Genes Required for Replication of the 15.5-Kilobase RNA Genome of a Plant Closterovirus

Abstract

A full-length cDNA clone of beet yellows closterovirus (BYV) was engineered and used to map functions involved in the replication of the viral RNA genome and subgenomic RNA formation. Among 10 open reading frames (ORFs) present in BYV, ORFs 1a and 1b suffice for RNA replication and transcription. The proteins encoded in these ORFs harbor putative methyltransferase, RNA helicase, and RNA polymerase domains common to Sindbis virus-like viruses and a large interdomain region that is unique to closteroviruses. The papain-like leader proteinase (L-Pro) encoded in the 5′-proximal region of ORF 1a was found to have a dual function in genome amplification. First, the autocatalytic cleavage between L-Pro and the remainder of the ORF 1a product was essential for replication of RNA. Second, an additional L-Pro function that was separable from proteolytic activity was required for efficient RNA accumulation. The deletion of a large, ∼5.6-kb, 3′-terminal region coding for a 6-kDa hydrophobic protein, an HSP70 homolog, a 64-kDa protein, minor and major capsid proteins, a 20-kDa protein, and a 21-kDa protein (p21) resulted in replication-competent RNA. However, examination of mutants with replacements of start codons in each of these seven 3′-terminal ORFs revealed that p21 functions as an enhancer of genome amplification. The intriguing analogies between the genome organization and replicational requirements of plant closteroviruses and animal coronavirus-like viruses are discussed.

The closterovirus family belongs to the Sindbis virus-like supergroup of positive-strand RNA viruses (35), representing filamentous plant viruses that possess the largest genomes of this supergroup (from ∼15 to 20 kb). In addition, closteroviruses exhibit striking similarities in genome organization to the phylogenetically remote animal coronavirus-like viruses. It was suggested that these similarities reflect parallel evolution toward large RNA genomes (17).

The available information concerning the functions of closterovirus proteins was inferred mainly from computer-assisted analysis. Nine open reading frames (ORFs) were found in the ∼15.5-kb genome of beet yellows virus (BYV), a prototype closterovirus (2–4). The 5′-terminal part of the BYV RNA harbors ORFs 1a and 1b, encoding papain-like leader proteinase (L-Pro), putative methyltransferase, RNA helicase, and RNA polymerase domains (Fig. 1A). Among the Sindbis virus-like viruses, closteroviruses are distinguished by a large interdomain region present in ORF 1a and by the apparent involvement of a +1 translational frameshift for expression of the ORF 1b product (4).

FIG. 1.

(A) Gene organization of BYV and summary of the introduced mutations. L-Pro, papain-like leader proteinase (the arrow shows the site of L-Pro-mediated cleavage); MET and HEL, putative methyltransferase and RNA helicase domains in the ORF 1a product; POL, putative RNA polymerase; HSP70h, an HSP70 homolog; CPm and CP, the minor and major capsid proteins, respectively; p6, p64, p20, and p21, the 6-, 64-, 20-, and 21-kDa products of ORFs 2, 4, 7, and 8, respectively. Asterisks denote replacement of the start codons in each of ORFs 2 to 8; fs, frameshift mutation. AN, SB, and SS, expanded deletions in the 3′-terminal part of the BYV genome. (B) Diagrammatic representation of the full-length BYV-Cal cDNA flanked by the SP6 RNA polymerase promoter in pBYV-NA. The most 5′-terminal and 3′-terminal trinucleotides in the BYV cDNA are shown. The SmaI restriction endonuclease site engineered downstream from the BYV cDNA (boldface) and selected sites used for cloning and mutagenesis are indicated above the diagram, together with their positions in the genome (kilobases). Four BYV cDNA fragments used to generate pBYV-NA are shown as double lines above the diagram. The boundaries of the five cDNA subclones engineered for site-directed mutagenesis are marked below the diagram.

The 3′-terminal part of the BYV genome encompasses ORFs 2 through 8, coding for a 6-kDa hydrophobic protein (p6), a homolog of cellular molecular chaperones from the HSP70 family (HSP70h), a 64-kDa protein (p64), two capsid proteins, a 20-kDa protein (p20), and a 21-kDa protein (p21; Fig. 1A) (2, 3, 7). Unlike all of the other elongated plant viruses assembled from a single type of capsid protein, closterovirus particles possess major and minor capsid proteins that form a long “body” and a short “tail” (CP and CPm, respectively; reference 5).

All ORFs located in the 3′-terminal parts of the related genomes of BYV and citrus tristeza closterovirus (CTV) are expressed via formation of a nested set of subgenomic RNAs (sgRNAs) (14, 22, 24). The levels and time courses of accumulation of different sgRNAs are controlled at the transcriptional level (41). The mechanism of closterovirus transcription appears to be similar to that of other Sindbis virus-like viruses (1, 20, 23, 29, 39, 40) and does not involve the leader priming that was found in coronavirus-like viruses (25, 37, 38, 50, 51).

In this study, we engineered a full-length 15.5-kb cDNA clone of BYV and demonstrated that the corresponding RNA transcripts were infectious after transfection into plant protoplasts. Site-directed mutagenesis was employed to reveal that the products of BYV ORFs 1a and 1b are involved in RNA replication. Among these products, L-Pro, encoded in the 5′-proximal region of ORF 1a, was required for efficient genome amplification, whereas products encoded by the rest of ORF 1a and by ORF 1b were essential for a basal level of RNA replication. In addition, p21, a product of the 3′-most BYV ORF, was identified as an enhancer of RNA accumulation.

MATERIALS AND METHODS

Nucleotide sequencing of the Californian isolate of BYV.

The Californian isolate of BYV (BYV-Cal) was obtained from Bryce W. Falk (University of California at Davis) and propagated on Tetragonia expansa plants. The virions and RNA were isolated as previously described (26). The double-stranded RNA (dsRNA) was purified by using CF11 cellulose (48). To determine the nucleotide sequences of the 3′ extremities of the minus and plus strands of BYV-Cal RNA, the virion RNA and BYV-specific dsRNA were polyadenylated in vitro by using yeast poly(A) polymerase in accordance with the manufacturer’s (U.S. Biochemical Corp.) protocol. The 911-nucleotide-long 5′-terminal region of the polyadenylated minus strand of the dsRNA was amplified via reverse transcription (RT)-PCR after denaturation by 20 mM methylmercuric hydroxide (Serva; reference 47). The RT and PCR were carried out by using SuperScript II reverse transcriptase (Gibco-BRL), Taq DNA polymerase, and Taq Extender (Stratagene) in accordance with the manufacturers’ protocols. The primers used for RT-PCR, dTSac (5′-GGTGAGCTC[T]₁₈) and 900Sph (5′-GTTGCATGCTTTATTTATCTTCCG), contained SacI and SphI sites for subsequent cloning of the amplified products into vector plasmid pGEM-4 (Promega). A similar protocol was used to clone the 440-nucleotide-long 3′-terminal fragment of polyadenylated virion RNA. The primers used were 500Sph (5′-GTGCATGCGTGGTAACTCCTAT) and dTSac. The cloned inserts were sequenced by the dye termination method using SP6 and T7 primers and an ABI Prism model 377 automated sequencer. The nucleotide sequence of the remaining part of the BYV-Cal genome was determined in a stepwise manner by using a full-length cDNA clone (see below) and sets of plus- and minus-strand oligonucleotide primers designed on the basis of previously determined fragments of the BYV-Cal nucleotide sequence.

Generation of a BYV genomic cDNA clone.

The first variant of the full-length BYV cDNA clone was assembled from three fragments that were amplified by RT-PCR with virion RNA of BYV-Cal as the template. These three fragments were delimited by the 5′ end, restriction endonuclease SacI and BamHI sites, and the 3′ end as shown in Fig. 1B. The 5′-terminal and 3′-terminal PCR products were inserted into pGEM-4-based plasmid pTL7SN (42) as EcoRI-SacI and BamHI-SmaI fragments, respectively. The EcoRI and SmaI sites were added to amplification products by inclusion into PCR primers, whereas the SacI and BamHI sites were naturally present in BYV-Cal cDNA. The unique SmaI was introduced downstream from the 3′ end of the viral cDNA to permit linearization of the plasmid prior to transcription (Fig. 1B). The junction between the SP6 RNA polymerase promoter and the 5′ terminus of the cDNA was repaired by site-directed mutagenesis (36) to ensure that the 5′ end of the corresponding in vitro transcripts would be identical to that of BYV RNA (Fig. 1B). Finally, the central part was inserted between the SacI and BamHI sites to form the full-length BYV clone. This strategy was similar to that used successfully to generate a full-length cDNA clone of tobacco etch virus (15).

The RNA transcripts derived from several tested RT-PCR full-length clones produced no infectivity in protoplast transfection experiments (data not shown), possibly due to errors introduced by PCR. Because of that, we subsequently replaced all but the 5′-terminal 840 nucleotides (position of the BglII site, Fig. 1B) of the original clone with DNA obtained via conventional cDNA cloning (the Gibco-BRL protocol for SuperScript II reverse transcriptase was used). Four partially overlapping fragments of double-stranded cDNA used for that purpose and delimited by BglII, EagI, SnaBI, NdeI, and SmaI sites are shown in Fig. 1B.

Eight individual full-length BYV cDNA clones of the second generation (with various combinations of cDNA fragments) were tested via transfection of corresponding RNA transcripts into tobacco protoplasts. Two of these transcripts were infectious, exhibiting similar levels of RNA accumulation (data not shown). Since the authentic 5′ end of BYV cDNA (5′-GTTT) is suboptimal for SP6 RNA polymerase, it was modified to the more optimal 5′-GAAT sequence in a clone designated pBYV-NA. It was found that the yield of corresponding in vitro RNA transcripts was substantially increased without affecting the specific infectivity of the transcripts, which was roughly similar to the infectivity of the virion RNA (data not shown). This apparently high infectivity of transcripts, compared to that of virion RNA, could be due to the relatively low infectivity of the latter. Even the best preparations of the virion RNA contain a substantial proportion of degradation products due to the fragility of long virions and RNA.

Mutagenesis of the BYV genome.

Since site-directed mutagenesis (36) of the ∼18.5-kb plasmid harboring ∼15.5 kb of the viral cDNA was impractical, five plasmids that contained different regions of the BYV genome were generated by subcloning of the cDNA fragments into pGEM-4 (Fig. 1B). Among those, p5′BYV harbored the genome region between nucleotides 1 and 3,472 (position of the SacI site), p65M contained the region between nucleotides 8,472 and 11,694 (delimited by two EcoRI sites), pSB-C was cloned in the opposite orientation and delimited by the SphI site at position 10,568 and the BamHI site at position 13,392, pNB contained the region from the NdeI site at position 11,309 to the BamHI site, and p3′-BYV harbored a fragment extending from the BamHI site to the SmaI site (Fig. 1B).

The p5′BYV clone was used to introduce mutations into the L-Pro-encoding region. The mutation designated dcl (for deletion of cleavage site) resulted in in-frame replacement of nucleotides 1,848 through 1,875 with the unrelated heptanucleotide GAGATCT harboring a BglII site. The replaced sequence included two glycine codons that form a scissile bond between L-Pro and the rest of the ORF 1a product (4). Four additional L-Pro mutations introduced via “loop-out” mutagenesis represented in-frame deletions of regions coding for putative structural elements of L-Pro, predicted by using the PHD program (EMBL, Heidelberg, Germany). The 1DL deletion removed BYV nucleotides 270 through 345, specifying a putative proline-rich, 26-residue-long loop. The 2DL deletion removed BYV nucleotides 493 through 564, corresponding to a 24-residue-long region containing a short α-helix. The 3DL deletion removed BYV nucleotides 937 to 1,007, specifying a putative 24-residue-long region with high solvent accessibility. The 4DL deletion removed BYV nucleotides 1,207 through 1,301, specifying a 32-residue-long region harboring a short α-helix (see Fig. 3A). In each case, the deleted region was replaced with a hexanucleotide comprising a BglII site.

FIG. 3.

Mutagenic analysis of L-Pro function in BYV RNA replication. (A) L-Pro coding region with a summary of the introduced deletions and the BglII site used to generate the 1-B and B-4 mutants. (B) Northern analysis of plus-strand RNA. (C) Northern analysis of minus-strand RNA. 1DL to 4DL, four relatively small (72 to 96 nucleotides) in-frame deletions introduced into the L-Pro-encoding region; dcl, deletion of nine codons, including two glycine codons that specify an L-Pro scissile bond; 1-4, 1-B, and B-4, expanded in-frame deletions in the L-Pro-encoding region outside of the proteinase domain; WT, wild type; fs, frameshift mutation.

Larger in-frame deletions were introduced into the L-Pro coding region by removing DNA fragments between the engineered BglII sites and a natural BglII site located at nucleotide 841. In the 1-B, B-4, and 1-4 mutants, regions located between nucleotides 270 and 841, 841 and 1,301, and 270 and 1,301, respectively, were deleted (see Fig. 3A). The DNA fragments from mutant derivatives of p5′BYV were cloned into the full-length clone by using two unique restriction sites, NheI (located in the vector part of the plasmid) and EagI (located at nucleotide 2,152).

The frameshift (fs) mutation upstream of the putative RNA helicase domain encoded by ORF 1a was introduced by filling (47) a unique AvrII restriction site located at nucleotide 6,420 (Fig. 1A). One large deletion (SS) in the 3′ half of the BYV cDNA was generated by removal of the ∼4,300-nucleotide fragment of pBYV-NA located between the unique SnaBI and StuI restriction sites. Another deletion (SB) spanned an ∼4,750-nucleotide region located between the SphI and BstEII sites (Fig. 1A). To generate a third large deletion, a unique ApaI site was engineered in p65M downstream from the ORF 2 start codon by site-directed mutagenesis. The mutant fragment was cloned into the full-length clone by using the SnaBI and NdeI sites, and a ∼5,600-nucleotide region between the ApaI and NsiI sites was deleted to form the AN mutant (Fig. 1A).

A series of start codon replacement mutants was generated by changing the first ATG codon of each of ORFs 2 through 8 by using subclones p65M, pSB-C, pNB, and p3′BYV (Fig. 1B). p65M was utilized to replace start codons in ORFs 2 and 3 with ATA. The start codon of ORF 4 was replaced with ACG by using pSB-C. Analogous mutation of the ORF 5 start codon was done by using pNB. Finally, start codons in ORFs 6, 7, and 8 were changed to ATA, GTA, and ACG, respectively, by using p3′BYV. Since ORFs 3 and 4, 4 and 5, 6 and 7, and 7 and 8 overlap, the start codon replacements were designed so as not to result in any amino acid changes in the products encoded in the alternative ORFs. The introduced mutations were verified by nucleotide sequencing. The sequences of the numerous primers used for engineering and mutagenesis of the BYV clone are available upon request.

Analyses of BYV mutants.

The genome size RNA transcripts were generated under capping conditions by using SmaI-linearized plasmids and SP6 RNA polymerase (Ambion; 16). These transcripts were transfected via electroporation into protoplasts obtained from a suspension culture of the Nicotiana tabacum cv. Xanthi nc cell line developed by D. Facciotti (Calgene, Inc.) and referred to as the DF line (18). Protoplasts were harvested at 86 h posttransfection, and the RNA samples were isolated by using TRIZOL reagent (Gibco-BRL). Northern analysis was conducted by using a Zeta-Probe membrane (Bio-Rad). RNA transfer from a 1% agarose gel onto a membrane was in 10× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) in accordance with reference 6a, whereas prehybridization and hybridization were conducted by using NorthernMax buffer (Ambion) at 50°C. ³²P-labeled positive- or negative-polarity single-stranded RNA probes were prepared by in vitro transcription using SP6 or T7 RNA polymerase (the T7 promoter is located downstream from inserts in all pGEM-4 derivatives) and p3′BYV linearized at the BstEII or BamHI site, respectively. The radiolabeled products were detected and quantified by using a PhosphorImager (Molecular Dynamics). At least four independent protoplast transfection experiments were conducted for each mutant; the means and standard deviations were used for comparative analysis of RNA accumulation.

The capped RNA transcripts of p5′BYV derivatives harboring mutations in the L-Pro coding region were generated in vitro by using SP6 RNA polymerase and KpnI-linearized plasmid DNA (the site is located at nucleotide 2,813). In vitro translations were conducted for 1 h by using wheat germ extracts (Promega). The products were labeled with [³⁵S]methionine (Amersham), separated in sodium dodecyl sulfate-containing 12% polyacrylamide gels, and electroblotted onto a PROTRAN nitrocellulose membrane (Schleicher & Schuell). The radioactivity measured in the bands corresponding to unprocessed or processed translation products by using a PhosphorImager was normalized to the number of methionine residues present in each product and used for comparative analysis of the proteolysis efficiency of mutant L-Pro variants.

Nucleotide sequence accession numbers.

The complete nucleotide sequence of the full-length BYV-Cal cDNA from which infectious RNA transcripts can be obtained was deposited into the GenBank (accession no. AF056575) database.

RESULTS

Sequence variability in the BYV genome.

Due to their critical role in RNA replication, we thoroughly characterized the 5′- and 3′-terminal regions of the BYV-Cal genome. The 3′ end of the minus strand of viral dsRNA was polyadenylated in vitro, amplified by RT-PCR, and cloned; the nucleotide sequences of six clones were determined. The inferred sequence of the 5′-terminal noncoding region of the plus strand was identical to that of the Ukrainian isolate (4) in five clones. One clone deviated by having an A instead of a U as the second nucleotide. In addition, all six clones harbored an extra G at the very 3′ end of the minus-strand RNA. The presence of an analogous nontemplate G in genomic minus strands was previously described for several viruses of plants (9, 12, 13), including CTV (27).

Unexpectedly, high sequence variation was found in the 3′-terminal noncoding regions of the two BYV isolates. In place of the very 3′-terminal heptanucleotide GGGGCCC_OH found in Ukrainian BYV (3), BYV-Cal harbored hexanucleotide GGGCCG_OH in 10 sequenced cDNA clones. In addition, the sequence 5′-C(A)₆(U)₅C(U)₅AUAUUAAG(A)₄(U)₅-3′, present in Ukrainian BYV (nucleotides 15,400 to 15,434), was replaced in BYV-Cal with the sequence (A)₆(U)_5–7(A)_7–14 (U)₅, starting at nucleotide 15,401. Among 10 sequenced independent cDNA clones, only 2 were identical in this hypervariable, AU-rich region of the BYV-Cal genome.

Overall, comparison of the Ukrainian and Californian isolates of BYV revealed two large genome regions with very different levels of nucleotide sequence similarity. The regions from the 5′ terminus to the end of ORF 1b (Fig. 1A) were 99.4% identical between the two isolates. In contrast, the 3′-terminal regions harboring ORFs 2 through 8 had 88.8% identity at the nucleotide level.

Identification of BYV genes essential for RNA replication.

The mapping of replication-associated functions in the BYV genome was conducted by site-directed mutagenesis of the full-length cDNA clone of BYV-Cal (Fig. 1). The roles of the putative RNA helicase and RNA polymerase domains in genome amplification were addressed by introduction of a frameshift mutation (fs in Fig. 1A and 2). This mutation introduced the stop codon into ORF 1a upstream from the RNA helicase domain. As expected, Northern hybridization analysis of protoplasts harvested at 3.5 days posttransfection detected essentially no BYV-specific RNA produced by the fs mutant (Fig. 2A). The input RNA transcripts were most likely degraded by that time. In all subsequent experiments, the fs mutant was used as a negative control representing a full-size, replication-deficient RNA transcript.

FIG. 2.

Northern hybridization analysis of the plus-strand (A) and minus-strand (B) BYV-specific RNAs in transfected tobacco protoplasts. The RNA transcripts used for protoplast transfections are indicated at the top of each lane. (−), no RNA added; WT, wild-type transcript; fs, frameshift mutation in ORF 1a; AN, SB, and SS, expanded deletions in the 3′-terminal part of the BYV genome illustrated in Fig. 1A; g, genomic RNA; int, intermediate-size RNA; numbers, sgRNAs that express corresponding BYV ORFs 2 through 8. The estimated sizes of BYV-Cal sgRNAs are as follows: 2/3, 6.1 kb; 4, 4.4 kb; 5, 2.6 kb; 6, 1.8 kb; 7, 1.2 kb; 8, 0.8 kb (22). The asterisks beside panel B show the positions of background bands.

The possible involvement of products encoded in the 3′-terminal part of the genome in RNA amplification was probed via introduction of a large deletion, AN, into a cDNA clone (Fig. 1A and 2). This deletion removed ORFs 2 through 7 and 55% of ORF 8. The ability of the AN mutant to accumulate shorter-than-wild-type RNA indicated that none of the eight 3′-terminal BYV genes is essential for RNA replication (Fig. 2A and B, lanes WT and AN). However, the levels of AN RNA accumulation were only 63% ± 11% and 59% ± 15% of the wild-type levels for plus- and minus-strand RNAs, respectively. These results suggested that the deleted region specifies a protein product(s) or RNA element(s) that enhances the accumulation of BYV RNA in protoplasts.

A second deletion mutant, SB, lacked a region extending from the middle of ORF 3 to nucleotide 15,315, thus reaching into the 3′-noncoding region (Fig. 1A). The ability of mutant SB to replicate in protoplasts was completely blocked (Fig. 2A, lane SB), suggesting that the region between the 3′-terminal boundaries of mutants AN and SB (nucleotides 15,061 to 15,315) is required for RNA amplification and is possibly involved in template recognition by the BYV RNA polymerase.

The third deletion mutant, SS, was also replication deficient (Fig. 2A, lane SS). This result was not unexpected, since the deletion affected ORF 1b, which encodes the RNA polymerase. The 80-amino-acid-long C-terminal region of the RNA polymerase lacking in the SS mutant is conserved among related plant RNA viruses (4). Examination of minus-strand RNA accumulation revealed a pattern similar to that of plus-strand RNAs for all four mutants (Fig. 2B). Since the amount of minus-strand RNA produced during infection is lower than that of the plus strands, longer film exposure resulted in more prominent appearance of the background bands most likely corresponding to plant rRNAs (marked by asterisks in Fig. 2B; reference 14).

Collectively, these experiments indicated that only the products of ORFs 1a and 1b are essential for amplification of BYV RNA, whereas proteins encoded in ORFs 2 to 8 are dispensable for RNA accumulation at the single-cell level.

Requirement of the leader proteinase for efficient RNA amplification.

A series of mutations was designed to test if the L-Pro encoded in the 5′-most part of ORF 1a is involved in RNA amplification. Previous analysis revealed that only the papain-like C-terminal domain of L-Pro, comprising about one-fourth of the molecule, is conserved among closteroviruses (Fig. 3A) (4, 27, 33). Since the remaining about three-fourths of L-Pro exhibited no conserved sequence motifs that could be targeted for mutagenesis, a computer analysis of the protein structure was conducted. Four structurally defined regions outside the proteinase domain were targeted for deletion: a 26-residue-long proline-rich loop in the 1DL mutant, a 24-residue-long hydrophilic element in the 3DL mutant, and 24- and 32-residue-long amino acid stretches containing short α-helical elements in the 2DL and 4DL mutants (Fig. 3A). The fifth deletion removed nine codons overlapping the L-Pro autocatalytic cleavage site (dcl in Fig. 3). Three larger in-frame deletions, 1-B, B-4, and 1-4, removed up to about four-fifths of the variable N-terminal domain of L-Pro (Fig. 3A).

It was found that the dcl mutant was replication incompetent (lane dcl in Fig. 3B), suggesting that the cleavage between L-Pro and the remainder of the ORF 1a product is essential for genome amplification. The 2DL, 3DL, and 4DL mutations had relatively modest effects on RNA accumulation (Fig. 3B and 4), whereas the 1DL mutation resulted in over twofold less RNA accumulation. The larger deletions 1-B, B-4, and 1-4 resulted in levels of RNA amplification reaching 20, 57, and 22% of the wild-type level, respectively (Fig. 3B and 4).

FIG. 4.

Effects of the mutations in the L-Pro coding region on the levels of RNA replication and proteolytic activity of L-Pro. The levels of plus or minus strands of genomic RNA in transfected protoplasts were quantified and compared to the level of proteolytic maturation of L-Pro in vitro for each mutant. Values expressed as percentages of the wild type (wt) are means and standard deviations from at least four independent experiments.

The relative effects of L-Pro mutations on minus-strand RNA accumulation were similar to those observed for the plus strands (Fig. 3C and 4). No specific effect on the formation of sgRNAs was revealed (Fig. 3B). These results suggested that L-Pro affects the general efficiency of RNA amplification and transcription.

Although all but dcl L-Pro mutations were introduced outside the proteinase domain, the replication defects of the mutants could be due to the impaired proteolysis. The proteolytic activity of the L-Pro variants was tested in vitro. The RNA transcripts corresponding to the wild-type and mutant derivatives of the 5′-terminal region of ORF 1a were translated in wheat germ extracts. It was revealed that the dcl mutant directed the formation of the expected unprocessed ∼100-kDa product (data not shown). Approximately 70% of the wild-type translation product was processed in vitro to yield ∼66-kDa L-Pro and an ∼34-kDa product (not shown). The processing efficiencies of mutants 1DL, 2DL, 3DL, and 1B were not significantly different from that of the wild type (Fig. 4). In contrast, mutants 4DL, B-4, and 1-4, in which deletions were located close to the proteinase domain, exhibited markedly lower proteolytic activity (Fig. 4).

However, no direct correlation between processing efficiency and RNA accumulation was found (Fig. 4). This is best exemplified by the 1DL and 1-B mutants, in which proteolytic activity was similar to that of the wild type, whereas RNA levels were only 49 and 19% of the wild-type level, respectively (Fig. 4). Two conclusions can be drawn from the mutagenic study of the L-Pro function. First, proteolytic processing of the ORF 1a product is strictly required for RNA replication. Second, L-Pro has an additional function in RNA amplification that is separable from its proteolytic activity.

Enhancer of RNA accumulation encoded in ORF 8.

The replication competence of the AN mutant lacking the 3′-terminal part of the genome indicated that none of ORFs 2 to 8 is essential for RNA amplification. However, this large deletion resulted in reduced amplification efficiency, despite the smaller size of the RNA. To test possible roles of ORFs 2 to 8 in replication, point mutations replacing start codons were introduced into each of these ORFs (Fig. 1A). The levels of RNA accumulation in protoplasts were similar to the wild-type level for all mutant RNA transcripts but one (Fig. 5A). The corresponding values were 103% ± 10%, 88% ± 17%, 99% ± 9%, 90% ± 10%, 90% ± 24%, 92% ± 11%, and 20% ± 9% for the variants harboring mutations in ORFs 2, 3, 4, 5, 6, 7, and 8, respectively. A fivefold reduction in RNA accumulation observed for the mutant with the replaced start codon in ORF 8 suggested that an important role is played by its product, p21, in RNA amplification (Fig. 5A). Examination of the minus-strand RNA revealed a similar reduction of the RNA level (24% ± 3%; Fig. 5B). The levels of sgRNAs were also decreased proportionally, suggesting that p21 has a general stimulating effect on RNA replication and transcription. Alternatively, p21 could increase RNA levels by protecting RNA from degradation.

FIG. 5.

Characterization of mutants with replaced start codons in ORFs 2 through 8. (A) Northern analysis of plus-strand RNA. (B) Northern analysis of minus-strand RNA. WT and fs, wild-type and frameshift mutant transcripts used for protoplast transfection, respectively. The numbers at the top correspond to mutant ORFs (cf. Fig. 1A). For designations of BYV-specific and background RNA bands marked by asterisks, see the legend to Fig. 2.

The mutant RNA lacking a start codon in CP-encoding ORF 6 failed to produce an RNA, designated int in Fig. 5A, that migrated between genomic RNA and the largest sgRNA. This latter RNA species may be a defective RNA whose formation or stability depends on CP.

DISCUSSION

Previous comparative analysis of the BYV and CTV genomes revealed that proteins encoded by ORFs 1a and 1b possess conserved domains characteristic of the replicases of Sindbis virus-like viruses (4, 17, 27). In contrast to the single-component genomes of BYV and CTV, the genome of lettuce infectious yellows closterovirus is divided among two RNAs (33). RNA 1 harbors ORFs 1a and 1b and an additional ORF that is unique to lettuce infectious yellows closterovirus. RNA 2 codes for homologs of BYV p6, HSP70h, p64, CP, CPm, and a protein with marginal similarity to p21 (33). It was demonstrated recently that RNA 1 can replicate in the absence of RNA 2 but not vice versa (34). This result further supported the involvement of the products of ORFs 1a and 1b in RNA replication, while the possible roles of several additional proteins encoded in one- or two-component closteroviruses remained elusive.

Generation of a full-length BYV clone permitted mapping of the genes involved in the replication and transcription of the 15.5-kb viral RNA. As expected, expression of intact ORFs 1a and 1b, encoding putative methyltransferase, RNA helicase, and RNA polymerase, was found to be essential for RNA replication. Deletion of the cleavage site located between L-Pro and the rest of the ORF 1a product abolished RNA replication, suggesting that autocatalytic release of L-Pro or maintenance of a proper protein structure around the cleavage site is critical for virus viability. In addition to the conserved papain-like proteinase domain, closterovirus L-Pro possesses a variable N-terminal domain (4, 27, 33). A function of this domain was probed by using a series of deletion mutants. Analysis of the mutant phenotypes revealed that the N-terminal domain is required for efficient amplification of the BYV genome. This function was separable from proteinase activity, since some mutants that exhibited no detectable processing defects accumulated reduced levels of RNA. The ability of an L-Pro mutant lacking most of the N-terminal domain to replicate, albeit inefficiently, suggested that the function of this domain in genome amplification is accessory rather than essential.

The possibility cannot be excluded that at least some of the deletions in the L-Pro-encoding region disturbed the RNA elements (e.g., RNA polymerase recognition signals) that are directly involved in the RNA accumulation process. These elements, however, are normally distinct for the plus and minus RNA strands. Since analyzed L-Pro mutations affected the accumulation of both strands to similar extents, it seems more likely that the observed effects were mediated by encoded protein products rather than by the RNA itself.

Interestingly, mutational analysis of the helper component proteinase (HC-Pro) encoded in plant potyviruses revealed a similar functional profile. Processing activity of the C-terminal papain-like proteinase domain of HC-Pro (8) was found to be indispensable for RNA replication (30), whereas the central domain functioned as a replicational enhancer (6, 31). Recent experiments suggested that HC-Pro activates RNA synthesis indirectly, possibly disarming the host defense system that limits the accumulation of viral RNA (31, 46).

Both potyvirus HC-Pro and closterovirus L-Pro belong to a large class of papain-like leader proteinases that provide various accessory functions in viral reproduction (19, 21). Additional examples include the aphthovirus L-proteinase that is involved in the shutoff of host protein synthesis and in virus pathogenicity (43, 45) and p29, a symptom determinant of the hypovirulence-associated virus of chestnut blight fungus (10). It appears that diverse plant, animal, and fungal viruses recruit papain-like proteinases to manipulate host functions in the course of an infection.

Six genes located in the 3′-terminal region of the BYV genome were found to be dispensable for virus RNA accumulation at the single-cell level. It seems likely that corresponding products p6, HSP70h, p64, CPm, CP, and p20 contribute to the processes of virion assembly, cell-to-cell and long-distance transport, and aphid transmission of BYV. On the other hand, p21, encoded by the 3′-most BYV gene, was identified as an enhancer of RNA amplification. The mutant lacking the p21 start codon exhibited a fivefold reduction in the accumulation of both the plus and minus strands of genomic RNA. The proteins related to p21 are conserved in other one-component closteroviruses, such as CTV and beet yellow stunt virus (17, 28, and data not shown). Although a database search did not reveal any non-closterovirus homologs of p21, the γb protein encoded in barley stripe mosaic hordeivirus seems to perform an analogous accessory function in RNA replication (44). It should be emphasized that none of the mutations in p21, in L-Pro, or in products of ORFs 2 through 7 specifically affected the synthesis of sgRNAs. These data suggest that the only closterovirus proteins required for genome transcription are the products of ORFs 1a and 1b.

Interesting analogies were observed in the organization and expression of closterovirus and coronavirus genomes (17). These analogies included unusually large replicases encoded in ORFs 1a and 1b, with ORF 1b expressed via translational frameshift; the presence of papain-like proteinases in the 5′-proximal part of ORF 1a; and the formation of multiple sgRNAs required for expression of non-replicase genes (11, 21, 37, 38). The functional analysis presented here revealed that closteroviruses, similar to coronavirus-like viruses (32), in addition to replicase, do require a product of the 3′-most ORF for efficient RNA amplification. Further mechanistic studies employing full-length cDNA clones of coronavirus-like viruses (49) and closteroviruses should explain why these two phylogenetically and biologically diverse virus groups convergently adopted this particular combination of genome replication and expression strategies in the course of evolution towards the largest known “RNA chromosomes.”