Cell, vol. 63, 363-366,

October

19, 1990, Copyright

0 1990 by Cell Press

Complete Replication of a Eukaryotic In Vitro by a Purified RNA-Dependent RNA Polymerase Robert J. Hayes and Kenneth W. Buck Department of Biology Imperial College of Science, Technology and Medicine London SW7 288 England

A soluble RNA-dependent RNA polymerase was isolated from Nlcotiana tabacum plants infected with cucumber mosaic virus (CMV), which has a genome of three positive-strand RNA components, 1, 2, and 3. The purlfied polymerase contained two virus-encoded polypeptides and one host polypeptide. Polymerase activity was completely dependent on addition of CMV RNA as template, and the products of reactlon were single-stranded (sa) RNA and double-stranded (ds) RNA, corresponding to RNAs 1, 2, and 3, and a subgenomic RNA (RNA 4) derived from RNA 3. The ratio of ssRNA to dsRNA was about 5:1, and the ssRNA was shown to be predominantly the positive strand. This demonstrates the complete replication of a eukaryotic virus RNA in vitro by a template-dependent RNA polymerase. Introduction Many important human and animal viruses, most plant viruses, and some bacteriophages have genomes of positive-strand (messenger-sense) RNA (Matthews, 1982). Examples include poliovirus, foot and mouth disease virus, tobacco mosaic virus, cucumber mosaic virus (CMV), and bacteriophages MS2 and 08. Replication of virus RNA is catalyzed by an RNA-dependent RNA polymerase (RdRp) and takes place in two stages: synthesis of a complementary (negative-strand) RNA using the virus genomic RNA as a template, and synthesis of progeny virus genomic RNA using the negative-strand RNA as a template. The best-characterized RdRp is that of bacteriophage Qp, which has been purified to homogeneity and shown to consist of one virus-encoded polypeptide and three host polypeptides (Blumenthal and Carmichael, 1979). This enzyme has a specific template requirement for 08 RNA and catalyzes both stages of the replication process. A number of template-dependent RdRp preparations have been obtained from cells infected by eukaryotic viruses, i.e., poliovirus (Van Dyke and Flanegan, 1980; Baron and Baltimore, 1982; Hey et al., 1987; Lubinski et al., 1987), black beetle virus (Saunders and Kaesberg, 1985) turnip yellow mosaic virus (Mouches et al., 1975) brome mosaic virus (Miller and Hall, 1983; Quadt et al., 1988) cowpea chlorotic mottle virus (Miller and Hall, 1984), and alfalfa mosaic virus (Houwing and Jaspars, 1988). However, none of these enzyme preparations was able to catalyze complete replication of its RNA template. In the present paper we describe the purification and

Virus RNA

properties of an RdRp isolated from plants infected by CMV. The genome of CMV is divided among three RNAs designated RNA 1 (3.4 kb), RNA 2 (3.0 kb), and RNA 3 (2.1 kb) (Gould and Symons, 1982; Rezaian et al., 1984, 1985). These RNAs serve as messenger RNAs for nonstructural proteins la, 2a, and 3a, respectively. RNA 3 also encodes the virus coat protein that is translated from a subgenomic RNA designated RNA 4 (1.0 kb). RNA 1, 2, and 3 are required to infect plants systemically, and the function of protein 3a is believed to be the potentiation of movement of virus particles from cell to cell. RNA 1 and RNA 2 together, but not separately, have been shown to induce the synthesis of a membrane-bound RNA polymerase in tobacco protoplasts (Nitta et al., 1988), suggesting a role for proteins la and 2a in RNA replication. However, attempts to produce a soluble, template-free CMV polymerase or to demonstrate the presence of proteins la and 2a in RNA containing membrane-bound polymerase preparations have hitherto been unsuccessful (Gordon et al., 1982; Jaspars et al., 1985). The CMV RdRp described here is soluble, completely dependent on addition of CMV RNA as a template, and contains proteins la and 2a, as well as a host polypeptide. Furthermore, the enzyme catalyzes the synthesis of both stages of the replication process, i.e., the synthesis of positive-strand RNA, as well as negative-strand RNA. This demonstrates the complete replication of a eukaryotic virus RNA by a purified template-dependent RNA polymerase. Results Isolation and Purification of CMV RdRp A membrane-bound polymerase (fraction 1) was obtained by differential centrifugation of extracts of CMV-infected tobacco leaves. Treatment with a nonionic detergent produced a soluble polymerase (fraction 2), from which endogenous RNA was removed by nuclease digestion to give fraction 3. More highly purified preparations were obtained by chromatography on an anion-exchange column (fraction 4) followed by separation on the basis of molecular size (fraction 5), and finally fractionation on a high resolution anion-exchange column (fraction 8). The specific activity of fraction 8 was 780,000 times that of fraction 1 (Table 1). RNA polymerase activities of fractions 4, 5, and 8 were completely dependent on addition of CMV RNA. No activity was observed in the absence of added RNA or upon addition of equivalent amounts of RNA from viruses in different taxonomic groups (Matthews, 1982) namely, tobacco mosaic virus, tomato bushy stunt virus, or red clover necrotic mosaic virus. Hence, the polymerase had a specific template requirement. Complete Replication of CMV RNA by the Purified RdRp The products of reaction programmed by CMV positivestrand RNAs were analyzed by gel electrophoresis (Fig-

Cell 364

Table

1. Activities

of RNA Polymerase

Fraction

Total Activitya (cpm x 10m7)

Total Protein

1 2 3 4 5 6

0.5 1.8 2.2 3.2 15.2 13.4

2580 168 184 14 2.4 0.08

a

Fractions

(mg)

Specific Activity (cpmlmg 21 1.1 1.2 2.3 6.3 1.7

x x x x x x

1

103 10s 105 10” 10’ 109

l.O>

ure 1). Fraction 4 gave rise to full-length double-stranded (ds) RNA corresponding to RNAs 1 to 4 (lanes 1 to 5). Fraction 5 gave additionally small amounts of RNA components, which were shown to be single-stranded (ss) by their susceptibility to ribonuclease A under conditions in which the dsRNA components were resistant (not shown). With fraction 6 the ssRNA components became the predominant product, the ratio of nucleotide incorporation into ssRNA and dsRNA being about 5:l (lanes 6 to 6). The ssRNA components comigrated with CMV RNAs 1 to 4 (not shown), demonstrating that they are full-length products. It is noteworthy that when RNA 3 was used as a template, both RNA 3 and the subgenomic RNA 4 were synthesized. To show that the ssRNA products were .evenly labeled along their length and to determine whether they were positive or negative strands, the ssRNA 1 product was extracted from a gel and hybridized with two oligonucleo-

12345676

-4

Agarose

-d-Fr.6-c

Gel Electrophoresis

of Products

5

0.8, 40.6

Pl

of

Lanes 1 to 5, products of RdRp fraction 4 reactions; lanes 6 to 8, products of RdRp fraction 8 reactions. Reactions were programmed with: lane 1, CMV RNA; lane 2, no added RNA; lanes 3 and 6, RNA 3; lanes 4 and 7, RNA 2; lanes 5 and 8, RNA 1. Double arrowheads indicate dsRNA bands. Single arrowheads indicate ssRNA bands. Bands were detected by autoradiography.

P2

u

(+I

Figure 1. Nondenaturing RdRp Reactions

4

41.6

b

Fr.4

3

3.3 k,

protein)

a RNA polymerase activity was assayed as described in the Experimental Procedures. The [32P]UMP incorporated (cpm) was multiplied by the ratio of the volume of each fraction obtained from 200 g of CMV-infected tobacco leaves to the volume of each fraction used for the assay.

-

2

u w

1.0

1.5

0.9 Figure

2. Analysis

1.8 of ssRNA

1 Synthesized

0.8

0.6 in an RdRp

Reaction

(A) The band corresponding to ssRNA 1 (Figure 1) was gel extracted, heated to 100°C for 30 s, annealed to oligonucleotides, and treated with ribonuclease H. The products were then electrophoresed through 1.2% agarose-formaldehyde gels and detected by autoradiography. Lane 1, ssRNA 1 reaction product; lane 2, ssRNA reaction products + ribonuclease H; lane 3, ssRNA 1 reaction product + Pl + P2 + ribonuclease H; lane 4, ssRNA 1 reaction product + P3 + P4 + ribonuclease H; lane 5, ssRNA 1 reaction product + Pl + P2 + P3 + P4 + ribonuclease H. Sizes in kilobases are shown on the side of the gel. The genomic RNA is indicated by a double arrowhead and ribonuclease H products by single arrowheads. (B) Diagram showing the expected products from ribonuclease H digestion of positive-strand RNA 1 annealed to Pl and P2 or negativestrand RNA 1 annealed to P3 and P4. The sequences of the oligonucleotides are: Pl, 5’-ATAGGTCATACCATTG-Y(nucleotides 988-1003); P2, 5’-GACAGCATGAAGTTTC-3’ (nucleotides 2488-2503); P3, B’GTCTTATGTTCACGAT9’(nucleotides 928-948); P4, 5’-AAGTGAGGAAGTCTGT-3’ (nucleotides 2717-2733).

tides, Pl and P2, with sequences complementary to internal sequences in RNA 1 (Figure 2b). Digestion with ribonuclease H hydrolyzed the RNA specifically at the sites of hybridization with Pl and P2 to produce fragments of the calculated size (1.0 kb, 1.5 kb, and 0.6 kb) (Figure 2a, lane 3) with nucleotide incorporation in proportion to their length. Hence, the ssRNA 1 reaction product consisted mainly of positive-strand RNA uniformly labeled along its length. The small amount of RNA undigested by ribonuclease H (Figure 2a, lane 3) was shown to be the RNA 1 negative

In Vitro Replication

of Eukaryotic

Virus

RNA

365

12345 205C ‘1 9 :

67

6

9

10

11

12

13

14

15

16

17

16

19

20

21

22

23

-

-.-

66,

-

SDS-PAGE

-Anti

Figure

3. Gel Electrophoretic

Analysis

of Proteins

WESTERN - Pl

of RdRp

SLOT Anti

-c- P2

-

ALPA Antl-Pl

Anti-P2

Fractions

Pmteins were electrophoresed in SDS-polyacrytamide gels and detected by silver staining (lanes 1 to 7) by Western blotting and probing with antiserum to protein la (lanes 6 to 11) or antiserum to protein 2a (lanes 12 to 15) or by blotting followed by antibody-linked polymerase assay (ALPA) with antiserum to protein la (lanes 16 to 19) or with antiserum to protein 2a (lanes 20 to 23). Lanes 1, 6, 12, 16, and 20, protein 2a expressed in E. coli; lanes 2, 9, 13, 17, and 21, protein la expressed in E. coli; lanes 3, 10, 14, 16, and 22, fraction 4 from healthy plant; lanes 4, 11, 15, 19, and 23, RdRp fraction 4 from CMV-infected plant; lane 5, RdRp fraction 5 from CMV-infected plant; lane 6, RdRp fraction 6 from (XV-infected plant; lane 7, inactive fraction 6 from CMV-infected plant. The M, of marker proteins are shown on the side of the gel: myosin (205 Kd), Bgalactosidase (116 Kd), phosphorylase 6 (97 Kd), bovine serum albumin (66 Kd), ovalbumin (45 Kd). The two faint bands in the M, range of 55 Kd to 60 Kd detected in all the silver-stained lanes (lanes 1 to 7) were also detected when only the buffer used to prepare the samples was applied to the gel (not shown); bands in this range are a commonly encountered artifact in silver-stained gels and have been attributed to traces of keratin-type proteins in the reagents used to prepare the samples for protein gel analysis (Ochs, 1963).

strand. In similar experiments using oligodeoxynucleotides P3 and P4 with sequences complementary to internal sequences of the RNA 1 negative strand (Figure 2b) and ribonuclease H digestion, most of the RNA (i.e., the positive strand) remained undigested, but products of the size (0.9 kb, 1.8 kb, and 0.6 kb) calculated for specific cleavage of RNA 1 negative strand at the sites of hybridization with P3 and P4 were formed (Figure 2a, lane 4). After hybridization with all four oligodeoxynucleotides, the ssRNA 1 product was completely cleaved by ribonuclease H, giving the expected mixture of the cleavage products of the positive and negative strands (Figure 2a, lane 5). The ratio of positive strand to negative strand was about 71. Similar results were obtained with the ssRNA 2 and ssRNA 3 reaction products (not shown). The results show conclusively that the most highly purified RdRp preparation (fraction 6) catalyzed the complete replication of CMV genomic RNA.

Virus-Encoded in Purified

RdRp

and Host Fractions

Polypeptides

To determine if the polymerase preparations contained proteins la and 2a, antibodies were raised against each of these proteins produced in Escherichia co/i from an expression vector. In Western blots (Figure 3, lanes 8 to 15) the la antibodies reacted specifically with the la protein expressed in E. coli (lane 9) and with a protein of the same electrophoretic mobility in RdRp fraction 4 (lane 11). Similarly, the 2a antibodies reacted specifically with the 2a protein expressed in E. coli (lane 12) and with a protein of the same mobility in RdRp fraction 4 (lane 15). The apparent M, of the proteins that reacted with the protein la or 2a antibodies were 98 K and 110 K, respectively, values similar to those obtained for the in vitro translation products of RNA 1 and RNA 2, respectively (Gordon et al.,

1982). Similar results were obtained with RdRp fraction 6. Hence, the purified RNA polymerase contained proteins la and 2a. Evidence that proteins la and 2a were both subunits of the RdRp, and had not just fortuitously copurified with it, was obtained from antibody-linked polymerase assays (Van der Meer et al., 1983; Candresse et al., 1986) (Figure 3, lanes 16 to 23). Proteins la and 2a produced in E. coli and RdRp fraction 4 were subjected to electrophoresis, blotted onto membranes, and incubated with la or 2a antibodies. The membranes, containing antibodies bound to proteins la or 2a, were then incubated with a solution of purified RdRp to allow the RdRp to bind to the second antigen binding site on the IgG antibody molecules. After further incubation with an RNA polymerase reaction mixture, the labeled products were detected by autoradiography. Sands in the positions of proteins la or 2a were detected using la antibodies (lanes 17 and 19) or 2a antibodies (lanes 20 and 23) respectively, but not with preimmune serum (not shown). Therefore, proteins la and 2a are both components of the RdRp. This conclusion was supported by the observation that la or 2a antibodies, but not preimmune serum, partially (about 40%) inhibited the activity of RdRp preparations. Comparison of the proteins of RdRp fractions 4, 5, and 6 by gel electrophoresis (Figure 3, lanes 4, 5, and 6) showed that fractions 4 and 5 contained a number of host proteins in addition to proteins la and 2a. However, fraction 6 contained only one major host protein of apparent M, of 50 K. Sometimes fraction 6 lost the ability to synthesize ssRNA but was still able to synthesize dsRNA with no apparent change in its protein composition. On a number of other occasions, the final stages of RdRp purification led to loss of the 50 Kd host protein (Figure 3, lane 7); such loss was invariably accompanied by complete loss of RdRp activity.

Cell 366

Discussion We have shown that the most highly purified RdRp preparation (fraction 6) catalyzed the complete replication of CMV RNA. Complete replication of viral RNA is likely to require not only polymerase activity to synthesize RNA, but also helicase activity to separate the template and product strands. Proteins la and 2a of CMV, which have been identified in the RdRp, contain sequence motifs characteristic of nucleic acid helicases and polymerases, respectively (Gorbalenya and Koonin, 1969; Habili and Symons, 1969). The presence of free negative-strand RNA in the ssRNA reaction product (Figure 2a, lane 4) confirms the ability of the RdRp preparation to separate the template and product strands. Soluble, template-dependent RdRp preparations have previously been obtained for very few viruses of eukaryotes, the best studied being those of poliovirus (Van Dyke and Flanegan, 1960; Baron and Baltimore, 1982; Hey et al., 1987; Lubinski et al., 1987; Plotch et al., 1989) and brome mosaic virus (Miller and Hall, 1983; Quadt et al., 1988). The inability of RdRp preparations to catalyze complete replication of RNA could be due to the absence of a helicase subunit or inhibition of helicase activity in partially purified preparations. In the case of poliovirus, sequence motifs characteristic of helicases and polymerases lie in proteins 2C and 3D, respectively (Gorbalenya and Koonin, 1989). Soluble, template-dependent preparations of poliovirus RdRp, either isolated from infected cells or produced in E. coli from cDNA clones, contain 3D, but not 2C, a protein known to be required for RNA replication in vivo (Li and Baltimore, 1988). Hence, the absence of a helicase subunit could be a factor in the inability of poliovirus RdRp preparations to catalyze complete RNA replication. In the case of brome mosaic virus, which has a similar genome organization to CMV, brome mosaic virus-encoded proteins la and 2a were detected in the RdRp preparations (Quadt et al., 1988). Hence, the helicase subunit was present, and inhibition of helicase activity could explain the ability of these enzyme preparations to produce only dsRNA. Inhibition of helicase activity probably also explains the ability of partially purified CMV RdRp (fraction 4) to synthesize only dsRNA (Figure 1, lanes 1 to 5) since further purification resulted in ability to synthesize free positive strands and negative strands. The CMV RdRp produced an excess of positive strands over negative strands. The ratio of positive to negative strands in the ssRNA product was 7:l. We have not measured this ratio in the dsRNA product, but even if nucleotide incorporation into dsRNA were completely in the negative strand, the overall ratio of positive to negative strands synthesized would be about 2.5 to 1. Since an excess of positive-strand templates was always present, the polymerase must utilize the negative-strand template preferentially, and clearly each negative-strand template was copied more than once. The reaction products using RNA 3 as a template included the subgenomic RNA 4, as well as full-length RNA 3 (Figure 1). This is consistent with data on brome mosaic

virus that RNA 4 is transcribed from a subgenomic promoter on the negative strand of RNA 3 upstream of the RNA 4 start site (Miller et al., 1985; Marsh et al., 1988). The purified CMV RdRp contained a host polypeptide, apparent M, of 50 K, in addition to the two virus-encoded polypeptides. This protein is apparently bound to the RdRp, because it could not be detected in material corresponding to fraction 6 from healthy plants. Furthermore, on those occasions when host polypeptide was lost in the final stage of RdRp purification, polymerase activity was concomitantly lost, suggesting that the host polypeptide is essential for activity. Viruses with small genomes often utilize host components for their replication. For example, the RdRp of bacteriophage Q6 consists of one virus-encoded and three host-encoded subunits, and an additional host protein is needed for negative-strand synthesis on a positive-strand template (reviewed by Blumenthal and Carmichael, 1979). A host protein has been shown to enable initiation of poliovirus negative-strand synthesis in vitro (Dasgupta, 1983; Hey et al., 1987). A highly purified preparation of an RdRp from plants infected by turnip yellow mosaic virus was shown to contain one virus-encoded polypeptide and one host polypeptide, although it was not established whether the host polypeptide was needed for activity, and the reaction products were not characterized (Candresse et al., 1986). A host RdRp is induced when tobacco is infected by CMV or other viruses (Fraenkel-Conrat, 1986). This enzyme consists of a single polypeptide, M, of about 130 K, has no template specificity, and synthesizes only short chains. It is clearly distinct from the RdRp described here, which contains two virus-encoded polypeptides and a host polypeptide, M, of about 50 K, is specific for CMV RNA, and catalyzes the complete replication of CMV RNA. The availability of a system for complete replication of a eukaryotic virus RNA in vitro opens up the way for investigations of the mechanism of RNA replication and the roles of the virus- and host-encoded subunits, which have not been possible hitherto. Such studies could lead to novel antiviral agents designed to inhibit different stages of the replication process. Furthermore, the insights gained in the purification of the CMV replicase could lead to development of in vitro systems for the replication of other animal and plant RNA viruses. Experimental

Procedures

Puriticatlon and Assay of CMV RdRp Leaves of young Nicotiana tabacum cv. Samsun plants were inoculated with CMV (Q strain), and 3 days later the infected leaves (200 g) were homogenized in TMDPG buffer (50 mM Tris-HCI [pH 8.21, 15 mM MgCls, 10 mM dithiothreitol [DTT], 0.1 mM phenylmethylsulphonyl fluoride, 18% glycerol) (4 ml buffer per gram of leaves) at 4OC. The homogenate was centrifuged at 3,000 x g for 30 min at 4OC. The supernatant was then centrifuged at 35,000 x g for 30 min at 4OC. The pellet was resuspended in TMDPG buffer (0.5 ml per gram of leaves) to give fraction 1. After addition of NP40 (final concentration, 0.75%) the mixture was stirred at 4OC for 1 hr and then centrifuged at 36,000 x g for 30 min. The supernatant was recentrifuged at 100,006 x g for 1 hr. The final supernatant (fraction 2) was treated with micrococcal nuclease (Miller and Hall, 1983) to yield fraction 3. Fraction 3 was applied to a DEAE-Biogel column (1 x 10 cm) at a flow rate of 1 mllmin. Unbound proteins were washed through with 30 ml of TMDPGN buffer (TMDPG

In Vitro Replication 367

of Eukaryotic

Virus

RNA

containing 0.75% NP40), and the RdRp activity was eluted with TMDPGN buffer containing 0.5 M KCI. Fractions of 1 ml were collected and assayed for RdRp activity. The pooled active fractions were termed fraction 4. Fraction 4 was passed through a Pharmacia fast protein liquid chromatography (FPLC) Superose 6 column (30 x 1 cm) at a rate of 0.25 mllmin with TMDPGN buffer containing 0.5 M KCI, and fractions of 1 ml were collected. Fractions containing RdRp activity were pooled (fraction 5). dialyzed against TMDPGN buffer, and then applied to a Pharmacia FPLC Mono 0 column (5 x 0.5 cm) at a flow rate of 1 mllmin. A linear gradient of 0 to 05 M KCI in TMPDGN buffer was then applied. Fractions of 1 ml were collected, and those with RdRp activity were pooled (fraction 6). Fraction 3 could be stored at -70°C for at least 1 month and thawed and refrozen several times without significant loss of RdRp activity. Fraction 4 could also be stored at -704C. but after thawing could not be refrozen without considerable loss of activity. The activities of fractions 5 and 6 were completely lost after freezing and thawing. These fractions were stored unfrozen at OOC and generally used within 1 hr of preparation. RdRp activity was assayed by mixing 25 ul of each fraction with 25 ul of 100 mM Tris-HCI (pH 6.2) containing 4% v/v glycerol, 20 mM MgCl*, 2 mM ATR 2 mM CTP, 2 mM GTR and 20 mM DTT [32P]UTP (10 mCi/ml; 1 rJ) and template RNA (IO pg) were then added, and the reaction mixture was incubated at 30°C for 5 min. After addition of 50 mM UTP (1 ut), incubation was continued for a further 25 min. Incorporation of the label into RNA was determined by spotting 5 ul aliquots onto DE52 discs (Whatman) (Sambrook et al., 1969). Protein concentrations were determined using a Bio-Rad kit according to the manufacturer’s instructions with bovine serum albumin as standard. RdRp reaction products were extracted with phenol-chloroform and precipitated by addition of 2.5 vol of ethanol in the presence of 2 M ammonium acetate. Vlrus RNA CMV RNA was extracted from purified CMV particles (Lot et al., 1974). RNA 1, RNA 2, and RNA 3 were synthesized by transcription in vitro using T7 RNA polymerase and vectors containing full-length clones of CMV RNA 1, RNA 2, and RNA 3 (pCMV1, pCMV2, and pCMV3, respectively). The capped transcripts initiated precisely at the 5’ termini of CMV RNAs. but contained up to 4 additional residues at the 3’termini. They have been shown to be infectious when inoculated together onto tobacco plants. Details of the construction of pCMV1, pCMV2, and pCMV3, in vitro transcription, and infectivity of transcripts will be published elsewhere (Hayes and Buck, 1990). Analysis of Nucleic Acldr RNA was electrophoresed through nondenaturing or denaturing (formaldehyde) agarose gels as described by Sambrook et al. (1969). Digestion with ribonuclease A to distinguish between ssRNA and dsRNA was by the method of Buck et al. (iQ7l). Annealing of oligonucleotides to ssRNA and digestion with ribonuclease H was as described by Hayes et al. (1966). Analyals of Proteins Proteins were electrophoresed in SDS-polyacrylamide gels (Laemmli. 1910) and detected by silver staining (Ochs, 1963) by Western blotting (Sherwood, 1967) or by blotting followed by antibody-linked polymerase assay (Van der Meer et al., 1963; Candresse et al., 1966). Ploductlon of ProBIns la and Pa Ndel sites were introduced into pCMV1 and pCMV2 (Figure 1) by in vitro mutagenesis (Kunkel et al., 1967) using oligonucleotides TAAAATTCATATGGCAACGTCCTC and CTTCTGTCATATGATAAGTCC, respectively. The complete la and 2a coding regions were then cut out with Ndel and BamHl and cloned into expression vector pET3a in E. coli BL21 cells (Rosenberg et al., 1967). Expression and purification of proteins la and 2a were essentially as described (Plotch et al., 1969). Electrophoretically homogeneous proteins were used to raise antisera in rabbits. Acknowladgmants We thank BP Nutrition for financial support. The costs of publication of this article were defrayed

in part by the

payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 16 U.S.C. Section 1734 solely to indicate this fact. Received

March

23, 1990; revised

August

7, 1990.

Refarancss Baron, M. H., and Baltimore, D. (1962). In vitro copying strand RNA by poliovirus replicase. Characterisation and its products. J. Biol. Chem. 257, 12359-12366. Blumenthal, T., and Carmichael, tion and structure of Q5-replicase.

of viral positive of the reaction

G. G. (1979). RNA replication: funcAnnu. Rev. Biochem. 48, 525-546.

Buck, K. W., Chain, E. B., and Himmelweit, F. (197t). Comparison of interferon induction in mice by purified Psnicillium chrysogenum virus and derived double-stranded RNA. J. Gen. Virol. 12, 131-139. Candresse, T, Mouches, C., and Bove, J. M. (1966). Characterisation of the virus-encoded subunit of turnip yellow mosaic virus RNA replicase. Virology 152, 322-330. Dasgupta, A. (1963). Purification of host factor required scription of poliovirus RNA. Virology 128, 245-251. Fraenkel-Conrat, H. (1966). RNA-directed CRC Crit. Rev. Plant Sci. 4, 213-226.

for in

RNA polymerases

vitrotranof plants.

Gorbalenya, A. E., and Koonin, E. V (1969). Viral proteins containing the purine NTP-binding sequence pattern. Nucl. Acids Res. 77, 64136440. Gordon, K. H. J., Gill, D. S., and Symons, R. H. (1962). Highly purified cucumber mosaic virus-induced RNA-dependent RNA polymerase does not contain any of the full length translation products of the genomic RNAs. Virology 123, 264-265. Gould, A. R., and Symons, R. H. (1962). Cucumber mosaic virus RNA 3. Determination of the nucleotide sequence provides the amino acid sequences of protein 3A and viral coat protein. Eur. J. Biochem. 726, 217-226. Habili, N., and Symons, R. H. (1969). Evolutionary relationship between luteoviruses and other RNA plant viruses based on sequence motifs in their putative RNA polymerases and nucleic acid helicases. Nucl. Acids Res. 17, 9543-9555. Hayes, R. J., and Buck, K. W. (1990). Infectious cucumber mosaic virus RNA transcribed in vitro from clones obtained from cDNA amplified using the polymerase chain reaction. J. Gen. Virol. 77, in press. Hayes, R. J., Brunt, A. A., and Buck, K. W. (1966). Gene mapping and expression of tomato bushy Stunt virus. J. Gen. Virol. 69, 3047-3057. Hey, T. D., Richards, 0. C., and Ehrenfeld, E. (1967). Host factorinduced template modification during synthesis of poliovirus RNA in vitro. J. Virol. 67. 602-611. Houwing, C. J., and Jaspars, in vitro transcription of virion 209, 264-266.

E. M. J. (1966). Coat protein blocks the RNAs of alfalfa mosaic virus. FEBS Lett.

Jaspars, E. M. J., Gill, D. S., and Symons, R. H. (1965). Viral RNA synthesis by a particulate fraction from cucumber seedlings infected with cucumber mosaic virus. Virology 744, 410-425. Kunkel, T A., Roberts, J. D., and Zakour, R. A. (1967). Rapid and efficient site-specific mutagenesis without phenotypic selection. Meth. Enzymol. 754, 367-362. Laemmli, U. K. (1970). Cleavage of structural proteins during of the head of bacteriophage T4. Nature 227, 660-665. Li, J., and Baltimore, D. (1966). Isolation of poliovirus tive in viral RNA synthesis. J. Virol. 62, 4016-4021.

assembly

2C mutantsdefec-

Lot, H., Harchoux, G., Marrou, J., Kaper, J. M., West, C. K., Van VlotenDoting, L., and Hull, R. (1974). Evidence for three functional RNA species in several strains of cucumber mosaic virus. J. Gen. Virol. 22, 61-93. Lubinski, J. M., Ransone, L. J., and Dasgupta, dependent synthesis of covalently linked dimeric poliovirus replicase. J. Virol. 67, 2997-3003.

A. (1967). PrimerRNA molecules by

Marsh, L. E., Dreher, T. W., and Hall, T C. (1966). Mutational analysis of the core and modulator sequences of the BMV RNA 3 subgenomic promoter. Nucl. Acids Res. 16, 961-995.

Cell 368

Matthews, R. E. F. (1982). Classification Intervirology 77 1-199.

and nomenclature

of vrruses.

Miller, W. A., and Hall, T. C. (1983). Use of micrococcal nuclease in the purification of highly template-dependent RNA-dependent RNA polymerase from brome mosaic virus-infected barley. Virology 125. 236241. Miller, W. A., and Hall, T. C. (1984). RNA-dependent isolated from cowpea chlorotic mottle virus-infected cific for bromoviral RNA. Virology 732, 53-80.

RNA polymerase cowpeas is spe-

Miller, W. A., Dreher, T., and Hall, T. C. (1985). Synthesis saic virus subgenomic RNA in vitro by internal initiation genomic RNA. Nature 373, 68-70.

of brome moon (-) sense

Mouches, C., Bove, C., Barreau, C., and Bove, J. M. (1975). TYMV RNA-replicase: formation of a complex between the purified enzyme and TYMV-RNA. Ann. Microbial. (inst. Pasteur) 727A, 75-90. Nitta. N., Takanami, Y., Kuwata, S., and Kubo, S. (1988). Inoculation with RNAs 1 and 2 of cucumber mosaic virus induces viral RNA replicase activity in tobacco mesophyll protoplasts. J. Gen. Virol. 69. 2695-2700. Ochs, D. C. (1983). Protein contaminants of sodium dodecyl polyacrylamide gels. Anal. Biochem. 135, 470-474.

sulphate-

Plotch, S. J., Palant. O., and Gluzman. Y. (1989). Purification and prop erties of poliovirus RNA polymerase expressed in Escherichia co/i. J. Viral. 63, 216-225. Cluadt, R., Verbeck, l-l. J. M., and Jaspars, E. M. J. (1988). Involvement of a nonstructural protein in the RNA synthesis of brome mosaic virus. Virology 765, 256-261. Rezaian, M. A., Williams, R. H. V., Gordon, K. H. J., Gould, A. R., and Symons, R. H. (1984). Nucleotide sequence of cucumber mosaic virus RNA 2 reveals a translation product significantly homologous to corresponding proteins of other viruses. Eur. J. Biochem. 743, 277-284. Rezaian, M. A., Williams, R. H. V., and Symons, R. H. (1985). Nucleotide sequence of cucumber mosaic virus RNA 1. Presence of a sequence complementary to part of the viral satellite RNA and homologies with other viral RNAs. Eur. J. Biochem. 750, 331-339. Rosenberg, A. H., Lade, B. N., Chui, D., Lin, S., Dunn, J. J., and Studier, F. W. (1987). Vectors for selective expression of cloned DNAs by T7 RNA polymerase. Gene 56, 125-135. Sambrook, J., Fritsch, E. F., and Maniatis. T. (1989). Molecular Cloning: A Laboratory Manual (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press). Saunders, K., and Kaesberg, merase from black beetle cells. Virology 744, 373-381.

P (1985). Template-dependent RNA polyvirus-infected Drosophila melanogaster

Sherwood, J. L. (1987). Comparison of a filter paper immunobinding assay, Western blotting and an enzyme-linked immunosorbent assay for the detection of wheat streak mosaic virus. J. Phytopathol. 778, 68-75. Van der Meer, J., Dorssers, L., and Zabel, P (1983). Antibody-linked polymerase assay on protein blots: a novel method for identifying polymerases following SDS-polyacrylamide gel electrophoresis. EMBO J. 2, 233-237. Van Dyke. T. A., and Flanegan, polypeptide P63 as a soluble Virol. 35. 732-740.

J. 8. (1980). ldentificahon of poliovirus RNA-dependent RNA polymerase. J.