Quasispecies nature of the genome of Grapevine fanleaf virus

Grapevine fanleaf virus (GFLV) is a bipartite, positive-sense, single-stranded RNA nepovirus which causes the most economically important disease of Vitis vinifera in the world (Goheen, 1977 , 1989 ). Extensive variability exists in the sequences of GFLV genomes (Serghini et al., 1990 ; Brandt et al., 1995 ; Esmenjaud et al., 1994 ; Sanchez et al., 1991 ). Nucleotide sequence differences were found to range from 8% to over 10%, with amino acid sequence differences in the range of 2 to 4%. This suggests that variability in GFLV symptomatology, from fanleaf to yellow mosaic or vein banding symptoms (Krake et al., 1999 ), may have a genetic component.

The level of genomic variation suggests that GFLV genomes may consist of a genetically diverse collection of mutants, the dominant members of which may vary during shifts among successive host varieties, in the manner of a quasispecies (Roossinck, 1997 ; Schneider & Roossinck, 2000 ). We have attempted a qualitative characterization of the GFLV genome, on the assumption that it does exist as a collection of sequences varying around its own consensus sequence. Gross variation in the viral RNA was visualized by restriction fragment length polymorphism (RFLP) analysis of sections of the unfractionated viral genome amplified by PCR. The variability was compared to variation found by sequence analysis of the same sections of the viral genome purified by cloning. GFLV isolates recovered from eight California vineyards were compared, and gross changes associated with the passage of these isolates in alternative hosts were visualized by this method.

Isolate collection and maintenance.
GFLV-infected grapevine samples were collected in California vineyards and propagated in greenhouses. Four GFLV isolates were from the Clonal Virus Collection (CVC) vineyard (Golino, 1992 ) of the University of California, Davis and were designated isolates AD. CVC plants were propagated from dormant wood. Ten uncharacterized isolates, designated EN, were collected from other vineyards by making one or two node cuttings from infected vines. These were propagated from green cuttings. Isolates G and H were collected from different blocks of one vineyard, J and K were from different blocks of a second vineyard, and the remainder were from separate vineyards.

Inoculations.
Chenopodium quinoa plants were used as a systemic, herbaceous host for GFLV. Plants were grown from seed to the six-leaf stage before inoculation. Five-hundred mg of grapevine shoot tip and leaf were ground in 5 ml 50 mM phosphate buffer, pH 6·5, containing 2% nicotine. All C. quinoa leaves were dusted with carborundum and inoculated with grape tissue extract using a gloved finger. Inoculated leaves were washed immediately with DI water.

RTPCR.
A modification of the method of Rowhani et al. (1995) was used for immunocapture of the virions into the wells of microtitre plates, after precoating plates with GFLV-specific γ-globulin. After washes with phosphate-buffered salineTween 20 (PBST) and phosphate-buffered saline (PBS), 3050 µl of warm 1% Triton X-100 was added to the wells, mixed by shaking, and then incubated at 65 °C for 10 min in a water-bath for virion disruption. Complementary DNA strand was synthesized by adding 5 µl of the above solution to 2·5 µl of 10 µM (RT) primer [5' AAAAATTTGCATAACAGTAAAAAG 3' (binding at positions 37543775 in the 3' UTR of viral RNA 2)] in a cocktail composed of 2 µl 10 mM dNTP mix, 2 µl 100 mM dithiothreitol, 4 µl 5xRTase buffer (250 mM TrisHCl pH 8·3, 375 mM KCl, 13 mM MgCl₂), 0·25 µl M-MLV RTase (200 Units/µl) (Life Technologies) and 4·25 µl DEPC-treated sterile water. The reverse transcription reaction was carried out using a top-heating thermal cycler for 1 h at 37 °C, and stopped by an incubation of 5 min at 99 °C, followed by 5 min at 5 °C.

Eighty µl of PCR reaction cocktail containing 10 µM C primer [5' CAAGGCAAGTGTGTCCAAA 3' (binding at positions corresponding to 37653724 in the viral coat protein (CP) gene sequence)], 2·5 µl 10 µM V primer [5' TGATGCTTATAATCGGATAACTA 3' (binding at genomic positions 22572270)], 0·25 µl Taq DNA polymerase (5 U/µl), 8 µl 10x PCR buffer, 2·5 µl 50 mM MgCl₂ and 44 µl of sterile water were added to each cDNA reaction for a final PCR reaction volume of 100 µl.

The PCR used a 2 min heating step at 95 °C, followed by ten cycles of 30 s melting at 95 °C, 1 min annealing at 63 °C, and 2 min elongation at 72 °C. These ten cycles were followed by 25 cycles of 30 s melting at 95 °C, 1 min annealing at 63 °C, and 2 min elongation at 72 °C with an incremental addition of 5 s of elongation per cycle to ensure full amplification.

RFLP analysis protocol.
After amplification, 5 µl of each reaction was run on a 1·5% agarose gel in 1xTrisacetate buffer (40 mM Trisacetate, 1 mM EDTA, pH 8·0) to confirm product synthesis and estimate DNA concentration by comparison with standards. The PCR product was ethanol precipitated. Half of the product was digested with 0·5 U AvaII restriction endonuclease (New England Biolabs) and the RFLP fragments were separated on a 20 cm 10% polyacrylamide gel in 1xTAE buffer overnight at constant 60 V (3 V/cm) using a ProteanII polyacrylamide gel system (Bio-Rad). DNA was visualized using a UV transilluminator following incubation of gels in 5 µl/ml ethidium bromide.

Cloning of 1557 bp products.
The 1557 bp product was gel-purified using the Prep-A-Gene DNA purification system (Bio-Rad) according to the manufacturers specifications. The fragment was cloned into the pCR2.1-TOPO vector using the TOPO TA cloning kit (Invitrogen) as per instructions.

Sequencing.
Sequence data were obtained with a Perkin Elmer/Applied Biosystems ABI 377 automated sequencer.

A 1557 bp fragment of the GFLV genome was amplified from the 3' end of RNA2 by nested RTPCR as shown schematically in Fig. 1. A similar 1557 bp product was recovered from all isolates (Fig. 2A); no band was produced from uninfected control material. The Genetics Computer Group (Madison, WI, USA) Map program was used to analyse virtual restriction fragment patterns generated by a panel of restriction endonucleases. AvaII was chosen for RFLP analysis based on its predicted ability to generate differential RFLP patterns from three GFLV CP gene sequences in the GenBank database [from Austria (accession no. U11768), France (F13, X16907) and California (GFLV 100, X60775)].

(16K): Fig. 1. Nested RTPCR and RFLP procedure. The CP gene is 3' proximal in GFLV RNA2, adjacent to the (putative) movement protein gene. cDNA was synthesized from immunocaptured virions followed by PCR. The RT primer was used to make first strand cDNA, then the C and V primers were used for production of a 1557 bp PCR product. The PCR product contains sequences from the 3'UTR in addition to CP region.

(91K): Fig. 2. (A) Agarose gel showing a specific 1557 bp RTPCR product amplified by C and V primers from the CP region of GFLV isolates present in some California vineyards. (B) Acrylamide gel showing RFLPs of the RTPCR products shown in (A). RTPCR products were digested by AvaII restriction endonuclease. Marker lanes contain DNA size standards (in bp) shown on the left for both images.

The RFLP pattern from 14 California isolates is shown in Fig. 2(B). The banding patterns produced were highly variable, and usually unique to each isolate. The sum of the fragment sizes added up to more than 1557 bp in each case, with the bands spanning a wide range of different intensities. The most numerous, faintest bands are the first ones lost during photoreproduction. Many of those faint bands are not visible in the final form of Fig. 2(B). Bands did not appear to arise from chance variation during the PCR stage. Thus, when each of the RNA templates was prepared at least three times from a given infected host (preparations separated by at least a month), the specific banding patterns consistently reappeared, suggesting that the bands specific to each isolate did not arise from chance variation during reverse transcription or PCR. The patterns did not appear to be generated by partial digestion by AvaII, since combinations of the small bands did not add up to equal the size of larger bands above them.

GFLV isolates were inoculated from the grapevine sources to C. quinoa, and serially passaged into healthy C. quinoa 1 day after the onset of symptoms, on a 1016 day cycle. RTPCR and RFLP analysis was performed on extracts from these plants, which showed systemic symptoms (grapevines grown under greenhouse conditions showed much reduced symptoms although they were also RTPCR positive for GFLV). The 1557 bp RTPCR product was obtained from each plant. In most cases, the RFLP banding pattern was simplified during passage. The RFLP profiles from isolate A infecting C. quinoa in sequential passages are shown separately in Fig. 3. A comparison of these profiles shows that the first passage from grapevine into C. quinoa did not cause much change in the RFLP pattern derived from the 1557 bp PCR product. On the second passage a simplified RFLP pattern was observed. This new pattern remained fairly stable through five passages. Bands appearing at approximately the 500 bp position in the second through fifth C. quinoa passages were not observed in the original host.

(47K): Fig. 3. RFLP analysis of RTPCR products derived from isolate A infecting C. quinoa. Separate profiles were made for each of five passages. RTPCR products were digested by AvaII restriction endonuclease. S designates the RFLP pattern produced from GFLV from the grapevine source material. The numbers above the lanes indicate passage number. Mobility of DNA size standards (in bp) is shown on the right. Fig. 4. RFLP analysis of clones from isolate G produced by AvaII digestion of plasmid inserts. Lane S shows RFLP of the RTPCR product from grapevine source S viral RNA. Marker lanes contained DNA size standards; specific fragment sizes (bp) are shown on the right.

The RTPCR products from some of the isolates were cloned into the pCR2.1-TOPO cloning vector and screened by RFLP analysis using AvaII after gel purification of the insert. As shown in Fig. 4, inserts from each clone were cleaved into fragments whose sizes summed to a close approximation of the initial 1557 bp PCR product. Thus, the RFLP banding pattern of each isolate was simplified by cloning, which reflects the isolation of single quasispecies members. As Fig. 4 shows, many of the bands appearing in the RFLP derived from grapevine were no longer present after cloning, while new bands not appearing in the source RFLP were seen. Isolate G produced an especially complex RFLP pattern. The genetic makeup of GFLV isolates D, E and G, showing distinct RFLP patterns in grapevine (Fig. 2b, lanes D, E and G), was assessed by sequence analysis after cloning in the pCR2.1-TOPO vector. Purified plasmid DNA from clones was analysed by RFLP and one clone from each isolate was chosen based on its distinctive RFLP pattern, and sequenced. These sequences have been deposited in the GenBank database (accession nos AF304013, AF304014 and AF304015). Sequence comparisons were made between the three clones.

The sequence data showed a variation from 11 to 13%, and from 4 to 9% at the nucleotide and amino acid level, respectively (data not shown). Changes in the AvaII sites predicted by these sequences correlated with the RFLP patterns observed. Specific regions of high variability (hypervariable regions) were not apparent.

The genomic variation in isolates of GFLV characterized here is portrayed by the RFLP patterns, as seen in Fig. 2. The bands specific to each isolate likely correspond to components of quasispecies populations. RFLP analysis reveals a gradation of banding intensities of great depth, the faintest of which are lost from the figure during photoreproduction. Though not used for identification of changes at the nucleotide level, RFLP nonetheless provides a qualitative portrayal of genomic variation. The viruses analysed here were isolated from across a wide geographical range, and were associated with variable infection symptomatologies. The commensurately variable RFLP data may reflect selection among mutant genomes, imposed by the different host varieties and other environmental conditions upon the variants in the cloud of GFLV sequences.

The RFLP bands cover a range of intensities broad enough to suggest that much of the complexity may reside with quasispecies members at concentrations below detectable thresholds. The immediacy of appearance, upon passage, of novel bands with no visible correlates in the original inoculum (Fig. 3) suggests that the newly appearing sequences were present in the original mixture, but at concentrations below the minimum threshold of visibility in RFLP analysis. The reproducibility of the patterns in each of the three replications of the procedure suggests that the variation does not arise from inaccuracies in the RTPCR analysis.

The primary bands in the RFLP patterns show wholesale changes between isolates, and upon passage of a given isolate. This is consistent with the variation seen in cDNA sequences of subclones analysed here, or from the database. Members of the quasispecies mix isolated by cloning differed by 11 to 13% in the 1557 bp genomic sequence amplified in this study. On statistical average, the GG^A/_TCC AvaII recognition site would occur three times in a 1557 bp sequence, producing four digestion fragments. The AvaII footprint would cover only 15 bp in that sequence. If the sequence were to vary randomly by 15%, approximately two positions in those 15 would change, destroying two of the three sites, and changing three of the four original restriction fragments. (Two new sites would, by chance, be created, further obscuring the original RFLP pattern.)

The number of bands from the California isolates visualized by RFLP analysis was seen to diminish upon serial passage within an herbaceous host. For example, banding patterns of isolates AD were simplified upon passage within C. quinoa (Fig. 3), a general host commonly used to culture plant viruses. This points up the extent to which diversity can be lost from a population of genomic sequences upon passage through a novel host background, suggesting that genomic variability should be characterized directly in the primary host.

The authors would like to thank Dr M. A. Walker for providing some of the virus isolates used in this study and Dr Yun-Ping Zhang for his assistance with this project. This work was funded by the California Grape Rootstock Improvement Commission.

Footnotes

^b Present address: Enterovirus Section, Centers for Disease Control and Prevention: NCID/DVRD/REVB, 1600 Clifton Road MS-G17, Atlanta, GA 30333, USA.