|Home | About | Journals | Submit | Contact Us | Français|
Brome mosaic virus (BMV) encodes two RNA replication proteins: 1a, which contains RNA capping and helicase-like domains, and 2a, which is related to polymerases. BMV 1a and 2a can direct virus-specific RNA replication in the yeast Saccharomyces cerevisiae, which reproduces the known features of BMV replication in plant cells. We constructed single amino acid point mutations at the predicted capping and helicase active sites of 1a and analyzed their effects on BMV RNA3 replication in yeast. The helicase mutants showed no function in any assays used: they were strongly defective in template recruitment for RNA replication, as measured by 1a-induced stabilization of RNA3, and they synthesized no detectable negative-strand or subgenomic RNA. Capping domain mutants divided into two groups. The first exhibited increased template recruitment but nevertheless allowed only low levels of negative-strand and subgenomic mRNA synthesis. The second was strongly defective in template recruitment, made very low levels of negative strands, and made no detectable subgenomes. To distinguish between RNA synthesis and capping defects, we deleted chromosomal gene XRN1, encoding the major exonuclease that degrades uncapped mRNAs. XRN1 deletion suppressed the second but not the first group of capping mutants, allowing synthesis and accumulation of large amounts of uncapped subgenomic mRNAs, thus providing direct evidence for the importance of the viral RNA capping function. The helicase and capping enzyme mutants showed no complementation. Instead, at high levels of expression, a helicase mutant dominantly interfered with the function of the wild-type protein. These results are discussed in relation to the interconnected functions required for different steps of positive-strand RNA virus replication.
In the early stages of positive-strand RNA virus infection, the viral genomic RNA is translated to yield the virus-encoded RNA replication proteins, which include an RNA-dependent RNA polymerase and commonly several additional factors. These act to recruit the viral RNA from translation to the replication complex, where it is used as a template for the synthesis of complementary negative strands. The negative strands, in turn, are utilized as templates for the production of progeny positive strands, which include new genomic positive-strand RNAs and, for several virus groups, additional subgenomic mRNAs.
Brome mosaic virus (BMV) is a representative member of the large alphavirus-like superfamily of positive-strand RNA viruses. All members of the alphavirus-like superfamily contain three homologous domains in their encoded replication proteins, which implies that these viruses share common mechanisms of RNA replication. The conserved domains are differently organized in different family members, and several subfamilies encode additional replication factors. The three conserved domains are a unique RNA capping enzyme domain, a superfamily I helicase-like domain, and a polymerase-related domain (30). Within the alphavirus-like superfamily, RNA-dependent RNA polymerase activity has been demonstrated for the bamboo mosaic potexvirus polymerase domain, expressed in Escherichia coli (35). Recombinant Semliki Forest virus (an alphavirus) nsP2 has both RNA helicase and nucleotide triphosphatase (NTPase) activities (15, 47), and rubella virus (19) and turnip yellow mosaic tymovirus (28) helicase-like domains possess NTPase activity. Capping-related activities, namely, methyltransferase and covalent binding of methylated guanylate, have been demonstrated for proteins encoded by alphaviruses (4, 32, 48), tobacco mosaic virus (36), and BMV (3, 29). The RNA-dependent RNA polymerase activity is presumably required for all steps of replication involving RNA synthesis, but what are the roles of the other enzymatic activities? Is the helicase or NTPase activity required for negative-strand or positive-strand synthesis, for both, or for some additional steps in the replication cycle? Is the capping enzyme active only in capping positive-strand genomic and subgenomic RNAs and not involved in negative-strand synthesis, as the negative-strand RNAs are not capped (37, 52)?
The BMV genome consists of three positive-sense RNA molecules. RNA1 encodes the 1a protein, which contains the capping enzyme and helicase-like domains, whereas RNA2 encodes the polymerase-like 2a protein. Both 1a and 2a are required for BMV RNA replication. The dicistronic RNA3 encodes the 3a protein required for cell-to-cell movement within infected plants, and the virus coat protein, which is translated from a subgenomic mRNA (RNA4) representing the 3′ end of RNA3 (reviewed in references 1 and 11). BMV proteins 1a and 2a are capable of catalyzing efficient replication and subgenomic mRNA transcription of BMV RNA3 in the yeast Saccharomyces cerevisiae (27). RNA3 can be introduced to the yeast cells either by RNA transfection or by RNA polymerase II-mediated transcription from a suitable DNA plasmid (24, 27). BMV RNA replication takes place in membrane-associated replication complexes located on the endoplasmic reticulum of both plant and yeast cells (44, 45). The yeast system can be used to study both host and viral functions involved in BMV RNA replication, and it has been shown that multiple yeast genes affect BMV replication (12, 23).
1a- and 2a-mediated RNA3 replication and subgenomic mRNA synthesis in yeast are specific for BMV RNAs and utilize previously characterized cis-acting RNA synthesis signals present in the 5′, 3′, and intergenic regions of RNA3 (27, 43, 49). Specifically, conserved tRNA-like 3′ end sequences and an intergenic replication enhancer (RE) are required for efficient synthesis of negative strands, the 5′ end sequence is required for positive-strand synthesis, and the subgenomic promoter, located in the intergenic region and partially overlapping the RE, is required for subgenomic mRNA synthesis. Genetic exchanges show that 1a and the RE regulate the selection of RNA3 templates for replication (41, 50). The 1a protein can dramatically stabilize RNA3 in yeast, in the absence of 2a protein and RNA replication (26). 1a-mediated RNA3 stabilization depends on the same RE sequences that are required for approximately 100-fold enhancement of negative-strand synthesis both in plant cells and in yeast (14, 43, 49). The 1a-stabilized RNA3 is also poorly translated in yeast (26). These combined results imply that 1a-mediated RNA3 stabilization represents an intermediate involved in RNA3 recruitment from translation to RNA replication.
As one step toward understanding the functions involved in various stages of positive-strand RNA virus replication, we have constructed defined point mutations at the predicted active sites of BMV 1a protein, designed to destroy individual activities. Unexpectedly, mutations at both the helicase and capping enzyme active sites of BMV 1a caused pronounced defects in the early step of 1a-mediated RNA3 stabilization or template recruitment for RNA replication. Additionally, the helicase domain was implicated in the synthesis of negative-strand RNA, and direct evidence for the importance of the viral mRNA capping function was obtained.
Yeast strain YPH500 (MATα ura3-52 lys2-801 ade2-101 trp1-Δ63 his3-Δ200 leu2-Δ1) was used in most experiments. Where indicated, its derivative YMI04, which contains chromosomally integrated derivatives of BMV RNA3 with URA3 and β-glucuronidase (GUS) open reading frames replacing the coat protein gene (23), was used. In other specified experiments, a YPH500 derivative with deletion of the XRN1 gene was used (33). Yeast cultures were grown at 30°C in defined synthetic medium containing 2% galactose or 2% glucose as a carbon source (7). Histidine, leucine, tryptophan, uracil, or combinations thereof were omitted to maintain plasmid selection. Yeast cells were transformed by the lithium acetate-polyethylene glycol method (25).
BMV 1a and 2a expression plasmids pB1CT19 and pB2CT15 have been described previously (27). They are 2μm plasmids containing the BMV 1a and 2a open reading frames flanked by constitutive ADH1 promoter and ADH1 polyadenylation sequences and containing HIS3 and LEU2 selectable markers, respectively. GAL1 promoter-driven centromeric 1a expression vectors pB1YT3 and pB1YT3H with URA3 and HIS3 selectable markers and 2a expression vector pB2YT5 with LEU2 selectable marker (Y. Tomita, M. Ishikawa, and M. Janda, unpublished data) were used to increase the levels of 1a and 2a expression in some experiments. Wild-type (wt) RNA3 expression was achieved using centromeric TRP1 marker-containing plasmid pB3RQ39 (24), in which RNA3 is flanked by GAL1 promoter and self-cleaving hepatitis delta ribozyme sequences. In assays of full RNA replication, an RNA3 derivative from which coat protein expression was abolished was transcribed from plasmid pB3MS82 (M. Sullivan, unpublished data). This plasmid contains a four-nucleotide insertion immediately following the coat protein gene initiation codon and an additional point mutation that changes a second in-frame AUG in the transcript to AUC, the same changes as described for pB3MS89 (49) but in the context of full-length RNA3. The use of pB3MS82 avoids any possible effects due to coat protein expression and RNA encapsidation. To study negative-strand synthesis in the absence of concomitant positive-strand synthesis, an RNA3 derivative lacking the 5′-end RNA replication signals was derived from plasmid pB3MS114 (Sullivan, unpublished). This construct removes 86 of the 91 nucleotides in the 5′ noncoding region of RNA3 (leaving the 5 nucleotides proximal to AUG) and replaces them with 38 nucleotides derived from the 5′ end of the yeast GAL1 message.
Point-mutated BMV 1a derivatives H80A, D106A, R136A, and K691A have been described previously (3). Mutations D755A and G781S were constructed in a similar manner, with the unique site elimination method (Chameleon kit; Stratagene) and oligonucleotides d(CATAGGCTGCTTGTTGCGGAGGCTGGTTTACTAC) and d(CAAGTTCTTGCCTTTTCGGACACAGAGCAGCAGATTTC), respectively. Mutation L52P arose spontaneously during PCR amplification of a part of the 1a open reading frame. It was initially detected due to its phenotype, the causative mutation was identified by sequencing, and a ClaI fragment of the 1a coding area of the mutant construct was used to transfer the mutation to pB1CT19. To construct URA3 marker-containing versions of pB1CT19 derivatives, the parent plasmids were cut at the unique BsiWI site within the HIS3 gene and ligated with a URA3 gene containing an HpaI-PvuII fragment derived from YEp352 (21).
Yeast cultures derived from single colonies of transformants on selective plates containing glucose were grown in selective liquid medium containing galactose and harvested in mid-logarithmic phase (optical density at 600 nm, 0.5 to 0.6). Total yeast RNA was isolated by extraction with hot acidic phenol, and concentrated by ethanol precipitation, as described elsewhere (34); 5-μg aliquots of total RNA were analyzed by formaldehyde-agarose gel electrophoresis as described elsewhere (39), followed by blotting onto Nytran nylon membranes (Schleicher & Schuell). Specific 32P-labeled hybridization probes were generated by transcription from plasmid restriction fragment templates with a Strip-EZ kit (Ambion). The probe used to detect BMV positive-strand RNAs was derived from fragment HindIII-EcoRI at the conserved 3′ end of RNA3, and the probe used to detect negative-strand RNA was derived from fragment SalI-XbaI in the coat protein coding region. Radioactive signals were detected and measured with a Molecular Dynamics PhosphorImager model 425 imaging system. All RNA analysis experiments were done with at least three independent yeast transformants, which gave reproducible results.
Oligonucleotide d(GCGGTCCAACGATTTCTGCG), complementary to nucleotides 64 to 83 of BMV subgenomic RNA4, was labeled with [γ-32P]ATP and T4 polynucleotide kinase; 106 cpm of labeled primer was annealed with 5 μg of yeast RNA, with 20 ng of BMV virion RNA, or with 10 ng of an in vitro-transcribed BMV RNA4 derivative, which was obtained by using a Megascript T7 kit (Ambion) with a PCR product containing RNA4 nucleotides 1 to 520 fused to T7 promoter as a template. The primer was extended in a 30-μl final volume containing 20 mM Tris-HCl (pH 8.3), 2.5 mM MgCl2, 7 mM dithiothreitol, 0.25 mM dATP, dGTP, dCTP, and dTTP, and 10 U of avian myeloblastosis virus reverse transcriptase (Promega) for 5 min at 42°C. Nucleic acids were precipitated with ethanol and analyzed in 6% polyacrylamide-urea gels. DNA sequencing ladder was generated with the same oligonucleotide with plasmid pB3MS82 as a template, using a Sequitherm cycle sequencing kit (Epicentre Technologies).
Preparation of total yeast membranes by flotation and assays for the methyltransferase and covalent guanylate binding activities of 1a have been recently described (3). For Western blotting, proteins separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) were transferred to a polyvinylidene fluoride membrane (Immobilon P; Millipore). After blocking with 5% nonfat dry milk and treatment with polyclonal rabbit anti-1a antiserum (1:6,000) (44), detection was performed with an Immun-Star kit (Bio-Rad) and Lumi-Imager luminescence imager (Boehringer Mannheim). GUS activity was measured with the chemiluminescence-based assay GUS-Light (Tropix Inc., Bedford, Mass.) according to the manufacturer's instructions.
Our aim was to construct defined single-point mutations in BMV 1a that would abolish either the capping-related functions or the putative helicase/NTPase activities of BMV 1a protein. The effects of the capping domain mutations H80A, D106A, and R136A (Fig. (Fig.1A)1A) on the enzymatic activities of 1a have been previously described (3). All three mutations reduced both the guanine-7-methyltransferase and covalent guanylate binding activities to very low or undetectable levels. The most significant activity remaining was the ability of mutant H80A to methylate GTP at 3% of the wt level. However, as this derivative was completely prevented from forming the covalent guanylate complex, a presumed intermediate in mRNA capping, it should also be blocked in the RNA capping reaction. Possible roles for these residues in enzymatic activities have been discussed previously (3, 5), and the homologous residues of Sindbis virus nsP1 are essential for virus replication (51). We included a further mutant, L52P (Fig. (Fig.1A),1A), located at the N-terminal end of the capping domain, in our analysis. This mutant may be more disruptive of the structure of BMV 1a, as it introduces a rigid proline residue into the sequence, whereas the other mutations are mainly alanine substitutions, which are generally presumed not to alter the overall structure of mutated proteins. The enzymatic activities of L52P were compared with those of wt BMV 1a: L52P had no detectable activity in covalently binding guanylate or in methyltransferase assays with GTP as the methyl acceptor substrate (Fig. (Fig.1B).1B). Thus, mutation L52P also disrupts the enzymatic activities of 1a involved in RNA capping.
For designing mutations in the helicase domain of BMV 1a, we relied on homology with better-characterized members of the immense family of helicase-like proteins (16). We individually mutated three of the most conserved residues in the C-terminal domain of BMV 1a: K691 and D755 to alanine, and G781 to serine (Fig. (Fig.1A).1A). K691 is located in helicase motif I. For the related protein Semliki Forest virus nsP2, it was shown previously that the corresponding lysine is needed for NTPase and helicase activities and for virus replication (15, 46, 47). A lysine residue in this position is universally conserved in helicase-like proteins and shown to be essential for NTP binding and hydrolysis, whereas the universally conserved aspartate corresponding to D755 in helicase motif II is involved in divalent cation coordination and either helicase activity or both helicase and NTPase activities (16). Motif III (containing G781 of BMV 1a) has been implicated in coupling of NTP hydrolysis with nucleic acid unwinding: proteins mutated in motif III often retain their NTPase activity and lose only the helicase activity (13, 16, 18). Thus, these three mutations are expected to inhibit the putative helicase and/or NTPase activities of BMV 1a in different ways.
When BMV 1a and 2a proteins are expressed in yeast together with BMV RNA3, they catalyze RNA3 replication and subgenomic RNA4 synthesis, which also leads to the expression of any open reading frame appropriately inserted on RNA4 (X in Fig. Fig.1C).1C). Such an open reading frame cannot be translated directly from RNA3 since it is downstream of the 3a gene (24, 27). Thus, translation products of RNA4 are indicative of RNA3 negative-strand production and subgenomic RNA4 synthesis. As a first step in studying the functionality of mutated 1a derivatives, we transformed the relevant plasmids together with a wt 2a-expressing plasmid into a yeast strain with chromosomally integrated cDNA expression cassettes for RNA3 derivatives with the coat protein gene replaced by the URA3 or GUS gene as a reporter (23) (Fig. (Fig.1C).1C). In the presence of wt 1a and 2a, this yeast strain can efficiently grow on plates lacking uracil, as the URA3 protein necessary for uracil biosynthesis is translated following BMV replication (Fig. (Fig.1D).1D). However, none of the yeast strains containing mutated 1a derivatives was able to grow on plates lacking uracil, and they closely resembled the yeast strain lacking 1a (Fig. (Fig.1D).1D). Also, the reporter strain containing wt 1a expressed high levels of GUS activity (300,000 arbitrary units), whereas no activity (<1,500 units; similar to strains without 1a expression) was detected in extracts of strains containing any of the mutated derivatives of 1a. Thus, all of the 1a mutants constructed appeared to be grossly defective at some stage of BMV RNA3 replication, subgenomic mRNA synthesis, or both.
We then studied the replication of BMV RNA3 directly by Northern blotting to detect both positive-strand and negative-strand RNA species (Fig. (Fig.2A).2A). It was verified that all of the mutant 1a derivatives were expressed at levels similar to that of wt 1a (Fig. (Fig.2A,2A, upper panel), with the exception of mutant L52P, which was reproducibly present at a reduced level (lane 3). In the presence of wt 1a, large amounts of positive-strand and negative-strand RNA3 and positive-strand subgenomic RNA4 accumulated (lane 2), whereas in the absence of 1a only the low level of positive-strand RNA3 produced by DNA-mediated transcription was detected (lane 1, where the RNA3 level is about 2% of that in lane 2). Positive-strand RNA3 accumulation for mutants L52P and H80A (lanes 3 and 4) was increased relative to the minus-1a control, whereas for the other mutants (lanes 5 to 9), there was little or no reproducible increase. Negative-strand synthesis was detectable at 8 to 10% of the wt level for mutant H80A (lane 4) and at 1 to 2% for mutants D106A and R136A (lanes 5 and 6). The helicase domain mutants as well as mutant L52P synthesized no detectable negative-strand RNA3. Mutant H80A accumulated subgenomic RNA4 at the low level of approximately 10% of wt. In some experiments, subgenomic RNA4 was barely detectable for mutants D106A and R136A (less than 0.5% of wt), whereas in other experiments, it was below the detection limit. The other mutants synthesized no detectable RNA4. It is notable that in the previous experiment, cells expressing mutant H80A produced no detectable GUS activity, even though this mutant is capable of RNA4 synthesis. As shown further below, this result is consistent with a defect in RNA4 capping, which would inhibit translation.
In yeast expressing wt 1a and 2a, the level of negative-strand RNA3 accumulation depends on RNA-dependent amplification of positive-strand RNA3 (24). To break this cycle of dependence (Fig. (Fig.1C)1C) and test more directly for defects in negative-strand synthesis, negative-strand synthesis was further studied under conditions where synthesis of progeny positive strands is prevented (Fig. (Fig.2B).2B). This was achieved by using an RNA3 derivative in which the virus-specific 5′ noncoding region sequences needed specifically for positive-strand synthesis were replaced by 5′ noncoding sequences of the yeast GAL1 mRNA. With this RNA3 derivative, negative-strand RNA3 synthesis catalyzed by wt 1a and 2a is between 15 and 20% of that detected with wt RNA3 (A. O. Noueiry, unpublished data). Under these conditions, negative-strand synthesis was again detected for the same set of mutants as previously: for H80A at approximately 20% of wt, and for D106A and R136A at 3% of wt (Fig. (Fig.2B).2B). These levels are now somewhat closer to wt than previously, as the amplification cycle using progeny positive strands to drive further negative-strand synthesis was inhibited in the wt case. These results point to an early defect in the replication cycle, at or preceeding the synthesis of negative-strand RNA3, for all of the mutants constructed. For the helicase mutants and for L52P, this defect was absolute; for the other capping domain mutations, it varied in severity.
1a alone, in the absence of 2a and RNA replication, can mediate dramatic stabilization of RNA3 in yeast. As described in the introduction, multiple results link this stabilization to the ability of 1a to recruit RNA3 from translation to replication. Therefore, this step would precede negative-strand RNA synthesis, which additionally requires the polymerase-like 2a protein. In our experiments, as in previous studies with 1a expressed from the ADH1 promoter (26), wt 1a caused RNA3 to accumulate to concentrations 8- to 10-fold higher than those observed in the absence of 1a expression (Fig. (Fig.3,3, lanes 1 and 2). The helicase domain mutants K691A, D755A, and G781S were defective in 1a-mediated RNA3 stabilization (lanes 7 to 9), showing no significant increase in RNA3 accumulation over the minus-1a control (lane 1). Surprisingly, the capping domain mutants divided into two groups. Mutants L52P and H80A displayed increased RNA3 stabilization approximately twofold higher than the wt 1a level (lanes 2 to 4), whereas mutants D106A and R136A showed very little if any increase in RNA3 accumulation over the minus-1a control (lanes 5 and 6), resembling the helicase mutants in this respect. Thus, the earliest observed defect for D106A and R136A capping domain mutants and for all helicase domain mutants was at the level of 1a-mediated RNA3 stabilization. Mutants L52P and H80A, despite showing increased RNA3 stabilization, were defective in negative-strand RNA synthesis.
The yeast chromosomal gene XRN1 encodes a 5′-3′ exoribonuclease specific for uncapped RNAs, which is centrally involved in the major pathway of mRNA degradation (22, 38). Specifically, shortening of the poly(A) tail of mRNA triggers decapping, which is followed by XRN1 nuclease-catalyzed degradation of the message body. Thus, in yeast strains devoid of XRN1, mRNAs lacking a cap structure are stabilized. Since in wt yeast uncapped viral RNA products will not accumulate, and failure to cap RNAs would appear similar to a failure to synthesize them, we used a Δxrn1 yeast strain to discern more directly the contribution of RNA capping defects to the poor RNA3 and RNA4 accumulation catalyzed by BMV 1a capping enzyme mutants. A subset of mutants was analyzed for RNA3 replication in Δxrn1 yeast compared with wt yeast (Fig. (Fig.4).4). The three 1a derivatives containing mutations at the capping enzyme active site residues were included in this analysis. Only one helicase domain mutant was included, since the three helicase mutants had shown similar phenotypes in earlier experiments. In the absence of 1a, DNA-derived RNA3 transcripts accumulated to approximately 12-fold-higher concentrations in Δxrn1 than in wt yeast (lanes 1 and 7). The increased accumulation presumably reflects an increase in the half-life of RNA3 due to accumulation of decapped RNA molecules, suggesting that the XRN1 pathway plays a role in RNA3 degradation in wt yeast cells. In the presence of wt 1a and 2a, under conditions of complete RNA3 replication, RNA3 accumulated twofold more in Δxrn1 than wt yeast (lanes 2 and 8). As the half-life of RNA3 in the presence of 1a is already very long in wt cells (26), XRN1 deletion has a proportionally much smaller effect in this situation. Subgenomic RNA4 accumulation was increased in Δxrn1 yeast to an even greater extent, four- to sixfold, than RNA3. In contrast, negative-strand RNA3 accumulation was twofold lower in Δxrn1 than in wt yeast, showing that under these conditions the elevated level of positive-strand RNA3 does not necessarily lead to increased negative-strand synthesis.
The helicase domain mutant K691A showed essentially no change in the accumulation of RNA3 replication products in Δxrn1 yeast and still closely resembled the situation where no 1a was present (Fig. (Fig.4,4, compare lanes 6 and 12 to lanes 1 and 7). The capping domain mutant H80A resembled wt 1a in positive-strand RNA levels: both RNA3 and RNA4 accumulated to higher levels in Δxrn1 yeast. The negative-strand levels of H80A were similar in wt and Δxrn1 yeast. In contrast, for mutants D106A and R136A, negative-strand RNA3 accumulation was approximately fourfold greater in Δxrn1 than wt yeast. Moreover, these mutants accumulated very large amounts of positive-strand RNA4 in Δxrn1 yeast, whereas little if any RNA4 was present in wt yeast. This dramatic increase in RNA4 may reflect both increased synthesis, due to increased amounts of template negative strands, and the stabilization of RNA4 in Δxrn1 compared to wt yeast. The extent of RNA4 stabilization in Δxrn1 may be even greater for mutants D106A and R136A than for wt 1a, since for these capping mutants, any RNA4 synthesized may be uncapped and thus is expected to be highly unstable in wt yeast cells.
To study the capping status of subgenomic RNA4 in cells expressing different 1a derivatives, a primer extension experiment was performed. RNA4 isolated from BMV virions gave rise to two prominent primer extension products, one corresponding to polymerase stopping at the 5′ end of RNA4 and the second product extending one nucleotide beyond the 5′ end RNA4 (Fig. (Fig.5,5, lane 2). The second product is due to elongation of the cDNA product by one nucleotide, with the capping G residue acting as a template (2). Accordingly, uncapped RNA4 produced by in vitro transcription gave rise to the shorter product corresponding to the 5′ end of RNA4 (Fig. (Fig.5,5, lane 1). RNA4 from wt yeast or from Δxrn yeast expressing wt 1a also gave two products, indicating that it was predominantly capped (lanes 4 and 6). In contrast, RNA4 isolated from any of the capping domain mutants, in either wt or Δxrn1 yeast, gave only the shorter product (lanes 5 and 7 to 9). These results indicate that 1a proteins mutated at the capping enzyme active sites are defective in capping subgenomic RNA4 in vivo.
Combining mutations may provide insight into the order or manner in which multiple functions are organized in a pathway. We combined mutation H80A, which showed greater than wt stabilization of RNA3 and retained partial functionality in negative-strand synthesis, with capping domain mutation R136A or with helicase mutation K691A, both of which on their own showed only minimal function (Fig. (Fig.22 and and3).3). The double mutants were compared with their parents with respect to both 1a-mediated RNA3 stabilization and full RNA3 replication (Fig. (Fig.6).6). In mutant combination H80A-K691, the nonfunctionality of mutant K691 was dominant, and this combination was completely defective in RNA3 stabilization and in negative-strand RNA3 synthesis (lane 5). Thus, the distinct function provided by the helicase-like domain is absolutely required also in this double-mutant context. The mutant combination H80A-R136A gave more complex results, showing phenotypes intermediate between its parents (lane 4). 1a bearing mutations H80A and R136A mediated an approximately 4-fold increase in RNA3 accumulation in the absence of 2a, less than H80A (18-fold) or wt 1a (8- to 10-fold) but clearly more than the nonfunctional single mutant R136A. During full replication, negative-strand synthesis of H80A-R136 was very close to that of H80A, a severalfold increase over R136A. Overall levels of positive-strand RNA3 and subgenomic RNA4 were intermediate between H80A and R136A. Thus, in the H80A-R136 combination, mutation H80A partially suppresses R136A, leading to increased function.
To study possible complementation or other types of genetic interaction between 1a mutants, we expressed two 1a derivatives simultaneously in yeast cells, using plasmids containing HIS3 and URA3 selectable markers. In a first series of experiments, we used plasmids similar to those used in all previous experiments, containing 1a derivatives expressed from the ADH1 promoter. Under these conditions, RNA3 replication was responsive to the gene dosage of 1a: two wt 1a-encoding plasmids caused a twofold increase in replication, as assessed by negative-strand RNA3 accumulation (Fig. (Fig.7A,7A, lanes 1 to 3). Expression of wt 1a together with nonfunctional variants, either capping domain mutant D106A or helicase mutant K691A, gave wt levels of negative-strand RNA synthesis (lanes 4 to 7). When the two mutants were expressed together, RNA3 negative-strand synthesis was very low (lanes 8 and 9), similar to that of mutant D106A on its own (Fig. (Fig.2A,2A, lane 5). Thus, there was no indication of intragenic complementation between these two helicase and capping domain mutants.
To detect possible weak genetic interactions, we wanted to increase the level of 1a protein expression, which was achieved by using a stronger, galactose-inducible GAL1 promoter. The same set of experiments was repeated with these GAL1-driven 1a expression constructs (Fig. (Fig.7B).7B). Under these conditions, increased 1a gene dosage gave only slightly increased replication, 120% of negative-strand synthesis compared to 1a expressed from HIS3 marker plasmid alone (lanes 1 to 3). This may in part be due to competition of transcription factors between the several GAL1 promoters present in the system (both 1a constructs, as well as 2a and RNA3, are expressed using a GAL1 promoter), which could limit their level of expression. Alternatively, the higher 1a levels produced may begin to saturate the system, and other host cell factors may become limiting for BMV replication. Similarly to the previous experiment, no complementation between mutants D106A and K691A was evident (lanes 8 and 9). However, a difference was observed in situations where wt 1a was expressed together with either of the mutants: in this case the level of replication, as assessed by negative-strand synthesis, decreased to approximately 60% of wt when D106A was coexpressed with wt 1a (lanes 4 and 6) and to 25% when K691A was coexpressed with wt 1a. The amounts of positive-strand RNA3 and RNA4 paralleled the amounts of negative-strand RNA3. Thus, the mutant proteins D106A and K691A dominantly interfered with the function of wt 1a, but only at high levels of expression. While the interference of D106A may be partially explained by promoter competition, other explanations need to be considered for the strong interference caused by helicase mutant K691A.
We have constructed point mutations at the capping enzyme and helicase active sites of BMV protein 1a (Fig. (Fig.1A),1A), in order to discern some aspects of the contribution of these activities to different stages of RNA replication. All three mutations constructed in the conserved residues of BMV 1a helicase motifs gave rise to indistinguishable phenotypes: no synthesis of negative-strand RNA was detected (Fig. (Fig.2).2). Furthermore, 1a-mediated stabilization of RNA3 derivatives, which reflects recruitment of RNA3 to the replication complex, was also abolished (Fig. (Fig.3).3). Thus, the presumed helicase/NTPase activity may function at or near the earliest steps of BMV RNA replication. It is possible to envision multiple ways in which a helicase/NTPase activity might facilitate RNA recruitment from translation to replication. It might be involved in RNA recognition or in modifying RNA structure to facilitate other recognition events. It might be involved in inhibiting translation or in mediating some kind of RNA transport process, which may be needed in formation of the replication complexes on the endoplasmic reticulum membrane. New assay systems need to be devised to distinguish between these and other possibilities.
It is noteworthy that the helicase mutants synthesized no detectable negative-strand RNA (detection limit is approximately 0.2% of wt). This was in contrast to capping mutants D106A and R136A, which synthesized low (1 to 2% of wt) levels of negative strands, although these capping enzyme and helicase mutants were similarly defective in 1a-mediated RNA3 stabilization. Even RNA3 derivatives lacking the intergenic RE sequences required for 1a-mediated RNA3 stabilization and concomitant enhancement of RNA synthesis are capable of mediating low (2 to 3%) levels of negative-strand RNA synthesis (43, 49). Presumably other RNA3 cis-acting replication signals such as the conserved tRNA-like 3′ end of RNA3 can mediate this low level of RE-independent RNA3 recruitment to replication. Thus, the helicase active site may be very strictly required for all kinds of RNA recruitment to replication, whether mediated by RE or by other sequences; alternatively, the helicase mutants may be additionally blocked in the synthesis of negative strands. For other positive-strand RNA viruses, there are also data implicating the virus-encoded helicases/NTPases in steps at or prior to negative-strand RNA synthesis. In the case of poliovirus, the guanidine-inhibited NTPase activity of replicase protein 2C appears to be required for initiation of negative-strand RNA synthesis (8, 42). In bovine viral diarrhea virus, a member of the family Flaviviridae, helicase-negative mutants fail to synthesize any detectable negative-strand RNA (17, 20).
Yet other functions for helicase-like domains within the alphavirus-like superfamily have been suggested based on studies of temperature-sensitive (ts) mutations. A strongly ts linker insertion allele of BMV 1a, containing a mutation within the helicase-like domain, was blocked in negative-strand, positive-strand, and subgenomic RNA synthesis at the restrictive temperature (31), implicating the helicase domain in all steps of RNA synthesis. On the other hand, experiments with ts point mutations within the helicase domain of alphavirus protein nsP2 have been interpreted to support a role for this domain in the conversion of replication complexes from negative-strand to positive-strand RNA synthesis (10). However, it should be noted that the effects of the ts mutations on the enzymatic activities of the helicase-like domain have not been studied, and it is possible that they alter other functions contained within the same domain. The same issue could of course be raised for the mutants studied here, although they were specifically designed to alter predicted active site residues. Thus, the helicase-like proteins of positive-strand RNA viruses may have multiple distinct functions at different stages of RNA replication. Our current results point to a function for BMV 1a helicase-like protein at an early step of RNA template recruitment; other viral helicases/NTPases may also function at this stage.
BMV 1a helicase mutant K691A dominantly interfered with the function of wt 1a, but only when both 1a constructs were expressed at high levels (Fig. (Fig.7).7). It is difficult to explain this result by considering the dimerization of 1a (40), since this might be expected to have similar effects at all levels of expression, if mutant and wt proteins randomly formed heterodimers. Instead, it is possible that the mutant proteins at higher levels of expression interact with another limiting component, such as a host protein or possibly a membrane component necessary for replication, since 1a is responsible for the membrane association of BMV RNA replication complexes (9, 45).
Our results indicate that the BMV 1a capping domain mutants previously shown to be defective in guanylate methylation and binding in vitro (3) were also defective in capping subgenomic RNA4 in vivo, as confirmed by primer extension (Fig. (Fig.5).5). As predicted, deletion of the yeast chromosomal gene XRN1, encoding the cap-sensitive 5′-3′ RNA exonuclease, was able to partially suppress the effects BMV 1a capping mutations D106A and R136A. In wt yeast with D106A or R136A, the level of RNA4 is extremely low, often undetectable (Fig. (Fig.2A2A and and4),4), whereas in Δxrn1 yeast these two mutants accumulated large amounts of uncapped RNA4 (Fig. (Fig.44 and and5).5). These results show that mRNA capping is normally an essential function for a positive-strand RNA virus utilizing capped RNAs and that BMV defective in RNA capping can replicate only in a suitably altered host genetic background. It is notable that mutant H80A, which is also defective in capping RNA4 (Fig. (Fig.5),5), was suppressed to a much lower level than D106A or R136A in Δxrn1 yeast (Fig. (Fig.4).4). This implies that H80 is also defective in steps other than RNA capping, as discussed below.
Our results have uncovered a function for the BMV 1a capping enzyme domain in the early step of 1a-mediated RNA3 stabilization, or template recruitment. This function is entirely distinct from RNA3 capping (plasmid-derived RNA3 is capped by cellular enzymes in the nucleus) or hypothetical recapping by 1a (26), since the engineered mutations that destroyed 1a capping functions (3) (Fig. (Fig.11 and and5)5) led to opposite effects on 1a-induced RNA3 stabilization: mutants L52P and H80A showed increased RNA3 stabilization, whereas mutants D106A and R136A were defective in RNA3 stabilization (Fig. (Fig.3).3). Yet all of these mutant proteins displayed poor synthesis of negative-strand RNA3 (Fig. (Fig.2),2), as if mutants L52P and H80A were frozen at the step of template recruitment and only rarely able to go beyond that. It also noteworthy that when mutations H80A and R136 were combined within the same protein, both parent mutant phenotypes contributed to the replication phenotype (Fig. (Fig.4).4). These results were unexpected, because mutations H80A, D106A, and R136A were designed to alter the active site residues involved in RNA capping reactions, and it appeared unlikely that such active site residues would have other functions directly involved in, for instance, RNA3 recognition.
It is possible that the engineered capping mutations may simply perturb 1a conformation and thereby disturb other functional sites in 1a protein involved in direct or indirect RNA3 recognition or manipulation. However, it is also possible that there is a connection between RNA capping and 1a-mediated RNA stabilization. RNA capping is expected to be coupled only to positive-strand RNA synthesis, since positive-strand RNAs need to be efficiently capped, and negative-strand RNAs are not capped. 1a may have distinct conformational states active in RNA3 stabilization, in negative-strand RNA synthesis, and in positive-strand RNA synthesis. The mutations that we have constructed might favor some of these conformations over others by altering the binding of substrates involved in RNA capping reactions or by related effects. Another possibility is that 1a-mediated RNA3 recruitment from translation to replication might involve 1a-mediated recognition of the RNA3 cap structure. Such a model would explain the ability of 1a-induced RNA stabilization to inhibit translation (26). It is also in keeping with results on the contribution of host gene LSM1 to BMV RNA replication, which suggest that 1a-induced RNA3 stabilization may depend on interactions with the RNA3 5′ end as well as the internal RE element (12). The capping mutations that we have constructed might alter the recognition of the reaction product, a cap structure. Too tight a recognition could lead to increased RNA3 stabilization, while a failure in recognition could lead to a failure to stabilize RNA3.
Our results highlight the multifunctionality of RNA virus replication proteins, which complicates the interpretation of mutational studies. It is instructive that 1a proteins mutated at capping enzyme active sites were also defective in RNA3 stabilization. In addition to their enzymatic functions, replication proteins are also involved in numerous binding interactions possibly involving recognition of host proteins required for RNA replication (23), direct or indirect recognition of the cis-acting elements in viral RNAs (49), and binding of the RNA replication complexes to host cell membranes (6, 45). The experiments described in this paper suggest that the mechanisms involved in 1a-mediated stabilization of RNA3 may be relatively complex and suggest several possibilities for steps involved in and functions required for this early replication stage.
We thank members of our laboratory for helpful discussions throughout the course of this work.
This research was supported by the National Institutes of Health through grant GM35072. P.A. is an investigator of the Howard Hughes Medical Institute.