Detection of Flavescence dorée and Bois noir phytoplasmas, Grapevine leafroll associated virus-1 and -3 and Grapevine virus A from the same crude extract by reverse transcription-RealTime Taqman assays

Authors


*E-mail: s.palmano@ivv.cnr.it

Abstract

Multiple detection of the phytoplasmas associated with Flavescence dorée (FD) and Bois noir (BN) diseases and of the viruses Grapevine leafroll associated virus -1 and -3 (Ampelovirus) and Grapevine virus A (Vitivirus) is described, using the same crude extract as template. Sap was prepared by semi-automatic maceration requiring minimal time and effort, consisting of a tissue grinding step in carbonate buffer and a boiling step in glycine buffer; two microlitres were used as template in each pathogen-specific assay. RealTime reverse transcription (RT)-PCR for FD phytoplasma detection was found to be five orders of magnitude more sensitive than the RT-PCR method described previously. However, the RealTime RT-PCR assay for the detection of BN phytoplasma needed a nested step to achieve high sensitivity, suggesting low concentration of template in the host. The viruses were detected by RealTime nested-PCR, which was more sensitive than the ELISA and RealTime RT-PCR assays previously described. The methods presented here have been successfully used to monitor infections in field and nursery samples during the 2008 grapevine growing season.

Introduction

Grapevine yellows diseases (GYs) caused by phytoplasmas (Boudon-Padieu, 2003) and virus diseases (Martelli, 2006) seriously threaten grape (Vitis vinifera) cultivation in Europe. Flavescence dorée (FD) and Bois noir (BN) are the most serious GYs and, although displaying similar symptoms in V. vinifera, are caused by genetically distinct organisms. FD has been reported in several European countries, where it is subject to quarantine restrictions (Maixner et al., 2006). This phytoplasma belongs to the 16SrV ribosomal group (elm yellows group; Lee et al., 2004) and is transmitted by the monophagous grapevine-limited leafhopper Scaphoideus titanus (Schvester et al., 1963).

Bois noir is the most widespread GY disease in Europe and in the Mediterranean Basin (Maixner et al., 2006). The aetiological agent belongs to 16SrXII ribosomal group (Stolbur group; Lee et al., 2000). The cixiid Hyalesthes obsoletus transmits BN phytoplasma to grapevine (Sforza et al., 1998), but widespread disease occurrence suggests the existence of other vectors (Weintraub & Beanland, 2006).

Among viruses, Grapevine leafroll associated virus -1 and -3 (GLRaV-1, -3) and Grapevine virus A (GVA) are considered to be the most important affecting the grapevines of Piedmont, northern Italy (Dr F. Mannini, CNR, personal communication). GLRaV-1 and -3 are classified in the genus Ampelovirus (Ling et al., 2004); leafroll diseases occur in all major grape-growing regions, reducing productivity and quality (Cabaleiro et al., 1999). GVA is a member of the Vitivirus genus and is associated with the rugose wood complex, reported from almost all grape-growing regions of the world, and mainly characterized by stem pitting and stem grooving (Martelli, 1997).

Efforts have recently been made to develop rapid and sensitive methods for phytoplasma and virus detection in grapevine. A protocol for rapid diagnosis of the FD phytoplasma (FDp) from crude extracts has been described (Margaria et al., 2007) and reverse transcription (RT)-PCR diagnosis of grapevine viruses from sap has been reported (La Notte et al., 1997; Rowhani et al., 2000; Dovas & Katis, 2003a,b; Nakaune & Nakano, 2006; Osman & Rowhani, 2006, 2008; Osman et al., 2007, 2008). In this study, a RealTime RT-PCR assay has been developed for detection of FDp and RealTime nested-PCR assays following RT-PCR for detection of BN phytoplasma (BNp), GLRaV-1 and -3, and GVA using the same sap extract as template. The protocols were used in the 2008 season to test grapevine samples for the five pathogens. As shown, the low-cost and rapid extraction, and fast detection procedures make the protocols highly suitable for use as diagnostic tools that can be used to reliably test up to 30 samples in a few hours. The methods are particularly recommended for large-scale screening of vineyards and nurseries, pathogen surveys, epidemiological studies, evaluation of propagation material and to support strategies for crop protection.

Materials and methods

Samples and reference strains

Initial screening for phytoplasmas was performed on sap extracts from periwinkle (Catharanthus roseus) plants infected by FD and Stolbur phytoplasmas from the authors’ collection. After validation, the assays were performed in routine conditions on 100 grapevine samples collected in various vineyards of typical wine areas or from nurseries in the Piedmont region.

Reference phytoplasma strains were maintained by grafting in periwinkle and were: American aster yellows (AAY, 16SrI), vaccinium witches’-broom (VWB, 16SrIII), Flavescence dorée (16SrV), Stolbur (16SrXII) and apple proliferation (AP, 16SrX), from sources previously described (Margaria et al., 2007).

RealTime nested-PCR for virus detection was validated on grapevine samples from a field collection infected by GLRaV-1, GLRaV-3 and GVA local strains. A large-scale analysis was carried out on 100 grapevine samples collected in the field in the 2008 season.

Negative controls consisted of healthy periwinkles or healthy grapevines maintained in the greenhouse; positive controls for phytoplasma assays were periwinkles infected with FDp or Stolbur phytoplasma maintained in the greenhouse by grafting.

Crude sap preparation

The template for RT-PCR and RealTime RT-PCR was prepared in Bioreba bags (Celbio) from 0·5–1 g midribs of leaves with symptoms, petioles or subcortical tissue by adding 20× (w/v) carbonate extraction buffer (15 mm Na2CO3, 34·9 mm NaHCO3, 2% polyvinylpyrrolidone 40, 0·2% bovine serum albumin, 0·05% Tween 20, 1% Na2S2O5, pH 9·6), and macerating with a rolling ball (RDM-50A press, Rexon). Four microlitres of the extract were boiled in 50 μL of denaturing buffer (0·1 m glycine, 0·05 m NaCl, 1 mm EDTA, 0·5% Triton X-100) to release the nucleic acid, and immediately chilled on ice; an aliquot of 2 μL was used as template for the RT-PCR or for the RealTime RT-PCR.

FDp detection by RealTime RT-PCR

A single-tube Taqman qRT-PCR assay was developed for FDp detection. The reaction mix of 25 μL included: 2 μL of template, 200 μm dNTPs, 5·5 mm MgCl2, 200 nm probe (Table 1), 300 nm of each primer (Table 1), 5 U MuLV reverse transcriptase (Applied Biosystems-AB), 2 U RNaseOUT (AB), 0·62 U AmpliTaqGold Polymerase (AB), 10× TaqMan® Buffer A (AB). The primers were designed based on the 16S rDNA sequence by alignment with homologous sequences of phytoplasmas belonging to several ribosomal groups. The 5′ and 3′ ends of the FD probe (Biosense Srl) were labelled with the fluorescent dyes FAM-6 (6-carboxyfluorescein) and TAMRA (tetra-methylrhodamin), respectively. RealTime RT-PCR reactions were performed in 96-well Optical Reaction plates using the StepOnePlusTM Real-Time PCR System (AB) and the data were analyzed using StepOneTM software v. 2·0. Amplification parameters were as follows: 52°C for 30 min (reverse-transcription), 95°C for 10 min (AmpliTaq activation) and 35 amplification cycles, consisting of 15 s at 95°C and 1 min at 58°C. Positive control (FDp infected periwinkle), healthy control (healthy grapevine) and water were included in each assay.

Table 1.   List of primers and probes used for detection of Flavescence dorée phytoplasma (FDp), Bois noir phytoplasma (BNp), Grapevine leafroll associated virus 1 and 3 (GLRaV-1, GLRaV-3), and Grapevine virus A (GVA)
PathogenReactionPrimer/probe namePrimer/probe sequenceReference
FDpRealTime RT-PCR395F
480R
FD-P
gccgcgtgaacgatgaa
gaataacgtcaagatagtttttccact
tttcggtatgtaaagttct
This study
BNpRT-PCR190F
660R
gagataagaaggcatcttctta
aacagtttttatagcatcacaa
This study
RealTime nested-PCR210F
280R
BN-P
cttcttatttttaaaagacctagcaatagg
gtcttggtaggccattaccc
ttagggaagagcttgcgtca
This study
GLRaV-1RT-PCRGLRaV1-ExtF
GLRaV1-ExtR
tgttgagggcgagtctaatcg
gatacataacgatacgagtgtgctga
This study
RealTime nested-PCRHsp70-149F
Hsp70-293R
Hsp70-225P
acctggttgaacgagatcgctt
gtaaacgggtgttcttcaattctct
acgagatatctgtggacgga
Osman et al., 2007
GLRaV-3RT-PCRGLRaV3-ExtF
GLRaV3-ExtR
gaccctgttattgctgttatgactgg
caccacccactgtgtatctgtga
This study
RealTime nested-PCR56F
285R
181P
aagtgctctagttaaggtcaggagtga
gtattggactacctttcgggaaaat
caggtaatagcggactgagactggtggaca
Osman & Rowhani, 2006
GVART-PCRGVA-ExtF
192R1
gcgaaagcccagccaactga
tttgctagctttaggacctactatatctacct
This study
Osman & Rowhani, 2008
RealTime nested-PCR77F1
GVA-R
104P
cgaccgaaatatgtacctgaatactc
cataatgaaccgcctttttgct
cttcgggtacatcgccttggtcgg
Osman & Rowhani, 2008
This study
Osman & Rowhani, 2008

In order to check the role of DNA as template in the extracts and evaluate the sensitivity of RT-PCR compared to PCR, 14 leaf samples of FDp-infected periwinkle plants maintained in the greenhouse for 2 months after grafting, were amplified with or without reverse transcriptase in the reaction mixture. Furthermore, as a control for the quality of the RNA in the crude extract, a fragment of the 18S rRNA gene of grapevine and periwinkle was also amplified using primers and conditions reported by Osman et al. (2007). Healthy periwinkle and grapevine, and water controls were included in the assay.

Specificity of the Taqman system was checked on total RNA extracts (RNeasy Plant Minikit, Qiagen) prepared from periwinkle plants singly infected with reference phytoplasma strains; RNA extracted from healthy periwinkles grown from seed and maintained in insect proof greenhouses were used as healthy controls. Two microlitres of RNA were used as template following the protocol described above.

The limit of detection (LOD) (Arinbruster et al., 1994; Boben et al., 2007) for RealTime RT-PCR was determined and compared to the values of the RT-PCR method described previously (Margaria et al., 2007). Plasmid pFD81-03, carrying the target sequence of the FDp 16S rRNA gene (Galetto et al., 2005), was linearized using SacI restriction enzyme (NEB) and the 16S rDNA fragment was transcribed using T7 RNA polymerase, following the MAXIscript Kit (AB) protocol. A DNase step was performed after transcription and RNA was purified by standard phenol-chloroform procedures (Sambrook et al., 1989). Purified RNA was quantified by spectrophotometry (model DU®530-Beckman) and used to prepare serial dilutions in crude sap of healthy grapevine. The series ranged from 1 ng μL−1 to 10−7 ng μL−1, which corresponded to 1·1×102 copies of transcript per microlitre. Three replicates per dilution were assayed in each RealTime RT-PCR run and three independent runs were performed. A run was considered positive when at least two of the three replicates were positive. The slope of the standard curve, the square regression coefficient (R2) and amplification efficiency were determined for each independent run (Wong & Medrano, 2005). RT-PCR was performed as previously described (Margaria et al., 2007), using the same dilutions as template. Three independent runs were performed. The amplification product was separated on a 2% agarose gel in TBE (Tris-Borate-EDTA) buffer (Sambrook et al., 1989) and visualized on a UV-transilluminator after staining in ethidium bromide. Crude sap from healthy grapevine was used as negative controls in both the assays.

BNp detection by RealTime nested-PCR

A preliminary test on 20 grapevines which tested negative for FDp but showed typical symptoms of phytoplasma infection (Margaria et al., 2007), was performed by RealTime RT-PCR as described above for FDp, using primers 210F-280R and a probe designed on the BNp 16S rRNA gene (Table 1). All of them gave negative results. To obtain the expected positive results, a two-step protocol was developed, consisting of RT-PCR from sap followed by a RealTime nested-PCR. One-tube RT-PCR reactions were performed using 2 μL of boiled extract (the same as prepared for FDp detection) as template, following the protocol previously reported (Margaria et al., 2007). Primers 190F and 660R (Table1) were used at an annealing temperature of 53°C. A 1:100 dilution in sterile double-distilled water of the first amplicon was used as template in the RealTime nested-PCR. Reactions were prepared in 10 μL total volume as follows: 2× Taqman Fast Universal PCR Master mix (AB), 1 μL template, 200 nm probe (Table 1), and 300 nm each primer (Table 1). The probe (Biosense Srl) was labelled with FAM-6 at the 5′-end as reporter and with black-hole-1 (BHQ-1TM) at the 3′-end as quencher. Amplification consisted of 20 s at 95°C and 35 cycles of 1 s at 95°C and 20 s at 58°C. The reaction was completed in 40 min. Positive controls, healthy controls and water were included in each assay.

Primer specificity for BNp was tested by RealTime RT-PCR on reference strains, as described above for FDp. LOD parameters of RealTime RT-PCR and RealTime nested-PCR for BNp detection were determined on transcripts prepared from a pGEMT-BN16S plasmid carrying the target sequence of the 16S rRNA gene of BNp (amplicon obtained using primers P1a/P7a - Lee et al., 2004), using NcoI restriction enzyme for vector linearization and SP6 RNA polymerase for the transcription reaction. Serial dilution series were prepared as described above for FDp: the tenfold dilution series ranged from 1 ng to 10−9 ng transcript per microlitre sap, corresponding to 2·3 copies of transcript per microlitre.

Virus detection

Preliminary RealTime RT-PCR assays were performed on a collection of GLRaV-1, -3 and GVA infected plants to test for efficient primer coupling for local isolates, and compared with DAS-ELISA (double sandwich-enzyme linked immunosorbent assay). One-tube RealTime RT-PCR reactions were performed as described above for FDp detection, using an annealing temperature of 58°C. Primers were: 149F-293R for GLRaV-1; 56F-285R for GLRaV-3 and 77F1-192R1 for GVA, in combination with the corresponding probes (Table 1). GLRaV-1, GLRaV-3 and GVA probes (Biosense Srl) were labelled at the 5′-end with FAM-6 as reporter and BHQ-1TM as quencher. Positive controls (total RNA purified from infected grapevines using RNeasy Plant Minikit, Qiagen), healthy controls and water were included in each assay.

ELISA was performed with a commercial kit (Agritest Srl), according to the manufacturer’s recommendation. Plant sap was extracted from 1 g of leaf midribs by homogenization in 10 mL of ELISA extraction buffer (0·5 M Tris-HCl pH 8·2, 0·14 m NaCl, 2% PVP MW 24000, 1% PEG MW 6000, 3 mm NaN3, 0·05% v/v Tween 20); 200 μL were loaded in each well. Signal was detected by measuring absorbance (Microplate reader 3550, Biorad) at 405 nm after 1 and 2 h incubation at room temperature with 1 mg mL−1 p-nitrophenyl phosphate. ELISA results for each sample were taken as the average absorbance value of two replicates per sample. In each assay, positive and negative controls (supplied with the kit) were included in the plate. Samples were considered positive when average absorbance was at least three times the average values for the healthy control.

Preliminary results, comparing ELISA and RT-PCR, underlined the need to achieve higher assay sensitivity by developing a virus-specific RealTime nested-PCR. RealTime nested-PCR was performed as follows: first step RT-PCR was prepared as previously described (Margaria et al., 2007), using an annealing temperature of 56°C, for the three combinations of primers (Table 1). The primers were first tested for specificity on RNA extracted from grapevine plants singly infected by each virus. Reaction mixtures and conditions of the RealTime nested-PCRs were as described for BNp, using primers and probes reported in Table 1.

RealTime nested-PCR was compared with DAS-ELISA in the summer of July 2008 on 42 plants which had previously been tested by ELISA during winter 2007.

The limit of detection was determined for each virus-specific ELISA, RealTime RT-PCR and RealTime nested-PCR on serial dilutions of three independent samples singly infected by each virus, as shown by preliminary ELISA tests. Six canes were collected from each plant. Two grams of bark tissue were prepared by cambial scraping of the canes, 1 g was used for ELISA and 1 g for preparation of crude sap for PCR assays. Tenfold dilutions were prepared in sap from cambial material of healthy plants. Each assay was performed as described above.

Results

FDp detection

The efficiency of the FDp assays was above 98%, square regression coefficient (R2) values were above 0·99 (theoretical optimal value = 1) and slope values ranged between −3·3 and −3·1 (optimal value = −3·3). Healthy and water controls always remained below the threshold. The specificity of the primers-probe set was tested on reference phytoplasma strains from periwinkle, belonging to ribosomal groups 16SrI, III, V, X and XII: fluorescent signal was detected only from FDp-infected samples (results not shown).

To check that RNA was acting as a template and not just residual DNA, the sensitivities of the RealTime RT-PCR and RealTime PCR assays were compared on 14 FDp-infected periwinkle samples by performing reactions with or without reverse-transcription enzyme in the reaction mix. Ct values for RealTime RT-PCR ranged from 18·7 to 25·7, whilst those for RealTime PCR ranged from 26·0 to 33·0, and the difference between values for an individual sample was between 5·2 and 9·1 (Table 2). Ct values of the 18S amplicon ranged between 17·4 and 20·3. No amplification was detected from healthy grapevine or the water control (Table 2).

Table 2.   Comparison between RealTime PCR and RealTime RT-PCR for detection of Flavescence dorée phytoplasma (FDp) 16S rRNA and plant 18S rRNA from crude sap of 14 periwinkle samples and negative controls
SampleFDp 16S rRNAPlant 18S rRNA
RealTime RT-PCRRealTime PCRCta differenceRealTime RT-PCR
  1. aCt = threshold cycle.

 124·130·76·619·7
 223·030·27·220·3
 325·733·07·319·6
 423·028·25·218·8
 519·826·66·817·4
 622·930·27·319·8
 722·530·27·720·0
 821·928·66·719·7
 923·832·68·819·6
1021·729·27·519·4
1121·128·17·018·6
1218·928·09·119·5
1318·727·08·317·9
1419·526·06·517·4
Healthy periwinkle18·9
Healthy grapevine19·2
Water

The results for the dilution series assays, in which RNA obtained from in vitro transcription of plasmid DNA was diluted in plant sap, are reported in Table 3 and show a LOD value of 10−5 ng μL−1 for the RealTime RT-PCR assay for FDp. In contrast, only the undiluted sample could be detected by RT-PCR in all three independent runs; the LOD in this case was 1 ng μL−1. The RealTime RT-PCR assay was therefore five orders of magnitude more sensitive than the RT-PCR assay.

Table 3.   Results of limit of detection (LOD) determination, between RT-PCR and RealTime RT-PCR for Flavescence dorée phytoplasma (FDp) detection, and RealTime RT-PCR and RealTime nested-PCR (following RT-PCR) for Bois noir phytoplasma (BNp) detection
ng transcripta per μL sapFDpBNp
RealTime RT-PCRRT-PCRRealTime RT-PCRRealTime nested-PCR
123123123123
  1. aTranscripts were prepared by in vitro transcription of pFD81-03 (for FDp) or pGEMTBN 16S (for BNp) plasmids carrying a fragment of the 16S rRNA gene.

  2. bnt = not tested.

0
1++++++++++++
10−1+++++++++
10−2+++++++++
10−3+++++++++
10−4+++++++++
10−5+++++++++
10−6+++
10−7+++
10−8ntbntntntntntntntnt+++
10−9ntntntntntntntntnt+++

BNp detection

The efficiency of the BNp assay by RealTime RT-PCR on 16S rRNA transcript serial dilutions was above 95·2%, R2 was above 0·99 and the slope gradient ranged between −3·4 and −3·3. Despite these good values, it was necessary to introduce a nested step in the Real Time RT-PCR for sensitive BNp detection. In fact, in a preliminary test on 20 grapevine plants, which tested negative for FDp but clearly showed typical phytoplasma symptoms, the BNp RealTime RT-PCR detected no infection, except for the periwinkle positive control. Higher sensitivity was obtained by performing a first RT-PCR followed by a RealTime nested-PCR, which allowed detection of BNp in all the samples, and there were no false positives.

The primers-probe set tested on reference phytoplasma strains from periwinkle gave a fluorescent signal from Stolbur (16SrXII) and AAY (16SrI) infected samples (results not shown) but was negative for all others. This positive AAY reaction is further discussed below.

The results for the dilution series assays, in which in vitro transcribed RNA was diluted in plant sap, gave a LOD value for RealTime RT-PCR of 10−5 ng μL−1 for BNp, which was the same as that obtained for FDp (Table 3). For RealTime nested-PCR, the level of detection was even lower at 10−9 ng μL−1 in all three independent runs (Table 3).

Virus detection

Primer pairs were checked on RNA extracted from grapevine plants singly infected by each virus and no cross-reactivity was observed.

In initial experiments ELISA and RealTime RT-PCR were compared: while ELISA detected GLRaV-1, -3 and GVA in 13/20, 20/20 and 20/20 samples, respectively, RT-PCR detected 0/20, 0/20 and 11/20, respectively. Moreover the Ct values for GVA ranged between 31·1 and 34·8 (results not shown), close to the detection limit of the assay. Negative and positive controls were included in each assay, and worked properly.

The lack of sensitivity of the molecular approach compared to the serological one indicated that it was necessary to develop a method able to detect the viruses with higher sensitivity. A RealTime nested-PCR assay was therefore designed, and results compared with ELISA on a collection of 42 samples (Table 4). These samples, when previously tested by ELISA during winter 2007, had given 20, 20 and 40 positives for GLRaV-1, GLRaV-3 and GVA, respectively; two samples had tested negative for all viruses. In summer 2008, ELISA tests on the same samples showed 100% correspondence for GLRaV-1, while two of the 20 samples for GLRaV-3 and six of the 40 samples for GVA were negative. RealTime nested-PCR gave 100% correspondence with 2007 ELISA results (Table 4). Negative controls and water controls tested negative in each assay and positive controls worked properly.

Table 4.   Comparison between enzyme-linked immunosorbent assay (ELISA) and RealTime nested-PCR for the detection of Grapevine leafroll associated virus 1 and 3 (GLRaV-1, GLRaV-3) and Grapevine virus A (GVA) in samples of grapevine
VirusSamples scoredWinter 2007Summer 2008
ELISA + ELISA−ELISA+
Nested+
ELISA−
Nested+
ELISA+
Nested−
ELISA−
Nested−
GLRaV-14220222022
GLRaV-342202218222
GVA424023462

To confirm these results, the limit of detection was determined for each of the three detection approaches. LOD determination of ELISA showed sensitivity up to the 10−1 dilution for 3/3, 3/3 and 2/3 samples infected by GLRaV-1, GLRaV-3 and GVA, respectively; one GVA-infected sample was detected up to the 10−2 dilution. RealTime RT-PCR detected GLRaV-1 up to the 10−2 dilution, and GVA up to 10−1 dilution; no amplification was observed for GLRaV-3. RealTime nested-PCR detected all the samples up to the 10−4 dilution.

Due to its higher sensitivity, RealTime nested-PCR was used during 2008 to test 100 grapevine samples. Analysis revealed co-infection of phytoplasmas and viruses in 30 of the 100 samples; mixed virus infections were found in 28 of the 42 virus-positive samples (Table 5).

Table 5.   Results of the multiple assays on 100 grapevine samples tested for Flavescence dorée phytoplasma (FDp), Bois noir phytoplasma (BNp), Grapevine leafroll associated virus 1 and 3 (GLRaV-1, GLRaV-3) and Grapevine virus A (GVA), using the same sap extract as template
PathogenPositive samples
FDp18
BNp10
GLRaV-11
GLRaV-31
FDp+BNp2
FDp+GLRaV-12
FDp+GLRaV-32
FDp+GVA3
FDp+GLRaV-1+GLRaV-31
FDp+GLRaV-1+GVA2
FDp+GLRaV-3+GVA4
FDp-GLRaV-1+GLRaV-3+GVA8
BNp+GLRaV-3+GVA1
BNp+GLRaV-1+GLRaV-3+GVA1
BNp+GLRaV-32
BNp+GVA3
GLRaV-1+GVA6
GLRaV-3+GVA3
GLRaV-1+GLRaV-3+GVA1
FDp+BNp+GLRaV-1+GLRaV-31
Negative for all pathogens28

Discussion

Assays that can detect pathogens present in mixed infection in the same plant at the same time are desirable, especially taking into account costs and labour. In this paper, a set of methods for detection of several grapevine pathogens using the same boiled sap extract as template is described. A Taqman fluorogenic assay was developed for detection of Flavescence dorée and Bois Noir phytoplasmas and optimized for GLRaV-1, GLRaV-3 and GVA detection by designing a RealTime nested-PCR, following a one-tube RT-PCR. As grapevine viruses have an RNA genome and are strictly localized to the phloem, like phytoplasmas, sampling could be done from the same tissue, and RT-PCR allowed use of the same extract, without the need to perform separate nucleic acid extractions.

RealTime PCR has already been applied to phytoplasma diagnosis in grapevine (Bianco et al., 2004; Galetto et al., 2005; Angelini et al., 2007; Hren et al., 2007) avoiding post-amplification steps and reducing contamination risks, but large scale applicability has been hampered by time-consuming protocols for template preparation. A rapid protocol for FDp detection (Margaria et al., 2007) was previously described, making the whole procedure more suitable for mass screening: a low-cost and rapid template preparation (crude extract in carbonate buffer) was combined with RT-PCR taking advantage of the higher number of copies of 16S rRNA present in active cells compared to the two copies of the 16S rRNA gene. The effective role of the 16S rRNA as template was demonstrated on crude sap prepared from FDp-infected periwinkle samples, which was used as template with or without the reverse-transcription enzyme in the reaction mixture. The difference in Ct values was from 5·3 to 9·1 points lower for RT-PCR, indicating that it takes advantage of up to 103 times more template (a Ct shift of three times corresponds to a tenfold dilution – Applied Biosystems StepOneTM and StepOne PlusTM Getting Started Guide). Ct values of the 18S rRNA amplicon were very consistent and indicated good quality RNA in the crude sap preparation.

The Taqman method described here combines the advantages of RT-PCR with the RealTime technique, avoiding use of ethidium bromide and reducing total testing time. Furthermore, Taqman technology introduces additional specificity and sensitivity compared to conventional RT-PCR; the new FDp assay was five orders of magnitude more sensitive than the RT-PCR previously described (Margaria et al., 2007), as shown by LOD values (Table 3), and could detect the infected samples in a Ct range of 20–32. As the sensitivity and efficiency of RT-PCR were already considered satisfactory (Margaria et al., 2007) and were further improved by using RealTime, a nested step was not included.

Attempts to develop a RealTime RT-PCR assay for BN phytoplasma detection failed, although primer efficiency tested on serial dilutions appeared to be over 95% and the LOD was similar to that for the FDp assay (Table 4). Better results were obtained by developing a RealTime nested-PCR following RT-PCR, which permitted detection of samples within a Ct range of 20–30, reaching a LOD close to two copies of transcript per microlitre of sap. The similarity of the LOD values of the phytoplasma RealTime RT-PCR assays (10−5 ng μL−1) but the need to introduce a nested step for efficient detection of BNp, may be due to some different biological properties of BNp from FDp, such as different replication rate or lower vitality of the cells, or to the behaviour of the BNp vector H. obsoletus which only occasionally feeds on grapevine, influencing pathogen accumulation.

The BNp primers also amplified AAYp, a phytoplasma of the16SrI group, due to the close phylogenetic relationship in the target sequence (Lee et al., 1998). However, members of this group have only rarely and sporadically been detected in grapevine in Italy (Alma et al., 1996) and no further reports are available. Thus the occasional AAYp infection is not in practice a problem. Moreover, the aim of the work was to develop a method to support the phytosanitary service, since FDp is subject to mandatory control in European countries, and the fundamental aim was discrimination between FDp and BNp for planning disease management. Nevertheless, if discrimination between BNp and AAYp is needed, it is suggested that the described method is used for rapid screening to distinguish between the two agents, and then RFLP analysis of 16S rRNA and ribosomal protein genes is performed (Lee et al., 1998) on the BNp positive samples.

Regarding viruses, rapid protocols for detection by RealTime RT-PCR from crude sap have recently been described (Osman & Rowhani, 2006, 2008; Osman et al., 2007, 2008). In this work a RealTime nested-PCR following RT-PCR from sap was designed, in order to reach high sensitivity throughout the season, taking into account erratic distribution in the host, variation in virus titre (Rowhani et al., 1997) and tissue sampling (fragments of midribs, petioles or subcortical tissue). LOD values showed that ELISA is as sensitive or more sensitive than RealTime RT-PCR. However, RealTime nested-PCR was from 2 to 5 orders of magnitude more sensitive than ELISA. These data are consistent with the preliminary comparison that was done between ELISA and RealTime RT-PCR on 20 samples, showing ELISA to be more sensitive, and justifying the need to introduce a nested step.

During the summer, ELISA failed to detect eight out of the 40 samples that the same test had found to be virus-infected in the previous winter. However, RealTime nested-PCR in summer found all 40 samples positive. Thus, the reliability of the serological test was dependent on season, while the RealTime nested-PCR appeared to be more consistent irrespective of the season. Similar results were reported by Rosa et al. (2007), comparing ELISA and RealTime RT-PCR for detection of Citrus tristeza virus (CTV) in citrus trees in a one-year study.

The results presented here revealed co-infection of phytoplasmas and viruses in 30% of the samples, and mixed virus infections in 69% of the virus-infected samples, suggesting that multiple infections are frequent in the field and that phytoplasmas and viruses often contribute together to the pathology of infected plants.

RealTime nested-PCR for detection of the viruses and BNp was sometimes carried out in the same 96-well plate, allowing up to 20 samples to be checked simultaneously for four pathogens including positive and negative controls. This aspect is a further recommendation for this method as a tool for mapping and screening over a wide range of pathogens.

Further advantages may derive by developing multiplex assays: availability of several fluorescent reporters with different emission spectra could make possible the simultaneous detection of several pathogens in the same well, leading to further savings in time and reagents. The method could also be adapted to detect phytoplasmas in single insects. Such an approach could be especially interesting for BNp, for which alternative vectors to H. obsoletus may exist.

Acknowledgements

This work was supported by a grant from ‘Servizio fitosanitario regionale’, Regione Piemonte, Italy. The authors also wish to acknowledge Dr R. G. Milne for editing the manuscript.

Ancillary