNA polymerase or linearized with EcoRI and transcribed with SP6 RNA polymerase to synthesize the minus- or plus-strand specific probe, respectively.

Infected Vero cells grown in 60-mm-diameter plates were radiolabeled for 1.5 h with [³⁵S]methionine (1,000 Ci/mmol; Amersham) and lysed in 1 ml of radioimmunoprecipitation buffer (26) supplemented with 74 μM antipain dihydrochloride protease inhibitor (Boehringer Mannheim). Aliquots (200 μl) of these lysates were mixed with 2 μl of a mixture of monoclonal antibodies E1-20, E2-1, C2, and C8 (35) to immunoprecipitate the SPs. Aliquots (800 μl) of the lysates were mixed with 7 μl of a mixture of rabbit polyclonal antisera GU1 and GU8 (5) to immunoprecipitate the NSPs. Immune complexes were recovered by using protein A-Sepharose beads (Pharmacia), boiled for 2 min in sodium dodecyl sulfate-containing sample loading buffer, and resolved by electrophoresis on a sodium dodecyl sulfate–10% polyacrylamide gel. Following autoradiography, the amount of radioactivity present in each protein or RNA band was quantitated with a Fujix BAS1000 Bio Imaging Analyzer (Fuji Photo Film, Tokyo, Japan), using software provided by the manufacturer. The cells in the eight-chamber slides were fixed and analyzed by IFA using monoclonal antibody E1-20 as described previously (26) to determine the percentage of infected cells.

Computer analyses.

Optimal and suboptimal RNA secondary structures were determined by using the FoldRNA and Mfold programs; the output files of these programs were plotted by using the Squiggles program (Genetics Computer Group, Madison, Wis.).

RESULTS

5′-terminal structure of the RUB genomic RNA.

The 5′-terminal stem-and-loop structure [5′(+)SL] shown in Fig. 1A was initially predicted by computer (3). Its existence in the RUB genomic RNA is supported by the finding that primer extension reactions near the 5′ end of the genomic RNA result in two strong-stop cDNA bands corresponding to the last residue of the 5′(+)SL (nt 65) and the 5′ end of the genome (3). The 5′(+)SL follows a 14-nt single-stranded leader (ss-leader). We call the vertical part of the stem-and-loop formed by pairing of nt 15 to 29 and nt 51 to 65 the stem; the region consisting of nt 30 to 50 is called the upper loop. The predicted secondary structure of the genomic 5′ end of SIN is shown for comparison in Fig. 1B. Also shown in Fig. 1A is the second element of interest in this study, the short ORF located between nt 3 and 54, which overlaps the start of the NSP ORF at nt 41 in an alternate frame.

FIG. 1.

Structures of the ends of the RNAs of RUB and SIN. The 5′ terminus of the RUB genomic RNA (A) contains a 14-nt ss-leader followed by a stem-and-loop structure [5′(+)SL] composed of nt 15 to 65. The 5′(+)SL of the genomic RNA of SIN (B) is located at the exact 5′ end (nt 1 to 44). Start and stop codons for the short ORF of RUB are underlined, and the initiation codons for the NSP ORF of both RUB and SIN are shaded. A potential pseudoknot formed by the RUB 5′ terminus is also shown (this pseudoknot was not taken onto consideration to determine the given ΔG value). The predicted complementary structure, 3′(−)SL, located at the 3′ end of the RUB minus-strand RNA, is shown in panel C (numbering of nucleotides is from the 3′ end).

The corresponding 3′(−)SL structure, complementary to the 5′(+)SL, is shown in Fig. 1C. This structure is not an exact reflection of the 5′(+)SL and is less stable than the 5′(+)SL (ΔG = −15.2 kcal/mol versus ΔG = −20.7 kcal/mol). Also, unlike the 5′(+)SL, which is preserved when longer stretches of the RUB genome are folded, the 3′(−)SL disappeared when longer 3′ segments of minus-strand RNA were folded (for instance, 150-, 200-, 400-, and 1,000-nt segments) due to alternate pairing with downstream sequences (data not shown).

Comparison of the 5′ ends of six RUB strains.

Previously, the sequence of the exact 5′ terminus of the RUB genome had been determined only for the w-Therien (wild-type [wt]) and RA27/3 (vaccine) strains (28, 34). We determined the 5′ sequences of four additional strains: f-Therien, a laboratory derivative of w-Therien selected for clear-plaque morphology (30); HPV-77, an independent vaccine strain (7); Hasnas, a wt strain isolated in the United States in 1990; and Machado, a wt strain isolated in Great Britain in 1992 (8). An alignment of these sequences is shown in Fig. 2. While the HPV-77 sequence is identical to that of w-Therien, the other strains differed by up to 3 nt within the 5′-terminal 100 nt. Interestingly, the Hasnas genome starts with an additional A residue. The Hasnas and Machado strains contain two of the three differences found between RA27/3 (G₇, U₃₄, and G₄₈) and w-Therien. The short ORF is preserved in all of these strains. However, the nucleotide differences within the short ORF would all result in amino acid substitutions in the predicted peptide (N₁₆→S in f-Therien; E₂→G, P₁₁→L, and N₁₆→D in RA27/3; and E₂→G and P₁₁→L in Hasnas and Machado).

FIG. 2.

Sequence alignment of the genomic 5′ ends of different strains of RUB. Abbreviations: w-Th, w-Therien; f-Th, f-Therien; RA, RA27/3; Has, Hasnas; Mach, Machado. Start and stop codons for the short ORF are boxed, and the initiation site of the NSP ORF is indicated by the arrow.

The predicted effects of the strain differences on the 5′(+)SL and 3′(−)SL were analyzed, with the result that the length of the ss-leader and the appearance of the stem of both structures (Fig. 1A and C) are invariant in all strains. The upper loop part of the 5′(+)SL (nt 30 to 50) and the corresponding part of the 3′(−)SL vary: they have fewer paired bases in RA27/3, Hasnas, and Machado than in w-Therien, f-Therien, and HPV-77 (data not shown). This relaxation in the upper loop in some of the strains results in a reduction in the estimated ΔG values of both the 5′(+)SL (−18.6 kcal/mol for RA27/3 and −18.5 kcal/mol for Hasnas and Machado versus −20.7 kcal/mol for w-Therien, f-Therien, and HPV-77) and 3′(−)SL structures (−13.7 kcal/mol for RA27/3 and −13.2 kcal/mol for Hasnas and Machado versus −15.2 kcal/mol for w-Therien, f-Therien, and HPV-77).

Mutagenesis of the ss-leader sequence.

A number of deletions of 1 to 6 nt (D1 to D11) and point mutations (S1 to S10) were introduced into the ss-leader of Robo302 (Fig. 3). Surprisingly, the most profound effects were produced by mutations that altered both A residues at nt 2 and 3. Deletions that led to removal of both A residues (D3, D4, D5, and D6), as well as substitution of the AA with CC (S1), were lethal. With the lethal mutants, no plaques were detected following transfection of Vero cells. Additionally, no virion RNA of mutants D3 and D4 was detected in the media by reverse transcription-PCR, and no intracellular RUB-specific RNA of mutants S1 and D6 was detected by Northern hybridization in Vero cells in which the media from the initially transfected cells had been passaged (data not shown). The other mutants were viable, and most grew to titers comparable to that of Robo302 (Table 1). Mutations which produced low titers following initial amplifications of plaques were reassayed by using the amplified stock to infect Vero cells, with harvest 2.5 days postinfection. Mutant D8 (nt 6 to 8 deleted) was found to produce titers 10-fold lower than those of Robo302. Roughly half of the mutants produced plaques with altered morphologies. With two exceptions (D8 and D10, in which nt 6 to 8 and nt 9 to 11, respectively, were deleted), all of these had mutations within the first 6 nt. Deletions within this region invariably gave rise to virus with altered plaque morphology, whereas many of the base substitutions produced virus with wt plaques.

TABLE 1.

Characterization of viable mutants

Virus	Verification of 5′ sequencea	Presence of the extra Gb	Titer Ic	Titer IId
D1	—	—	—	—
D2	—	—	—	—
D7	OK	4/4	1 × 105	3.3 × 106
D8	OK	0/4	3 × 105	7.7 × 105
D9	Duplication of the first 7 nt in 3 of 4 clones	0/4	2.5 × 105	4.3 × 106
D10	OK	3/5	6 × 106	—
D11	OK	3/4	1 × 107	—
S2	Mutated residue (C2) deleted in all clones	4/4	6.4 × 106	8.4 × 106
S3	C1G2 residues deleted in all clonese	8/8	4 × 106	1.44 × 106
S4	OK	6/6	3 × 106	—
S5	OK	3/3	6 × 106	—
S6	OK	4/4	4.5 × 106	—
S7	OK	4/4	1.3 × 106	—
S8	OK	3/4	1 × 106	—
S9	OK	2/2	2.6 × 106	—
S10	OK	0/3	1 × 107	—
DSL1	OK (4 clones)i	—	4.4 × 105	2.4 × 106
DSL2	OK (3 clones)i	—	2.8 × 105	3.9 × 106
DSL3	OK	0/2	No plaquesf	—
DSL4	OK (3 clones)i	—	1.2 × 106	—
DSL5	OK (3 clones)i	—	3.4 × 106	—
SSL1	OK (3 clones)i	—	1 × 106	—
SSL2	OK	3/4	2 × 106	2.4 × 106
SSL3	OK	0/4	5 × 104	9 × 105
SSL4	OK	1/2	1 × 106	6.5 × 106
ISL1	OK (6 clones)i	—	4 × 105	3.7 × 106
ISL1dg	wt-specific U31 (downstream from the insert in ISL1 mutant) deleted (3 clones)i	—	1 × 106	6.5 × 106
Robo302	OK	3/5h	6 × 106	8 × 106

^aMutations were confirmed (OK) by 5′-terminal sequencing of virion RNA, unless found otherwise as indicated; — in all columns, not determined. 

^bThe extra G (G₀) residue is encoded in the in vitro RNA transcripts by the last nucleotide of the SP6 promoter; numbers of clones containing G₀ out of total clones sequenced are given. 

^cVero cells were infected with viruses eluted from plaques (unknown MOI), and titers in the media were determined when cytopathic effect was observable (6 ± 1 day postinfection). 

^dVero cells were infected at an MOI of 0.5 PFU/cell, and titers in the media on day 2.5 postinfection were determined; mean values of titers obtained in two independent experiments are given. 

^eIt is unknown whether C₁G₂ or G₀C₁ were removed in this virus; in either case, the first nucleotide is G. 

^fThis virus is viable but does not form plaques. 

^gSelected during passage of I_SL1 mutant (see text for details). 

^hVerification of Robo302 5′ sequences after one passage was done previously (27). 

ⁱThese viruses were verified by the second sequencing procedure that used an oligonucleotide for PCR complementary to the 5′ end; therefore, the presence of the extra G at the 5′ end could not be resolved; the numbers of plasmid clones sequenced to confirm the mutations are given. 

The presence of the mutations in the generated virus samples was tested by 5′-terminal sequencing of virion RNA (Table 1). While most of the mutations were confirmed, viruses recovered from two mutations intended to substitute the first residue of the AA dinucleotide at nt 2 and 3 were found to contain altered sequence: S2 virus RNA started with 5′-G₀CAUGG instead of the intended 5′-G₀CCAUGG, indicating that one of the two C residues had been removed (G₀ designates the additional residue encoded in the in vitro RNA transcripts by the last nucleotide of the SP6 promoter); and S3 virus RNA started with 5′-GAUGG instead of the expected 5′-G₀CGAUGG, indicating that either G₀C₁ or C₁G₂ were deleted. Additionally, mutant D9 virus was expected to start with 5′-G₀C₁AAUG₅A₁₂UC (nt 6 to 11 are deleted). Instead, three of the four sequenced 5′ clones started with 5′-C₁AAUG₅A₁₂UC₁AAUG₅A₁₂UC, indicating that a duplication of the first 7 nt had occurred; the fourth clone started with the expected sequence lacking G₀. Presumably, these alterations in expected sequence were due to in vivo mutations that occurred during virus replication following transfection. To eliminate the possibility that the alterations were introduced by SP6 RNA polymerase during in vitro transcription, the corresponding in vitro RNA transcripts were sequenced, with the result that all transcripts were found to start with the expected sequence. Additionally, the relative infectivity of S2 and S3 transcripts was roughly 10 to 20 times lower than the infectivity of Robo302 transcripts, while all of the other mutant transcripts [including mutations in the 5′(+)SL] had infectivities comparable to that of Robo302. This finding also suggests that the initial S2 and S3 transcripts mutated following transfection. The D9 genotype appears to be unstable, and the D9 virus used in this study is represented by a mixture of the two variants.

In Robo302 virus harvested after one passage following transfection, roughly 60% of the 5′-terminal clones contain the additional G₀ residue, which disappears after five passages (27). As shown in Table 1, different proportions of the 5′-terminal clones from some of the mutants also contain the G₀. Whether the G₀ was maintained and contributed to viability of some of the mutants (S2, S3, S4, etc.) is unknown and remains to be investigated further.

Mutagenesis of the AUG initiation codon for the short ORF (nt 3 to 5) to CUG, ACG, and AUA (mutants S6, S7, and S8, respectively) or replacement of the third (G₉CU) codon within the ORF to UAA termination codon (S10) yielded viable viruses with wt plaque morphology. These mutants grew to titers similar to Robo302 titers (Table 1). Thus, the short ORF is not required for replication. A potential pseudoknot interaction can be configured between nt 2 to 6 and nt 33 to 37 (7) (Fig. 1A). Several of the mutations which would alter the stability of this pseudoknot gave rise to virus with wt plaque morphology, while others yielded virus with altered plaque morphology. One of these, S4 (A₃→G substitution), led to opaque plaques. However, the S5 mutant was constructed to change the complementary U₃₆ residue in the pseudoknot to C. S5 virus formed opaque plaques indistinguishable from those formed by S4, and thus the opaque-plaque phenotype is independent of the putative pseudoknot.

Mutagenesis of the 5′(+)SL.

Five deletion mutations which were intended to make gross changes in the structure were introduced into the left side of the 5′(+)SL (Fig. 3): nt 15 to 21 at the bottom of the stem (D_SL1), nt 25 to 29 at the top of the stem (D_SL2), nt 15 to 29 (the entire left side of the stem; D_SL3), nt 30 to 35 in the lower part of the upper loop region (D_SL4), and nt 36 to 40 in the bulge of the upper loop region preceding the A₄₁UG initiation codon for the NSP ORF (D_SL5). No mutations were introduced downstream from the AUG initiation codon for the NSP ORF (nt 41 to 43) situated at the top of the 5′(+)SL, because the choice of mutations in this region would have been restricted to a few 1-nt silent substitutions to prevent alteration of the NSP amino acid sequence. Also, a 3-nt substitution (G₂₂CU→AUC) was made to sew up the bulge at the middle of the stem (S_SL1), and a 4-nt (GUUU) insertion was introduced between nt 29 and 30 to straighten and sew up the upper part of the 5′(+)SL (I_SL1). The computer-predicted effects of the mutations on the 5′(+)SL are shown in Fig. 4A. As expected, the mutations that eliminated bulges are stronger than the wt structure (e.g., −30 kcal/mol for I_SL1 versus −20.7 kcal/mol in Robo302). Interestingly, the mutants that deleted regions of the 5′(+)SL formed alternate stem-and-loop structures utilizing nucleotides from the ss-leader. However, these structures were weaker as well as unstable: the 5′-terminal sequences were paired with different downstream regions when longer 5′-terminal segments were folded (data not shown).

FIG. 4.

Predicted 5′-terminal structures of RNAs containing mutations within the 5′(+)SL. (A) Effects of the five deletions (D_SL1 to 5), the mutation that sews up the bulge in the stem (S_SL1), and the 4-nt insertion in the upper loop (I_SL1) are shown. I_SL1d virus is a clear-plaque derivative selected during passage of I_SL1 mutant in Vero cells. (B) Effects of the three multiple-point mutations are shown. In both panels, the localization of deletions is indicated by arrowhead and inserted or substituted nucleotides are shaded. The NSP ORF initiation codon is boxed.

All of these mutants were viable, and the presence of the intended mutations was confirmed in each case. With exception of the D_SL4 mutant, which produced wt plaques, these mutants had altered plaque morphologies. Mutants D_SL1 and D_SL2 formed plaques that were barely detectable. D_SL3 did not form plaques; however, a slight RUB-like deterioration of cell monolayers was observed when cells were infected with undiluted stocks of this virus. RUB-specific RNA was detected in D_SL3-infected cells by Northern hybridization (the amount of the D_SL3 RNA was 80 times lower than in Robo302-infected cells) and thus this virus was viable, although no SPs or NSPs were detectable in D_SL3-infected cells by radioimmunoprecipitation (data not shown). After mutant I_SL1, which forms opaque plaques, was amplified in Vero cells, a small fraction of clear plaques was observed. The virus from one of these plaques, subsequently called I_SL1d, was purified and its 5′ end was sequenced. Surprisingly, it still contained the entire insertion; however, one of the authentic nucleotides located downstream from the insertion (U₃₁) was deleted. This deletion led to a slight relaxation in the upper part of the mutant 5′(+)SL (Fig. 4A). These mutants replicated less well than Robo302; the titer range was from 2- to 20-fold lower when plaque isolates were initially amplified but only 2- to 4-fold when these amplified stocks were used in an experiment with a controlled MOI (Table 1).

As an alternative approach to alter the stem-and-loop structure, we also made three different multiple-point mutations, S_SL2, S_SL3, and S_SL4 (Fig. 3), that caused a profound destabilizing effect in both the 5′(+)SL (Fig. 4B) and the complementary 3′(−)SL as predicted by computer (data not shown) without changing the overall length of the 5′ untranslated region. All three of these mutants were viable, and all three produced opaque plaques. Two of these viruses, S_SL2 and S_SL4, grew to titers comparable to that of Robo302, while S_SL3 produced somewhat lower titers (Table 1). S_SL3 also produced smaller plaques that did the other two mutants.

Effects of selected mutations on the virus macromolecular synthesis.

Monolayers of Vero cells were infected at an MOI of 0.5 PFU/cell with the following mutants: S2, S3, D7, D8, and D9, which contain mutations in the ss-leader (Fig. 3); and I_SL1, I_SL1d, D_SL1, D_SL2, S_SL2, S_SL3, and S_SL4, which contain mutations in the 5′(+)SL. Virus RNA and protein synthesis were analyzed at both 48 and 72 h postinfection, with similar results at both time points; analysis at 72 h postinfection is presented here.

Production of the plus-strand genomic and subgenomic RNAs is shown in Fig. 5A. As can be seen, with none of the mutants was synthesis of these RNAs significantly altered in comparison with Robo302. The amount of RNA produced was roughly correlated with the percentage of cells infected at the time of analysis as determined by IFA, which varied from 40 to 50% for Robo302, S3, D7, D8, D_SL1, S_SL2, and S_SL4 to 60 to 70% for S2, D9, S_SL3, I_SL1, I_SL1d, and D_SL2. The viruses which made the most RNA had the highest infected cell percentage (e.g., D9 and S2). The genomic RNA/subgenomic RNA ratios among these viruses were also comparable, varying in the range of ±30% from the ratio for Robo302. Structural protein synthesis is shown in Fig. 5B. As with RNA synthesis, SP synthesis did not vary markedly among these viruses, and the amount of radioactivity in the SP bands was proportional to the intensity of the subgenomic RNA bands.

FIG. 5.

Effects of selected mutations on virus macromolecular synthesis. Vero cells were infected with viruses indicated above each lane at an MOI of 0.5 PFU/cell. At 72 h postinfection, total intracellular RNAs were extracted and plus-strand RNAs were assayed by Northern hybridization (A) or the cells were radiolabeled with [³⁵S]methionine and the SPs (B) and NSPs (C) were immunoprecipitated and resolved by polyacrylamide gel electrophoresis. In panel A, G signifies position of the genomic RNA band and SG signifies the subgenomic RNA band. In panels B and C, positions of the RUB-specific proteins and molecular weight markers (in kilodaltons) are indicated on the left and right, respectively.

The most marked difference between the mutants was in NSP production, as shown in Fig. 5C. Specifically, NSP synthesis was lower in the 5′(+)SL mutants than in Robo302 or the ss-leader mutants. NSP synthesis normalized to SP synthesis is shown in Fig. 6. Particularly dramatic was the decrease in NSP synthesis observed in mutants D_SL1, D_SL2, and S_SL4, whose NSP/SP ratios were approximately four times lower than that of Robo302.

FIG. 6.

Relative rates of the NSP synthesis in cells infected with the selected mutants. Amounts of radioactivities in the SP and NSP bands in Fig. 5B and C were quantitated with a phosphorimager. Total radioactivities in the NSP bands normalized to the radioactivity in the SP bands for each virus are given (SP synthesis was found to be roughly equivalent in all mutants).

DISCUSSION

In this study, we report the first use of the RUB infectious clone for major genetic manipulation. Extensive mutagenesis was done on the 5′ genomic sequences, and this mutagenesis was tolerated by the virus in that the majority of the mutants were viable. Thus, the infectious clone will be useful for site-directed mutagenesis studies.

As the first step in this study, we determined the genomic 5′ sequences of four additional strains of RUB (f-Therien, HPV, Hasnas, and Machado) for comparative purposes since the 5′ ends extending to the 5′-most nucleotides of only two RUB strains, w-Therien and RA27/3, had been sequenced previously (28, 34). The two major elements of the 5′-terminal secondary structure, the 14-nt ss-leader and the 5′(+)SL, are present in all these strains. The stem part of the 5′(+)SL was completely conserved in that no differences were found within the regions (nt 15 to 29 and nt 51 to 65) that compose the stem. The six strains differed at up to 3 nt localized in the ss-leader and the upper loop part (nt 30 to 50) of the 5′(+)SL, and the differences in the upper loop led to predicted structural differences. This finding is consistent with results of Johnstone et al. (10), who sequenced nt 18 to 540 of seven different strains and found no difference in the stem but variation in the upper loop that led to variability in the predicted structure. The maintenance of the stem and the variability of the upper loop correlated with the mutagenesis results in that mutations in the stem resulted in more dramatic effects on the virus growth and plaque morphology compared to deletions in the upper loop.

In addition to the secondary structure, an interesting feature at the 5′ terminus addressed in this study is the short ORF located between nt 3 and 54 of the RUB genome that overlaps the NSP ORF in an alternate translation frame. We found that this ORF was preserved in the six independent strains; however, its coding sequence was not as highly conserved as would be expected of a short peptide that plays a significant function in the virus replication cycle since up to three substitutions within the encoded 17 amino acid residues were found in some strains. Subsequently, we made mutations that interfered with or excluded translation of this ORF, and the resulting mutant viruses were found to be viable and have wild-type growth and plaque morphology. Therefore, this short ORF is not required for the virus replication.

Mutagenesis of the 5′(+)SL was initially approached by deleting portions of the stem and upper loop and yielded the overall result that none of these mutations, including a 15-nt deletion of the entire left side of the conserved stem (mutant D_SL3), were lethal, although this latter mutant was severely crippled. The deletions in the stem had a greater effect on virus titer and plaque morphology than did the deletions in the upper loop. Computer-assisted refolding of these deletions indicated that stem structures could still be formed, sometimes with nucleotides in the ss-leader; however, these structures were both weaker and less stable. The mutations that resulted in the weakest stem structure had the greatest effect on plaque morphology. To profoundly modify the appearance of the 5′ secondary structure without changing the length of the 5′ untranslated region, we made three multiple-point mutations (S_SL2, S_SL3, and S_SL4) which were predicted to completely alter the 5′(+)SL. All three of these mutants exhibited altered plaque morphology, and two grew to titers that approached Robo302 titers. The mutant that grew roughly 10 times less well than Robo302 and had the severest alteration in plaque morphology (S_SL3) had the least stable 5′ structure. Therefore, there appears to be a requirement for a stem structure with some degree of stability at the 5′ end of the RUB genome for efficient replication. Overall, these findings are in accordance with the results of a similar study conducted on the 5′ end of SIN in which it was shown that various deletions within the 5′(+)SL of SIN of up to 15 nt in length introduced downstream from nt 5 were not lethal but had variable effects on virus replication (19). In the SIN study, deletions within the first 5′ nt of the 5′(+)SL were lethal (except for deletion of nt 1, which was viable). However, as discussed below, we believe that this was due to the fact that these nucleotides are the first 5 nt of the SIN genome and not because they are part of the 5′(+)SL.

At the molecular level, the only parameter that differed among different mutants was production of NSPs, which was reduced by the mutations in the 5′(+)SL but not in the ss-leader. Thus, the 5′(+)SL is important for efficient initiation of translation of the NSP ORF. This finding is consistent with previous observations of a stimulatory effect of this structure on translation of reporter genes in vitro and in vivo (18, 24, 25). The La autoantigen has been shown to bind the 5′(+)SL, and some evidence indicates that La binding increases the efficiency of translation of the NSP ORF (4). It will be interesting to measure the effect of our 5′(+)SL mutations on La binding. Especially profound suppression of NSP synthesis was observed in mutants D_SL1 and D_SL2, which formed barely detectable plaques, and no viral proteins were detectable in mutant D_SL3, which failed to form plaques. This observation is consistent with our recent findings indicating that the level of NSP synthesis can be an important determinant of RUB cytopathogenicity (27).

Computer folding demonstrated that the 3′(−)SL at the 3′ end of the minus-strand RNA of RUB is weaker than the 5′(+)SL and was destabilized by the mutations introduced into the 5′(+)SL. Since these mutations did not preclude viability and did not significantly affect plus-strand RNA synthesis, we conclude that the 3′(−)SL is not a negative-strand promoter necessary for initiation of genomic plus strands. As discussed above, large alterations in the 5′(+)SL of the SIN infectious clone were tolerated, although the effect of these mutations on plus-strand RNA synthesis was not studied (19). The other published data that argue against the idea of the 3′(−)SL being a promoter are the findings that the 5′ noncoding sequences in defective interfering RNAs of both Semliki Forest virus (12, 23) and SIN (16, 17, 32, 33) are highly rearranged, with large portions of the 5′(+)SL being deleted and/or replaced by heterologous sequences of both viral and cellular origin. Computer analysis failed to recognize any common 5′-terminal secondary structure among such defective interfering RNAs of SIN (33). In this regard, minus-strand RNA is completely complexed in double-stranded replicative forms and replicative intermediates in togavirus-infected cells which are the templates for plus-strand RNA synthesis (11). It is not known if these RNAs are unwound by themselves to any extent, and thus it is possible that formation of any secondary structure by the minus-strand RNA, such as the 3′(−)SL, does not occur. In this case, initiation of plus-strand RNA could occur by template switching by the replicase from the plus-strand to the minus-strand template or by recognition of a double-stranded RNA feature. However, if such unwinding indeed takes place, then the possibility remains that a specific, more complex tertiary structure of a larger 3′ segment of the minus-strand RNA serves as a promoter for the plus-strand RNA.

Perhaps the most intriguing finding in this study was that only a few nucleotides at the extreme 5′ end of the genome were essential for RUB viability. Specifically, mutations that removed or altered the two A residues at nt 2 and 3 simultaneously (but not mutations that altered only one of these residues) were lethal. Mutations that changed the first A residue resulted in viable viruses, which however contained secondary mutations that moved an A residue within the first 3 nt of the genome. This finding coincides with the SIN mutagenesis study in which deletions in the first 5 nt were found to be lethal (19). Since the crucial AA dinucleotide in the RUB genome is located in the ss-leader, this indicates that in both RUB and SIN the 5′-most nucleotides are critical in virus replication, but not because in SIN they belong to the bottom of the 5′(+)SL as was previously suggested (19). It is possible that these residues constitute a core element of a short signal at the 5′ end of genomic RNA (or 3′ end of the minus-strand RNA) which is absolutely necessary for RNA replication. However, the signal is not as simple as one or more A’s near the 5′ end of the genome, since D5, which moved the A₇A₈ dinucleotide to the exact 5′ end, was lethal. It should be noted that D8, which lacked G₆A₇A₈, gave rise to a small-plaque phenotype virus, while D9, which lacked nt 6 to 11, gave rise to a virus with a duplicated 5′ end. Thus, nt 6 to 8 as well as some other surrounding nucleotides may also play a role in this putative signal. Alternatively, it has been shown that alphavirus genomic RNAs cyclize through interactions of nucleotides near the 5′ and 3′ ends of the molecule (6). Although the function of cyclization, as well as whether or not the RUB genomic RNA cyclizes, is not known, the exact 5′ nucleotides of the togavirus genomes could be involved in cyclization. Further analysis is required to resolve these possibilities.