3'-coterminal subgenomic RNAs and putative cis-acting elements of Grapevine leafroll-associated virus 3 reveals 'unique' features of gene expression strategy in the genus Ampelovirus

By analogy with BYV [4] and CTV [6], the two well studied members of the genus Closterovirus, ORFs 2 through 12, covering the 3' half of the GLRaV-3 genome (Fig. 1) would be expected to be expressed via eleven 3'-coterminal sgRNAs. As a first step towards comparative exploration of replication strategy of viruses in the genus Ampelovirus, we investigated the presence of sgRNAs in grapevine naturally infected with GLRaV-3. Total RNA preparations from scrapings of bark tissues were analyzed by Northern blot hybridization with positive-stranded RNA-specific riboprobes corresponding to nts 17,899 to 18,498 at the 3' end of GLRaV-3 genomic RNA. As shown in Fig. 2, four sgRNAs were present at higher levels and they were putatively identified as specific to p20B (ORF 10), p20A (ORF 9), p21 (ORF 8) and CP (ORF 6) genes. The two sgRNAs for p4 (ORF11) and p7 (ORF12) were not resolved due to the small differences in their sizes and appeared as a single moderately expressed band in Northern blots. The three barely visible bands were putatively identified as sgRNAs corresponding to CPm (ORF 7), p55 (ORF 5) and p5+HSP70 h (ORFs 3 and 4) genes. The specificity of the abundantly-accumulated sgRNAs to CP, p21, p20A and p20B genes was further confirmed by hybridization with riboprobes prepared using gene-specific sequences (Fig. 2). The riboprobe specific to the CPm hybridized weakly with the corresponding sgRNA band. Since the riboprobe showed strong hybridization with sgRNA of the CP, the observed weak signal further confirms that the sgRNA of CPm is poorly expressed. Among the four sgRNAs that accumulated at higher levels, the sgRNA corresponding to p20B gene accumulated at the highest level, followed by sgRNAs for p21, p20A and CP, respectively (Fig. 2). These results suggest that 3'-coterminal sgRNAs accumulate at variable amounts, reflecting differences in their expression levels and/or turnover rates in infected grapevine tissues. None of the sgRNAs were detected with a riboprobe specific to the 5'NTR (data not shown) further confirming that they are 3'-coterminal to the virus genome.

In order to characterize the sgRNAs further and map their locations in the virus genome, we needed to obtain the sequence of the Washington isolate of GLRaV-3. Although sequence is available for South Africa, Chile and New York isolates of GLRaV-3, considerable variation in their genome size between 17,919 and 18,498 nt warranted generating full genome sequence of the Washington isolate. In addition, having genome sequence information for the parental isolate of GLRaV-3 would enable mapping the 5'-transcription start site of sgRNAs more precisely in the cognate viral genome sequence. Due to its large size, the entire genome of GLRaV-3 was amplified into seven segments using virus-specific primers (Additional file 1, Figure S1 and Table S1). The cDNA clones representing each of the genomic segments were sequenced by directed sequencing protocol ("DNA walking") using progressive sequence-specific primers designed based on the partial nucleotide sequence obtained. This strategy, instead of cloning and sequencing by random oligonucleotide primers, decreased the number of steps required for determining the complete genome sequence of the virus and assembling the consensus sequence into a full-length genomic RNA sequence.

The RNA genome sequence of Washington isolate of GLRaV-3 was determined to be 18,498 nt long and it was deposited in GenBank under the accession no. GU983863. The genome contains thirteen putative ORFs with 737 nt long 5'NTR and 277 nt long 3'NTR (Fig. 1). The genome organization of Washington isolate was identical to GLRaV-3 isolates from New York [26], Chile [27] and South Africa [23]. The sizes of different ORFs and the 3'NTR were similar between all isolates (Additional file 1, Table S2). However, the size of the 5'NTR was significantly different, with New York and Chile isolates containing 158 nt, and South Africa and Washington isolates having 737 nt. In general, the genome of Washington isolate of GLRaV-3, downstream of 5'NTR sequence, showed higher level of nucleotide sequence identity with corresponding sequence of virus isolates from New York (~97%) and Chile (~99%) than with South Africa isolate (~92%). Overall, higher sequence identity values indicate that Washington isolate is closely related to GLRaV-3 isolates from New York and Chile than to the South Africa isolate (Additional file 1, Table S2).

Due to the discrepancy in the size of 5'NTR of the four GLRaV-3 isolates, we examined the sequence of 5'NTR of several isolates from six cultivars: four wine grape cultivars (Cabernet Sauvignon, Syrah, Merlot, Chardonnay), one table grape cultivar (Thomson Seedless) and one juice grape cultivar (Concord) planted in geographically widely separated regions in the US. The 5'RACE system was employed to verify the exact size of 5'NTR using two gene-specific downstream primers complementary to 860 to 883 nt (primer M1012) and 1034 to 1059 nt (primer M1013) of ORF1a of the Washington isolate. The expected DNA fragments from RT-PCR amplification would be ~304 nt and ~480 nt, if the 5'NTR is 158 nt in size as reported in New York and Chile isolates or it would be ~883 nt and ~1059 nt, if the 5'NTR is 737 nt in size as found in isolates from Washington and South Africa (Fig. 3a). Using the abridged anchor primer supplied with the 5'RACE kit as an upstream primer (primer AAP), a single product of ~883 bp and ~1059 bp were amplified with virus-specific primers M1012 and M1013, respectively (Fig. 3b). Sequence analysis of eight independent clones for each isolate showed that the size of 5'NTR is 737 nt with 98-100% sequence identity with corresponding sequence of the Washington isolate. The 5'NTR is A-U rich (22.12% As and 47.49% Us) and showed 83% nucleotide identity with the 5'NTR of the South Africa isolate. The 158 nt 5'NTR sequence of the Washington isolate immediately upstream of ORF1a showed 100% identity with corresponding 5'NTR sequences of New York and Chile isolates. From these results it is clear that the 737 nt 5'NTR is indeed authentic and an unusually long nontranslated sequence could be characteristic of GLRaV-3 including New York and Chile isolates.

Although the 5'NTRs of several GLRaV-3 isolates from the US and South Africa were of the same size, pairwise comparison showed non-uniform sequence identity distributed across the entire sequence (Fig. 4a). An unusually long stretch of 65 nt tandem repeat was observed between nucleotides 187 to 315 in the 5'NTR of all isolates of GLRaV-3 from the US sequenced in this work, where the first repeat was found between nucleotides 187-250 and the second between 251-315. Four nucleotide differences were observed in the tandem repeat sequences and these differences were distributed randomly in the entire length of the repeat (Fig. 4b). Whereas, such a signature tandem repeat was absent in 5'NTR of the South Africa isolate. However, the South Africa isolate has an additional 65 nt sequence that maintained 737 nt size of its 5'NTR (Fig. 4a). Alignment of 5'NTR sequences showed high sequence identity in the end sequences and in the middle portion with two distinct, highly variable regions in between. To further examine differences in the 5'NTR of GLRaV-3 isolates, we compared their predicted secondary structure using computational calculations at the MFOLD web server [30]. The 5'NTR of Washington and South Africa isolates folded into a complex structure consisting of a long SL structure with several substructural hairpins of variable lengths (Fig. 5a). This indicated that, although both isolates of GLRaV-3 have similar size 5'NTRs, the primary sequence and the predicted secondary structural architecture differed between them. In contrast, the 277 nt 3'NTR of Washington, New York, Chile and South Africa isolates of GLRaV-3 showed >97% similarity and folded into identical secondary structures (complementary sequence) consisting of two long SL structures (Fig. 5b). The 5' most SL consisted of 166 nt (18,333-18,498) and the 3'most structure with 83 nt (18,240-18,322) with the 5' most one forming a complex structure containing four substructural hairpins of variable length.

The members of the genera Closterovirus and Crinivirus employ the production of sgRNAs to serve as messengers for specific genes as one of the adaptive strategies to express their polycistronic genomes in their hosts [4‐6]. As a first step toward understanding strategies underlying production of the sgRNAs of viruses in the genus Ampelovirus, we mapped the 5' termini of the four most abundantly expressed genes in relation to the genomic RNA of GLRaV-3. For this purpose, the 5' ends of CP, p21, p20A and p20B sgRNAs were RT-PCR amplified by the 5'RACE system from total RNA isolated from grapevine tissue infected with GLRaV-3 using a combination of an abridged anchor primer and gene-specific primer, and the amplicons were cloned into pGEM-T vector. Sequences obtained from six independent clones for each of the sgRNAs were identical and were subsequently used to map the exact nucleotide position of the 5'-end of each sgRNA. The results showed that the length of sequence between the 5'-end and the putative start codon of each ORF (sgRNA leader sequence) is 48, 23, 95 and 125 nt, respectively, for CP, p21, p20A and p20B sgRNAs (Fig. 6a). This data demonstrated that the four sgRNAs have distinctly different sizes of mRNA leader sequences that are collinear with the genomic RNA. All four sgRNAs started with an adenylate, similar to the 5'-end of the genomic RNA. Based on this information, the transcription start site (TSS) for CP, p21, p20A and p20B was located at 13,800, 16,273, 16,755 and 17,265 nt, respectively, in the genome sequence of the Washington isolate of GLRaV-3 (Additional file 1, Table S3). The 48 nt leader sequence of CP sgRNA is located entirely in the intergenic region (IGR) between p55 and CP, the 23 nt leader sequence of p21 encompass 13 nt C-terminus of CPm ORF and 10 nt IGR between CPm and p21, the 95 nt leader sequence of p20A overlaps with the C-terminus of p21, and the 125 nt leader sequence of p20B encompass 119 nt C-terminal portion of p20A and 6 nt IGR between p20A and p20B. Using the location of TSS, we estimated the size of sgRNA for CP, p21, p20A and p20B (Fig. 6a) as 4,699, 2,226, 1,744 and 1,234 nts, respectively (Additional file 1, Table S3). The TSS for CP, p20A and p20B sgRNAs match with those reported for South African isolate of GLRaV-3 [31]. The study by Maree et al. [31] also indentified TSS for the other 3' sgRNAs. However, we could not amplify 5'-end sequences for the other sgRNAs (ORFs 2, 3, 4, 5, and 7), despite several attempts, possibly due to their low abundance in infected grapevine tissue.

The 5' leader sequences of CP, p21, p20A and p20B of the Washington isolate were more similar (98-100% identity) to the corresponding sequences in the New York and Chile isolates than with South Africa isolate (88-94% identity) (Additional file 1, Table S4). Except for the 5'-end nucleotide, the leader sequences of the four sgRNAs did not reveal any shared sequence motifs, in contrast with the presence of a conserved heptanucleotide in some sgRNAs of BYV and CTV [16]. Since conserved nucleotide sequences around the TSS of sgRNA were implicated in the synthesis of sgRNAs of Tobacco mosaic virus (TMV) [32] and Citrus tatter leaf virus (CTLV) [33], we compared a 50 nt region around the TSS (-25 to +25 relative to the initiation site of each sgRNA) of each of the four sgRNAs of GLRaV-3. The results revealed no conserved sequences surrounding the TSS, in contrast to the presence of conserved octanucleotide sequences in TMV and CTLV.

The sequences upstream of the ORFs of CTV were shown to control production of sgRNAs [13, 14]. In general, the minus strands corresponding to these regions could be folded into one or two SL structures with the first nucleotide (A) of the sgRNA leader a few nucleotides downstream of the last SL structure [34]. To predict the possible involvement of RNA secondary structures in the sgRNA synthesis analogous to TMV, CTLV and CTV, the 50 nt sequence corresponding to the minus strand sequence around the TSS of the four sgRNAs was analyzed by MFOLD [30]. The results showed that, although the 50 nt sequence corresponding to the minus strand sequence of sgRNAs for CP, p21, p20A and p20B folded into predicted SL structures, conservation of the predicted secondary structures was not observed (Fig. 6b). The 5'end nucleotide of sgRNA specific to CP and p20A was located outside the SL structure, whereas that specific to p21 sgRNA was located on the stem of the SL structure and it was in the loop in the case of p20B sgRNA. The integrity of these secondary structural features were maintained in the leader sequences of all GLRaV-3 isolates, despite low (84-94%) sequence identity especially between Washington and South Africa isolates, suggesting a requirement for the putative secondary structure in sgRNA synthesis. The genome sequences immediately ~100 nt upstream from the 5' terminus of each sgRNA were also compared to verify the possible conservation in the putative control elements for sgRNA production. The data revealed no similarity between the sequences or their predicted structures and no structures similar to those found for CTV.

GLRaV-3 isolates were collected from five wine grape cultivars (Vitis vinifera, cvs. Cabernet Sauvignon, Syrah, Merlot, Chardonnay), one table grape cultivar (V. vinifera, cv. Thomson Seedless) and one juice grape cultivar (V. labruscana 'Concord'). Cambial scrapings of canes from a single Merlot grapevine were used for isolating genomic-length replicative-form double-stranded (ds) RNA essentially as described by Valverde et al. [44]. The integrity of dsRNA preparations were verified by resolving in 0.8% agarose gels and using dsRNA-enriched preparation from citrus (Sour Orange) infected with Citrus trizteza virus (CTV). Unless otherwise mentioned, the same dsRNA preparation was used for cloning and sequencing of the entire genome of GLRaV-3.

The denatured dsRNA served as a template for cDNA synthesis. The reaction was carried out in 25 μL reaction mixture in the presence of gene-specific complementary primer using Superscript III reverse transcriptase kit (Invitrogen Corp, Carlsbad, CA) by following the manufacturer's instructions. The cDNA preparation was used as a template for subsequent PCR amplification of different portions of the virus genome (Additional file 1, Figure S1) using primers listed in Additional file 1, Table S1). These primers were initially designed based on nucleotide sequence information available for New York isolate of GLRaV-3 (AF037268), [26]. Primers for subsequent experiments were designed based on the sequence obtained for GLRaV-3 isolate from Washington (accession no. GU983863). Unless stated otherwise, locations of all primers were indicated using the entire genome sequence of GLRaV-3 isolate from Washington. Each PCR reaction in 20 μL consisted of a final concentration of 2-4 ng of template cDNA, 0.3 μMoles each of sense and antisense primers, 0.2 mM of each dNTP, 1X concentration of reaction buffer consisting 1 mM Mg²⁺ and 0.025 U of high fidelity Takara Prime Star DNA polymerase (Takara Bio Inc., Japan). The temperature profile for the amplification included denaturation at 94°C for 30 sec followed by 35 cycles of denaturation at 98°C for 10 sec, annealing at 60°C for 5 sec and extension at 72°C for 60 sec per 1 kb size of amplicon and a final extension at 72°C for 10 minutes.

The amplicons were cloned in pUC-119 based vector and two independent clones specific to each amplicon were sequenced in both orientations at the Molecular Biology Core Instrumentation Facility, Washington State University, Pullman, WA. Wherever necessary, additional clones were sequenced to resolve nucleotide sequence ambiguities. Nucleotide and predicted amino acid sequences were analyzed using Vector NTI Advance11 software (Invitrogen Corp, Carlsbad, CA).

The exact 5' end sequence of GLRaV-3 was determined using a commercially available 5'RACE system for rapid amplification of cDNA ends kit (Version 2.0, Invitrogen, Carlsbad, CA). Total RNA was isolated from bark scrapings of canes collected from GLRaV-3-infected Merlot vines based on a protocol reported by Tattersall et al.[45]. First-strand cDNA was synthesized using the gene-specific primer AR (5'-CATTAAGGGCCCTGTTAAAC-3', complementary to nucleotides 833 to 852 of the Washington isolate, GU983863). The purified first-strand cDNA was 'dC' tailed and the following primer combinations were used to amplify virus-specific DNA fragments: virus-specific primer M1012 (5'-AAGTCCGACAACTTCACGTTCCCT-3', complementary to nucleotides 860 to 883 in GU983863) and the abridged anchor primer (AAP) supplied with the kit and virus-specific primer M1013 (5'-AAGTTGAGGTCCTTGCCTCCATCAAG-3', complementary to nucleotides 1034 to 1059 in GU983863) and the AAP supplied with the kit. The temperature profile for the amplification of DNA included denaturation at 94°C for 3 min followed by 35 cycles of denaturation at 94°C for 10 sec, annealing at 56°C for 30 sec and extension at 72°C for 60 sec and a final extension at 72°C for 5 min.

For verification of 3'end sequence of GLRaV-3, total RNA was poly-adenylated with Yeast Poly(A) polymerase (USB, Cleveland, OH) and used for cDNA synthesis using SuperScriptIII reverse transcriptase (Invitrogen, Carlsbad, CA). The first-strand cDNA was used as a template to amplify virus-specific DNA using an oligo dT primer M111 (5'-GGTCTCGAG(T)₁₈-3') and a virus-specific forward primer GF (5'-ATTAGCATATGTAGAAAAGGGGAAG-3', nucleotides 18174 to 18198 in GU983863). The 5' and 3' RACE PCR products were cloned into pGEM-T vector (Promega, Madison, WI) and six independent clones specific to each PCR product were sequenced in both orientations. Sequences were analyzed as described above.

Total RNA extracts prepared from cambial scrapings collected from GLRaV-3-infected Merlot grapevines were separated by electrophoresis in a formaldehyde denaturing 1.1% agarose gel in MOPS buffer, and transferred onto a nylon membrane using an electrotransfer unit (Hoefer Pharmacia Biotech, San fransisco, CA) as described by Lewandowski and Dawson [46]. Northern blot hybridization and probing of membranes for the presence of genomic RNA and subgenomic RNAs of GLRaV-3 with nonradioactive digoxigenin (DIG)-labeled riboprobes was carried out according to Tatineni et al. [33]. The probe for detecting all sgRNAs includes 17,899-18,498 nt at the 3' terminus of the Washington isolate of GLRaV-3 (Additional file 1, Table S5). DNA template for gene-specific riboprobe synthesis were amplified from virus-specific clones in pUC119 using the primers listed in Additional file 1, Table S5 and Table S6. T7 and SP6 polymerase promoter sequences were included in the forward and reverse primers, respectively, and the PCR amplified products were agarose-gel eluted and used as a template to generate positive-stranded RNA-specific probes by using T7- or SP6-RNA polymerase and nucleotides containing DIG-labeled UTP.

Total RNA isolated from cambial scrapings of GLRaV-3-infected Merlot grapevine was used to map the transcription start sites of four sgRNAs, namely, CP, p21, p20A and p20B. The gene specific primers (Additional file 1, Table S7) were used to synthesize the first-strand cDNAs. The first-strand cDNAs were column purified and 'dC' tailed at the 5' end using terminal deoxynucleotidyl transferase, and amplified by using 2.5 U of Taq DNA polymerase (New England Biolabs, Ipswich, MA) in a 50 μL reaction volume with an abridged anchor primer (supplied with kit) together with gene specific nested primer. The following conditions were used for PCR: 1 cycle at 94°C for 2 min, followed by 35 cycles at 94°C for 20 sec, 56°C for 20 sec and 72°C for 1 min, and one cycle at 72°C for 5 min. The PCR products were ligated into pGEM-T Easy vector (Promega, WI) and the inserts were sequenced at the DNA sequencing Core, Interdesciplinary Center for Biotechnology Research, University of Florida, Gainesville, FL.

Sequences were edited and assembled using ContigExpress module in the VectorNTI sequence analysis software package (Invitrogen Corp, Carlsbad, CA). Nucleotide and amino acid sequence identity levels were calculated using Vector NTI Advance program (Invitrogen Corp. CA). Secondary structure analysis of the 5'NTR and 3'NTR sequences was carried out by using MFOLD software [30].