Characterization of the AdoMet-Dependent Guanylyltransferase Activity That Is Associated with the N Terminus of Bamboo Mosaic Virus Replicase

Abstract

Bamboo mosaic virus (BaMV), a member of the potexvirus group, infects primarily members of the Bambusoideae. Open reading frame 1 (ORF1) of BaMV encodes a 155-kDa polypeptide that has long been postulated to be a replicase involved in the replication and formation of the cap structure at the 5′ end of the viral genome. To identify and characterize the enzymatic activities associated with the N-terminal domain of the BaMV ORF1 protein, the intact replicase and two C-terminally truncated proteins were expressed in Saccharomyces cerevisiae. All three versions of BaMV ORF1 proteins could be radiolabeled by [α-³²P]GTP, which is a characteristic of guanylyltransferase activity. The presence of S-adenosylmethionine (AdoMet) was essential for this enzymatic activity. Thin-layer chromatography analysis suggests that the radiolabeled moiety linked to the N-terminal domain of the BaMV ORF1 protein is m⁷GMP. The N-terminal domain also exhibited methyltransferase activity that catalyzes the transfer of the [³H]methyl group from AdoMet to GTP or guanylylimidodiphosphate. Therefore, during cap structure formation in BaMV, methylation of GTP may occur prior to transguanylation as for alphaviruses and brome mosaic virus. This study establishes the association of RNA capping activity with the N-terminal domain of the replicase of potexviruses and further supports the idea that the reaction sequence of RNA capping is conserved throughout the alphavirus-like superfamily of RNA viruses.

Bamboo mosaic virus (BaMV), a member of potexvirus group, has a plus-strand RNA genome (∼6.4 kb) with a 5′ m⁷G(5′)ppp(5′)G cap structure and a 3′ poly(A) tail (18). The 4.1-kb open reading frame 1 (ORF1) gene of BaMV encodes a 155-kDa polypeptide (17). The presence of conserved motifs such as GKS and GDD, signatures of helicase (6, 10) and polymerase (6, 13), respectively, led to the prediction that the 155-kDa protein may have an RNA helicase activity in the middle region and a polymerase activity in the C-terminal domain. The RNA-dependent RNA polymerase activity of the 155-kDa viral protein was recently corroborated by showing that the Escherichia coli-expressed viral protein was able to synthesize complementary RNA molecules using 3′ end fragments of either the plus or minus strand of a BaMV RNA transcript as a template and that this polymerase activity was abolished when the GDD motif was deleted (16). The N-terminal region of the ORF1 product of potexviruses also shows distant similarity to the putative Sindbis virus-like methyltransferase (24), suggesting that the N terminus of the BaMV 155-kDa protein may be responsible for the cap formation at the 5′ ends of genomic and subgenomic RNA transcripts. Figure 1 shows the assumed domain organization of the 155-kDa BaMV replicase and the conserved His, Asp, and Arg residues found in the Sindbis virus-like methyltransferase.

FIG. 1.

(A) Domain organization of the replicase (1,365 amino acids) of BaMV. A highly hydrophilic region and a proline-rich stretch are thought to divide the 155-kDa replicase into methyltransferase, helicase, and RNA-dependent RNA polymerase (RdRp) domains. B1, B2 (908 amino acids), and B3 (442 amino acids) represent the intact BaMV replicase and two C-terminally truncated proteins expressed in yeast in this study. The conserved residues (His, Asp, Arg, and Tyr) of the Sindbis virus-like methyltransferase found in the N terminus of the BaMV replicase are indicated. (B) Alignment of partial sequences of the N termini of BaMV replicase, 1a protein of brome mosaic virus (BMV), and nsP1 proteins of Semliki Forest virus (SFV) and Sindbis virus (SIV). Asterisks, conserved residues including His, Asp, and Arg. The association of mRNA capping activities with nsP1 of Semliki Forest virus and Sindbis virus and 1a of brome mosaic virus has been established (2, 4, 12, 14, 19).

The 5′-terminal cap structure m⁷G(5′)pppN is a characteristic of eukaryotic mRNAs that is required for translation and stability. The formation of the cap requires three consecutive enzymatic activities according to the studies of capping reactions in mammalian and Saccharomyces cerevisiae cells and several viral systems. The proposed reaction scheme is summarized as the following (21, 28):

1

2

(The reaction at step 2 can be divided further into two half-reactions: enzyme + ^γp^βp^αpG → enzyme-^αpG + ppi and enzyme − ^αpG + ^β′p^α′pN(pN)_n → G^αp^β′p^α′pN(pN)_n + enzyme.)

3

where AdoMet is S-adenosylmethionine and AdoHcy is S-adenosylhomocysteine. RNA 5′-triphosphatase removes the 5′ γ phosphate of nascent mRNA (step 1), after which mRNA guanylyltransferase donates a GMP moiety, derived from GTP, to form a 5′–5′ triphosphate linkage, typical for a cap structure (step 2). In the step 2 reaction, the GMP moiety is linked first to the enzyme as a covalently bound intermediate (the first half-reaction) and then transferred to the 5′ diphosphate terminus of the RNA molecule (the second half-reaction). Thereafter, the cap is methylated by AdoMet at position 7 of the terminal guanosine, yielding a cap 0 structure and AdoHcy (step 3). A variety of RNA viruses have evolved diverse pathways for cap formation (29). For example, GTP is methylated before the transguanylation reaction in alphaviruses (1) and brome mosaic virus (4, 12).

Because the capping of cellular mRNAs is a nuclear function, it is not accessible to cytoplasmic viruses. Thus, BaMV, replicating in the cytoplasm of host cells, should be equipped with its own mRNA capping system. It was also interesting to know whether the capping activity of BaMV has characteristics similar to those of alphaviruses and brome mosaic virus since these viruses all belong to the large alphavirus-like superfamily of RNA viruses. To test the hypothesis that the N-terminal domain of the 155-kDa polypeptide harbors RNA capping activity, the ORF1-encoded polypeptide was expressed, in different lengths, in Saccharomyces cerevisiae and the activities associated with the viral proteins were characterized. The results show that the N-terminal domain of the 155-kDa viral protein indeed harbors an AdoMet-dependent guanylyltransferase activity. The viral protein forms a covalent complex with m⁷GMP as for alphaviruses and brome mosaic virus. This discovery further supports the idea that the order of the capping reaction is conserved among members of the alphavirus-like superfamily, although there are limited sequence similarities among them.

MATERIALS AND METHODS

Chemicals.

Nucleotides and nucleotide analogs such as GTP, dGTP, GDP, GMP, m⁷GMP, and guanylylimidodiphosphate (GIDP) were purchased from Sigma. AdoMet was from Boehringer Mannheim, and ³²P-labeled nucleotides and Ado[methyl-³H]Met were from NEN.

Plasmid construction.

Plasmid pYES2 (Invitrogen) was used to carry the BaMV cDNAs for viral protein expression in S. cerevisiae INVSc1 (MATα his3-Δ1 leu2 trp1-289 ura3-52). Primer pair 5′-CCCAAGCTTATGGCACTCGTTTCTAAAGTCTTTGAC and 5′-GCCAGATCTAGAGAGTAGGTCAGTTATCCG was used to amplify the full-length cDNA of BaMV ORF1 (4,095 nucleotides [nt]) in a 50-μl PCR mixture that contained 1 ng of pBL (carrying the full-length cDNA of BaMV), 0.32 μM (each) primer, 0.2 mM (each) deoxyribonucleotide, and 2.5 U of Pfu polymerase. Sequences AAGCTT and TCTAGA represent the engineered cutting sites of HindIII and XbaI, respectively, and the sequences in italics are within the coding region of BaMV ORF1. The PCR was carried out for 30 cycles (94°C for 45 s, 65°C for 45 s, 72°C for 8 min), followed by a 10-min extension at 72°C. Another set of primers, 5′-CCCAAGCTTATGGCACTCGTTTCTAAAGTCTTTGAC and 5′-GCTCTAGAGCTCATTTGGGCTCCAAGGGTTCATC, was used to amplify a 3′-end-deleted BaMV ORF1 (2,724 nt) that encodes both the putative methyltransferase and helicase domains, whereas 5′-CCCAAGCTTATGGCACTCGTTTCTAAAGTCTTTGAC and 5′-GCTCTAGAGCTCATTCGGTAGTTGCTGCGTCTGT were used for the amplification of an even shorter 3′-end-deleted fragment of ORF1 (1,326 nt) that encodes only the putative methyltransferase domain under the PCR conditions described above. The PCR-amplified cDNAs were then digested with HindIII and XbaI, ligated with pYES2, and transformed into E. coli Top10F (Invitrogen) to obtain expression vector pYEB1 (containing a 4,095-nt cDNA), pYEB2 (containing a 2,724-nt cDNA), and pYEB3 (containing a 1,326-nt cDNA). The expressed viral proteins encoded by plasmids pYEB1, pYEB2, and pYEB3 are abbreviated B1 (1,365 amino acids), B2 (908 amino acids), and B3 (442 amino acids), respectively (Fig. 1).

Plasmid pET29 (Novagen) was used for the expression of the N-terminal domain of the BaMV ORF1 product in E. coli. Primer pair 5′-GTGCGGCACATATGGCACTCGTTTCTAAAGTCTTTGAC and 5′-CTTGCGAAGCTTAAGCCGCTTTGCATTCTGGT was used to amplify a 1,449-nt cDNA fragment containing the 5′ end of BaMV ORF1 in a PCR. CATATG and AAGCTT represent the engineered cutting sites of NdeI and HindIII, respectively. The sequences in italics are within the coding region of BaMV ORF1. The conditions of PCR were as described above. The endoribonuclease-digested cDNA fragment was inserted into pET29 and transformed into E. coli BL21(DE3). This construct (pEBM2984) allows the production of the N-terminal portion (the first 483 amino acids) of the 155-kDa protein in E. coli cells.

Expression of recombinant viral proteins in yeast cells.

To express the viral proteins in yeast cells, plasmids pYEB1, pYEB2, and pYEB3 were transferred separately into S. cerevisiae cells by electroporation (pulse controller; Bio-Rad) under conditions of 1.5 kV, 25 μF, and 200 Ω. The transformed yeast cells were selected by growing them in glucose-containing SC agar (2% glucose, 0.67% yeast nitrogen base without amino acids, 0.2% Kiwibrew solution, 2% agar). The absence of uracil in SC agar provides a selective pressure for the cells carrying plasmid derivatives of pYES2. The acquisition of the desired plasmid by yeast cells was confirmed by the amplification of the respective cDNA fragment by PCR.

To induce the BaMV proteins in yeast, recombinant cells grown in 10 ml of glucose-containing SC medium were pelleted down, washed, and resuspended in 10 ml of galactose-containing SC medium (2% galactose). The cultivation was continued for another 24 h at 30°C. The harvested pellets were suspended in 0.25 ml of ice-cold protein extract buffer (40 mM Tris [pH 8.0], 80 mM NaCl, 1.6 mM EDTA, 4 mM dithiothreitol [DTT], 4 mM EGTA, 4 mM phenylmethylsulfonyl fluoride), and disrupted at 4°C by vigorous mixing (10 times for 30 s each) with 0.45 g of acid-treated glass beads (0.3 to 0.45 mm) in a 1.5-ml Eppendorff tube. The cell extract was then subjected to centrifugation at 18,000 × g for 10 min at 4°C. The pellet fraction (P18) was further fractionated by sucrose discontinuous-gradient centrifugation. A 2-ml sample, which contained 60% (wt/wt) sucrose, was first overlaid on top of a 1.3-ml 67% sucrose solution in an SW41Ti (Beckman) ultracentrifuge tube. Then 7.1- (50%) and 1.6-ml (10%) sucrose solutions were laid onto the tube. All sucrose solutions contained 100 mM NaCl and 50 mM Tris (pH 7.5). Centrifugation (35,000 rpm) was carried out at 4°C for 20 h.

Preparation of antibodies against the N-terminal portion of the BaMV ORF1 product.

The N-terminal domain of the 155-kDa viral protein was produced by E. coli BL21(DE3) cells harboring pEBM2984 under conditions described previously (16). The insoluble E. coli-expressed truncated protein was thereafter purified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The electroeluted protein from the corresponding protein band was then injected into rabbits to raise antibodies. The antibodies were purified by passing the rabbit serum through a DEAE Affi-Gel blue gel column (Bio-Rad).

Immunoprecipitation and Western blotting.

For immunoprecipitation, the SDS-denatured sample of the guanylyltransferase assay was diluted 20-fold with TET buffer (1% Triton X-100, 50 mM Tris [pH 7.5], 150 mM NaCl, 5 mM EDTA) and then mixed with antibodies conjugated to CNBr-activated Sephadex beads. After 2 h of gentle mixing, the beads were washed six times with TET buffer and subjected to SDS-PAGE analysis. The antibodies, which are specific for the N terminus of the ORF1 product of BaMV, were also used in Western blot analysis.

Activity assay.

The guanylyltransferase activity was detected by the formation of ³²P-labeled proteins after incubation of proteins with [α-³²P]GTP (9, 27). Unless otherwise stated, the standard reaction was carried out at 30°C for 1 h in a 20-μl reaction buffer that contained 5 μl of enzyme preparation, 10 μCi of [α-³²P]GTP (3,000 Ci/mmol), 50 mM Tris (pH 7.5), 5 mM DTT, 2 mM MgCl₂, 10 mM KCl, 1.2% n-octyl-β-d-glucopyranoside, and 100 μM AdoMet. The reaction was stopped by adding SDS (final concentration, 2%) and was followed by 3 min of boiling. The reaction products were analyzed by SDS-PAGE and visualized by autoradiography or phosphorimager.

The methyltransferase activity was detected by the transfer of [³H]methyl from Ado[methyl-³H]Met to a methyl acceptor (14). The reaction was carried out at 30°C for 50 min in a 25-μl solution that contained 50 mM Tris (pH 7.5), 2 mM MgCl₂, 2 mM DTT, 1.2% n-octyl-β-d-glucopyranoside, 10 μM AdoMet, 0.75 μCi of Ado[methyl-³H]Met (80 Ci/mmol), 10 μl of enzyme solution, and 10 mM methyl acceptor. At the end of the reaction, 1 ml of 10 mM ammonium acetate (pH 8.5) was added, and the reaction products were adsorbed to 1 ml DEAE-Sephadex in a Pasteur pipette, washed with the same buffer containing 100 mM NaCl, and then eluted with the same buffer containing 500 mM NaCl. The incorporated ³H on the methyl acceptor was measured by liquid scintillation.

The covalent linkage of [³H]methyl to BaMV protein was performed at 30°C for 20 min in a 30-μl solution that contained 50 mM Tris (pH 7.5), 10 mM KCl, 2 mM MgCl₂, 5 mM DTT, 1.2% n-octyl-β-d-glucopyranoside, 5 μCi of Ado[methyl-³H]Met (80 Ci/mmol), and 10 μl of enzyme solution. To determine the effect of GTP on methylation of proteins, 100 μM GTP was included in the reaction mixture. The reaction was stopped by adding SDS to a final concentration of 2% and was followed by boiling for 3 min. The sample was analyzed by SDS-PAGE in a 10% polyacrylamide gel, and products were visualized by fluorography.

Thin-layer chromatography (TLC) analysis of guanylate moiety.

The reaction products of the guanylyltransferase assay were separated by SDS-PAGE, and the ³²P-labeled proteins were eluted from the gel and concentrated. The protein samples were then incubated at 65°C for 10 min with 0.5 N HCl or 0.5 N NaOH or incubated at 37°C for 20 min with 3.8 M acidic hydroxylamine (pH 4.3) as described by Jose et al. (9). The treated samples were then spotted on a polyethylenimine cellulose plate, developed with formic acid buffer (0.5 M formic acid, 0.5 M LiCl₂), and visualized by autoradiography.

RESULTS

Viral protein expression in yeast cells.

In order to study the inherent activities associated with the N-terminal region of the BaMV 155-kDa protein, we constructed three yeast-expression vectors in this study. Plasmid pYEB1 encodes the entire 155-kDa replicase (B1), while pYEB2 and pYEB3 encode C-terminally truncated proteins with deduced sizes of 100 (B2) and 50 kDa (B3), respectively. The positions of truncation were based on the assumption that the long hydrophilic stretch between amino acids 406 and 520 and the proline-rich segment from amino acid 895 to 910 may represent the borders of functional domains. The expression of B1, B2, and B3 proteins in yeast was determined by Western blotting analysis (Fig. 2). After centrifugation of cell extracts at 18,000 × g, B1 and B2 were found in both soluble (S18) and insoluble (P18) fractions, while B3, the putative methyltransferase domain, was present mostly in the P18 fraction.

FIG. 2.

Expression of the BaMV replicase (B1) and two C-terminally truncated proteins (B2 and B3) in S. cerevisiae. Yeast cells harboring the desired plasmid were disrupted, and the supernatant (S18) and pellet (P18) after centrifugation at 18,000 × g were analyzed by SDS–10% PAGE. The BaMV proteins were detected by rabbit antibodies directed against the N-terminal portion of BaMV replicase. S2, extracted proteins from yeast cells harboring plasmid pYES2 (Invitrogen).

Detection of guanylyltransferase activity.

The formation of the covalently bound GMP-enzyme complex was used as an index for guanylyltransferase activity. In the initial trial we incubated protein samples with [α-³²P]GTP with or without AdoMet and analyzed the reaction products by denaturing SDS-PAGE. The results indicate that B1, B2, and B3 proteins from the pellet fraction (P18) formed covalently bound complexes with [³²P]GMP (Fig. 3). The proteins from the supernatant fraction (S18), however, did not show the corresponding bands (data not shown). The complex formed only when AdoMet was included in the reaction buffer. The band that migrated slightly above the B3 protein and that appeared in every sample including S2 (the yeast background control) was suspected to be the cellular mRNA guanylyltransferase of yeast (53 kDa) according to its apparent molecular mass (26). The labeling reaction of the yeast cellular mRNA guanylyltransferase does not require AdoMet. Therefore, the AdoMet dependence of activity distinguishes BaMV guanylyltransferase from the yeast counterpart. In order to alleviate the interference of the yeast enzyme in the activity assay, we further purified the BaMV protein in the P18 fraction with a discontinuous sucrose gradient. Western blotting analysis of the fractionated samples (B3 in this case) showed that a portion of the viral protein floated to the interface between the 10 and 50% sucrose solution (Fig. 4), suggesting that this part of the viral protein is membrane associated. Similar to the B3 protein, the B1 and B2 proteins could also be found in the membrane fraction (Fig. 5A). All three membrane-associated viral proteins exhibited guanylyltransferase activity only in the presence of AdoMet, and obviously B3 had stronger activity than B2 and B1 (Fig. 5B). The result also shows that the background activity caused by the yeast protein was alleviated, so the membrane fraction was used in a subsequent guanylyltransferase assay. To confirm the ³²P-labeled protein as a BaMV protein, the reaction products of the guanylyltransferase assay were immunoprecipitated using antibodies against the N-terminal portion of the BaMV replicase and then the labeling of B2 and B3 by [α-³²P]GTP was examined by SDS-PAGE analysis (Fig. 6). The requirement of AdoMet to form the covalent ³²P-labeled proteins was observed again. No yeast cellular capping enzyme was found after immunoprecipitation. The aforementioned data show that the N-terminal portion of the BaMV replicase indeed harbors a guanylyltransferase activity and that the presence of AdoMet is crucial for this activity.

FIG. 3.

AdoMet-dependent guanylyltransferase activity. The yeast-expressed BaMV proteins from the pellet fraction (P18) were incubated with [α-³²P]GTP under the conditions described in Materials and Methods and analyzed by SDS-PAGE. The viral proteins (B1, B2, and B3) were radiolabeled only when AdoMet (100 μM) was included in the reaction buffer (arrows). The band immediately above B3 and shown in every lane probably represents the yeast cellular mRNA guanylyltransferase (53 kDa). S2, proteins from the P18 fraction of yeast cells harboring plasmid pYES2.

FIG. 4.

Partition of B3 protein in a discontinuous sucrose gradient. The truncated protein (B3) from the P18 fraction was loaded onto a discontinuous sucrose gradient and subjected to ultracentrifugation as described in Materials and Methods. After centrifugation, the samples from the membrane fraction (interface between 10 and 50% sucrose), intermediate fraction (50% sucrose), sample loading layer (60% sucrose), and pellet fraction (bottom of tube) were collected and analyzed by Western blotting.

FIG. 5.

(A) Western blotting analysis of the viral proteins (B1, B2, and B3) in the membrane fraction. The yeast membrane fraction containing each viral protein was obtained from the discontinuous sucrose gradient and was analyzed with rabbit antiserum directed against the N-terminal portion of the BaMV replicase. (B) Guanylyltransferase assay of the membrane-associated viral proteins. The assay buffer was as described in Materials and Methods and included 100 μM AdoMet and 5 μl of enzyme solution. The intensity of each viral protein in the Western blotting analysis (A) suggests that approximately equal amounts of viral protein were used in each assay. S2, proteins from the P18 fraction of yeast cells the harboring plasmid pYES2.

FIG. 6.

Immunoprecipitation of the ³²P-labeled B2 and B3 proteins. The B2 or B3 protein from the membrane fraction was incubated with [α-³²P]GTP in the presence or absence of AdoMet (100 μM). The reaction mixtures were then incubated with antiserum against the N-terminal portion of the BaMV replicase, and the precipitated samples were analyzed by SDS-PAGE and autoradiography. S2, proteins from the membrane fraction of yeast cells harboring plasmid pYES2.

Characteristics of BaMV guanylyltransferase activity.

Other nucleoside triphosphates including [α-³²P]ATP, [α-³²P]UTP, and [α-³²P]CTP were used to verify the nucleotide selection in the guanylyltransferase assay. The results showed that both B2 and B3 were labeled by [α-³²P]GTP but not by either [α-³²P]ATP, [α-³²P]UTP, or [α-³²P]CTP (Fig. 7). The time course of the ³²P labeling of the viral proteins showed that the extent of labeling increased with the incubation time and reached a plateau after 30 min (data not shown). The activity was also dependent on the concentration of AdoMet; the more AdoMet, the stronger was the activity (data not shown). EDTA (5 mM) abolished the activity, suggesting that Mg²⁺ is crucial. AdoHcy (100 μM) or pyrophosphate (5 mM) was also included in the reaction buffer to determine the effects on the product. The result showed that AdoHcy was unable to stimulate the activity and, in fact, that it had a slightly inhibitory effect against the function of AdoMet. On the other hand, pyrophosphate (5 mM) annihilated transguanylation completely (data not shown). This suggests strong binding of pyrophosphate to the enzyme, and the loss of activity may be caused by the competition between pyrophosphate and GTP binding.

FIG. 7.

Nucleotide selection of the BaMV guanylyltransferase activity. The B2 or B3 protein from the membrane fraction was incubated with [α-³²P]GTP, [α-³²P]ATP, [α-³²P]UTP, or [α-³²P]CTP under standard reaction conditions as described in Materials and Methods. The reaction products were analyzed by SDS-PAGE and autoradiography. S2, proteins from the membrane fraction of yeast cells harboring plasmid pYES2.

TLC analysis of the guanylate moiety.

The AdoMet dependence of guanylyltransferase activity could have different implications. AdoMet could act as an effector that induces protein conformational changes and stimulates transguanylation. Alternatively, GTP could be methylated first by AdoMet and subsequently transferred as an m⁷GMP moiety onto the active site of the enzyme as for alphaviruses and brome mosaic virus. Two approaches were employed to clarify this uncertainty. In the first attempt, we looked into whether the methyl group from AdoMet would covalently link to the BaMV protein by incubating the B3 protein with Ado[methyl-³H]Met in the presence or absence of GTP. The results showed that the B3 protein was actually labeled by ³H and that this reaction occurred only when GTP was present (Fig. 8). The second attempt was to identify the guanylate moiety released from the ³²P-labeled BaMV protein. The ³²P-labeled B3 was isolated by SDS-PAGE and concentrated by ultracentrifugation. The protein was then treated with HCl, hydroxylamine, or NaOH, and the products were separated by TLC. The linkage of the ³²P-labeled moiety to B3 was susceptible to acidic conditions but resistant to alkaline conditions, indicating a phosphoamide type of chemical bonding between the guanylate moiety and the protein (Fig. 9). The migration of the acid-released ³²P-labeled moiety suggests that the guanylate moiety is identical to the standard m⁷GMP. It should be noted that the slightly higher mobility of m⁷GMP in HCl than in acidic hydroxylamine was also found in the standards. Taken together, the above data suggest that GTP is first methylated by the methyl donor, AdoMet, and then covalently linked to an active-site residue of the BaMV capping enzyme through a phosphoamide bond. The active-site residue therefore most probably is lysine, histidine, or arginine. In the subsequent reaction, the m⁷GMP is transferred to the 5′-diphosphate terminus of the RNA to complete the capping process.

FIG. 8.

Methylation of B3 protein by Ado[methyl-³H]Met. The B3 protein from the membrane fraction was incubated with Ado[methyl-³H]Met in the presence or absence of 100 μM GTP as described in Materials and Methods. The reaction products were then analyzed by SDS-PAGE and fluorography. S2, proteins from the membrane fraction of yeast cells harboring plasmid pYES2.

FIG. 9.

TLC analysis of the guanylate moiety released from the ³²P-labeled B3 protein. The ³²P-labeled B3 protein was treated with either HCl, NaOH, or acidic hydroxylamine, and the treated samples were then analyzed by TLC and detected by autoradiography as described in Materials and Methods. The standards GTP, GDP, GMP, m⁷GMP (in hydroxylamine), and m⁷GMP (in HCl) were run along with the analyzed products on a same TLC sheet and visualized by UV illumination (spots A to E, respectively).

Detection of methyltransferase activity.

The methylation of B3 by Ado[methyl-³H]Met and the identification of m⁷GMP linked to B3 point to the existence of a methyltransferase activity on B3. To verify this possibility, B3 was incubated with Ado[methyl-³H]Met along with several potential methyl acceptors. The potential acceptors were then recovered with DEAE-Sephadex, and the labeling of ³H on the acceptors was counted by liquid scintillation. The average results of three independent experiments are shown in Fig. 10. Although the incorporation of ³H was low, GTP could be methylated by Ado[methyl-³H]Met in a statistically significant way. GIDP, a nonhydrolyzable analog of GTP, was a better acceptor than GTP by a factor of approximately fourfold under the reaction conditions described in Materials and Methods. The methylation of GMP, m⁷GMP, and dGTP was insignificant.

FIG. 10.

Methyltransferase activity of the B3 protein. Various potential methyl acceptors (10 mM) were incubated with B3 protein and Ado[methyl-³H]Met. The methylated acceptor was recovered by a DEAE-Sephadex column and counted by liquid scintillation. CL, acceptor was omitted from the reaction mixture; S2, proteins from the membrane fraction of yeast cells harboring plasmid pYES2. The results are averages from three independent experiments.

DISCUSSION

ORF1 of potexviruses encodes a polypeptide whose size is larger than 150 kDa. This viral protein has long been proposed to be a replicase with polymerase activity residing at the C terminus and methyltransferase activity residing at the N terminus. The association of RNA-dependent RNA polymerase activity with the C-terminal domain has been demonstrated previously (16). In this study, the full-length 155-kDa protein of BaMV and two C-terminally truncated proteins were expressed in S. cerevisiae. The activity assay and immunoprecipitation indicate that both the full-length and the truncated viral proteins have guanylyltransferase activity. The methyltransferase assay also showed that the N-terminal portion has methyltransferase activity that catalyzes the transfer of a methyl group from AdoMet to GTP or GIDP. Therefore, our present study proves that the N terminus of the 155-kDa protein, from amino acid 1 to 442, harbors both guanylyltransferase and methyltransferase activities. Among the three recombinant proteins, B3 showed the strongest guanylyltransferase activity (Fig. 5). It is possible that the helicase domains in B2 and B1 hydrolyze GTP and, consequently, reduce the activity of transguanylation. The N-terminal portion of the BaMV replicase was also expressed in E. coli cells with and without an S tag, a 15-amino-acid peptide specifically interacting with S protein derived from pancreatic RNase A, fused at the N terminus. No activity was found in either of the E. coli-expressed proteins (data not shown). A problem of viral protein folding in E. coli or interference caused by the S tag might be the cause for enzyme inactivity.

Most replicase activities of the alphavirus-like superfamily, including alphaviruses and various plant viruses, were isolated from membrane fractions (11, 15, 23, 25). For Semliki Forest virus, membrane attachment has been demonstrated to be through the interaction of nsP1 (3). Moreover, the capping activity of nsP1 requires the association with anionic membrane phospholipids. The RNA-dependent RNA polymerase activity capable of synthesizing the BaMV genomic and subgenomic RNA transcripts was also identified in the membrane fraction from the BaMV-infected plant cell extract (C. H. Tsai, unpublished data). The B3 protein expressed in yeast was found mainly in the P18 fraction, whereas B1 and B2 were found in both the P18 and S18 fractions. Fractionation of viral proteins by sucrose discontinuous-gradient centrifugation supports the association of B3 protein with the membrane. Similar to what was found for nsP1 of Semliki Forest virus, membrane association could be crucial for BaMV capping activity since activities were found in the P18 fractions but not in the S18 fractions. BaMV capping activity may require a membrane environment to resume a correct protein conformation or orientation. The competition of yeast proteins in S18 for the utilization of GTP and AdoMet should offer another explanation. BaMV proteins need to be purified from the S18 fraction to further address the necessity of membrane association for transguanylation activity.

In general, the capping reaction at the 5′ end of eukaryotic mRNA is accomplished by three consecutive enzymatic reactions. They involve RNA triphosphatase, mRNA guanylyltransferase, and RNA (guanine-7-) methyltransferase. The order of these three catalytic reactions in alphaviruses and brome mosaic virus varies in that methylation of GTP occurs prior to transguanylation (1, 4, 12). It was therefore interesting to know whether potexviruses have similar characteristics with respect to the capping reactions. In the present study, we found that the guanylyltransferase activity of BaMV is AdoMet dependent and that m⁷GMP is the moiety linked to the enzyme. The evidence provided in this study supports the conservation of the capping reaction throughout members within the alphavirus-like superfamily regardless of sequence dissimilarity among these capping enzymes. The N terminus of the BaMV replicase also exhibited methyltransferase activity. GIDP is a better methyl acceptor than GTP, and the preference was also found in nsP1 of Sindbis virus (20). For the 1a protein of brome mosaic virus, the formation of the adduct of m⁷GMP and 1a protein inhibited its own methyltransferase activity (4). It is possible that a similar phenomenon occurred in the BaMV capping enzyme when GTP was used as the methyl acceptor. GIDP is nonhydrolyzable; therefore it cannot link covalently to the B3 protein, and this may presumably render GIDP a better methyl acceptor. dGTP, a very good methyl acceptor in the case of the 1a protein, was a poor acceptor in this study. Despite the conservation of the main characteristics of the capping reaction, other subtle properties such as substrate specificity may differ between members of the alphavirus-like superfamily. Indeed, the overall sequence similarity between the first 450 amino acids of the BaMV replicase and the first half of 1a of brome mosaic virus is no more than 10% based on the Clustal method.

Notwithstanding these distant relationships, several amino acid residues were found to be conservative among the capping enzymes of members within the alphavirus-like superfamily (24). The alignment of partial amino acid sequences of the capping enzymes from alphaviruses, brome mosaic virus, and BaMV is shown in Fig. 1. The importance of the conserved His residue for guanylyltransferase activity has been established by mutation analyses for Sindbis virus (30), Semliki Forest virus (2), and brome mosaic virus (4). In contrast, its role in methyltransferase activity is controversial in that the activity decreased in Sindbis virus and brome mosaic virus, but actually increased in Semliki Forest virus, when the His residue was mutated. Mutation of the conserved Arg residue resulted in loss of methyltransferase activity and viral infectivity in Sindbis virus (30) and concomitant loss of guanylyltransferase and methyltransferase activities in Semliki Forest virus (2) and brome mosaic virus (4). Despite the above efforts, the in-depth function of these conserved residues remains obscure and requires extensive study. For the capping enzyme of yeast (5) and some well-studied viruses such as vaccinia virus (22) and Chlorella virus (7, 8), the GMP moiety is attached to the enzyme via a lysine residue within a conserved active site motif (KXDG). No such conserved motif was found in the capping enzymes in the alphavirus-like superfamily. Based on a mutational result, Ahola and Ahlquist inferred that the conserved His residue links to m⁷GMP (4). Further physical evidence such as peptide mapping and X-ray structure certainly would help to resolve these uncertainties.