Function and Structure of RNA Polymerase from Vesicular Stomatitis Virus*

The RNA-dependent RNA polymerase associated with vesicular stomatitis virus was isolated to apparent homogeneity by a newly developed procedure, which includes stepwise removal of proteins from virions by successive treatme,;. with high concentrations of cesium sulfate and cesium chloride, followed by glycerol gradient centrifugltion or chromatography on phosphocellulose or DEAE-Sephadex column. The polymerase thus purified contained L (large protein) and NS proteins as the intrinsic subunits and multiple species of enzyme were found which differ in the molar ratio of L to NS. Since the enzyme with the highest activity was composed of equimolar amounts of the two subunits and exhibited the sedimentation coefficient of approximately 11 S in a buffer containing 0.2 M NaCl, the structure of active protomer was suggested to be (L) ,(NS),. In accordance with this conclusion, enzyme preparations deficient in tlie content of NS protein, were activated by the addition of NS protein. The purified RNA polymerase catalyzed the synthesis of poly(A), which was covalently attached to the 3’ termini of RNA products, and RNA, only in the presence of all 4 substrates. The present finding might be the first which indicates that the transcriptase itself catalyzes post-transcriptional modification of mRNA by adding poly(A) sequences to the 3’-OH termini. The molecular mechanism of the switch from transcription to poly(A) synthesis, however, remains to be investigated.

The RNA-dependent RNA polymerase associated with vesicular stomatitis virus was isolated to apparent homogeneity by a newly developed procedure, which includes stepwise removal of proteins from virions by successive treatme,;. with high concentrations of cesium sulfate and cesium chloride, followed by glycerol gradient centrifugltion or chromatography on phosphocellulose or DEAE-Sephadex column. The polymerase thus purified contained L (large protein) and NS proteins as the intrinsic subunits and multiple species of enzyme were found which differ in the molar ratio of L to NS. Since the enzyme with the highest activity was composed of equimolar amounts of the two subunits and exhibited the sedimentation coefficient of approximately 11 S in a buffer containing 0.2 M NaCl, the structure of active protomer was suggested to be (L) ,(NS),. In accordance with this conclusion, enzyme preparations deficient in tlie content of NS protein, were activated by the addition of NS protein.
The purified RNA polymerase catalyzed the synthesis of poly (A), which was covalently attached to the 3' termini of RNA products, and RNA, only in the presence of all 4 substrates. The present finding might be the first which indicates that the transcriptase itself catalyzes post-transcriptional modification of mRNA by adding poly (A) sequences to the 3'-OH termini. The molecular mechanism of the switch from transcription to poly (A) synthesis, however, remains to be investigated.
Vesicular stomatitis virus, the prototype of the rhabdovirus group, contains a single strand RNA of 3.6 to 4.0 x lo6 daltons as the viral genome which is complementary to messenger RNA synthesized in virus-infected cells (1,2). Upon infection tQ susceptible cells, VS virus' exhibits two types of RNA metabolism: mRNA transcription, and replication of the viral require the continuous supply of unstable protein(s) synthesized in infected cells (14,15) and the product is 40 S negative-strand RNA which is not attached with poly(A) sequence (16).
genome. It has been established that the synthesis of mRNA is catalyzed by a virion-associated RNA-dependent RNA polymerase (ribonucleoside triphosphate: RNA nucleotidyltransferase (EC 2.7.7.6)), at least during the initial stage of virus multiplication. The products transcribed are monocistronic, and two species of mRNA with the sedimentation coefficients of 12 to 18 S and 28 to 31 S have been identified as the major components (3)(4)(5). Like most eukaryotic mRNA, they possess polyadenylic acid (poly (A)) sequences covalently linked to the 3'-OH termini (6)(7)(8). Since RNA synthesized in uitro in purified virions is attached with poly(A) sequences (9, lo), and since viral RNA contains no poly(U) stretches to code poly (A) (ll-13), an enzyme activity is expected to exist in virions 4307 which catalyzes polymerization of adenylate in the absence of template.
On the other hand, replication of the genome RNA seems to * This work was supported in part by grants from the Ministry of Education of Japan and the Asahi Press.
In order to reveal the molecular mechanism underlying the multiplication of VS virus-RNA as outlined above, we tried to purify the enzyme system which catalyzes transcription and poly(A) synthesis from the New Jersey serotype of VS virus, and to identify the molecular structure as well as the enzymatic properties. It has been well established that VS virions contain at least five separate structural polypeptides: L (large protein; M, = 190,000); G (glycoprotein; M, = 69,000); N (nucleoprotein; M, = 50,000); NS (Mr = 40,000 to 45,000); and M (membrane protein; M, = 29,000) (17). Since a viral subparticle which is composed of the RNA, N, L, and NS proteins and is capable of synthesizing RNA in oitro was isolated (18), these three proteins have been believed to be sufficient for transcription, whereas the two envelope proteins, G and M, are not essential. The dissociation and reconstitution studies by Wagner and his colleagues (19) revealed that the L protein and the N protein.RNA complex were required for transcription.
In this report, we propose that both of the polypeptides L and NS are required for N-RNA complexdirected RNA synthesis and, moreover, the RNA polymerase itself catalyzes post-transcriptional addition of poly (A) to the product RNA. During the preparation of the manuscript, Emerson and Yu (20) and Imblum and Wagner (21)  A preliminary report of this study has been published previously (22). MATERIALS AND was allowed to proceed for 3 hours at room temperature.

Requirements for RNA Polymerase Reaction
In the course of the systematic search for the optimum conditions of RNA synthesis by RNA polymerase in VS virus, a few new features were found: the maximum activity was observed after treatment of virus preparations with 0.1% Nonidet P40 (NP40); the optimum pH for the reaction was 8.3, which was higher than the value (pH 7.3) reported by Huang et al. (26); although Mn'+ iOn at Concentrations over 2 mM was unable to substitute for Mg *+ ion, but rather inhibited the Mg*+-dependent reaction as reported by others (26,27), the substitution with Mn2+ at concentrations below 2 mM permitted a considerable level of RNA synthesis. The maximum activity obtained with 1 mM Mn'+, however, was only one-third of that with 5 mre Mg*+ ion. Such Mn*+-dependent reaction has been demonstrated independently for the in vitro synthesis of RNA by disrupted virions of Newcastle disease virus (28), and ribonucleocapsids isolated from rabies virus-infected cells (29). The reaction was very sensitive to salt (NaCl) concentration with a sharp optimum at 0.07 M, and such salt sensitivity was observed even with the purified RNA polymerase (DEAE-Sephadex fraction). Thus, the concentrations employed in other reports, 0.1 M (30) or 0.15 M NaCl (18), lead to considerable suppression of enzyme activity. Since the polymerase activity was rather reduced at temperatures higher than 30", all the reactions described were performed at 30".

Solubilization of RNA Polymerase from Virions
Treatment of VS virions with nonionic detergent, Triton X-106, in the presence of dithiothreitol at low ionic strength leads to solubilization of viral envelopes. When disrupted virions were centrifuged on a discontinuous gradient of cesium sulfate, solubilized envelope lipids and proteins remained on the top of gradient while the viral core particles sedimented on the cushion of 1 M C&O,.
Among the Cs,SO, fractions thus obtained, RNA polymerase activity was found only in the core from Vesicular Stomatitis Virus 4309 fractions indicating that RNA polymerase protein(s) remained bound in core particles. When the polypeptide composition of the core particles as well as of the original virus preparation was analyzed by polyacrylamide gel electrophoresis in the presence of SDS, the proteins L, N, NS, and a portion of M were found in the core particles ( Fig. 1).
The cores were further dissociated by exposure to a high ionic strength buffer containing 1.5 M.C!SC~ and subjected to centrifugation in a discontinuous gradient of CsCl from 1.75 to 3 M; the dissociated proteins were separated from ribonucleoproteins and recovered in the top fraction. Neither the top fraction nor the ribonucleoprotein fraction alone exhibited RNA polymerase activity when tested separately. However, upon combining the two fractions, the activity was restored quantitatively (Fig. 2). When the ribonucleoprotein template containing 23 pg of N protein was assayed by adding the top fraction, the activity increased linearly at least up to 10.3 pg of the dissociated protein. Polyacrylamide gel electrophoresis revealed that the dissociated protein fraction contained the L, NS, M proteins, and a small amount of N protein, while the majority of N protein remained attached in the ribonucleoprotein (RNP) fraction as a sole component (Fig. 1). A trace amount of L protein often accompanied the RNP fraction which seemed to cause a low but significant level of endogenous RNA polymerase activity.
In the enzyme reconstitution experiment as noted above, no activity was found when viral RNA purified by SDS-phenol extraction method was used as a template in place of the RNP fraction (Table I) suggesting that the N protein is absolutely required for RNA to be transcribed by the enzyme. Neither synthetic polyribonucleotides including poly(A), poly(U) ( Table I), poly(C), and poly(G) (data not shown), nor natural nucleic acids as Escherichia coli ribosomal RNA and T7 DNA exhibited template activity (Table I). Moreover, attempts to replace the N protein by polyamines or histones have also been unsuccessful.

Purification of RNA Polymerase
Glycerol Gradient Centrifugation-In order to identify the polypeptide component(s) necessary for RNA polymerase activity, the proteins dissociated by CsCl were further fractionated by centrifugation through a 15 to 35% glycerol gradient in 0.1 M sodium phosphate buffer (pH 7.0) containing 1 M NaCl, 1 mM dithiothreitol and 0.2% Triton X-100. As shown in Fig. 3, a peak of enzyme activity was found only when assayed by adding the RNP template. The peak fraction of enzyme activ-  (11.4 c(g of protein); (B) core particle (5.9 pg of protein); (C) RNA polymerase (4.7 rg) purified by glycerol gradient centrifugation; and (D) ribonucleoprotein (5.4 clg of protein) purified by two cycles of CsCl centrifugation. Samples were treated with 0.5% SDS and 25 rn~ dithiothreitol for 10 min at 37' and analyzed by polyacrylamide gel (7.5%) electrophoresis in the presence of SDS. Gels were stained with Coomassie brilliant blue and after destaining, scanned with Joyce-Loebl microdensitometer MKIII. Electrophoresis was from right to left. FIG. 2 (center). Cesium chloride centrifugation of dissociated core. Core particles (3.1 mg of protein) were treated with 1.5 M CsCl and centrifuged in a discontinuous gradient of CsCl as described under "Materials and Methods." After centrifugation, RNA polymerase activity was determined with 5 ~1 ea'ch of the fraction (0.. . .O), 5 ~1 each of the fractions supplemented with 5 ~1 of the fraction number 3 or 5 ~1 each of the fractions supplemented with 5 pl of the fraction number 10 (A-.-. A). RNA synthesis was performed for 1 hour at 30". Centrifugation was from right to left. FIG. 3 (right). Glycerol gradient centrifugation of solubilized RNA polymerase. RNA polymerase was solubilized from core particles as described in Fig. 2, and dialyzed to remove CsCl against Buffer V. Two milliliters of the dialyzed enzyme containing 1.76 mg of protein was centrifuged on 58 ml of 15 to 35% glycerol gradient in 0.1 M sodium phosphate buffer (pH 7.0) containing 1 M NaCl, 0.2% Triton X-100,. and 1 mM dithiothreitol in a Spinco SW25.2 rotor for 78 hours at 24,000 rp'm at 4". After centrifugation, RNA polymerase activity (O--O) of 25 ~1 each of the glycerol fractions was determined using ribonucleoprotein containing 23 pg of N protein as the template. protein distribution was analyzed by SDS-polyacrylamide gel electrophoresis of 50 rl each of the glycerol fractions (L protein, A---A; NS protein, 04; M protein, 04).  Fig. 4A, there appeared three separate peaks. The first peak (peak I) contained the L and NS proteins at an approximate molar ratio of 1:0.2 to 0.5. The second peak (peak II) contained the two proteins at a molar ratio of 1:O.g to 1.0, and the third peak (peak III) contained mainly the NS protein.
Since  (04) as well as the molar ratio of NS to L (x . . . . X) were plotted. NaCl concentration (----) of the eluate was determined by measuring conductivity. B, peak lof the eluate was combined (total protein, 199 pg) and rechromatographed on a small column (bed volume, 1 ml) of DEAE-Sephadex A50. C, peak II of the eluate was combined (total protein, 422 pg) and rechromatographed on DEAE-Sephadex column (bed volume, 1 ml). D, peak III of the eluate was combined (total protein, 152 pg) and rechromatographed.
the enzyme purification are eluted in peaks I and III. In fact, rechromatography of each peak fraction on second DEAE-Sephadex columns yielded a main peak, which was eluted at the salt concentration similar to that required for the first chromatography, but in addition shoulders and small peaks of crosscontaminants or dissociated products (Fig. 4 of lower NS content. However, the amounts of NS protein required for maximum stimulation were more than those expected on the basis of stoichiometry of native enzyme (L:NS = l:l), presumably because some of the isolated NS was inactive or the reassociation was incomplete, or both.
The time course, as well as the poly(A) content of product RNA (Table III) of the reaction carried out in the presence of NS protein addition, are indistinguishable from those of the reaction in the absence of added NS. These observations support the conclusion that the NS protein is one of the intrinsic subunits of VS virus transcriptase.

Poly(A) Synthesis by RNA Polymerase
Although purified virions are able to synthesize RNA containing poly (A) which is covalently attached to the 3'-OH terminus of product RNA (8), no poly (A) is synthesized in the presence of ATP as a sole substrate (9,10). With use of disrupted virions, 30 to 35% of radioactivity of the product RNA formed remained acid-insoluble after treatment with   Fig. 4, while NS protein was purified by phosphocellulose column chromatography as described in the text. Stimulation by NS protein (4.9 rg/ml) of RNA synthesis catalyzed by 0.03 ml of peak I enzyme (15.3 pg of L and 1.7 pg of NS protein per ml; NSiL molar ratio = 0.47), or 0.02 ml of peak II enzyme (23.5 pg of L and 5.2 rg of NS protein per ml; NS/L molar ratio = 0.93) was examined in the standard reaction mixture containing urea-treated RNP template (18.4 (22). It has been believed that such poly(A) sequences covalently attached to RNA are synthesized by an enzyme system different from the RNA Polymerase from Vesicular Stomatitis Virus transcriptase, and a poly(A) polymerase was indeed isolated from vaccinia virus (31,32). In contrast, the purified VS virus-RNA polymerase itself exhibited poly(A) synthesizing activity. The synthesis of poly(A) sequences by the purified RNA polymerase also required the presence of all four triphosphates. It is worthwhile to note that the ribonuclease-resistant fraction of product RNA synthesized in the presence of Mn2+ ion was 10 to 15%, which is considerably lower than that of RNA synthesized in the regular reaction mixture containing Mg2+ as a divalent metal cofactor (Fig. 7). The kinetic profile of the appearance of ribonuclease-resistant poly(A) sequence by the purified enzyme was essentially identical to that catalyzed by disrupted virions (data not shown).
Product RNA synthesized by the purified RNA polymerase exhibited a sedimentation coefficient of approximately 16 S as a main component and the ribonuclease-resistant poly(A) was also found in this peak (Fig. SA) indicating that the poly(A) is associated to this high molecular weight RNA. In fact, the poly(A) gave a peak of 4 to 5 S when RNA was centrifuged after treatment with ribonuclease (Fig. 8B). The molecular size and poly(A) content of product RNA synthesized on the urea- RNA polymerase and NS protein were purified by phosphocellulose column chromatography as described in the text. RNA synthesis was performed with 0.05 ml of RNA polymerase (11.4 pg of L and 2.2 pg of NS protein per ml) in the presence or absence of 0.42 Kg of NS protein. After incubation at 30" for 2 hours, the reaction mixtures were divided into two equal portions, one of which was treated with ribonuclease for the determination of poly(A) content. treated RNP template were similar to those formed on the untreated RNP (data not shown).
Location of Poly (A) in Product RNA An attempt was then made to determine the position of the poly(A) sequences in the RNA products synthesized by the purified enzyme system. Banerjee et al. (8) reported experiments which indicated that the poly (A) sequences were present at the 3' end of RNA products synthesized in vitro with disrupted VS virions. To test this possibility, RNA was synthesized by the nurified VS virus-RNA polymerase in the presence of [y-""P]A,'P, to label the 5' terminus. Then it was labeled at the 3' terminus with the periodate oxidationtritiated borohydride reduction technique, and the doubly labeled RNA was isolated by sucrose gradient centrifugation (Fig. 9A). When the sucrose gradient fractions were digested with a mixture of ribonuclease A and Tl, the major portion of 3zP radioactivity (attached at the 5'-terminal nucleotide) was rendered acid-soluble. In contrast, most of the 3H radioactivity (attached to the 3'-terminal nucleotides) remained acid-insoluble (Fig. 9B). From these experiments, it can be concluded that the poly(A) sequences are covalently attached to the 3' end of RNA products.

DISCUSSION
Of the several types of virion-associated nucleic acid polymerases, extensive purification and analysis of enzymatic properties as well as molecular structure have been performed only on the reverse transcriptase (the RNA-dependent DNA polymerase) from oncomaviruses (33,34). In contrast, however, little is known about the DNA-and RNA-dependent RNA polymerases present in wide varieties of viruses including poxviruses, diplomaviruses, myxoviruses, paramyxoviruses, and rhabdoviruses mainly due to the difficulty of obtaining highly purified enzymes in quantities.
Isolation of nucleocapsid cores containing RNA polymerase activity from VS virus was first achieved by either polyethylene glycoi-dextran phase separation or agarose gel chromatography of virions disrupted with a nonionic detergent Triton N-101 (35). Similar ribonucleoprotein, which retained infectivity as well as RNA polymerase activity, was obtained by disruption of virions with Triton N-101 in the presence of 0.5 M CsCl followed by isolation of the ribonucleoprotein by centrifugation in sucrose gradient (36). According to these procedures the two envelope proteins, G and M, were completely solubilized and could be removed from the ribonucleocapsid.
On the other hand, Emerson and Wagner (18) reported that the transcriptase activity was dissociated from virions of the Indiana serotype VS virus by treatment with high ionic strength buffer containing Triton X-100 and was separated from the template ribonucleocapsid by centrifugation through sucrose gradients. Solubilized proteins they obtained, however, contained not only the polymerase but also all the envelope proteins, and it seemed difficult to isolate pure enzyme by subsequent purification employing conventional techniques of enzyme purification (19).
As reported previously (37-39)) equilibrium centrifugation in CsCl or C&SO, is useful for dissociating RNA polymerase from DNA so long as they are not in the process of RNA synthesis. Moreover, it has been established that CsCl is more potent than C&O, in this dissociation since a part of the polymerase, which was bound to DNA regions with weak affinity but were still recovered in the cesium.sulfate complex, was further dissociated from the DNA by exposure to CsCl. FIG. 8 (left). Sucrose gradient centrifugation of product RNA synthesized by purified VS virus-RNA polymerase. RNA synthesis was carried out in a 0.5ml reaction mixture containing 7.2 pg of purified RNA polymerase (glycerol gradient fraction) and ribonucleoprotein template (45.6 pg of N protein). After incubation for 1 hour at 30", the reaction mixture was divided into two equal portions; one' part was treated with 0.4% SDS at 37" for 3 min (A), while the other portion was treated with ribonuclease prior to SDS treatment (B). Samples were centrifuged on 12 ml of 15 to 35% sucrose in 0.01 M Tris-HCl (pH 7.4 at 4")/0.15 M NaCl buffer in a Spinco SW41 rotor at 40,000 rpm for 18 hours. The sucrose fractions of Experiment A were divided into equal halves, and acid-insoluble radioactivity was determined before (O-O) or after (O--O) ribonuclease treatment, while in Experiment B acid-insoluble radioactivity was determined with whole fractions. As the reference for determination of sedimentation coefficient of the products, "C-labeled Escherichia coli total RNA was centrifuged in a separate tube. Centrifugation was from right to left. FIG. 9 (right). Sucrose gradient centrifugation of terminally labeled RNA. RNA synthesis was carried out in 0.5 ml using 0.84 c(g of purified RNA polymerase (glycerol gradient fraction) and ribonucleoprotein template (15.8 Kg of N protein). The reaction mixture contained 0.21 mM [y-SZP]ATP (specific activity, 7 x 10' cpm/nmol) in place of [3H]ATP. After incubation for 3 hours at 30", 5 ~1 of 10% SDS was added and the reaction mixture was dialyzed first against 0.05 M sodium acetate buffer (pH 5.5) containing 0.1% SDS, and then against sodium acetate buffer containing 0.05% NaIO, in the dark. After removal of excess NaIO, by dialysis against sodium acetate buffer, RNA was treated with 2 mM [3H]NaBH, (specific activity, 7 x 10' cpm: nmol) for 3 hours in the dark. Labeled RNA was centrifuged on 11 ml of 15 to 35% sucrose in 0.01 M Tris-HCl (pH 7.4 at 4"C)/O.15 M NaCl buffer in a Spinco SW41 rotor at 35,000 rpm for 16 hours. The sucrose gradient was fractionated into 32 tubes and acid-insoluble radioactivities in 0.15 ml portions were determined before (A) or after (B) ribonuclease treatment. As the reference for determination of sedimentation coefficient of the products, 3H-labeled Escherichia coli total RNA was centrifuged in a separate tube. Centrifugation was from right to left. 3H radioactivity, GO; szP radioactivity, l ---0.
These two lines of investigation led us to develop the enzyme purification procedure described in this report, such that the proteins in viral particles are dissociated stepwise on the basis of their affinity to the genome RNA. Recently a procedure based on the similar principle was reported independently for the isolation of reverse transcriptase from avian myeloblastosis virus (40). In contrast to the finding by Emerson and Wagner (19), the enzyme solubilized by the present procedure was not unstable, provided that the appropriate concentrations of Triton X-100 (0.2%), dithiothreitol (1 mM), and glycerol (30%) were included in all buffers used during the enzyme purification. Thus, further purification could be carried out by several conventional techniques of enzyme purification such as glycerol gradient centrifugation or chromatography on DEAE-Sephadex and phosphocellulose columns, as employed here. Purified RNA polymerase preparations with the highest activity contained equimolar amounts of the L and NS proteins; the activity of enzyme preparations containing less NS protein (NS/L < 1) was always lower and was considerably stimulated by the addition of NS protein, which alone was inactive in RNA synthesis. Thus, the enzyme activity expressed per unit amount of L protein is proportional to NS content up to 1:l in molar ratio (Fig. 5). On the basis of these observations, we tentatively propose that both the L and NS proteins are the intrinsic subunits of VS virus-RNA polymerase and the native enzyme consists of equimolar amounts of the two polypeptides. Although Emerson and Wagner (19) succeeded in preparing NS-free L protein as well as pure NS protein from the Indiana serotype of VS virus, our attempts have been unsuccessful to isolate L protein from the New Jersey serotype employed in this research, supposedly due to differences in the conformation of L.NS complexes between the two strains of VS virus. The difference in the structure of the genetic elements has also been proposed between the two strains because six complementation groups were identified for the New Jersey serotype of VS virus (41) whereas the Indiana serotype contains only five complementation groups (42). Sedimentation velocity analysis indicated that the functional form in low ionic strength buffers is apparently (L) ,(NS) ,, associating to form dimer (L),(NS)2 in the presence of high concentrations of salt. In concert with this interpretation, approximately equal amounts of the L and NS proteins were found in purified virions. Although a number of enzyme proteins are known to form polymers in the absence of salt such as the DNA-dependent RNA polymerase from E. coli (43), it is also known that enzymes such as the Q,-RNA replicase, the RNA-dependent RNA polymerase from Qa phage-infected E. coli cells, dissociate at decreased ionic strength (44).
During the purification of VS virus-RNA polymerase, it was recognized that the poly(A) synthesizing activity could never be separated from the transcriptase activity and that the purified RNA polymerase itself directed the synthesis of RNA containing poly(A) sequence. The content of poly (A) in product RNA synthesized by enzymes of different purification steps was always constant indicating that the transcriptase itself catalyzes the post-transcriptional addition of poly(A) sequence. If this be the case, the transcriptase must recognize a unique structure of the 3'-OH terminus of product RNA or of the signal on the termini of transcription unit in the template RNA. It should also be considered that poly(A) could be synthesized by repeated reading of oligo(U) stretches if such sequences exist at the termini of transcription unit on the template RNA (8). Such a'template-slippage mechanism of poly(A) synthesis was proposed for the DNA-dependent poly(A) synthesis by the bacterial RNA polymerase (45).