The complete nucleotide sequence of the rice grassy stunt virus genome and genomic comparisons with viruses of the genus Tenuivirus

Transcription

1 Journal of General Virology (1998), 79, Printed in Great Britain The complete nucleotide sequence of the rice grassy stunt virus genome and genomic comparisons with viruses of the genus Tenuivirus Shigemitsu Toriyama, 1 Takao Kimishima, 1 Mami Takahashi, 1 Takumi Shimizu, 1 Nobuhiro Minaka 2 and Katsumi Akutsu 3 1, 2 Division of Microbiology 1, and Laboratory of Statistics 2, National Institute of Agro-Environmental Sciences, Kannondai 3, Tsukuba, Ibaraki , Japan 3 Faculty of Agriculture, Ibaraki University, Ami-machi, Ibaraki , Japan Rice grassy stunt virus (RGSV, IRRI isolate) has six genomic RNA segments. The nucleotide (nt) sequences of RNAs 1 4 were determined. The cumulative length of the RGSV genome, including RNAs 5 and 6, was nt. All six RNA segments had an ambisense coding strategy and almost identical terminal sequences over 17 nt. The virus complementary (vc) sequence of the largest segment, RNA1, had an open reading frame encoding a protein of M r (the 339 1K protein), while the virus sense (v) sequence encoded a protein of M r in the 5 -proximal region. The predicted 339 1K protein contained the highly conserved motifs of the RNA-dependent RNA polymerase and a short but distinct Arg/Gly-rich stretch at the C terminus. The putative RNA polymerase showed strong similarity with that of rice stripe tenuivirus (RSV); they shared 37 9% amino acid identity over 2140 residues. The predicted proteins of M r on vrna2 and on vcrna2 were only slightly similar in sequence to the proteins encoded by vrna2 and vcrna2 of other tenuiviruses. The predicted proteins encoded by RNA3 and RNA4 did not show significant similarity to any database proteins. Only the putative RNA polymerase encoded on RNA1 was well-conserved between RGSV and RSV. The low sequence similarities in proteins encoded by RNAs 2, 5 and 6, together with the unique RNA segments 3 and 4, indicate that RGSV may be distinct from other tenuiviruses. Introduction Rice grassy stunt virus (RGSV) is circulatively and propagatively transmitted by the brown planthopper, Nilaparvata lugens (Sta l), and is classified as a member of the genus Tenuivirus (Hibino, 1986; Murphy et al., 1995). The genome of the tenuiviruses rice stripe virus (RSV) and rice hoja blanca Author for correspondence: Shigemitsu Toriyama. Fax storyam niaes.affrc.go.jp Present address: Graduate School of Ibaraki University, Ami-machi, Ibaraki , Japan. The nucleotide sequence data reported in this paper have been submitted to DDBJ, EMBL and GenBank under the accession numbers AB (RNA1), AB (RNA2), AB (RNA3) and AB (RNA4). virus (RHBV) consists of four RNA segments, while that of maize stripe virus (MStV) is made up of five segments. The terminal sequences of each RNA segment are complementary and can form a panhandle-like secondary structure. Other features of the tenuivirus genome include the ambisense coding strategy of RNA segments 2, 3 and 4 and RNA5 of MStV and RNA1 of RSV are negative-stranded (reviewed by Ramirez & Haenni, 1994; Toriyama et al., 1994). The nucleocapsid proteins and the major nonstructural proteins encoded by RNAs 3 and 4 of tenuiviruses are very similar in sequence in RSV, MStV and RHBV. The RNA2 segment of tenuiviruses encodes a 23 kda protein on its virus sense (v) sequence and a 94 kda protein on its virus complementary (vc) sequence. The amino acid sequences of these proteins are highly conserved among RSV, MStV and RHBV. Although the function of these proteins has not been determined experimentally, the 94 kda protein is SGM CAFB

2 S. Toriyama and others vaguely similar in sequence to the virion membrane glycoproteins of phleboviruses (Takahashi et al., 1993; Estabrook et al., 1996; de Miranda et al., 1996). RGSV particles are pleomorphic; they can appear as thin filamentous particles or circular filaments, and often form spiral configurations. The filamentous nucleoprotein particles contain RNA-dependent RNA polymerase activity similar to that of RSV (Toriyama, 1986, 1987). So far, only the RSV genome has been entirely sequenced and its largest segment, RNA1, encodes a putative RNA-dependent RNA polymerase. The nucleotide (nt) sequences of tenuiviruses share some features with those of phleboviruses of the family Bunyaviridae. In particular, the terminal sequences and amino acid sequences of the RNA polymerase gene are conserved between tenuiviruses and phleboviruses (Toriyama et al., 1994). Recent studies of the RGSV genome have revealed that this virus has six genomic RNA segments. The terminal sequences and ambisense coding strategy of RGSV RNAs 5 and 6 are characteristic of tenuiviruses. Protein species encoded by RGSV RNAs 5 and 6 correspond to those encoded by RNAs 3 and 4 of other tenuiviruses. Furthermore, the predicted proteins encoded by RNAs 5 and 6 are only distantly similar in sequence to the corresponding proteins of other members of the genus Tenuivirus (Toriyama et al., 1997). Here, we report the complete nucleotide sequences of RGSV RNA segments 1, 2, 3 and 4. Thus, the entire nucleotide sequence of the RGSV genome is complete, and the relationship of RGSV to the other tenuiviruses is discussed. Methods Virus and RNAs. The IRRI isolate of RGSV was used. The procedures for purification of RGSV and each double-stranded (ds) RNA of segments 1, 2, 3 or 4 were as previously described (Toriyama et al., 1997). To detect smaller RNAs with the RNA1 sequence, we extracted RNA from the upper fraction above the B1 component band of RGSV after sucrose gradient centrifugation (Toriyama et al., 1997). cdna synthesis and cloning. Whole viral RNA preparations were extracted from purified RGSV and used as templates. Each of the purified dsrna1, dsrna2, dsrna3 or dsrna4 was used after denaturation with methylmercuric hydroxide (Boehringer Ingelheim). cdna synthesis was performed using the M-MLV reverse transcriptase (Gibco BRL) and the synthetic oligonucleotide primer P1 (5 ACTAG- TCGACACACAAAGTC 3 ); the 10 nt sequence at the 3 end was obtained by two-dimensional sequence analysis (Takahashi et al., 1990). The synthesized second-strand DNA was blunt-ended with T4 DNA polymerase and ligated into SmaI-digested puc18. Recombinant plasmids were transformed into competent Escherichia coli JM109 (Wako) (Toriyama et al., 1997). We obtained 13 cdna clones ranging in size from 500 to 1500 nt that together covered about 5000 nt in the 5 -terminal region and 2000 nt in the 3 -terminal region of RGSV RNA1. The DNA in the remaining interior region of RNA1 was synthesized by PCR amplification using Ex Taq polymerase (Takara) and primer sequences derived from predetermined nucleotide sequences. To clone RNA2, we isolated two clones (clones 33 and 15) that had about 2000 nt from internal nucleotide sequences of RNA2. From the 10 nt common terminal sequences and the predetermined nucleotide sequences of clone 33, we designed PCR primers that led to the isolation of RNA2 cdna clones; the end result consisted of four overlapping cdnas that covered the entire RNA2, each consisting of about 1000 nt. At least two templates, each newly prepared from RGSV RNAs, were used to produce PCR-derived cdna clones. Clone Sm-60, which had almost full-size cdna from RNA3, and clone D36, which covered twothirds of the RNA3 sequence, were obtained. Two attempts at cloning the 3 -terminal region (about 300 nt) failed, so this region was amplified by PCR, resulting in three PCR clones (6, 7 and 8). PCR amplification of the internal region spanning nt yielded clones PCR-2 and PCR-4. Clones D4-13, D4-14 and D4-4 were obtained from RGSV RNA4, of which clones D4-13 and D4-14 contained full-size cdna of RNA4. For sequence confirmation of the 5 -terminal region of RNA4, a PCR clone (A-1) was obtained and sequenced. We confirmed ambiguous nucleotide sequences by sequencing other PCR-derived clones and using Cy5 primers (Pharmacia). For each sequence determination, both strands of more than two cdna clones were sequenced. When sequencing PCR clones, nucleotide sequences of five to ten cdna clones were determined. DNA sequence. The cdna clones of RNAs 1, 2, 3 and 4 were subcloned into puc18 and sequenced using the Pharmacia DNA Sequencing kit and an ALFred DNA sequencer (Pharmacia). The sequence data were assembled and analysed using the DNASIS (Macintosh) program (Hitachi). GenBank EMBL, NBRF and PIR databases were searched for nucleic acid and amino acid sequence similarity. Northern blot analysis. Total plant RNA was extracted from RGSV-infected or non-infected rice leaves (frozen under 80 C) with guanidine HCl (Sambrook et al., 1989). These RNAs, and RNA extracted from the upper component of RGSV, were separated by electrophoresis on a 1% formaldehyde MOPS agarose gel and transferred to a Hybond- N membrane (Amersham). The DNA fragments for two probes specific to RNA1 sequences were cut from the cdna clones containing the RNA1 sequence with restriction enzymes AflII AseI and AflII BalI, respectively. Probe A consisted of 636 bp (nt ) of the RNA1 sequence for specific detection of the 18 9K open reading frame (ORF). Probe B consisted of 690 bp (nt ) for detection of another ORF (encoding a protein of M r 18275) present in nt of the vrna1 sequence. Each DNA fragment was labelled and used as a specific probe using the Gene Image labelling and detection system (Amersham). The RNA molecular markers comprised 1 6, 1 0, 0 6, 0 4 and 0 3 kb (Boehringer Mannheim). Phylogenetic analysis. Amino acid sequences were aligned using CLUSTAL V (Higgins et al., 1992). Phylogenetic analysis based on these aligned sequences was carried out using PAUP version (Swofford, 1993). The searching option of PAUP is the branch-and-bond exhaustive method under the PROTPARS stepmatrix weighing of amino acid changes (Felsenstein, 1993). Bootstrap confidence probabilities were calculated for internal branches by 1000 replicates. Results and Discussion Terminal sequences The 5 - and 3 -terminal 10 nt of RGSV RNAs 2, 3, and 4 were determined by two-dimensional mobility shift analysis (Takahashi et al., 1990) and were the same as the terminal nucleotide sequences of other tenuiviruses. The 10 nt terminal sequence was used as a primer for cdna synthesis or PCR CAFC

3 Complete sequence of rice grassy stunt virus genome 16 nt from the 3 end were identical in RNAs 2, 3 and 4. Beyond nt 18, the sequences were unique to each segment. In RNA1, just one bp (A U) differed out of the 17 nt common sequence. A similar single base mismatch is observed in nt 6 at the 3 end of RSV RNA1 (Takahashi et al., 1990). The basepaired region of the panhandle structure could act as a recognition site for the RNA polymerase (Honda et al., 1987). This mismatch is unique to RNA1, which encodes the RNAdependent RNA polymerases of RSV and RGSV, and may be involved in replicative and transcriptional control. Fig. 1. The 5 - and 3 -terminal sequences of the six RNA segments of RGSV. Complementary nucleotides at the 5 and 3 ends are linked with short lines. Underlined regions are identical among all segments, except for a base pair (A U; boxed) in segment RNA1. The terminal sequences of RNA5 and RNA6 were obtained from previous data. amplification of genomic RNAs of RGSV. Fig. 1 shows the terminal sequences obtained by direct sequencing and analysis of the cdna sequences. The first 17 nt from the 5 end and Sequence of RNA1 The RGSV RNA1 segment was 9760 nt and had the following base composition: A, 28 87%; U, 34 47%; C, 20 48%; and G, 16 16%. This nucleotide sequence is 790 nt longer than the 8970 nt sequence of RSV (Toriyama et al., 1994). RGSV RNA1 contains a long ORF in the vc sequence that begins with an AUG initiation codon at position and continues to a UAG stop codon at position The predicted translation product should contain 2925 amino acids with an M r of (the 339 1K protein) (Fig. 2), which is similar in size to the RNA polymerase protein which has 2919 amino acid residues and is encoded by RSV RNA1 (Toriyama et al., 1994). An ORF, found at nt of the 5 end of vrna1, had the capacity to encode a protein of 166 amino acids with a predicted M r of (the 18 9K protein). An additional ORF encoding a predicted protein with an M r of was found at nt (in a 1 frameshift) of the vrna1 sequence. To detect candidate RNAs or mrnas encoding the two ORFs present on the vrna1 sequence, we performed Fig. 2. Schematic representation of ORFs on the v and vc sequences of RNA1, RNA2, RNA3 and RNA4. The ORFs are displayed on vrnas as a black bar and on vcrnas as a shaded bar. The proteins deduced from each ORF are shown as boxes, indicating the molecular size of the predicted proteins. CAFD

4 S. Toriyama and others (a) (b) Fig. 3. Northern blot analysis of RNAs encoding the vrna1 18 9K ORF using specific probe A. (a) The DNA fragment (636 bp; 10 pg) was dissected to prepare probe A (lane 1) and purified dsrna1 (lane 2); RNAs were extracted from the upper component of RGSV preparation after sucrose density gradient centrifugation (lanes 3, 4 and 5) and from RGSVinfected rice tissue (lanes 6, 7 and 8). For each extracted RNA, samples were loaded in increasing amounts, 1, 3 and 10, respectively. S-1 indicates a candidate RNA which may encode the vrna1 18 9K ORF. (b) The 636 bp DNA fragment is in lane 1 and RNAs extracted from the upper component of the RGSV preparation ( 1, lane 2; 3, lane 3) are in lanes 2 and 3. The positions of RNA molecular size markers (kb) are indicated on the right. Northern blot analysis using specific probes. As shown in Fig. 3, a discrete RNA band was detected by probe A in RNAs of both the upper component of the RGSV preparation and RGSV-infected rice tissue. This RNA, of approximately 650 bases, may encode the 18 9K ORF present in the 5 -proximal sequence of RNA1. Detection of the small RNA (S-1) suggests that this 18 9K ORF on vrna1 is functional. The RNA seemed to be encapsidated rather than mrna, since it was extracted from the RGSV preparation separated by sucrose density gradient centrifugation. A discrete signal was not found with probe B, although a distinct reaction was found with RGSV RNA1 itself and some smear was visible in RNA extracted from the upper component of the RGSV preparation (data not shown). The presence of the ORF encoding the 18 9K protein at the 5 end indicated that the RGSV RNA1 has an ambisense coding strategy. A short intergenic region was present between the two ORFs encoding the predicted 18 9K and 339 1K proteins. So far, an ambisense coding strategy has been reported in L RNAs of lymphocytic choriomeningitis virus (LCMV) (Salvato & Shimomaye, 1989) and Lassa fever virus (Djavani et al., 1997), which belong to the Arenaviridae, but has not been observed in the polymerase-encoding segments of other negative-stranded RNA viruses. The 11 kda protein encoded in vrna L of LCMV has a zinc finger motif, which is believed to be an RNA-binding protein with a regulatory role. We were unable to observe such a zinc finger motif or other distinct motifs in the RGSV 18 9K protein. No protein with a similar amino acid sequence was found in the database homology search. A database search for the nucleotide sequence and the predicted 339 1K protein encoded by RGSV RNA1 revealed a distinct similarity in the nucleotide and amino acid sequences of RSV and phleboviruses of the family Bunyaviridae. Fig. 4 compares the putative RNA polymerases encoded by RGSV RNA1, RSV RNA1 and the L RNAs of other related viruses. Between RGSV and RSV, a comparison of 2140 amino acids revealed an amino acid identity of 37 9% and a similarity of 74 4%. The area within that region that showed the highest similarity was the amino acid residues from the N- terminal region, which shared 54 0% identity. The RGSV amino acid sequence of this region showed 36% identity with those of phleboviruses [Uukuniemi virus (UUKV), Rift Valley fever virus (RVFV) and Toscana virus (TOSV)] (Fig. 4). The compared region contains the major RNA-dependent RNA polymerase module, motifs A E (Poch et al., 1989) and motif pre-a proposed by Mu ller et al. (1994); the amino acids in these motifs are highly conserved among RGSV, RSV and UUKV. A low but distinct similarity was also observed between the L proteins of hantaviruses Hantaan virus (HTNV) and Seoul virus (SEOV) of the family Bunyaviridae and Tacaribe virus (TACV) of the family Arenaviridae. On the other hand, no similarity was found between RGSV and the tospoviruses tomato spotted wilt virus (TSWV) and impatiens necrotic spot virus (INSV), or the bunyaviruses La Crosse virus and Bunyamwera virus (BUNV). As shown above, the putative RNA polymerases of RGSV and RSV showed strong similarity, and the amino acid sequence of this region (shaded region of RSV polymerase box in Fig. 4) was almost the same size as the entire RNA polymerase molecules of phleboviruses and other related viruses. However, the similarity between the polymerase amino acids of RGSV and RSV at their termini (approximately 500 N-terminal and 300 C-terminal amino acids, shown as unshaded areas on the RSV RNA polymerase box) was only about 20%. The nucleotide sequences of these extra terminal CAFE

5 Complete sequence of rice grassy stunt virus genome RGSV RSV UUKV RVFV TOSV HTNV SEOV TACV Fig. 4. Amino acid sequence homology between the putative RNA polymerases of RGSV, RSV, UUKV, RVFV, TOSV, HTNV, SEOV and TACV. Database accession numbers are as follows: RSV, D31879; UUKV, D10759; RVFV, X56464; TOSV, X68414; HTNV, D25531; SEOV, X56492; and TACV, J Sequences that showed identity and/or similarity with those of RGSV were aligned with the corresponding amino acid sequence of the 339 1K protein of RGSV and shaded. RNA polymerase motifs (pre-a, A E) are depicted as vertical bars in the RGSV polymerase box. regions might be different in origin or have been introduced by recombination (Goldbach & de Haan, 1994). Another prominent feature of the 339 1K protein was a short but distinct Arg Gly (R G)-rich domain (underlined), which is most likely to be an RNA-binding motif at the C terminus: PDSQFPTYNPPSSRGRGRRGRGRSYMF- (C terminus). The Arg Gly-rich sequences of RGSV looked like a variation of the RGG box, which is defined as consisting of closely spaced Arg-Gly-Gly (RGG) repeats interspersed with other, often aromatic, amino acids that specifically bind RNA (Burd & Dreyfuss, 1994). We found that the short RGGlike domain located in the 339 1K protein of RGSV was not made up of distinct repeats of RGG, but had an Arg Gly-rich domain that was surrounded by Tyr (Y), Phe (F), Ser (S) and Thr (T) residues. Analysis of the secondary peptide structure of the 339 1K protein using the GCG software package (Wisconsin) also showed that the 27 C-terminal amino acid residues had high hydrophilicity and a distinct β-turn stretch, indicating that this part of the protein is exposed on the surface. The C terminus of the RSV RNA polymerase protein lacks such a basic domain but instead is rich in acidic amino acids Asp (D) and Glu (E) (Toriyama et al., 1994). Similar acidic clusters were found in the RGSV 339 1K protein at and residues from the C terminus. Such sequences may be essential for interaction with RNA templates that are encapsidated with nucleocapsid proteins or for the folding of the RNA polymerase itself. The L protein of TSWV (BR-01) has an extremely acidic C terminus, five Asp and five Glu residues in a 15 amino acid stretch, but this region is deleted in the L protein of INSV. It is therefore unlikely that the acidic region is a common essential sequence (van Poelwijk et al., 1997). Thus, the RNA-binding motif in the RNA polymerase gene of RGSV seems to be specific to that virus, as is its ambisense RNA structure. Fig. 5. Amino acid identity between proteins encoded by RGSV vrna2 and vcrna2 ORFs and by RNA2 ORFs from other tenuiviruses over the amino acid sequences indicated. Figures in parentheses indicate percentage similarity. Sequence data for RSV, MStV and RHBV were determined previously (Takahashi et al., 1993; Estabrook et al., 1996; de Miranda et al., 1996). Similarities between RSV, MStV and RHBV are shaded. Sequence of RNA2 The consensus nucleotide sequence of RNA2 comprised 4056 nt and had the following base composition: A, 33 35%; U, 31 9%; C, 18 44%; and G, 16 29%. One ORF was found on vrna2 and another on vcrna2. The vrna2 ORF encoded a protein of 200 amino acids with an M r of (the 23 3K protein), whereas the ORF on vcrna2 encoded a protein of 822 amino acids with an M r of (the 93 9K protein) (Fig. 2). Between these two ORFs was a long intergenic region of 568 nt (nt from the 5 end) that was rich in oligo(a). A database search with the proteins encoded by RGSV CAFF

6 S. Toriyama and others Table 1. Coding capacity of the rice grassy stunt virus genome Abbreviations used in the table are as follows: NCR, noncoding region; IR, intergenic region between two ORFs; Pol, RNA polymerase; N, nucleocapsid; and NS, major nonstructural protein. No. of nucleotides RNA segment Total 5 NCR ORF on vrna* IR ORF on vcrna* 3 NCR (18 9) (339 1; Pol) (23 3) (93 9) (22 9) (30 9) (19 4) (60 4) (21 6) (35 9; N) (20 6; NS) (36 4) 79 Total * Numbers in parentheses are the molecular masses of the genome products in kda. RNA2 revealed a similarity to the proteins encoded by RNA2 of tenuiviruses (Takahashi et al., 1993; de Miranda et al., 1996; Estabrook et al., 1996) (Fig. 5). The putative 23 3K protein showed a weak but significant similarity (22 30% identity and about 65% similarity) with the nonstructural proteins of other tenuiviruses. On the other hand, RSV, MStV and RHBV showed high identity (40 61%) and similarity (82 91%). The putative 93 9K protein on vcrna2 showed about 21% identity and 67% similarity to the putative proteins encoded by vcrna2 of other tenuiviruses, between which identity was 42 53% and similarity was 82 87% (Fig. 5). The putative 93 9K protein was also weakly similar to the virion membrane glycoproteins of UUKV (Ro nnholm & Pettersson, 1987), RVFV (Collett et al., 1985) and Punta Toro virus (PUTV) (Ihara et al., 1985) of the genus Phlebovirus, with identities of 19% (over 237 amino acid residues), 16% (213 amino acids) and 19% (228 amino acids), respectively; these viruses and RGSV shared degrees of similarity that were almost the same as those observed between these viruses and the other tenuiviruses. Sequences of RNA3 and RNA4 The complete nucleotide sequences of RNA segments 3 and 4 comprised 3123 and 2915 nt, respectively. Like the other RGSV segments, each had an ORF on the vrna and one on the vcrna (Fig. 2). The intergenic region between the two ORFs of RNA3 was extremely long (1382 nt). We therefore confirmed the length of the intergenic region through repeated sequencing. The 5 half of the intergenic region of RNA3 contained several oligo(a) and 11 AU repeats, but no obvious oligo(a) sequence. The intergenic region of RNA4 contained two oligo(g) and two oligo(u). Database searches of the putative product of the 22 9K ORF of vrna3 revealed a weak similarity, 25 7% identity over 144 amino acids, with the predicted product (the 21 6K protein) encoded by vrna5 of RGSV (Toriyama et al., 1997). No other significant matches were found with sequences of the putative products encoded by vrna3, vrna4 and vcrna4. Some very weak similarities were observed in database searches, but overall these products did not display a consistent similarity to any proteins of the other tenuiviruses. The similarity observed between the 22 9K protein of vrna3 and the 21 6K protein of vrna5 suggests that RNA3, or even RNA4, may have been produced partly by recombination with some other RGSV RNA segments and or unknown RNAs (Goldbach & de Haan, 1994). However, we have no experimental data to indicate either a role or an origin for these RNA segments. RGSV genome sequence The entire nucleotide sequence of the RGSV genome, six RNA segments, is summarized in Table 1. The total number of nucleotides in the genome was and all segments had an ambisense coding strategy. Terminal nucleotide sequences (Fig. 1) were conserved with the other tenuiviruses and had the capacity to form secondary panhandle-like structures. All these molecular characteristics are shared with the tenuiviruses. RNA segments 1, 2, 5 and 6 of RGSV were larger, in terms of nucleotides, than those of the corresponding RNA segments of the other tenuiviruses, yet each of the ORFs was almost the same in size. The size difference for segments 2, 5 and 6 resulted from their longer noncoding regions, but for RNA1 it was the result of the second ORF on the vrna1 sequence. All or part of the nucleotide sequence in the intergenic regions of RNA segments 1 4 could fold into secondary structures, but no distinct hairpin structure was found. Nucleocapsid and major nonstructural proteins of RGSV are only distantly related to the corresponding proteins of other tenuiviruses. RGSV has six genomic RNA segments, in contrast to the other tenuiviruses which have four or five RNA segments and which share highly conserved nucleotide CAFG

7 Complete sequence of rice grassy stunt virus genome Fig. 6. The most parsimonious tree of RGSV, tenuiviruses (RSV, MStV and RHBV), phleboviruses (UUKV, RVFV and PUTV) and TSWV. BUNV was used as an outgroup virus. Input data were the combined matrix of nucleocapsid proteins and virion membrane glycoproteins or membrane glycoprotein-like proteins, each of which was aligned using CLUSTAL V. Gaps were treated as new character states in tree search. Figures beside internal branches indicate bootstrap probabilities (P 50% only) for 1000 replications. Accession numbers for sequence data of the nucleocapsids are as follows: RSV, X53563; MStV, M57426; RHBV, L07940; RGSV, AB000403; UUKV, 33551; RVFV, X53771; PUTV, K02736; TSWV, D00645; and BUNV, D Accession numbers for sequence data of the virion membrane glycoproteins and related proteins are as follows: RSV, D13176; MStV, U53224; RHBV, L54073; RGSV, AB010376; UUKV, M17417; RVFV, M11157; PUTV, M11156; TSWV, S48091; and BUNV, M sequences. Previously, we suggested that RGSV should be classified into a new genus (Toriyama et al., 1997). As described above, the predicted proteins encoded by RNA2 were similar to the proteins encoded by RNA2 of the other tenuiviruses. The level of sequence similarity was very low, as observed between RGSV RNA segments 5 and 6 and RNAs 3 and 4 of other tenuiviruses. Phylogenetic analysis of the nucleocapsid protein and the 94 kda virion membrane glycoprotein-like protein of RGSV, RSV, MStV and RHBV indicated that RGSV is a monophyletic group and apparently apart from the other tenuiviruses (Fig. 6). The phylogenetic tree indicated that these tenuiviruses, together with RGSV, form a sister group to the phleboviruses, as first observed in sequence comparisons between RSV and phleboviruses (Kakutani et al., 1991; Takahashi et al., 1993). On the other hand, results presented here indicate that the RNA polymerase gene (RNA1) at least is well-conserved between RGSV and RSV, the prototype of the genus Tenuivirus. Zanotto et al. (1996) noted that the overall sequence similarity of the RNA-dependent RNA polymerases is insufficient to support their use as phylogenetic markers for many RNA viruses, particularly those with low sequence similarity. The sequence similarity of the RNA polymerase genes of RGSV and RSV seems to be high (Fig. 4); the RNAdependent RNA polymerase encoded by RNA1 is wellconserved between RGSV and RSV with 37 9% amino acid identity over 2140 amino acid residues. The putative RNAdependent RNA polymerase gene, L RNA, is the most conserved RNA segment between the tospoviruses TSWV and INSV, whereas the S RNA is the least conserved. The level of amino acid identity of the nucleocapsid protein varies between 26 and 81% in the genus Tospovirus (de A vila et al., 1993; Satyanarayana et al., 1996; van Poelwijk et al., 1997). The example of tospoviruses suggests that RGSV should not be too readily classified as a new genus. However, RGSV RNA1 has an additional ORF for a putative 18 9K protein on the vrna1 sequence. Furthermore, the presence of two additional RGSV RNA segments, RNAs 3 and 4, might be a decisive factor in whether RGSV should be regarded as a new genus distinct from other tenuiviruses. In RSV, Ishikawa et al. (1989) found the fifth RNA segment (S5; M r ) in the top component after sucrose density centrifugation; it was abundant in circular filaments 290 nm in length and encapsidated by RSV nucleocapsid. Recently, S5 was shown to be a subgenomic RNA with the part of the vrna4 sequence, but lacking the 3 half sequence (Ishikawa & Omura, 1996). In the present study, RGSV RNA segments 3 and 4 were revealed to be independent molecules and not subgenomic RNA. All these findings together suggest that although RGSV is most closely related to the tenuiviruses, it should probably be placed in a taxonomically separate genus in the future. The biological properties of RGSV differ from those of other tenuiviruses (Toriyama et al., 1997). Furthermore, the brown planthopper, N. lugens (Sta l), a major natural vector of RGSV, is able to hibernate only in tropical and subtropical zones. The rice plant, Oryza sativa, is the only substrate of brown planthoppers which cannot survive in other grasses or in temperate zones where the rice plant does not overwinter (Kisimoto, 1981). Thus, RGSV has survived and evolved in an extremely limited niche comprising just two hosts, the rice plant, O. sativa, and the brown planthopper, N. lugens. Its narrow host specificity may be part of the reason why RGSV has evolved in the tropical zone in a way that is different from the other tenuiviruses. The host specificity of other tenuiviruses is broader, enabling them to extend their niche and become widely distributed, particularly in rice-growing areas of the world. Tenuiviruses have been shown to be more closely related to phleboviruses than to tospoviruses (Toriyama et al., 1994); this was confirmed by our phylogenetic tree (Fig. 6). As suggested by van Poelwijk et al. (1997), tospoviruses and tenuiviruses might have descended from the animal-infecting Bunyaviridae by two independent evolutionary pathways. Tospoviruses and tenuiviruses have survived in completely different niches, the former in thrips dicotyledons and the latter in planthoppers monocotyledons. This is a case in which insects have played in crucial role in virus evolution, as pointed out by Goldbach (1986). We would like to thank H. Koganezawa (formerly of IRRI) and K. Ishikawa (National Agriculture Research Center) for the gift of RGSV materials, and also N. Watanabe (Institute of Physical and Chemical Research) for analysis of the secondary structure of the RNA polymerase CAFH

8 S. Toriyama and others proteins. This work was supported in part by grants-in-aid (to S. T.) from the Ministry of Agriculture, Forestry and Fisheries [Biocosmos Program 98-I-C-(1)]. References Burd, C. G. & Dreyfuss, G. (1994). Conserved structures and diversity of functions of RNA-binding proteins. Science 265, Collett, M. S., Purchio, A. F., Keegan, K., Frazier, S., Hays, W., Anderson, D. K., Parker, M. D., Schmaljohn, C. S., Schmidt, J. & Dalrymple, J. M. (1985). Complete nucleotide sequence of the M RNA segment of Rift Valley fever virus. Virology 144, de A vila, A. C., de Haan, P., Kormelink, R., Resende, R. de O., Goldbach, R. W. & Peters, D. (1993). Classification of tospoviruses based on phylogeny of nucleoprotein gene sequences. Journal of General Virology 74, de Miranda, J. R., Mun oz, M., Wu, R., Hull, R. & Espinoza, A. M. (1996). Sequence of rice hoja blanca tenuivirus RNA-2. Virus Genes 12, Djavani, M., Lukashevich, I. S., Sanchez, A., Nichol, S. T. & Salvato, M. S. (1997). Completion of the Lassa fever virus sequence and identification of a RING finger open reading frame at the L RNA 5 end. Virology 235, Estabrook, E. M., Suyenaga, K., Tsai, J. H. & Falk, B. W. (1996). Maize stripe tenuivirus RNA2 transcripts in plant and insect hosts and analysis of pvc2, a protein similar to the Phlebovirus virion membrane glycoproteins. Virus Genes 12, Felsenstein, J. (1993). PHYLIP: phylogeny inference package, version 5.3c. Department of Genetics, The University of Washington, Seattle, WA, USA. Goldbach, R. W. (1986). Molecular evolution of plant RNA viruses. Annual Review of Phytopathology 24, Goldbach, R. & de Haan, P. (1994). RNA viral supergroups and the evolution of RNA viruses. In The Evolutionary Biology of Viruses, pp Edited by S. S. Morse. New York: Lippincott Raven. Hibino, H. (1986). Rice grassy stunt virus. In CMI AAB Descriptions of Plant Viruses, no Higgins, D. G., Bleasby, A. J. & Fuchs, R. (1992). CLUSTAL V: improved software for multiple sequence alignment. Computer Applications in the Biosciences 8, Honda, A., Ueda, K., Nagata, K. & Ishihama, A. (1987). Identification of the RNA polymerase-binding site on genome RNA of influenza virus. Journal of Biochemistry 102, Ihara, T., Smith, J., Dalrymple, J. M. & Bishop, D. H. L. (1985). Complete sequences of the glycoproteins and M RNA of Punta Toro phlebovirus compared to those of Rift Valley fever virus. Virology 144, Ishikawa, K. & Omura, T. (1996). Characteristics of segment 5 of rice stripe virus which encodes the non-capsid protein (NCP) (abstract). Annals of the Phytopathological Society of Japan 62, 340. Ishikawa, K., Omura, T. & Hibino, H. (1989). Morphological characteristics of rice stripe virus. Journal of General Virology 70, Kakutani, T., Hayano, Y., Hayashi, T. & Minobe, Y. (1991). Ambisense segment 3 of rice stripe virus: the first instance of a virus containing two ambisense segments. Journal of General Virology 72, Kisimoto, R. (1981). Development, behaviour, population dynamics and control of the brown planthopper, Nilaparvata lugens Sta l. Review of Plant Protection Research 14, Mu ller, R., Poch, O., Delarue, M., Bishop, D. H. L. & Bouloy, M. (1994). Rift Valley fever virus L segment: correction of the sequence and possible functional role of newly identified regions conserved in RNA-dependent polymerases. Journal of General Virology 75, Murphy, F. A., Fauquet, C. M., Bishop, D. H. L., Ghabrial, S. A., Jarvis, A. W., Martelli, G. P., Mayo, M. A. & Summers, M. D. (editors) (1995). Virus Taxonomy. Sixth Report of the International Committee on Taxonomy of Viruses, pp Vienna & New York: Springer Verlag. Poch, O., Sauvaget, I., Delarue, M. & Tordo, N. (1989). Identification of four conserved motifs among the RNA-dependent polymerase encoding elements. EMBO Journal 8, Ramirez, B.-C. & Haenni, A.-L. (1994). Molecular biology of tenuiviruses, a remarkable group of plant viruses. Journal of General Virology 75, Ro nnholm, R. & Pettersson, R. F. (1987). Complete nucleotide sequence of the M RNA segment of Uukuniemi virus encoding the membrane glycoproteins G1 and G2. Virology 160, Salvato, M. S. & Shimomaye, E. M. (1989). The completed sequence of lymphocytic choriomeningitis virus reveals a unique RNA structure and a gene for a zinc finger protein. Virology 173, Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory. Satyanarayana, T., Michell, S. E., Reddy, D. V. R., Brown, S., Kresovich, S., Jarret, R., Naidu, R. A. & Demski, J. W. (1996). Peanut bud necrosis tospovirus S RNA: complete nucleotide sequence, genome organization and homology to other tospoviruses. Archives of Virology 141, Swofford, D. L. (1993). PAUP: Phylogenetic analysis using parsimony. Version 3.1. Computer program. Smithsonian Institute, Washington, DC, USA. Takahashi, M., Toriyama, S., Kikuchi, Y., Hayakawa, T. & Ishihama, A. (1990). Complementarity between the 5 - and 3 -terminal sequences of rice stripe virus RNAs. Journal of General Virology 71, Takahashi, M., Toriyama, S., Hamamatsu, C. & Ishihama, A. (1993). Nucleotide sequence and possible ambisense coding strategy of rice stripe virus RNA segment 2. Journal of General Virology 74, Toriyama, S. (1986). An RNA-dependent RNA polymerase associated with the filamentous nucleoproteins of rice stripe virus. Journal of General Virology 67, Toriyama, S. (1987). Ribonucleic acid polymerase activity in filamentous nucleoproteins of rice grassy stunt virus. Journal of General Virology 68, Toriyama, S., Takahashi, M., Sano, Y., Shimizu, T. & Ishihama, A. (1994). Nucleotide sequence of RNA1, the largest genomic segment of rice stripe virus, the prototype of the tenuiviruses. Journal of General Virology 75, Toriyama, S., Kimishima, T. & Takahashi, M. (1997). The proteins encoded by rice grassy stunt virus RNA5 and RNA6 are only distantly related to the corresponding proteins of other members of the genus Tenuivirus. Journal of General Virology 78, van Poelwijk, F., Prins, M. & Goldbach, R. (1997). Completion of the impatiens necrotic spot virus genome sequence and genetic comparison of the L proteins within the family Bunyaviridae. Journal of General Virology 78, Zanotto, P. M. de A., Gibbs, M. J., Gould, E. A. & Holmes, E. C. (1996). A reevaluation of the higher taxonomy of viruses based on RNA polymerases. Journal of Virology 70, Received 12 March 1998; Accepted 17 April 1998 CAFI