Specific interaction between the classical swine fever virus NS5B protein and the viral genome Ming Xiao 1,2 , Jufang Gao 2 , Wei Wang 2 , Yujing Wang 2 , Jun Chen 2 , Jiakuan Chen 1 and Bo Li 1 1 Ministry of Education Key Laboratory for Biodiversity Science and Ecological Engineering, The Institute of Biodiversity Science, Fudan University, Shanghai, China; 2 College of Life and Environment Sciences, Shanghai Normal University, China The NS5B protein of the classical sw ine fever virus ( CSFV) is the RNA-dependent RNA polymerase of the virus and is able to catalyze the viral genome replication. The 3¢ un- translated region is most likely involved in regulation of t he Pestivirus genome replication. However, little is known about the i nteraction between the CSFV NS5B p rotein and the viral genome. We used different RNA templates derived from the plus-strand viral genome, or the minus-strand viral genome and the C SFV NS 5B protein obtained f rom t he Escherichia coli expression system to address th is problem. We first showed that the viral NS5B protein formed a complex with the plus-strand g enome t hrough t he genomic 3¢ UTR and that the NS5B protein was also able to bind the minus-strand 3¢ UTR. Moreover, it was found that viral NS5B protein bound the minus-strand 3 ¢ UTR more effi- ciently than the plus-strand 3¢ UTR. Furth er, we obse rved that the plus-strand 3¢ UTR with deletion of CCCGG or 21 continuous nucleotides a t its 3¢ terminal had no b inding activity and also l ost the activity for i nitiation of m inus- strand RNA synthesis, which similarly occurred in the minus-strand 3¢ UTR with CATATGCTC or the 21 nuc- leotide f ragment deleted from the 3¢ terminal. Therefore, it is indicated that the 3¢ CCCGG sequence of the plus-strand 3¢ UTR, and t he 3¢ CATATGCTC f ragment o f the minus- strand are essential to in vitro syn thesis o f t he minus-strand RNA a nd the plus-strand RNA, respectively. The same conclusion is also appropriate for the 3¢ 21 nucleotide terminal site of both the 3¢ UTRs. Keywords: CSFV; RdRp; replication; RNA synthesis; 3¢ UTR. Classical swine fever virus (CSFV) is the causative agent of swine fever, which is a h ighly contagious and f atal viral disease o f pigs. CSFV, bovine v iral diarrhea virus (BVDV), and Border Disease virus (BDV) are members of the Pestivirus genus within the Flaviviridae family. BVDV a nd BDV can infect both ruminants and pigs. The h epatitis C virus (HCV), an etiological agent of non-A, non-B hepatitis, also belongs to the Flaviviridae family. Pestiviruses are small, enveloped, plus-strand RNA viruses, similar t o HCV. The RNA genome is  12.5 kb in length, consisting of a large and continual open r eading frame (ORF), a 5¢ untranslated region (5¢ UTR) and a 3¢ untranslated region (3¢ UTR). The ORF is translated into a polyprotein, which is further processed into 12 mature proteins b y viral and host cell proteases. The 12 proteins comprise four structure proteins (C, E rns , E1, and E2) and eight nonstructure proteins (N pro , P7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B). In the CSFV genome, the genes encoding N pro ,C,E rns , E1, E2, p7, and NS2 have proved to be dispensable for RNA replication [1]. The 3¢ UT R and the 5¢ UTRarebelievedtoregulatePestivirus genome replica- tion [2,3]. The Pestivirus genomic replication c onsists of several c onsecutive processes. Repliase first recognizes and binds the 3¢ UTRandstartsRNAsynthesis,inwhicha minus-strand RNA is produced with the plus-strand genomic RNA as a template. Then, a progeny plus-RNA is produced with the novel minus-RNA as a template [4]. The 5¢ UTR is also the site fo r initiating t ranslation of the v iral genomes, at which an internal ribosomal entry site (IRES) is observed [5]. Short 3¢ terminal extensions do not interfere with infectivity of in vitro transcript whereas 5¢ extensions sometimes do and sometimes do not [6]. The CSFV NS5B gene is located a t the 3¢ end of the genome adjacent to the 3¢ UTR. The CSFV NS5B protein has an RNA-dependent RNA polymerase (RdRp) activity, and thus plays a central role in viral RNA replication [7–10]. Even NS5B as a fusion protein with the green fluorescent p rotein still displays an RdRp activity [11]. The NS5B proteins of BVDV and HCV have been expressed in different systems and their biochemi- cal properties h ave been s tudied [12–17]. N S5B protein is able to catalyze RNA elongation b y a primer-dependent o r copy-back mechanism, and can initiate RNA synthesis from the 3¢ end of different RNA templates in vitro [7,9,15]. I t is reported t hat t he mechanism for de novo initiation of RNA synthesis is a lso a ssociated with t he NS5B proteins [17–21]. Moreover, the crystal structure of HCV NS5B protein has been characterized [22]. Correspondence to B. Li, Ministry of Education Key Laboratory for Biodiversity Science and Ecological Engineering, The Institute of Biodiversity Science, Fudan University, Shanghai, 200433, China. Fax: +86 21 6564246, Tel.: +86 2 1 65642178, E-mail: bool@fudan.edu.cn Abbreviations: BDV, Border disease virus; BVDV, bovine viral diarrhea virus; CSFV, classical swine fever virus; EMSA, electro- phoretic mobility shift assay; HCV, hepatitis C virus; IRES, internal ribosome entry site; RdRp, RNA-dependent RNA polymerase; TNTase, terminal nucleotidyl transferase. (Received 25 June 2004, revised 28 July 2004, accepted 6 August 2004) Eur. J. Biochem. 271, 3888–3896 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04325.x Studies of pestiviral replication have been hampered b y the lack of efficient culture systems. Most of the informa- tion on cis-acting sequences of these v iral genomes comes from the research o n H CV [23 –26]. Recently, the studies of the cis-acting sequences of the CSFV genome have been performed in vitro. A 21 nucleotide fragment at the 3¢ terminal of the 3¢ UTR m ay be essential to initiation of RNA s ynthesis [7,27]. Furthermore, i t i s found that the 12 nucleotide insertion ÔCTTTTTTCTTTTÕ present in the 3¢ UTR of the CSFV HCLV strain [28], an efficient vaccine strain, might be responsible for a virulence [29]. But, the binding of the CSFV NS5B t o the cis-acting sequences relative to the viral replication has not yet been character- ized. In this study, we have performed competitive electrophoretic mobility shift assays (EMSA) and R dRp assays, and examined the influence of d ifferent cis-acting elements on binding of NS5B protein and RNA synthesis in vitro. Materials and methods Expression and purification of NS5B proteins The recombinant plasmid for expression of CSFV NS5B protein was constructed as described pr eviously [7]. Total RNA was extracted from CSFV Shimen s train. A f ull- length NS5B cDNA was obtained by RT-PCR, and cloned into the pET28(a) vectors. A methionine codon for initiating translation was added to the 5¢ end of the NS5B coding sequence. Additional sequences coding for six histidines at the C-terminus were engineered to facilitate the purification of NS5B protein. The i nserted regions of all clones were sequenced through d ideoxynucleotide sequen- cing and no c hanges were found. These resulting plasmids were introduced into Escherichia c oli strain BL21DE3 for expression driven by the bacteriophage T7 RNA poly- merase. T he E. coli strain BL21DE3 cells were cultured in M9ZB media with 40 lgÆmL )1 kanamycin at 37 °C. Expression was induced by addition of iso propyl thio- b- D -galactoside. Extraction and purification were per- formed as described previously [17]. Briefly, the bacterial cell culture was harvested and washed w ith phosphate- buffered saline ( NaCl/P i ). The cells from 1000 mL were resuspended in 20 mL of the buffer containing 50 m M Na-phosphate (pH 8.0), 300 m M NaCl, 10 m M imidazole, 10 m M 2-mercaptoethano l, 10% ( v/v) glycerol, 1 % (v/v) Nonidet P-40, supplemented with 1 m M phenylmethylsulfo- nyl fluoride and 10 m M leupeptin. After undergoing freezing and thawing once, c ells were subjected to sonica- tion. The cleared lysate was obtained by centrifugation at 35 000 g for 15 min. The cleared lysate containing the recombinant protein was purified using Ni-nitrilotriacetic acid–Sepharose resin ( Gibco BRL). Briefly, t he CSFV NS5B with a polyhistidine tag was bound to the Ni-nitrilotriacetic acid resin pre-equilibrated w ith the above buffer, and then w ashed with buffer containing 50 m M imidazole. The bound NS5B was eluted with buffer containing different c oncentrations of imidazole (100 m M to 500 m M ). The N S5B protein was c ollected, combined and dialyzed in buffer A [50 m M Tris/HCl (pH 8.0), 1 m M dithiothreitol, 50 m M NaCl, 5 m M MgCl, 10% (v/v) glycerol]. NS5B proteins were quantified as described previously [15]. In brief, NS5B protein solutions and dilutions of bovine serum albumin with known concentra- tion were subjected to SDS/PAGE. The gels with the samples were staine d with Coomassie B rilliant Blue. T he amount of NS5B protein was determined by densitometric scanning and comparing the two samples on the same gel. Purified proteins were separated by SDS/PAGE, and immunoblotted with a nti-His6 tag monoclonal antibody. RNA preparation The RNAs were generated as describe d previously [7]. In brief, cDNA fragments containing complete CSFV 3¢ UTR, 5¢ UTR and random coding sequences were initially cloned into the pGEM-T vector (Promega) f rom total C SFV RNA by RT-PCR. These nucleotide sequences were verified. A pair o f primers were designed on either side of the expected mutant fragment, or the desired wild-type sequence. The standard PCR m ethod based on the primers was u sed. The P CR product was obtained, treated with E. coli Klenow fragmen t, t hen with T4 DNA ligase, cloned into the pGEM-T vector and transformed into E. coli BL21 (DE3) cells. P lasmids w ere e xtracted and sequenced. The plasmids containing expected mutations were verified b y sequencing. Wild-type and m utant RNA templates were synthesized by PCR and subsequent in vitro transcription based on these RT-PCR products. A DNA Vent poly- merase and the primer containing bacteriophage T7 promoter were used in the PCR. After the sequence was verified, the resulting PCR products were used as the template for the subsequent in vitro transcription. The in vitro transcription was performed in 50 lL o f reaction mixtures following the standard method: 20 lLof5· tran- scription buffer, 2 lL of RNasin (20–40 UÆmL) (Promega), 5 lL of each dNTP ( 2.5 m M ), 5 lgofthetemplate,2lLof T7 R NA polymerase ( 10–20 U ÆmL) (Promega). T he mix- ture was incu bated at 37 °Cfor2h.DNaseI(10lL; Takara) was added to the mixture and incu bated at 37 °C for 15 min. The mixture w as extracted with phenol/chloro- form. After ethanol precipitation, the RNA was d ried, and redissolved in 20 lL of double distilled H 2 O. Labeled RNA fragments were produced in the analogous way with [ 32 P]UTP[aP]. Integrity of the RNA was analyzed by denaturing formaldehyde-agarose gel electrophoresis. The concentration of R NA was determined b y measuring its absorbance at 260 nm. Competitive electrophoretic mobility shift assays The [ 32 P]UTP[aP]-labeled RNA fragment containing the 3¢ UTR of plus-strand or minus-strand genome was used as the probe for the competitive electrophoretic mobility shift assays ( EMSA). In e ach assay, unless otherwise specified, 1 p mol of labeled RNA was incubated w ith 400 ng of CSFV NS5B protein in a buffer containing 20 m M HEPES (pH 7.3), 5 m M MgCl 2 ,7.5m M dithiothreitol, 5% (v/v) glycerol, 125 m M NaCl, 100 lgÆmL )1 bovine s erum albu- min, 1 U of RNasin, and various amount of competitor RNA. The reactions were performed at room temperature for 3 0 m in. The reaction products were analyzed on a native 6% polyacrylamide g el. T he gel was dried a nd subjected to autoradiography. Ó FEBS 2004 Interaction between CSFV NS5B and the genome (Eur. J. Biochem. 271) 3889 RdRp assays Total volume f or RNA polymerization t o determine the activity of RNAs or RdRp was 50 lL, containing the following supplements: 50 m M HEPES (pH 8.0), 5 m M MgCl, 10 l M dithiothreitol, 25 m M KCl, l m M E DTA, 20 U RNasin, 50 lg actinomycin D ( Sigma), 200 l M each dNTP (including a single radiolabeled CTP, [ 32 P]CTP[aP]), 1 lL of RNA template (250 ngÆmL )1 )and50n M NS5B proteins. The mixture was incubated at 37 °C for 2 h , and the reaction was stopped by addition of 2 lLofEDTA (200 m M ). The reaction samples were extracted with ph enol/ chloroform, a nd RNAs were precipitated with i sopropyl alcohol. Precipitates w ere resolved i n 2 5 lL of g el buffer [40 m M MOPS (pH 7.0), 10 m M sodium acetate, 1 m M EDTA, 50% (v/v) formamide, 2.2 M formaldehyde], heated to 55 °C f or 15 min, then chilled o n i ce with addition of 1 lL o f e thidium b romide (10 mgÆmL )1 ). Af ter a 10 min incubation at room temperature, 5 lL o f loading buffer [50% (v/v) glycerol, 0.25% (v/v) bromphenol blue, 0.25% (v/v) xylenecyanol, 1 m M EDTA] w as added, and samples were loaded onto 1.5% agarose gel containing 2.2 M formaldehyde, 40 m M MOPS (pH 7.0), 10 m M sodium acetate, and 1 m M EDTA. Electrophoresis was performed at 5 VÆcm )1 . Terminal nucleotidyl transferase activity assays Terminal n ucleotidyl t ransferase activity was determined in the same way as the RdRp assay. In brief, 10 l M cold UTP, GTP, CTP, or ATP mixed with 10 lCi of the ir a 32 P-labeled equivalent, was u sed a s the single ribonucleotide triphos- phate in the p resence of the CSFV native 3 ¢ UTR, as a template. T he reaction products were separated o n a 7 M urea/20% polyacrylamide gel. At the same time, the control experiments containing fractions isolated from untrans- formed E. coli lysate were carried out in the same way as above. Results The RdRp activity of the full-length CSFV NS5B protein To evaluate the biological activity of CSFV RNA poly- merase, a full-length NS5B cDNA was obtained. The cDNA was cloned into t he pET28(a) vector, which allowed expression of the full-length NS5B protein with a polyhis- tidine tag in E. coli. F ollowing induction with isopropyl thio-b- D -galactoside, e xtraction of bacterial lysate was performed w ith the lysis buffe rs containing high concentra- tions of salt, g lycerol and detergent. P urification was preformed with N i-nitrilotriacetic acid resin. A protein with the molecular mass equivalent to t he full-len gth CSFV NS5B was detected in the expression product, but was not present in a n e quivalent fraction obtained from the control experiment. The protein was able to be immunoblotted with anti-His6 tag monoclonal antibody, which was not found in the control e xperiment (Fig. 1A,B). Taken together, these results indicated that the expression product was the recombinant NS5B protein of CSFV. When this protein was i ncubated with the R NA templates containing the full- length 3¢ UTR of plus-strand or minus-strand CSFV genome ( a 6 03 or 701 nucleotide fragment, respectively) under the conditions for RNA polymerization, newly synthesized RNA products were detected, indicating that the f ull-length NS5B protein was able to catalyze RNA 123 12 3 45678 UACGUACG AC D B M12 50 80 97 600600 Fig. 1. Expression and RdRp-TNTase assays of the full-length CSFV N S5B protein. ( A) The N S5B proteins o btained th ro ughout expression and extraction and purification were analyzed on SDS/PAGE and revealed by C oomassie staining. (B ) The pro teins were immunoblotted with anti-His 6 tag monoclonal antibody. Lane 1: the full-length NS5B protein. Lane 2: E. coli lysate as control. Lane M: Molecular mass markers. (C) The products from RdRp assays at 37 °C for 2 h containing the native 3¢ UTR of the plus-strand (lane 1) or minus-strand (lane 2) RNA genome were loaded onto 1.5% agarose gel containing 2.2 M formaldehyde, 40 m M MOPS (pH 7.0), 10 m M sodium acetate, and 1 m M EDTA. Lane 3 was for no expre ssion p rodu ct. N um bers to the left re fer t o the position of RNA c ontain ing 600 nucleotides. (D) TN Tase ac tivity assays of the N S5B protein were condu cted with a single ribonucleotide triphosphate mixed with its equivalent labeled with [a 32 P] as described i n Materials and methods. The r esults were shown o n a 7 M urea/20% polyacrylamide gel. 3890 M. Xiao et al.(Eur. J. Biochem. 271) Ó FEBS 2004 synthesis with the native 3¢ UTR of the plus-strand or minus-strand RNA genome as templates (Fig. 1C). There- fore, t he recombinant C SFV NS5B protein expressed in E. coli had a RNA-dependen t RNA polymerase activity. The above experiments demonstrate that RdRp activity is associated with the E. coli-expressed CSFV NS5B protein. RNA f ragments longer than the templates were found in RNA synthesis by the N S5B protein (Fig. 1C) [30]. In addition to the R dRp activity, a terminal nucleotidyl transferase (TNTase) function could also contribute to the formation of these products. A TNTase could add non- templated terminal nucleotides to either the newly synthes- ized RNA or the input template. T herefore, we f urther examined whether the purified CSFV NS5B had such a TNTase activity. With the full-length 3¢ UTR as a template, TNTase activity was tested i n the presence of four different radiolabeled ribonucleotide triphosphates to see if there was any end-labeling activity at the 3¢ terminal of the temperate. As shown i n Fig. 1D, the radiolabeled products were detected in all t he reactions containing the NS5B protein (lanes 1–4) but not in the control experiments containing fractions isolated from untransformated E. coli lysate (lanes 5–8), i ndicating that the C SFV NS5B protein did have a TNTase activity. The 3¢ UTR is believed to be the first entry site for viral replicases to initiate RNA genome replication. To charac- terize the binding activity of the CSFV RdRp, estimation of the binding affinity of t he purified NS5B to the 3¢ UTR was performed by EM SA. The 603 nucleotide RNA template consisting of the full-length CSFV 3¢ UTR and the s ubse- quent coding sequence from the plus-strand genome, designed as +RNA0 (also see Fig. 4A), was labeled w ith [ 32 P]UTP[aP]. The 701 nucleotide fragment co ntaining t he full-length minus-strand 3¢ UTR and the subsequent upstream sequence, designed as –RNA0, was a lso labeled with [ 32 P]UTP[aP]. The purified NS5B proteins at various concentrations were incubated with a fixed amount of the radiolabeled +RNA0, and the reaction products were resolved on a polyacrylamide gel under nondenaturing conditions. The gel was subjected to autoradiography. As shown i n F ig. 2A, the amount of RNA–NS5B complex retarded at the loading wells was in agreement with increasing amount of purified NS5B protein. At the same time, the EMSA was carried out under the condition of a fixed amount of the NS5B and the radiolabeled +RNA0 in the presence of increasing a mount of competitor, the unlabeled +RNA0. The EMSA showed that RNA of the RNA–NS5B complex specifically competed with + RNA0. These r esults clearly d emonstrate an i nteraction between CSFV NS5B p rotein and the +RNA0 (Fig. 2B). In t he same way, the E MSA containing a fixed a mount of the NS5B and the radiolabeled –RNA0 in the presence of increasing amounts of c ompetitor, the unlabeled –RNA0 also demonstrated an interaction between CSFV NS5B protein and the –RNA0 (results not shown). Mapping of the NS5B-binding site on the CSFV genome The NS5B protein is the replicase for the C SFV genome. Therefore, all sites of the full-length CSFV genome could be bound for the NS5B protein. Some bindings were weaker and some stronger. Only the stronger bindings might be efficient to initiate genome replication. To define more precisely the site of the viral genome essential to efficient binding of CSFV NS5B protein, a series of c ompetition experiments were performed with the radiolabeled +RNA0 and various RNA competitors derived from different sites of the CSFV plus-strand genome. In addition to th e above +RNA0, these R NA competitors contained plus-strand 5¢ UTR, +RNAr1 (a 375 nucleotide RNA fragment corresponding to the c oding sequence n ext to 3 ¢ UTR), and +RNAr2 (an random sequence corresponding to positions 1245–1700 of the p lus-strand genome). As shown in Fig. 3A, the complex formed b etween the r adiolabeled +RNA0 and the NS5B protein was almost completely abolished by 25-fold molar excess of the unlabeled +RNA0 (lane 5). In contrast, the +5¢ UTR, +RNAr1 and +RNAr2 had n o effect on the binding (lanes 6 –15). The fact that the +RNAr1 without 3¢ UTR had no effect on the complex formed between the radiolabeled +RNA0 con- taining 3 ¢ UTR and the NS5B p rotein, showed that the s ite of interaction between them is 3¢ UTR. Moreover, our results suggest that the NS5B protein bound more 3 ¢ UTR than other regions of the plus-strand genome. The Pestivirus genomic replication consists of two consecutive processes. Replicase first recognizes an d binds the 3¢ UTR and starts RNA s ynthesis, in which a minus- RNA is produced with the plus-genomic RNA as a template. Then, a progeny plus-RNA is produced with the novel minus-RNA as a template [4]. Therefore, t he specific cis-element and the structure for direction of a plus-RNA synthesis might be present in the minus-RNA, as well as i n the plus-RNA. To investigate the binding activity of the CSFV NS5B protein on the minus-RNA, the competitive EMSA was performed with the R NA competitors 1234567 1234567 8 ABNS5B +RNA0 Fig. 2. Formation of the complex between CSFV and +RNA0. (A) The EMSA was performed with purified NS5B proteins at various con- centrations (lanes 2–7 for 90, 100, 200, 300, 400, 500 ng, respectively) incubated with a fi xed amount of the radiolabeled +RNA0. +RNA0 consisted of t he 228 nucleotide full-length viral 3¢ UTR and subse- quent 375 n ucleotid e co ding re gion. Lane 1 represents f ree + RNA0 probe to which no NS5B protein was added. (B) The EMSA was carried out with a fixed amount of the NS5B in t he presence of increasing amount of competitor +RNA0 (lanes 1–8 fo r 0, 0.5, 1 , 5 , 10, 25, 50, 100 pmol, respectively). Ó FEBS 2004 Interaction between CSFV NS5B and the genome (Eur. J. Biochem. 271) 3891 derived from the minus-RNA of the g enome. These RNA competitors were –RNA0, –RNAr1 (a 328 nucleotide fragment adjacent to the minus-strand 3¢ UTR), –RNAr2 (a random sequence corresponding to positions 2033–2500 of the minus-strand g enome) and minus-strand 5¢ UTR. Similarly, we found that the –RNA0 competed most effectively with the labeled RNA for binding of the CSFV NS5B protein among the competitors, and that –RNAr1, –RNAr2 and )5¢ UTR had no effect on the binding at all (results not shown), i ndicating that a n interaction occurred between the NS5B protein and the native minus-strand 3¢ UTR. It is known that both the plus-genomic RNA and the minus-genomic RNA are present in viral host cells when the replication of positive-se nse RNA virus o ccurs. It is necessary to compare the binding activities o f the plus- strand 3¢ UTR–NS5B with that of the minus-strand 3¢ UTR–NS5B. The purified recombinant NS5B was incu- bated with the radiolabeled –RNA0 in the presence of unlabeled +RNA0 or unlabeled –RNA0. Interestingly, RNA0 was stronger than +RNA0 in the interaction with the N S5B protein (Fig. 3B), i ndicating that NS5B bound the minus-strand 3¢ UTR was more efficiently than the plus- strand 3¢ UTR. Mapping of the specific NS5B-binding sequence on the 3¢ UTR From the a bove experimental r esults, it i s d erived that a specific element recognized by the viral RNA polymerase to initiate RNA replication might be harbored within the plus- strand 3¢ UTR a nd the minus-strand 3¢ UTR. To detect the specific NS5B-binding sequence, a further competition experiment was conducted. Various RNAs were used as the competitor, resulting from deletion of the n ative 3 ¢ UTR with PCR and subsequent in vitro transcription. The plus- strand genome was first addressed: +RNA1 t o +RNA5 represent, respectively, the RNA templates containing the 3¢ UTR with deletion of ÔCÕ, ÔCCÕ, ÔCCCÕ, ÔCCCGGÕ and the 21 nucleotide fragment at t he 3¢ terminal of the plus-strand genome (Fig. 4A). The radiolabeled +RNA0 was used as the substrate for the NS5B protein for t he competition experiments. Among these RNAs, +RNA1 c ompeted most efficiently (Fig. 4B, lanes 2–4), followed by +RNA2 and +RNA3 (lanes 5–10). In contrast, +RNA4 and +RNA5 w ere poor competitors (lanes 11–16). These results indicated that the CSFV NS5B protein bound the 3¢ UTR mainly through interaction with the  21 nucleotide bases at the t erminal. The ÔCCCGGÕ at the 3¢ terminal of the plus- strand 3¢ UTR was important for binding th e NS5B protein. Similarly, the binding experiment was p erformed containing the RNA competitors f rom the mutation of the minus-strand 3¢ UTR: –RNA1 to –RNA5 represent, respectively, the RNA templates containing the minus- strand 3¢ UTR with deletion of ÔCÕ, ÔCATATGÕ, ÔCATA TGCTÕ, ÔCATATGCTCÕ and the 21 nucleotide fragment at the 3¢ terminal. We found that –RNA1 was the m ost efficient competitor, then –RNA4, a nd –RN A5 w as the poorest among these minus-strand RNA fragments (Fig. 4 C). Therefore, t he 3¢ ÔCATATGCTCÕ sequence of 3¢ terminal of the minus-strand genome is essential to binding of the CSFV NS5B protein. The minus-strand 3¢ UTR with deletion of the 21 nucleotide fragment at the terminal did not interact with the N S5B protein at all. Characterization of the cis -acting sequence at the plus-strand or minus-strand 3¢ UTR for RNA synthesis To characterize the cis-acting s equence a t the plus-strand 3¢ UTR for RNA synthesis, RdRp assays containing the 123456 – +RNA0 +RNAr1 +RNAr2 +5 UTR ' – 78 12345678 9 1011121314 A – –RNA0 – B +RNA0 Fig. 3. Mapping of the NS5B-binding site on the CSFV genome. (A) Competitive EMSA was performed with CSFV NS5B protein and the 32 P-labeled +RNA0 in absence (lane 2) or the presence (lanes 3–14) of increasing amounts (1, 5, 25 pmol, respectively) of various competitors (+RNA0, +RNAr1, + RNAr2, and +5¢ UTR, as indicated). +RNAr1 i s a 375 nucleotide RNA fragment correspond - ing to the co ding sequence next to 3¢ UTR, and +RNAr2 i s a 456 nucleotide random sequence of coding region . Lane 1 represents free +RNA0 pro be to which no NS5B protein was added. (B) The results were obtained from com petitive EMSA with CSFV NS5B protein and the 32 P-labeled –RNA0 in the presence of unlabe led +RNA0 ( lanes 6–8 for 10, 25, 50 p mol, respectively) or –RNA0 (lanes 3–5 for 10, 25, 50 pmo l, respectively ). –RNA0 consist s of a 373 nucleotide minus- strand 3¢ UTR and a 328 nucleotide subsequent up stream sequence. Lanes 1 and 2 represent no N S5B protein and n o unlabeled –RNA0, respectively. 3892 M. Xiao et al.(Eur. J. Biochem. 271) Ó FEBS 2004 above five mutant RNAs were performed. The CSFV NS5B protein was incubated with +RNA1, +RNA2, +RNA3, +RNA4 and +RNA5, respectively, for 2 h in the presence of radiolabeled CTP. The reaction prod ucts were analyzed on an agarose g el. As expected, the newly synthesized RNA from the RNA polymerization containing +RNA1 and +RNA2 was de tected (Fig. 5A, lanes 1 and 2 ). +RNA3, +RNA4 and +RNA5 were unable to direct RNA synthesis (Fig. 5A, lanes 3–5). These results showed that the three continuous ÔCÕ at the 3 ¢ terminal of the plus-strand 3¢ UTR were essential to the initiation of RNA synthesis. Similarly, the cis-acting sequence n ecessary for RNA synthesis on the minus-strand CSFV g enome was examined as described above. The RNA templates with different lengths were formed f rom the minus-strand CSFV genome through PCR and subsequent in vitro transcription. The RdRp reactions containing these R NA tem plates ( Fig. 4A; –RNA1, –RNA2, –RNA3, –RNA4 and –RNA5) were, respectively, performed in t he presence of the CSFV NS5B protein. It was observed t hat –RNA1 and –RNA2 were efficient templates a nd yielded a major product (Fig. 5B, lanes 1 and 2). In contrast, – RNA3, – RNA4 and the RNA with deletion of the 21 nucleotide fragment at the 3¢ terminal of the minus-RNA genome did not have any template activity (F ig. 5A, lanes 3–5), indicat ing that several con- tinuous nucleotide b ases at the 3¢ ter minal of the minus- strand CSFV genome were important to initiating synthesis of the plus-strand RNA. Comparatively, we found that these results from the RdRp assays were essentially in agr eement with the data from the c ompetition EMSA. T aken together, t hese results suggested that the  21 nucleotide fragment l ocated at the 3¢ end of plus-strand 3¢ UTR might be the first site for the CSFV NS5B protein to initiate R NA genome replication, whereas the promoter sequence for NS5B protein to synthesize the plus-strand RNA was 21 continuous nucleo- tide bases at the 3 ¢ terminal of the minus-strand C SFV genome. Discussion Using E. coli cells as expression systems, we obtained the purified full-length CSFV NS5B protein. The NS5B protein is able to synthesize minus-strand RNA with plus-strand 12345 12345 AB 600 600 Fig. 5. RdRp assays containing various RNA templates with mutant 3¢ UTR. The products from RdRp assays at 37 °C for 2 h containing the R NA templates with mutant 3¢ UTR w ere loaded onto 1.5% agarose gel containing 2.2 M formaldehyde, 40 m M MOPS (pH 7.0), 10 m M sodium acetate, and 1 m M EDTA. ( A) Mutant plus-strand RNA 3¢ UTR (lanes 1–5; +RNA1 to +RNA5). ( B) RNA templates with the mutant minus-strand 3¢ UTR (lanes 1–5; –RNA1 to –RNA5). Numbers to the left refe r to t he position o f RNA co ntainin g 600 nucleotides. 11615141312111023 45 6 789 11615141312111023456789 B A CGGCCC +RNA0 – +RNA1 +RNA2 +RNA3 +RNA4 +RNA5 C – –RNA1 –RNA2 –RNA3 –RNA4 –RNA5 CGGCC +RNA1 CGGC +RNA2 CGG +RNA3 C +RNA4 +RNA5 ACCTCGTATAC –RNA0 ACCTCGTATA –RNA1 ACCTC –RNA2 ACC –RNA3 AC –RNA4 –RNA5 Fig. 4. Map ping of t he specific NS5B-binding s e quenc e on t he plus- strand 3¢ UTR or minus-strand 3¢ UTR. (A) Various RNA t emplates containing the wild-t ype an d the m utant plus- strand 3 ¢ UTR or minus- strand 3¢ UTR. )RNA1 to )R NA5 represe nt, respect ively, the RNA templates containing the plus-strand 3¢ UTR with deletion of ÔCÕ, ÔCCÕ, ÔCCCÕ, ÔCCCGGÕ and t he 21 nucleotide fragment at the 3¢ terminal. –RNA1 to –RNA5 represent, respectively, the RNA templates con- taining the m inus-strand 3¢ UTRwithdeletionofÔCÕ, ÔATATGÕ, ÔATATGCTÕ, ÔATATGCTCÕ and the 21 nucleotide fragment at the 3 ¢ terminal. (B) The c o mpetitive EMSA was preformed with CS FV NS5B protein and the 32 P-labeled +RNA0 in the absence (lane 1) or presence (lanes 2–16) of increasing amount (10, 25, 50 pmol) of various competitors (+ R NA1 to + RNA5). (C) The competitive EMSA w as preformed with CSFV NS5B protein and the 32 P-labeled -RNA0 in absence (lane 1) or the presence (lanes 2–16) of increasing amount (10, 25, 50 pmol, respectively) of various competitors (– RNA1 to –RNA5, as indicated). Ó FEBS 2004 Interaction between CSFV NS5B and the genome (Eur. J. Biochem. 271) 3893 native CSFV 3¢ UTR a s a template and to synthesize plus- strand RNA with minus-strand CSFV native 3¢ UTR as a template. Under RdRp assay conditions, the protein operates highly processively on these RNA templates up to several hundred nucleotides long. Therefore, the CSFV NS5B protein from prokaryotic cell expression systems is demonstrated to have RdRp activity, as reported previously [7]. This enzyme does not require an exogenous primer, and can recognize t he specific sequence for initiation of RNA synthesis. In addition to the RdRp activity, we showed that the CSFV NS5B protein possessed the TNTase activity. A TNTase protein does not proceed in elongating the primer after a ddition of the first residue, and is able to catalyze the addition of a single nucleotide residue to the 3¢ terminal of an RNA template, which was observed in our preparation of purified CSFV NS5B p rotein. In some p revious reports, HCV NS5B has been shown to possess the T NTase activity in addition to the RdRp activity [14,15,31], whilst in others it has not [17]. It has been observed that the TNTase activity is associated with BVDV NS5B protein [31]. Although CSFV NS5B proteins have been demonstrated to have the RdRp activity, little is known a bout whether t hese enzymes possess TNTase activity before the current rep ort. Th e fact that the CSFV NS5B protein was shown to have the TNTase activity in addition to an RdRp activity in this research supports the proposal that a copy-back mechanism is associated with initiation of RNA synthesis. It is assumed that the synthesis of a template-independent 3¢ extension is first performed by the TNTase activity, f ollowed b y the looping back and priming. A copy-back mechanism has been reported to be associated with initiation of RNA synthesis in HCV and BVDV [12,14,15], however, it has recently been found that the mechanism of de novo initiation is preferred f or the HCV NS5B pr oteins [17,19–21]. This mechanism has been observed in BVDV NS5B for initiation of RNA synthesis [21]. It was also found that both the copy- back mechanism and the de novo mechanism for initiation of RNA synthesis m ight be p resent in the NS5B protein of CSFV and BVDV [7,13]. The fact that the CSFV NS5B has a T NTase a ctivity, toget her with other data i n t his r esearch and previous reports, increases this possibility that two mechanisms, copy-back a nd de novo, are compatible with each other in the CSFV viral genomic replication. To characterize the i nteraction between the CSFV NS5B protein and different sites of the genome, competitive electrophoretic mobility shift assays were performed. Firstly, the plus-strand genome was addressed. The NS5B protein was incubated in the presence of the r adiolabeled 3¢ UTR and different site sequences as competitor. It was observed that a specific complex was formed between CSFV NS5B protein and its full-length 3¢ UTR. The 5¢ UTR formed the complex poorly as the 3¢ UTR was a superior competitor relative to the 5¢ UTR. The rando m sequence of coding region had no effect o n the formation of the complex. Our conclusion is consistent with that drawn from a similar study by using EMSA, in which specific interaction between the H CV NS5B protein and the noncoding region oftheviralgenomewasalsoaddressed[32].Inour competitive E MSA towards the minus-strand genome, the minus-strand 3¢ UTR was the strongest c ompetitor, fol- lowed b y t he random sequence of c oding region and t he minus-strand 5¢ UTR, similar to the competitive experiment towards the plus-genome. Interestingly, when the plus- strand 3¢ UTR and the minus-strand 3¢ UTR were used at the s ame t ime, the NS5B p roteins bound to the minus- strand 3¢ UTR more strongly than to the p lus-strand 3¢ UTR. Although we have not yet found sufficient evidence, it is proposed that th e activity of the minus-strand 3¢ UTR is stronger than the plus-strand 3¢ UTR in initiation of RNA synthesis. Indeed, it i s observed that a greater amount of plus-strand RNA than minus-strand RNA is d etected when the BVDV was replicating in its native cells [4]. This obse rvation was reported in the ce lls harboring the HCV RNA replicon, in which the plus-strand RNA was more abundant than the minus-strand RNA [33]. These results agree with the fact that the HCV RNA- dependent RNA polymerase replicates in vitro the 3¢ terminal region of the minus-strand viral RNA more efficiently than the 3¢ terminal region of the plus-RNA [26]. Further, we found that the mutated plus-strand 3¢ UTR with deletion of 5 or 21 nucleotide bases at the 3 ¢ terminal lost binding activity, and that the m utated minus- strand 3¢ UTR with deletion of a 9 or 21 nucleotide sequence at the 3¢ terminal also did not have the b inding activity, indicating that the CSFV NS5B p rotein binds the 3¢ UTR mainly through interaction with the s everal nucleotide bases at the t erminal of the 3¢ UTR. Therefore, the 3¢ UTR, specifically the 3 ¢ terminal region, is very important fo r binding of the CSFV NS5B protein, whereas the sequences of the c oding region have no effect on the binding at all, irrespective of w hether it is plus-strand genome or not; t his is not in agreement with t he previous observation from HCV in which the partial NS5B-coding sequence and subsequent partial 3¢ UTR are the necessary sites for binding o f NS5B protein [25]. But, our data is consistent with Cui and his colleagues’ reports that the 3 ¢ end o f encephalomyocarditis virus is involved in the binding of RdRp [34,35]. In addition to NS5B protein, other viral proteins are a ble to b ind the 3¢ UTR, e.g. the H C V NS3 protein can bind the 3¢ UTR in vitro and displayed a protease and a helicase activity [36]. Some cellular proteins arealsoabletobind3¢ UTR, such as the heterogeneous nuclear ribonucleoprotein C [37] and r ibosomal proteins [38]. To analyze the function of the NS5B-binding sequence i n the initiation o f RNA synthesis, we performed the RdRp assays, i n which the mutated 3¢ UTRs with deletions of different l engths at the 3¢ terminal of the C SFV p lus-strand genome were addressed. The plus-strand 3¢ UTR with deletion of 3, 5 or 21 nucleotide bases at the 3¢ terminal was observed to lose the activit y for i nitiation o f minus-RNA synthesis. When the mutated minu s-strand 3¢ UTRs with deletions of different lengths at the 3¢ terminal were used in the RdRp assays, similar results were observed, i.e. the minus-strand 3¢ UTR with 8, 9 or 21 nucleotide bases deleted from its 3 ¢ terminal had no template a ctivity. Together with the above binding data, it is s hown that the sequence approximately b etween position 1 and 2 1 at the CSFV genome, irrespective of whether it is plus-strand genome or not, might be the cis-acting signals for i nitiation of the C SFV R NA synthesis, which is compatible w ith ou r earlier conclusion that the sequence of the 3¢ terminal of t he 3¢ UTR is essential to initiation of RNA synthesis [7,9]. In this study, we noted that the template activity o f t he RNAs 3894 M. Xiao et al.(Eur. J. Biochem. 271) Ó FEBS 2004 was well in agreement with their binding activity, which is consistent with the earlier reports that the t emplate activity of RNAs for Qb replicase i s correlated with their binding affinity [39]. A similar conclusion has also been drawn from the HCV [25]. Our results differ f rom a nother study in which t he template activity of RNAs for the HCV N S5B i s inversely correlated with the binding activity of these RNAs to the N S5B [15]. This discrepancy may be partially due to the f act t hat homopolymeric RNAs were used as templates in that study and we used heteropolymeric RNAs as templates. Previous studies have demonstrated specific sequences and s tructures at the 3¢ terminal of plus-strand RNA v iruses involving the binding of RNA polymerase and initiation of replication, e.g., a 3¢ terminal 98 nucleotide X-region of HCV [25,32], the secondary structure and the single-strand region a t the 3¢ end of BVDV 3¢ UTR [3], a nonbase-paired 3¢ ÔACCÕ sequence o f a tRNA-like s tructure of turnip yellow mosaic v irus [40], a psudoknot structure of polioviruses [41], and a stem-loop structure of turnip crinkle virus [ 42]. Our experiments show ed that the 3¢ ÔCCCGGÕ sequence at the plus-strand 3¢ UTR and the 3¢ ÔCATA TGCTCÕ fragment at the minus-strand 3¢ UTR is essential to the CSFV R NA synthesis. The 3¢ UTR with the above deletion has no template activity, and also no b inding activity to NS5B. The predicted secondary structure of these mutant 3¢UTRs is not more stable than that of their wild- type equivalent [7]. The f act that an internal single-strand region is necessary for binding of HCV NS5B protein and initiating RNA s ynthesis fuels the possibility that internal initiation rather than initiation from the very 3¢ terminal is associated with HCV [25], in contrast to our conclusion that the mechanism of initiation from the very 3 ¢ termina l of 3¢ UTR might be one of peculiarities of CSFV replication because the sequence of the 3¢ terminal of 3¢ UTR is necessary fo r binding of NS5B protein and initiation of RNA synthesis. Our previous report shows that the CSFV RdRp originating from expression in prokaryotic cells has high template specificity. Replacement and deletion in the middle part of the viral 3¢ UTR has little influence on initiation of RNA synth esis. The 3¢ terminal of the 3 ¢ UTR is much more important to initiation of RNA synthesis than its m iddle p art [7]. The importance of the 3¢ terminal of the 3¢ UTR is reinforced in this report. Acknowledgements This work was s upported by National Natural S cience Foundation of China (30170214) and Science and Technology Foundation of Shanghai Higher Education. References 1. Mose r, C., S tettler, P., Trutschin, J .D. & H ofma nn, M.A. ( 1999) Cytopathogenic and n oncytopathoge nic RNA replicons o f clas- sical s wine fever virus. J. Virol. 73, 7784–7794. 2. Behrens, S.E., Grassmann, C.W., Thiel, H.J., Meyers, G. & Tautz, N. (1998) Charac terization of an autonomous subgenomic pesti- virus R NA replicon. J. Virol. 72, 2364–2372. 3. Yu, H ., Grassmann, C.W. & Behrens, S.E. ( 1999) Sequence and structural elemen ts at the 3¢ terminus of bovine viral diarrhea virus genomic RNA: f unctio nal role d uring RNA replication. J. Virol. 73, 3638–3648. 4. Gong, Y., Trowbridge, R., Macnaughton, T.B., W estaway, E.G., Shannon, A.D. & Gowans, E.J. (1996) Characterization of RNA synthesis during a one-step growth curve and of the replication mechanism of bovine viral diarrhea virus. J. Gen. Virol. 77, 2729– 2736. 5. Flet che r, S.P. & Ja ckson, R .J. (200 2) Pestiv irus intern al ribosome entry s ite (IRES) structu re and function: Elem ents in the 5 ¢ untranslated region important for IRES function. J.Virol. 76, 5024–5033. 6. Boyer, J.C. & Haenni, A.L. (1994) Infec tious transcripts and cDNA clones of RNA viruses. Virology 198, 4 15–426. 7. Xiao, M., Wang, Y ., Chen, J. & Li, B. (2003) Characterization of RNA-dependent RNA polymerase activity o f C SFV N S5B pro- teins expressed i n Escherichia coli. Virus Genes 7, 67–74. 8. Lohmann, V., O verton, H. & Bartenschlager, R. (1999) Se lective stimulation o f hepatitis C v irus and pestivirus N S5B RNA poly- merase activity by GTP . J. Biol. C hem. 274, 10807–10815. 9. Xiao,M.,Chen,J.&Li,B.(2003)RNA-dependentRNApoly- merase activity of NS5B proteins expressed in pestiviral host cells. Acta Virol. 47 , 17–26. 10. Steffen, S., Thiel, H.J. & Behrens, S.E. (1999) The RNA-dependent RNA polymerase of different members of the f amily Flaviviridae exhibits similar p roperties in vitro. J. Gen. Virol. 80, 2583–2590. 11. Xiao, M., Zhang, C.Y., Pan, Z.S., Wu, H.X. & G uo, J.Q. (20 02) Classical swine fever v irus NS5B-GFP fusion possesses an RNA- dependent RNA polymerase activity. Arch. V irol. 147, 1779–1787. 12. Zhong, W., Gutshall, L.L. & D el Ve cchio, A.M. (1 998) Iden tifi- cation and characterization of an RNA-dependent RNA poly- merase activity within the nonstructural protein 5B region of bovine vi ral diarrhea virus. J. Virol. 72, 936 5–9369. 13. Lai, V.C.H., Kao, C.C., Ferari, E., Prak, J., Uss , A., Wright- Minogue, J., H ong, Z. & L au, J.Y.N. (1999) Mutational analysis of bovine viral diarrhea virus RNA-dependent RNA polymerase. J. Virol. 73, 10129–10136. 14. Be hrens, S.E., Tomei, L. & Francesco, R.D. (1996) Identification and Properties o f the RNA-dependent R NA polymerase of hepatitis C virus. EMBO J. 15 , 12–22. 15. Lohmann, V., Ko ¨ rner,F.,Herian,U.&Bartenschlager, R. (1997) Biochemical Properties of hepatitis C virus NS5B RNA- dependent RNA p o lymerase a nd Ide ntification o f a mino ac id sequence motifs essential for enzymatic activity. J. Virol. 71, 8416– 8428. 16. Ferrari, E., Wright-Minogue, J ., Fang, J.W.S., Barrudy, B.M., Lau, J.Y.N. & Hong, Z. (1999) Characterization of soluble hepatitis C virus RNA-dependent RN A polymerase e xpressed in Escherichia coli. J. Viro l. 73, 1649–1654. 17. Oh,J W.,Ito,T.&Lai,M.M.(1999) A recombinant hepatitis C virus RNA-dependent RNA polymerase capable of copying the full-length viral RNA. J. Virol. 73, 7 694–7702. 18. Yi, G.H., Zhang, C.Y., Cao, S., Wu, H.X. & Wang, Y. (2003) De novo RN A synthesis by a recombinant classical swine fever virus RNA-dependent RNA polymerase. Eur. J. Biochem. 270, 4952–4961. 19. Zh ong, W., Uss, A.S., Ferrari, E., Lau, J.Y.N. & Hong, Z. (2000) De nov o initiation of RNA synthesis by hepatitis C virus n on structural protein 5B polymerase. J. Virol. 74, 2017–2022. 20. Ranjith-Kumar, C.T., Kim, Y.C., Gutshall, L., Silverman, C., Khandekar, S., Sarisky, R .T. & Kao, C.C. (2002) Mechanism o f de novo initiation by the hepatitis C virus RNA-Depen de nt R NA plymerase: Role of divalent me tals. J. Vi rol. 76, 1 2513–12525. 21. Kao, C.C., Del Vecchio, A.M. & Zhong, W. ( 1999) De novo initiation of RNA synthesis by a recombinant flaviridae RNA- depentent RNA polymerase. Virology 253, 1–7. 22. Bressanelli, S., Tomei, L ., Roussel, A., I ncitt, I., Vitale, R.L., Mathieu, M., De Francesco, R. & Rey, F. (1999) Crystal structure Ó FEBS 2004 Interaction between CSFV NS5B and the genome (Eur. J. Biochem. 271) 3895 of the RNA-Dep endent RNA plymerase of he patitis C virus. Proc.NatlAcad.Sci.USA96, 13034–13039. 23. Shim, J.H., Larson, G., W u, J.Z. & Hong, Z. (2002) Selection of 3¢-template bases a nd initiating nucleotides by hepatitis C virus NS5B RN A-depende nt RN A polyme rase. J. Virol. 76, 7030– 7039. 24. Ran jith -Ku mar , C.T. , Gutsha ll, L., Kim , M.J. , Saris ky, R.T . & Kao, C.C. (2002) Requirements for de novo initiation of RNA synthesis by recombinant flaviviral RNA-dependent RNA poly- merases. J. Virol. 76, 1252 6–12536. 25. Oh, J W., Sheu, G T. & Lai, M.M. (2000) Template requirement and initiation site selection by he patitis C virus polymerase on minimal viral template. J. Biol. Chem. 27 5 , 17710–17717. 26. Reigadas , S., Ventura, M., S a rih-Cottin, L., Castroviejo, M., Litvak, S. & Astier-Gin, T . (2001) HCV RNA-dependent RNA polymerases replicates in vitro the 3¢ terminal region of the minus- strand v i ral RNA more efficien tly than t h e 3 ¢ terminal region of the plus-strand RNA. Eur. J. Biochem. 268, 5 857–5867. 27. Xiao, M., Zhi, Z.Z., Jueping, L. & Zhang, C.Y. (2002) Prediction of recognition s ites of c l assical swine fever virus genomic replica- tion with i nformatio n analysis. Mo l. Biol. 36, 34–43. 28. Wu,H.X.,Wang,J.F.,Zhang,C.Y.,Fu,L.Z.,Pan,Z.S.,Wang, N., Zhang, P .W. & Zhao, W .G. (2001) Attenuated l apinized Chinese strain of classical swine f ever virus: complete nucleotide sequence and character of 3¢-noncoding region. Virus Genes 23 , 69–76. 29. Xiao, M., Zhi, Z.Z. & Zhang, C.Y. (2001) Qualitative, quantita- tive and structural analysis of non-coding regions of classical swine fever virus genome . Chinese S ci. Bull. 46, 1251–1257. 30. Kao, C.C., Yang, X., Kline, A., Wang, Q.M., Barket, D. & Heinz, B.A. (2000) Te mplate requirem e nts f or RN A synthesis by a recombinant hepatitis C virus RNA-dependent RNA polymerase. J. Virol. 74, 11121–11128. 31.Ranjith-Kumar,C.T.,Gajewski,J.,Gutshall,L.,Maley,D., Sarisky, R.T. & Kao, C.C. (2001) Terminal nucleotidyl transferase activity of recombinant Flaviviridae RNA-dependent RNA plymerases: Implication f or vira l RNA synthesis. J. Virol. 75 , 8615–8623. 32. Cheng, J.C., Chang, M.F. & Chang, S.C. (1999) Specific inter- action between the hepatitis C virus NS5B RNA polymerase and the 3¢ end of t he viral RNA. J. Virol. 73, 7044–7049. 33. Lo hmann, V., Ko ¨ rner, F., Koch, J., Herian, U., Theilmann, L. & Bartenschlager, R. (1999) Replic ation o f subgenomic h epatitis C virus R NAs in a hepat oma cell line. Science 285, 110–113. 34. Cui, T. & Porter, A.G. (19 95) Localization of binding site for encephalomyocarditis virus RNA polymerase in the 3¢-non coding region of the viral RNA. Nucleic Acids Res . 23, 377 –382. 35. Cui, T., S ankar, S. & Porter, A.G. (1993) Binding of encephalomyocarditis virus RNA polymerase to the 3¢-noncoding region of the viral RNA is specific and requires the 3¢-poly(A) tail. J. Biol. Chem. 268 , 26093–26098. 36. Banerjee, R. & Da sgupta, A. ( 2001) Specific interaction o f hepa- titis C virus protease/helicase NS3 with the 3¢-terminal sequences of viral positive- and negative-strand RNA. J. Virol. 75, 1708– 1721. 37. Gontarek, R.R., Gutshall, L.L.,Herold,K.M.,Tsai,J.,Sathe, G.M., Mao, J., Prescott, C. & Delvcchio, A .M. (1999) hnRNP C and po lypyrimidine tract-binding p rotein specifically interact with the pyimidine-rich region within the 3¢ NT R of the HCV genom e. Nucleic A cids Res. 27 , 1457–1463. 38. Wood, J., Frederickson, R.M., Field, S. & Patel, A.H. (2001) Hepatitis C virus 3¢ X region interacts with hu man ribosomal protein. J. Virol. 75, 1348–1358. 39. Brown, D. & Gold, L. (1996) RNA replication by Qb replicase: a working model. Proc . Natl A cad. Sci. USA 93, 11558–11562. 40. Deiman, B.A.L.M., Koenen, A.K., Verlaan, P.W.G. & Pleiji, C.W.A. (1998) Minimal template r equiremen ts for initiation of minus-strand synthesis in vitro by the RNA-dependent RNA polymerase of turnip yellow mosaic virus. J. Virol. 72, 3965–3972. 41. Jacobson, S .J., Konings, D .A.M. & Sarnow, P. ( 1993) Biochem- ical a nd genetic e viden ce for a pseudoknot structure at the ter- minus of the poliovirus RNA genome and its role in viral amplification. J. Virol. 67, 2961–2971. 42. Song, C . & Simon, A.E. (1995) Requirement of a-3¢-terminal stem-loop in in vitro transcription by an RNA-dependent RNA polymerase. J. Mol. Biol. 254, 6–14. 3896 M. Xiao et al.(Eur. J. Biochem. 271) Ó FEBS 2004 . Specific interaction between the classical swine fever virus NS5B protein and the viral genome Ming Xiao 1,2 , Jufang Gao 2 ,. known about the i nteraction between the CSFV NS5B p rotein and the viral genome. We used different RNA templates derived from the plus-strand viral genome, or the