Development of a polyvalent RT-PCR detection assay covering the genetic diversity of Cherry capillovirus A




The variability of Cherry capillovirus A (CVA) was analysed using a short, 275-bp region of the viral RNA-dependent RNA polymerase gene amplified by a polyvalent RT-PCR assay. As for other members of the family Betaflexiviridae, CVA appears to show significant diversity, with an average pairwise nucleotide divergence of 9·4% between isolates in the analysed region. Phylogenetic analyses provide evidence for the existence of at least five clusters of CVA isolates, one of which is associated with noncherry hosts of the virus, providing evidence that transmission of CVA isolates between cherry and noncherry hosts is probably rare. Comparison of existing detection techniques using a panel of CVA isolates representative of the various phylogenetic groups indicated that dot-blot hybridization assays show high polyvalence but may lack the sensitivity to detect CVA in some samples. On the other hand, available detection primers failed to amplify a wide range of CVA isolates. Partial genome sequencing of two divergent isolates allowed the identification of conserved genomic regions and the design of new primer pairs with improved polyvalence. These new primer pairs were used to develop PCR assays allowing the reliable detection of CVA isolates belonging to all phylogenetic clusters.


Cherry virus A (CVA) is a member of the genus Capillovirus in the family Betaflexiviridae. It was first described in Germany in a sample of sweet cherry (Prunus avium) in a mixed infection with Little cherry virus 1, one of the agents of little cherry disease (Jelkmann, 1995). It has also been observed in sour cherry [Prunus cerasus (Sabanadzovic et al., 2005)], Japanese flowering Kwanzan cherry [Prunus serrulata (Eastwell & Bernardy, 1998)] and also in noncherry hosts such as apricot and peach (James & Jelkmann, 1998; Barone et al., 2006), plum (Svanella-Dumas et al., 2005; Barone et al., 2006) and, more recently, Japanese apricot [Prunus mume (Marais et al., 2008a)]. CVA seems to be very widely distributed in cherry on a worldwide basis, but infection of noncherry hosts appears to be less frequent (Marais et al., 2011). As for other capilloviruses, there is no information to date about the possible existence of a vector or vectors (Jelkmann, 1995; Adams et al., 2005).

There is so far no clear evidence for any association of CVA infection with specific disease symptoms in any of its hosts, but the situation is often complicated by the frequent existence in woody hosts of mixed infections with other viruses such as Prune dwarf virus (PDV), Plum bark necrosis and stem pitting-associated virus (PBNSPaV), Apple chlorotic leaf spot virus (ACLSV), Apricot pseudo chlorotic leaf spot virus (APCLSV), Cherry green ring mottle virus (CGRMV), Cherry necrotic rusty mottle virus (CNRMV), Little cherry virus 1 (LChV-1) or Little cherry virus 2 (LChV-2) (Jelkmann, 1995; Eastwell & Bernardy, 1998; Isogai et al., 2004; Sabanadzovic et al., 2005; Svanella-Dumas et al., 2005; Barone et al., 2006; Mandic et al., 2007). As a consequence, CVA is generally considered as a virus causing only latent infections. However, possible synergistic effects of CVA on the severity of the symptoms caused by other viruses or the possibility that particular CVA isolates could cause symptomatic infections in some of their hosts cannot be excluded.

The genome of the reference CVA isolate is a single-stranded, positive-sense RNA molecule of 7383 nucleotides (nt), excluding the polyA tail (Jelkmann, 1995). Its genomic organization, consisting of two open reading frames (ORF) (Fig. 1), is similar to that of Apple stem grooving virus (ASGV), the type species of the genus Capillovirus. ORF 1 encodes a polyprotein of 266 kDa, with typical motifs for viral RNA replicase (RNA-dependent RNA polymerase, RdRp) in the central part of the protein. Alignments of the C-terminal part of ORF 1 showed significant homologies with the coat proteins (CP) of members of the genera Capillovirus and Trichovirus. The CP seems to be expressed from the genomic RNA as a fusion protein to the RdRp and also to be independently expressed from a subgenomic messenger RNA (Tatineni et al., 2009; Hirata et al., 2010). ORF 2 is nested within ORF 1 and encodes, in another reading frame, a 52 kDa protein homologous to the movement protein (MP) of ASGV and ACLSV.

Figure 1.

 Schematic representation of the genome organization of the Cherry capillovirus A (CVA) X82547 reference isolate and positions of the different primer pairs used in the study. The two open reading frames are indicated: ORF 1 (55–7081) encoding the RNA-dependent RNA polymerase/coat protein fusion protein; and ORF 2, encoding the movement protein (5400–6790). The positions of the products amplified using primer pairs are: CVA-1/CVA-2, 4621–5454; CVA-fw1/CVA-rev1, 5986–6287; CVA-fw1/CVA-rev1b, 5986–6428; CVA-fw2/CVA-rev2, 6907–7249.

In contrast to other common fruit-tree-infecting viruses like ACLSV (Candresse et al., 1995; Foissac et al., 2005), and despite its high prevalence in cherry trees, only very limited information is available on the genetic diversity of CVA or the suitability and polyvalence of existing CVA detection assays (Jelkmann, 1995; James & Jelkmann, 1998). However, the limited partial sequence information obtained for the viral RdRp gene through the use of the polyvalent degenerated oligonucleotides nested RT-PCR assay (PDO RT-PCR) indicated that the genetic diversity of CVA could be significant (Foissac et al., 2005), as is the case for many other members of the family Betaflexiviridae (Adams et al., 2004).

In the present study, the molecular variability of CVA was investigated. A preliminary account of this diversity analysis was reported in a previous study (Marais et al., 2008b). The results show a substantial level of variability in the short RdRp fragment analysed and allow the identification of several distinct clusters of isolates, one of which corresponds to noncherry isolates of CVA. The polyvalence of the currently available CVA detection tests was then evaluated on representative CVA isolates, revealing that no existing assay was able to detect all studied CVA isolates. The partial sequencing of the genome of divergent CVA isolates allowed the design of new PCR assays which extend the range of detectable CVA isolates.

Materials and methods

Virus isolates

The CVA isolates included in this study are listed in Table 1. Most of the isolates from cherry were obtained from sweet cherry trees from southern France collected during the 2003, 2004 and 2005 summers during surveys performed to characterize a new cherry decline disease. The noncherry isolates were collected from plums or apricots during surveys in the Czech Republic or during a survey of old Italian fruit tree varieties (Barone et al., 2006). The CVA-PF strain was originally isolated from plum (Prunus domestica) in France (Svanella-Dumas et al., 2005) and was propagated by chip-budding on GF305 peach seedlings under greenhouse conditions. The CVA-V590 isolate was obtained from a P. avium tree affected by the cherry necrotic crook disease (Foissac et al., 2005). It was maintained in a greenhouse on P. avium cv. Sam.

Table 1.   List of Cherry capillovirus A (CVA) isolates included in the present study, indicating their host and country of origin, together with the relevant sequence accession numbers
Virus isolateOriginal host (country)CultivarAccession numberReference
  1. ND, not determined.

A1Prunus avium (France)EnjidelHQ267818This work
A2Prunus avium (France)GardelHQ267819This work
A4Prunus avium (France)EnjidelHQ267820This work
A5Prunus avium (France)EnjidelHQ267821This work
C3Prunus avium (France)EnjidelHQ267822This work
V3654Prunus avium (France)V3654HQ267823This work
V3692Prunus avium (France)V3692HQ267824This work
Bal1Prunus avium (France)EnjidelHQ267825This work
Bal8Prunus avium (France)EnjidelHQ267826This work
Bal9Prunus avium (France)EnjidelHQ267827This work
Bal10Prunus avium (France)FerpactHQ267828This work
Bal11Prunus avium (France)FerpinHQ267829This work
Bal19Prunus avium (France)LempereurHQ267830This work
Bal21Prunus avium (France)FerlizacHQ267831This work
12517Prunus avium (France)NDHQ267832This work
P1Prunus avium (France)NDHQ267833This work
943Prunus avium (France)NDHQ267834This work
944Prunus avium (France)MagarHQ267835This work
947Prunus avium (France)FerpactHQ267836This work
948Prunus avium (France)FerpinHQ267837This work
949Prunus avium (France)Sweet ValentineHQ267838This work
952Prunus avium (France)BurlatHQ267839This work
955Prunus avium (France)EnjidelHQ267840This work
967Prunus avium (France)EnjidelHQ267841This work
972Prunus avium (France)FertilleHQ267842This work
978Prunus avium (France)BurlatHQ267843This work
983Prunus avium (France)RainierHQ267844This work
1021Prunus avium (France)M1HQ267845This work
1024Prunus avium (France)NDHQ267846This work
1025Prunus avium (France)NDHQ267847This work
1026Prunus avium (France)SonnetHQ267848This work
J1Prunus avium (ND)NDHQ267849This work
J2Prunus avium (ND)NDHQ267850This work
J3Prunus avium (ND)NDHQ267851This work
J4Prunus avium (ND)NDHQ267852This work
J5Prunus avium (ND)NDHQ267853This work
R5BPrunus avium (Germany)NDAF413926(Foissac et al., 2005)
R1Prunus avium (Germany)NDAF413925(Foissac et al., 2005)
R5APrunus avium (Germany)NDAF413924(Foissac et al., 2005)
T6Prunus avium (France)NDAF413923(Foissac et al., 2005)
V590Prunus avium (France)NDAF413922(Foissac et al., 2005)
ColtPrunus avium (USA)ColtAY944065(Sabanadzovic et al., 2005)
BingPrunus avium (USA)BingAY944066(Sabanadzovic et al., 2005)
 Prunus avium (Germany)SamX82547(Jelkmann, 1995)
C20Prunus avium (Italy)Sangue di bufaloDQ445286(Barone et al., 2006)
 Prunus avium (India)ManigamFN691959(Noorani et al., 2010)
KwanzanPrunus serrulata (USA)KwanzanAY944064(Sabanadzovic et al., 2005)
S68/32Prunus domestica (Italy)TurconaDQ445284(Barone et al., 2006)
PFPrunus domestica (France)Mirabelle doréeAY792509(Svanella-Dumas et al., 2005)
RT2Prunus domestica (Czech Republic)Cerncicka domaciHQ267854This work
RT3Prunus domestica (Czech Republic)Merunkova renklodaHQ267855This work
AL27/1Prunus armeniaca (Italy)Z. luisaDQ445279(Barone et al., 2006)
AL17/19Prunus armeniaca (Italy)CerasielloDQ445287(Barone et al., 2006)
A2/6Prunus armeniaca (Italy)ScassulilloDQ445288(Barone et al., 2006)
A3/7Prunus armeniaca (Italy)Boccuccia lisciaDQ445289(Barone et al., 2006)
A5/14Prunus armeniaca (Italy)StradonaDQ445291(Barone et al., 2006)
AL37/38Prunus armeniaca (Italy)PaolonaDQ445278(Barone et al., 2006)
PM1Prunus mume (China)NDEU730949(Marais et al., 2008a)
PM2Prunus mume (China)NDEU730950(Marais et al., 2008a)
PM3Prunus mume (China)NDEU730951(Marais et al., 2008a)

Nucleic acid extraction from plant samples

For RT-PCR analyses, total nucleic acids (TNA) were extracted using the procedures described by Foissac et al. (2005). Extraction procedure 1 was used for cherry flower samples and extraction procedure 2 for all other samples. For Smart™ Long Distance (LD) PCR, total RNAs were extracted from CVA-PF-infected peach leaves and from CVA-V590-infected cherry leaves using the RNeasy Plant Minikit (QIAGEN).

CVA detection by dot-blot hybridization

For molecular hybridization, cherry flowers were ground (1/10 w/v) in PBS-Tween-PVP-DIECA buffer [137 mm NaCl, 3 mm KCl, 1·5 mm KH2PO4 and 8 mm Na2HPO4 pH 7·2 supplemented with 0·05% Tween 20, 20 mm sodium diethyldithiocarbamate (DIECA) and 2% polyvinylpyrrolidone (PVP K25)]. After centrifugation at 13 000 g for 10 min, 100 μL of the supernatant were mixed with 100 μL denaturation buffer [2·4 m NaCl, 240 mm sodium citrate, 15% (v/v) formaldehyde] and incubated for 45 min at room temperature. Alternatively, TNA samples (50 μL) prepared as described above were directly mixed with an equal volume of denaturation buffer. After incubation, the mixture was centrifuged for 10 min at 13 000 g and the supernatant blotted onto N+ charged nylon membrane (Roche Diagnostics) using standard procedures (Sambrook et al., 1989). Nucleic acids were finally fixed on the membranes by UV-crosslinking (1200 mJ cm−2) (UV Stratalinker 1800; Stratagene).

Two riboprobes were used in the hybridization experiments. The CVA-39 probe was synthesized from plasmid p39, which contains almost all of the CVA movement protein gene plus the 3′ noncoding region (3′NCR) (X82547; Jelkmann, 1995). The CVA-105 probe, derived from plasmid p105, covers the 5′NCR and one-third of the RdRp gene (X82547; Jelkmann, 1995). The plasmids were linearized with SmaI (New Englands Biolabs) and used as template for the synthesis of digoxigenin-labelled cRNA probes using the DIG-labelling kit (Roche Diagnostics) as recommended by the supplier. The prehybridization step was performed in hybridization buffer (1 mL/10 cm2) [5× SSC buffer (1× SSC is 0·3 m NaCl, 30 mm sodium citrate pH 7), 0·02% SDS, 0·1% N laurylsarcosyl, 2% blocking reagent (Roche Diagnostics) and 50% formamide] at 65°C for 2 h. Hybridization was carried out in the same buffer at 65°C for 16 h in the presence of the probe (100 ng mL−1). After three washes at room temperature in 2× SSC buffer containing 0·1% SDS, the membranes were washed twice for 10 min each time at 65°C in 0·5× SSC, 0·1% SDS. Detection was carried out using an anti-digoxigenin alkaline phosphatase conjugate and the CDP-Star as chemiluminescent substrate following the protocol recommended by the supplier (Roche Diagnostics).

CVA detection by RT-PCR amplification

The CVA-specific RT-PCR assay described by James & Jelkmann (1998), using the CVA-1 and CVA-2 primers (Fig. 1 and Table 2) and allowing the amplification of a 834-bp fragment of the CVA genome, was performed as described by these authors. Alternatively, three primer pairs (CVA-fw1/CVA-rev1, CVA-fw1/CVA-rev1b and CVA-fw2/CVA-rev2, Fig. 1 and Table 2) were evaluated in either one-step or two-step RT-PCRs. The RT-PCR tests were repeated three times.

Table 2.   Sequences and position on the Cherry capillovirus A (CVA) genome of primers used for the specific PCR amplification of CVA isolates
Primer namePrimer sequence (5′–3′)aPositionbReference
  1. aNucleotide degeneracy codes used: M = A or C; Y = C or T; R = A or G.

  2. bPositions provided in reference to the X82547 reference sequence (Jelkmann, 1995).

  3. cPositions provided in reference to the PDO fragment of CVA PF sequence (AF792509).

  4. dPositions provided in reference to the PDO fragment of CVA V590 sequence (AF413922).

CVA-1TTGATTCGTCTCCTGCGACT5454–5435(James & Jelkmann, 1998)
CVA-2GAATACTCAAGCTTACTGAAG4621–4641(James & Jelkmann, 1998)

The one-step RT-PCR assays were performed using 5-μL TNA samples in a 50-μL reaction volume containing 10 mm Tris–HCl pH 8·8, 1·5 mm MgCl2, 50 mm KCl, 0·1% Triton X-100, 250 μm dNTPs, primers at 1 μm each, 0·5 U reverse transcriptase (AMLV RTase, Abgene) and 0·4 U Extra Pol 1 DNA Polymerase (Eurobio). The reverse transcription was carried out at 42°C for 45 min and the RTase was then denaturated at 95°C for 5 min. Forty amplification cycles of 95°C for 30 s, 54°C for 30 s and 72°C for 1 min, followed by a final extension step at 72°C for 10 min were then applied. All amplification products were analysed by nondenaturing agarose gel electrophoresis.

For the two-step RT-PCR assays, 3 μL of TNA samples were submitted to reverse transcription initiated by a dT18 primer and using Supercript®II Reverse Transcriptase according to the manufacturer’s recommendations (Invitrogen). In a second step, the PCR was carried out using 3 μL cDNA in a 50-μL reaction volume as described above, using each primer at 1 μm and 1 U DyNAzymeII DNA polymerase (Finnzymes). Amplification was performed with the same cycling scheme as described above, but for only 35 cycles.

Polyvalent degenerated oligonucleotides (PDO) nested RT-PCR amplification, cloning of amplification products and nucleotide sequence analysis

The PDO nested RT-PCR was applied to TNA extracts as described by Foissac et al. (2005). Amplification products were either directly sequenced (GATC Biotech or CoGenics) or, in cases of mixed infection, sequenced following cloning into the pGEM-T Easy vector (Promega). All sequences were deposited in GenBank and the relevant accession numbers are listed in Table 1.

Multiple sequence alignments were performed using the clustal w program (Thompson et al., 1994) as implemented in mega version 4 (Tamura et al., 2007). Phylogenetic trees were constructed using the neighbour-joining technique using strict nucleotide identity distances and randomized bootstrapping for the evaluation of branching validity. Mean diversities, genetic distances (p-distances calculated on nucleotide or amino acid identity) and Nei-Gojobori synonymous/nonsynonymous substitution rates (Nei & Gojobori, 1986) were calculated using mega version 4 (Tamura et al., 2007).

Smart™ LD-PCR amplification and partial genome determination for CVA isolates V590 and PF

The long-distance (LD) RT-PCR amplifications were performed as recommended by the kit supplier (BD Biosciences). The cDNAs were synthesized from purified total RNAs using a LD-polyT primer (5′-CACTGGCGGCCGCTCGAGCATGTACT25NN-3′) in the presence of Supercript®II Reverse Transcriptase as recommended (Invitrogen). The LD-PCRs were then performed using the LD primer (5′-CACTGGCGGCCGCTCGAGCATGTAC-3′) and the CVA-F-pol-bis primer or the CVA-V590-F primer (Table 2) for the CVA-PF and CVA-V590 isolates, respectively. The PCR conditions followed the manufacturer’s recommendations with 40 cycles using an annealing temperature of 55°C and an elongation period of 4 min at 68°C.

The LD-PCR products were purified using a Nucleotrap kit (BD Biosciences) and cloned into the pGEM-T Easy vector. The sequences of the cloned amplification products were determined by CoGenics. The partial genome sequences were finally assembled using mega version 4 (Tamura et al., 2007).


Analysis of CVA variability

The genetic variability of CVA was evaluated by the analysis of the short sequence around the viral RdRp active site amplified using the polyvalent PDO nested RT-PCR (Foissac et al., 2005). In total, the 275-nt sequence was determined for 38 CVA isolates (Table 1). These sequences were deposited in GenBank under accession numbers HQ267818 to HQ267855. In addition, the corresponding sequence could be retrieved from GenBank for a further 22 isolates, to yield a final dataset consisting of 60 isolates. Using this dataset, average pairwise divergence levels of 9·4% and 2·9% were obtained when considering the nucleotide sequences or the encoded amino acid sequences, respectively (results not shown). Most interesting was the finding that this diversity could be structured into five distinct clusters, as illustrated by the unrooted neighbour-joining tree constructed based on the nucleotide sequences and shown in Figure 2. Besides the major group of isolates (group 3) which contains also the original CVA type isolate (X82547) (Jelkmann, 1995), four additional groups of isolates can be defined with high (89–100%) bootstrap support.

Figure 2.

 Unrooted neighbour-joining phylogenetic tree calculated from the sequence of a partial fragment (275 nt) of the RNA-dependent RNA polymerase gene (Foissac et al., 2005) obtained from a range of Cherry virus A (CVA) isolates. The isolates analysed are listed in Table 1. Isolates for which the corresponding sequence was available in GenBank were included in the analysis and the relevant accession numbers are indicated in Table 1. The host and geographical origin of the isolates are indicated in italics as follows: SC, sweet cherry; PA, Prunus armeniaca (apricot); PM, Prunus mume (Japanese apricot); PD, Prunus domestica (plum); PS, Prunus serrulata; Fr, France; It, Italy; Ch, China; CR, Czech Republic; Ge, Germany; USA, United States of America; In, India. The tree was constructed by the neighbour-joining method from strict nucleotide identity distances and the statistical significance of branches was evaluated by bootstrap analysis (1000 replicates). Only bootstrap values higher than 70% are indicated. The scale bar represents 2% nucleotide divergence. The five identified phylogenetic groups are indicated on the right side of the figure. The isolates indicated by stars were included in the experiments to evaluate the polyvalence of the molecular detection assays.

Besides the high bootstrap values, these five phylogenetic groups were also supported by the differences observed between the intra- and inter-group diversity levels (Table 3). Intra-group average nucleotide pairwise divergence (diversity) ranged from 0·5% for group 5 to 2·9% for group 4. These values are to be compared with inter-group diversity levels ranging from 9·1% (group 1 vs. group 2) to 20·6% (group 1 vs. group 5) (Table 3). Such values are within the range of diversity observed in the same genome region for another fruit tree virus belonging to the family Betaflexiviridae, ACLSV, the type member of the genus Trichovirus (Foissac et al., 2005). When considering amino acid sequences, inter-group diversity levels fell to values between 1·1% and 8·3% (Table 3), indicating that most of the nucleotide changes were silent. This was confirmed by the observation that the average sequence diversity for the complete dataset for the first, second and third positions of codons were 3·0%, 2·7% and 21·9%, respectively, and by the comparison of the mean genetic distances between isolates calculated for synonymous (28·0%) and nonsynonymous (1·6%) mutations.

Table 3.   Inter-group average nucleotide (below diagonal) and amino acid (above diagonal, in italics) diversities in the polyvalent degenerated oligonucleotide fragment of the RNA-dependent RNA polymerase gene for the five phylogenetic clusters of Cherry capillovirus A isolates
 Group 1 (%)Group 2 (%)Group 3 (%)Group 4 (%)Group 5 (%)
Group 1 3·83·03·48·3
Group 29·1 2·21·17·1
Group 312·89·9 1·96·6
Group 418·216·618·1 7·1
Group 520·618·715·917 

Remarkably, group 2 clustered the CVA isolates from noncherry hosts [apricot, Japanese apricot (P. mume) and plum]. There were only two exceptions to this observation: isolate P1 from sweet cherry which, although somewhat isolated, clustered in group 2, and isolate PM3 obtained from P. mume in China (EU730951, Marais et al., 2008a), which clustered within group 1 (Fig. 2). The association of isolate PM3 with group 1 was, however, somewhat distant, as its average pairwise nucleotide divergence with group 1 isolates reached 7·4%, compared with the corresponding overall intra-group 1 value of 1·9%.

Evaluation of the polyvalence of available CVA detection assays

Considering the molecular variability observed in the conserved region of the RdRp gene, the ability of detection assays to detect members of the five CVA lineages was evaluated. Three techniques were compared to the polyvalent PDO nested RT-PCR: the CVA-specific RT-PCR assay developed by James & Jelkmann (1998) and the dot-blot hybridization assays described by the same authors and using probes hybridizing either to the 3′ end of the genome (from the movement protein gene to the 3′ NCR) or to the 5′ part of the genome (including the 5′NCR and the beginning of the RdRp gene) (Jelkmann, 1995). This comparative study was performed using a selection of 36 CVA isolates representative of the five identified clusters (Fig. 2 and Table 4).

Table 4.   Comparison of four RT-PCR assays for the detection of isolates belonging to the five phylogenetic groups of Cherry capillovirus A (CVA)
  1. +, positive signal; −, negative signal; nt, not tested.

  2. aMolecular hybridization using the two CVA probes.

  3. bRT-PCR 1/2: RT-PCR assay using the CVA1 and CVA2 primers (James & Jelkmann, 1998).

  4. cRT-PCR A: RT-PCR assay using the CVA-fw1 and CVA-rev1 primer pair (Table 2).

  5. dRT-PCR B: RT-PCR using the CVA-fw1 and CVA-rev1b primer pair (Table 2).

  6. eRT-PCR C: RT-PCR using the CVA-fw2 and CVA-rev2 primer pair (Table 2).

Group 1944+++++
Group 2A5/14++++
Group 3J4+++++
Group 4J2+++++
Group 5V590++++

Except for one isolate (944), there was a perfect coincidence between the hybridization results obtained with the two probes. Moreover, 93% (29/31) of the isolates tested were detected by the dot-blot molecular hybridization assays (Table 4). By contrast, only 59% of the tested isolates (20/34) were detected by RT-PCR using the CVA-1 and CVA-2 primers. In addition, this rate of detection varied considerably depending on the phylogenetic group of isolates considered: all group 3 (13/13) and group 4 (1/1) isolates tested were detected, but only 6/10 members of group 1 were detected, and none of the nine isolates tested from group 2 were amplified. Similarly, the sole isolate tested for group 5 was not amplified (Table 4). These results indicate that dot-blot hybridization shows a generally broad polyvalence, allowing the detection of isolates from all phylogenetic groups, but may lack somewhat in sensitivity compared with a RT-PCR assay. They also indicate that the CVA-1/CVA-2 primer pair failed to detect a number of CVA isolates. In order to be able to develop new, more polyvalent primers for CVA detection, partial genome sequences were obtained for two divergent CVA isolates.

Partial genomic sequencing of the divergent PF and V590 CVA isolates

Two isolates representative of group 2 (CVA-PF) and group 5 (CVA-V590), which were not detected using the CVA-1/CVA-2 primer pair, were selected for partial genomic sequencing. For each isolate, a specific forward primer (CVA-F-pol-bis or CVA-V590-F, respectively, Table 2) was selected in the PDO fragment and used to amplify the 3′ genome half by Smart Long Distance PCR (LD-PCR). The sequences of the ∼3·2 kb amplified fragments were determined and compared with each other and with that of the reference CVA isolate (X82547; Jelkmann, 1995). The partial CVA-PF and CVA-V590 sequences were deposited in GenBank under accession numbers HQ267856 and HQ267857, respectively.

Sequence comparisons were performed on the 3208-nt fragment available for all three isolates and comprising the 3′NCR and part of the ORF encoding the RdRp/CP fusion protein. Excluding the polyA tail, the 3′NCR was 297 nt (CVA-V590) and 299 nt (CVA-PF) long, values that closely match the 302-nt size of the 3′NCR of the reference isolate. Overall, the CVA-V590 and CVA-PF nucleotide sequences diverged by 14·4%. They diverged from the reference sequences by 14·3% and 12·3%, respectively (Table 5). However, this variability was not equally distributed along the sequenced fragment. Although it was the only part of the sequenced genome showing indel (insertion-deletion) polymorphism (data not shown), the 3′NCR was significantly more conserved, showing only 5·7–8·7% divergence between isolates. Conversely, the coding region upstream of the 3′NCR was more variable and showed 12·7–15·2% divergence between isolates (Table 5). This region in fact harboured two ORFs, the long ORF encoding the RdRp-CP typical of capilloviruses and, located internally and in another reading frame, the shorter ORF encoding the movement protein of 463 amino acids (Yoshikawa et al., 1992; Jelkmann, 1995).

Table 5.   Pairwise nucleotide and amino acid sequence divergence observed between the three Cherry capillovirus A (CVA) isolates CVA-PF, CVA-V590 and the CVA reference isolate (X82547) in the sequenced 3′ terminal 3·2-kb fragment
Isolates comparedaORF 1 (partial)b (%)ORF 2 (%)3′ NCRc (%)Overalld (%)
  1. aV590, CVA-V590 sequence; PF, CVA-PF sequence; X82547, CVA reference isolate, accession number X82547 (Jelkmann, 1995).

  2. bAnalysis of the partial open reading frame 1, containing the RdRp gene fused at its 3′ end with the CP gene.

  3. c3′ NCR: 3′ noncoding region.

  4. dAnalysis using the complete 3·2-kb sequence.

  5. eNucleotide sequence divergence.

  6. fAmino acid sequence divergence.


As expected, the variability was mostly concentrated on the third base of the codons, as shown by the separate calculations of mean sequence diversity on the three bases of the codons, which gave values of 2·1%, 0% and 34·8% for the first, second and third bases, respectively. However, this pattern was shifted for the overlapping ORF 2 which accumulated mutations on the second base of the codons. This region of overlap was expected to be more constrained from an evolutionary point of view because selection pressures acting on the two overlapping ORFs are at least partially compounded. As a consequence, this region was expected to accumulate fewer mutations and therefore to show a reduced variability. This was indeed the case, since the pairwise nucleotide divergence between the three isolates in this region was between 8·4% and 9·4%, as compared to ∼50% higher values of 12·7–15·2% observed for the whole coding region (Table 5).

Development of polyvalent CVA-specific RT-PCR assays

The identification of conserved domains in the CVA genome allowed the design of three primer pairs targeting conserved regions between the three isolates for which sequences were available, with the hope that these primers would allow the detection of all known CVA isolates. Three primers (CVA-fw1, CVA-rev1 and CVA-rev1b, Fig. 1 and Table 2), allowing the development of two primer pairs sharing the same forward primer (CVA-fw1), were selected in the region of overlap between ORF 1 and ORF 2. They amplified fragments of 302 or 443 bp, depending on the reverse primer used. A third primer pair (CVA-fw2/CVA-rev2, Fig. 1 and Table 2) targeted the 3′ end of the CP gene and a part of the 3′NCR and yielded an amplification product of ∼340 bp, depending of the CVA isolate considered. When possible, the sequences of the primers ended at the second base of the codon in ORF 1, which is the most conserved position. When necessary, primers were degenerated at variable positions (twofold degeneracy), but the primers were selected so that these variable positions were always close to their 5′ end.

Initial attempts to use the CVA-fw1/CVA-rev1 primer pair in a one-step RT-PCR assay indicated a lack of sensibility or reproducibility of the technique (data not shown). A two-step RT-PCR was therefore carried out, yielding the amplification products of the expected size (302 bp) for both CVA-PF and CVA-V590 (Fig. 3). The two other primer pairs were therefore evaluated using the same two-step RT-PCR procedure, again yielding the expected amplification results (Fig. 3). In all cases, sequencing of the amplified fragments verified the specificity of the amplification.

Figure 3.

 Detection of CVA-PF and CVA-V590 isolates by RT-PCR using the three new primer pairs. Lanes 1, 4, 7 and 10: amplification products using the CVA-fw1/CVA-rev1 primer pair. Lanes 2, 5, 8 and 11: amplification products using the CVA-fw1/CVA-rev1b primer pair. Lanes 3, 6, 9 and 12: amplification products using the CVA-fw2/CVA-rev2 primer pair. Lanes 1, 2 and 3: healthy control. Lanes 4, 5 and 6: CVA-PF (propagated in GF305 peach seedlings). Lanes 7, 8 and 9: CVA-PF (from the original host). Lanes 10, 11 and 12: CVA-V590 (propagated in sweet cherry plants). L: molecular weight markers. Arrows on the right side of the figure indicate the size of representative marker bands.

In order to confirm the ability of the new RT-PCR assays to broadly detect CVA isolates, amplifications were again performed on a range of CVA isolates representing all phylogenetic groups and the results are presented in Table 4. Overall, the RT-PCR assay using CVA-fw1 and CVA-rev1 primers (RT-PCR A) seemed to be the most polyvalent compared to the other two assays, as it was able to detect all tested isolates, whatever the phylogenetic group to which they belonged. The RT-PCR B and RT-PCR C assays were found to be slightly less polyvalent, with a detection rate of 95·6%. Nucleotide alignment of the different primer pairs on the recently published sequence of an Indian CVA isolate (GenBank accession number FN691959; Noorani et al., 2010) confirmed this observation. Indeed, whereas the sequences of primers CVA-fw1 and CVA-rev1 (RT-PCR A) were found to be 100% identical to the Indian isolate sequence, the CVA-fw2 and CVA-rev2 primers showed one mismatch each. These results confirm the ability and the improved polyvalence of the new assays and, in particular, their ability to detect isolates of groups 1, 2 and 5 that were poorly detected or not detected at all by the previously available assay.


The results presented here provide evidence for the existence, within the CVA species, of a significantly wider diversity than was previously known (Foissac et al., 2005). Part of the reason for this under-representation of CVA diversity might in fact stem from the use of assays of limited polyvalence for CVA detection, with the result that some phylogenetic clusters were either under-represented or altogether absent from previous studies. The existence of several phylogenetic groups of isolates, one of which seems to be specifically associated with noncherry hosts, is an interesting finding, and indicates that there is probably very limited movement of isolates between cherry and noncherry hosts. Like all other viruses in the genus Capillovirus (Jelkmann, 1995; Adams et al., 2005) and many other agents in the closely related genera Trichovirus and Foveavirus (Adams et al., 2005; Martelli et al., 2007), CVA has no known vector and its transfer from plant to plant is considered to occur essentially through vegetative propagation techniques. In this context, it is noteworthy that cherry species are very rarely grafted, if at all, on noncherry species, which might largely explain the barrier to CVA transmission between these two groups of hosts. It is, however, interesting to note that for another widespread virus in the family Betaflexiviridae with no known vector and infecting these hosts, ACLSV, previous studies have failed to provide evidence for host-based differentiation of isolates (Al Rwahnih et al., 2004). In a similar fashion, Magome et al. (1997) failed to find evidence for host-specific clusters of isolates in the case of the type member of the genus Capillovirus, ASGV, despite the fact that it has been reported from very diverse and unrelated hosts such as apple and pear, citrus and lily.

The results obtained with a large panel of CVA isolates representative of the species’ diversity indicate that dot-blot molecular hybridization with either of the two tested probes shows high polyvalence, as over 90% of tested isolates were detected. Given that representative members of each phylogenetic cluster could be detected, the few isolates that were not detected likely reflect a sensitivity problem rather than a lack of polyvalence. The two cRNA probes used were relatively long ones (about 1·8 and 2·5 kb for probes CVA-39 and CVA-105 respectively), so that their overall identity to specific CVA isolates was expected to reflect the overall diversity of the virus, estimated by the 13·7% average nucleotide divergence between the two partially sequenced isolates analysed here and the reference CVA isolate. Such a divergence level between a cRNA probe and its target is not expected to abolish hybridization, but can reduce sensitivity 10- to 100-fold (Wetzel et al., 1990).

The average 13·7% divergence between isolates translates into an average of two to three mismatches on a 20-mer PCR primer. Given the fact that existing PCR primers were selected in the absence of information on the virus’ variability (James & Jelkmann, 1998), it is not altogether surprising to discover that a range of CVA isolates were not amplified using them. The new primers reported here were selected in regions of the genome that had been identified as being under stronger conservative selection pressure and, specifically, in short regions shared by some of the most divergent CVA isolates known. While one isolate was missed by two of the new primer pairs (CVA-fw1/CVA-rev1b and CVA-fw2/CVA-rev2), the third primer pair, CVA-fw1/CVA-rev1, detected all isolates tested. The detection assay based on them therefore represents a large improvement over pre-existing detection techniques in terms of sensitivity and of polyvalence.

Capilloviruses are unique in the family Betaflexiviridae as having a large ORF which encodes a fusion protein of the viral RdRp and CP (Adams et al., 2004). These two domains are joined by a more variable linker domain which overlaps with the ORF 2 encoding the viral movement protein in another reading frame. The precise genome expression strategy for capilloviruses is still somewhat unclear, but recent results demonstrate that ASGV produces a CP subgenomic messenger RNA (Tatineni et al., 2009) and that artificial truncation of ORF 1 so that the CP can no longer be expressed as a fusion to the RdRp does not abolish infectivity or the ability to systemically invade host plants (Hirata et al., 2010). As a consequence of these findings, the expression model proposed by Hirata et al. (2010) proposes that ORF 1 is expressed as three proteins: the viral RdRp alone, produced through ribosomal frameshifting leading to premature termination of ORF 1 translation, the CP alone, produced from a subgenomic messenger RNA, and the RdRp-CP fusion, produced by translation of ORF 1. Although this last protein is dispensable for host invasion, it appears to modulate viral accumulation and, as a consequence, pathogenicity. These results led Hirata et al. (2010) to speculate that trichoviruses might have evolved from a capillovirus ancestor. Such a scenario seems unlikely, however, since evolution of overlapping ORFs is generally considered to happen mostly through invasion of more evolutionary ancient ORFs by the incremental increase in overlap between originally contiguous genes and in which, therefore, a nonoverlapping or short-overlap organization is the ancestral state (Keese & Gibbs, 1992; Lartey et al., 1996; Belshaw et al., 2007). In this more likely scenario, capilloviruses would have evolved from a trichovirus by an increase of the regions of overlap between the ORF 1 (RdRp)/ORF 2 (MP) and between the ORF 2 and ORF 3 (CP) to reach a stage in which ORF 1 and ORF 3 are finally fused together, as in the current Capillovirus genetic organization. Such a scenario is consistent with the MP encoded by ORF 2 being more widespread (i.e. shared between different evolutionary lineages) and more conserved than the overlapping RdRp-CP linker sequence, and therefore ancestral. This is also in line with the very different evolutionary pressures acting on the two ORFs, as indicated by the ratio of nonsynonymous over synonymous mutations (dNS/dS, modified Nei-Gojobori distances), with values of 0·23 and 0·733 for ORF 2 and for the overlapping part of ORF 1, respectively, a situation similar to the one reported by Magome et al. (1997) for ASGV.

One central remaining question in the understanding of CVA biology concerns its potential ability to cause symptomatic infection in its host plants. So far, CVA has not been associated with specific symptoms and is largely considered to induce only latent infections. The objectives of the surveys used in the present work to gather CVA isolates included identifying the agent responsible for a new cherry decline disease in southern France (Marais et al., 2008b). The results obtained showed that CVA had high prevalence and diversity but also that it was not statistically associated with plants with symptoms. The new detection assays reported here extend the ability to detect divergent isolates of CVA in cherry and also in noncherry hosts and should prove useful in the future to evaluate any new results trying to link CVA or at least some of its isolates or strains to specific symptomatology in cherry or in noncherry hosts.


The authors thank Dr M. Navratil (Palacky University, Olomouc, Czech Republic) for providing some CVA samples and Dr W. Jelkmann (Julius Kühn Institute, Dossenheim, Germany) for providing the p39 and p105 plasmids and for useful discussions.