Detection of ‘Candidatus Phytoplasma solani’ in roots from Bois noir symptomatic and recovered grapevines

‘Candidatus Phytoplasma solani’ is the causal agent of Bois noir (BN) in grapevine (Vitis vinifera). It is usually detected in leaves, where typical disease symptoms are seen. However, little information is available on the presence of this phytoplasma in grapevine roots. Here, we investigated ‘Ca. P. solani’ in roots collected from 28 symptomatic, 27 recovered and eight asymptomatic grapevine plants. Protocols based on high-resolution melting (HRM) combined with real-time quantitative PCR (qPCR-HRM) and nested-qPCR-HRM were developed to identify ‘Ca. P. solani’ tuf-type variants with single nucleotide polymorphisms. In all, 21.4% of roots from symptomatic plants were positive to ‘Ca. P. solani’ using qPCR-HRM, and 60.7% with nested-qPCR HRM. Also, 7.4% of roots from recovered plants were positive using qPCR-HRM, which reached 44.4% using nested-qPCR HRM. These analyses identified tuf-type b1 on 88.2% of the positive samples from symptomatic grapevines, and 66.6% from recovered grapevines, with all other samples identified as tuf-type a. This study reports the presence of ‘Ca. P. solani’ in the roots of both symptomatic and recovered grapevines. These qPCR-HRM and nested-qPCR-HRM protocols can be applied to increase the sensitivity of detection of, and to simplify and speed up the screening for, ‘Ca. P. solani’ tuf-types.


Results
set-up of qpCR-HRM and nested-qpCR HMR for 'Ca. p. solani' detection. Different trials to optimise the qPCR-HMR started from the different matrices (i.e., leaves, roots) spiked with serial dilutions of 'Ca P. solani' PCR tuf fragments. These revealed that, related to DNA from roots, at concentrations >25 ng/reaction, the PCR was inhibited, while this not was shown with DNA from leaves until 100 ng/reaction. For the leaves, PCR inhibition was observed at 500 ng/reaction (Table 1). In particular, for 'Ca. P. solani' detection in root samples, 5 ng/reaction DNA target provided the appropriate dilution (data not shown). No amplification was observed in the negative controls. The limit of quantification (LOQ) of PCR tuf fragments corresponded to around 40 copies/reaction of tuf-PCR fragment for both purified PCR fragments alone or combined with root and leaf DNA ( Table 1). All of the standard curves performed according to samples artificially spiked with 'Ca P. solani' PCR tuf fragments, P7 and 19-25 calibrators, and Sy5/4-infected samples indicated that the assay was operating at 100% ± 10% efficiency, except for the Sy5/4 roots, which showed poor mean efficiency (135.2%) (Tables 1 and 2). A similar limit of detection (LOD) was observed among the samples tested, which ranged from mean Cq of 35.28 to 37.19 (Tables 1 and 2). The Cq values of all of the samples confirmed the reproducibility within a low coefficient of variation (CV) of between 0.36%-3.8% (CV <25%) 37 (Tables 1 and 2). For the nested qPCR-HRM set-up, the optimal cycle number for the first PCR was 35, because the C q showed an elevated concentration that remained proportional to the differences between all of the starting quantities (Table 3). For the nested-qPCR-HRM assays, the PCR product diluted at 1/200 showed the characteristic melting temperature peak for all samples analysed. Therefore, 35 cycles was adopted as the optimal cycle number for the PCR. The HRM assay applied to the dilutions of the calibrator samples (i.e., P7, [19][20][21][22][23][24][25] and the PCR purified fragment (Fig. 1A,B), as well as the control samples from the BN symptomatic leaves (Table 4 and Fig. 1C,D), distinguished two different clusters, in agreement with the PCR-RFLP assays 29 (data not shown). When the artificial samples created by mixing the P7 and 19-25 calibrator samples (representative of two tuf types) were analysed by qHMR, an additional cluster was shown that was different from that obtained when these were analysed as 100% calibrator samples for P7 and 19-25. (Fig. 2).
Detection and characterisation of 'Ca. p. solani' on grapevine roots. The qPCR-HMR assay detected 'Ca. P. solani' in six root samples from 28 symptomatic grapevines (21.4%). The nested-qPCR-HMR assay detected 'Ca. P. solani' in 17 root samples out of the 28 symptomatic grapevines (60.7%) (   Table 4). 'Ca. P. solani' was not detected in the roots of the asymptomatic plants (Table 4). 'Ca. P. solani' tuf types were the same in root and leaf tissues tested from the same plant (Table 4). Moreover, the qPCR-HMR assay detected 'Ca. P. solani' in two root samples out of 27 recovered plants (7.4%). The nested-qPCR-HMR assay detected 'Ca. P. solani' in 12 root samples (44.4%) ( Table 4). 'Ca. P. solani' was detected in root samples from plants recovered from 1 year (40.0%), 2 years (57.1%), 3 years (25.0%), 4 years (57.1%) and 5 years (25.0%), respectively ( Table 4). The estimated copy numbers of the detected tuf gene ranged from means of 82.3 to 604.2 copies/5 ng DNA in the root samples of the symptomatic plants, from means of 44.1 to 79.1 copies/5 ng DNA in the root samples of recovered plants, and these ranged from means of 573 to 15032 copies/5 ng DNA in the symptomatic control leaf samples ( Table 4).
The TaqMan qPCR assay 21 used as the reference tool detected 'Ca. P. solani' in seven out of 28 symptomatic (25%) and two out of 27 recovered plants (7.4%) ( Table 4). On the other hand, the conventional protocols for nested PCR 38 only detected the phytoplasma in two root samples of the 28 symptomatic plants, and one root sample of the 27 recovered plants (Fig. 4).
The HRM software defined two different clusters that related to these samples: one was linked to the 19-25 calibrator for tuf-type a, and the other to the P7 calibrator for tuf-type b1 (Fig. 5). Of the 17 samples positive for 'Ca. P. solani' in the roots of symptomatic plants, 15 clustered with P7 and two with 19-25. Instead, of the 12 positive samples detected in the roots of the recovered plants, eight were linked to P7 and four to 19-25 (Table 4). No other HRM profiles were detected according to the sequence amplicons analysed.

Discussion
In this study, we report the presence of 'Ca. P solani' in root samples collected from recovered and BN symptomatic grapevines, where the presence of symptoms had been observed in the canopy of vines for at least 1 year and up to 5 years 28,39 .
The HRM test associated to RT-qPCR technology set-up in this study provides a simple and rapid resource for screening for the presence and relative abundances of tuf-type a and tuf-type b1 variants of 'Ca. P solani' in grapevine leaf and root tissues, which can be validated through analysis of the melting curves of the amplicons produced by PCR, without the need for PCR-RFLP 29 or sequences analysis. These data are further supported by sequence analysis of the PCR amplicons from selected samples. In addition, the qPCR-HMR tests (i.e., represented as a mix of different concentrations of tuf-type a:tuf-type b1), emphasise that these procedures can be used to discriminate between the different tuf types, while also simultaneously analysing other molecular variants. However, this study underlines the need to find appropriate DNA template dilutions, in particular for DNA extracted from roots that often included the PCR related to the humic acids in the soil 40 .  Table 4). Two typical genotyping patterns as normalised melting curves (A,C) and normalised difference plots (B,D) are shown. Different colours indicate distinct clusters (green, tuf-type; red, tuf-type b1). RFU: relative fluorescence units.   Tuf-type variants are associated with this BN epidemiology, including their specific association to H. obsoletus haplotypes and the life strategy of these insect vectors on their plant hosts 3 . The presence of tuf-type a and tuf-type b1 were detected in the roots from both symptomatic grapevines, where their symptoms were clearly visible on the canopy, and in recovered grapevines, which did not show any leaf symptoms. In contrast, the tuf-b2 variant, which is also known as tuf-type ab 41 and has been found only in Austrian vineyards, was not detected here 4 .
The roots and leaves of the same symptomatic plants were shown to be infected by the same 'Ca. P solani'tuf type, although the root samples were collected in 2014 and the leaf samples were harvested over the previous 5 years 28,39 . Thus, this study confirms that the phytoplasma in these plants remains the same over time, even across different organs, such as roots and leaves. The prevalence of tuf-type b1 in the samples of this study agrees with previous studies of symptomatic grapevine leaves in the Mediterranean basin 39 . A similar result was observed in the roots of recovered plants, although with a relatively higher proportion of positive tuf-type a. Further studies that can analyse greater numbers of infected roots from recovered plants are needed to determine whether there is any epidemiological significance associated to this aspect. Although the plants that show recovery from   phytoplasma are less likely to become re-infected 13,42 , the presence of a reservoir of 'Ca. P. solani' in the roots might lead to the reappearance of symptoms in such recovered plants.
The analysis by qPCR-HRM of the root and leaf samples extracted from the same plants showed that the copy numbers of the tuf gene were higher in the leaves than in the roots, regardless of type. Therefore, we hypothesise that the main difficulty for detecting this pathogen in grapevine roots will depend on the low phytoplasma concentrations for this organ. The nested-qPCR-HRM improved the phytoplasma detection in roots.  Knowledge of the distribution of phytoplasma across the various plant organs is usually essential for better understanding of the interactions between phytoplasma and their plant host. Typically, phytoplasma diagnosis for grapevines is carried out in a restricted seasonal period, from June to September, when the phytoplasma symptoms are clearly expressed in the leaf tissue. The possibility to test roots and to successfully detect the phytoplasma can expand the time-frame in which phytoplasma testing can be done.
Phytoplasma move within plants through the phloem, from source to sink, and they can pass through sieve-tube elements in phloem tissues [43][44][45] . Previous studies performed on apple trees on established rootstock that have recovered from apple proliferation have shown that the root systems of these trees remain colonised for the lifetime of the tree 17 .
These data show the presence of 'Ca. P. solani' in roots from both symptomatic and recovered plants, which suggests that the concentration and location of the pathogen affects the appearance of BN. In addition, the present study shows that all root samples of the plants that were symptomatic for five consecutive previous years were positive for 'Ca. P. solani'; the phytoplasma was detected in 50% of these samples with the qPCR-HRM test, without the following nested qPCR-HRM step. These data demonstrate that the accumulation of phytoplasma in the roots is higher in plants infected over several years. Furthermore, our investigation suggests that the 'Ca. P. solani' levels in the roots of recovered plants is lower compared to the roots of symptomatic plants; moreover plants recovered over 5 years maintained the phytoplasma in the roots. The potential role of the pathogen in the recovered plant is not completely clear; however, previous studies have shown the induction of defence mechanisms in recovered plants and in asymptomatic parts of infected plants 16,46-48 . In conclusion, we propose these rapid and easy molecular approaches for detection of 'Ca. P. solani' tuf types in grapevine roots. In particular, we propose the more sensitive nested-qPCR-HRM method, which can be applied to detect phytoplasma at low titres for plant organs such as roots. This might also be useful for the selection of healthy propagation material without the need for the canopy, such as during the winter. These data underline the presence of 'Ca. P. solani' in roots from both symptomatic and recovered plants, also highlighting that the phytoplasma can persist in the roots irrespective of the presence of disease symptoms on the plant. However, the relatively low number of root samples that were positive to 'Ca. P. solani' here, as well as the low titres of the phytoplasma detected in the recovered plants compared to the symptomatic plants, indicate that phytoplasma disappearance in grapevine roots is possible. On the other hand, the titre of the pathogen in the roots might affect the balance between appearance and disappearance of symptoms.  (Table 1). Two sub-samples of secondary roots fragments (length, 10-15 cm; diameter, 3-5 mm) were collected from about 20 cm in depth. After washing in tap water, the root sub-samples were put into 0.05% (v/v) Tween 20 in 50-mL tubes (Falcon) and sonicated for 10 min. The roots were rinsed in distilled water, and kept at −20 °C until DNA extraction. DNA extraction. Total DNA was extracted from roots using the cetyl trimethyl ammonium bromide (CTAB) procedure 49 . For each sub-sample, 2 g of pooled roots was ground in liquid nitrogen, and 200 mg of the pulverised materials was added to 2-mL microcentrifuge tubes with 1 mL extraction buffer (3% CTAB, 100 mM Tris-HCl, pH 8.0, 20 mM EDTA, 1.4 M NaCl, 2% [w/v] soluble PVP-40), and 1% (w/v) metabisulphite was added. After incubation at 68 °C for 30 min, purification with chloroform/ isoamyl alcohol (24:1), and precipitation with 0.6% isopropanol were conducted. Finally, the DNA was dissolved in 50 μL pure water. The DNA purity and quantity was also determined (BioPhotometer plus; Eppendorf Inc., Westbury, NY, USA) and was assessed on at least 100 ng/µL DNA, with the absorption ratios at 280/260 in the range of 1.6-1.8, and at 260/230 in range of 1.3-2.0. To increase the chance of detection of the phytoplasma, the DNA obtained from the two root sub-samples per plant were merged and analysed. set-up of qpCR-HRM and nested-qpCR-HMR for 'Ca. p. solani' detection. Detection and characterisation of 'Ca. P. solani' was carried out in the grapevine root samples using the phytoplasma tuf gene, which encodes the translation elongation factor Tu. For testing the reproducibility and sensitivity to detect tuf-type variants for both qPCR-HRM and nested-qPCR-HRM protocols, several parameters were evaluated. For the primers, the forward Tuf-U/f (5′-GATCCAGTGCGTGAAGTTGA-3′) and reverse Tuf-U/r (5′-ATTCCACGCAACAAAGCTCC-3′) primers were designed using the Primer3 software (http://www.ncbi. nlm.nih.gov/tools/primer-blast/), and the specificity of primers for 'Ca. P. solani' tuf gene sequence was verified using the BLAST programme (http://blast.ncbi.nlm.nih.gov/Blast.cgi). These primers identified a 242-bp amplicon that included the nucleotide substitutions of C → T (position 63; tuf-type a → tuf-type b1, b2) and A → G (position 124; tuf-type b1 → tuf-type a, b2) (Fig. 6). Total plant DNA from the 'Ca. P. solani' isolates 19-25 (tuf-type a) and P7 (tuf-type b), used as calibrator samples, was extracted from phytoplasma-inoculated periwinkle plants, kindly provided by Dr. Xavier Foissac (INRA and University of Bordeaux, France). As positive controls, leaf tissue DNA of symptomatic grapevines, previously analysed 28 were also included ( Table 4).
The qPCR inhibitors, the optimal concentration of DNA template, and the limits of quantification (LOQ) and detection (LOD) 50 , estimated from analysis of replicate standard curves, were determined. Firstly, to calculate 'Ca P. solani' copy number, the purified tuf PCR fragments amplified from calibrators by qPCR-HRM were used. The molecular weight (daltons) was determined for a single PCR fragment (http://www.bioinformatics. org/sms2/dna_mw.html), and converting from daltons to nanograms (http://www.unitconversion.org/weight/ daltons-to-nanograms-conversion.html). Finally, the number of copies was calculated according to eq. (1): Copy number quantity (ng)/PCR fragment molecular weight (ng) (1) The LOQ and possible inhibitors of the different matrices (i.e., leaves, roots) with the detection of 'Ca. P. solani' by qPCR-HRM was investigated, with artificial positive samples generated. In detail, the DNA pool of healthy grapevine root matrix, (500 ng, 100 ng, 75 ng, 25 ng, 5 ng/reaction,) and leaf matrix, (500 ng, 100 ng, 5 ng/ reaction) were spiked with the 10-fold serial dilutions purified P7 'Ca P. solani' tuf PCR fragment (from 5 × 10 −5 to 5 × 10 −10 ng/reaction; corresponded to 4.01 × 10 5 to 4.01 tuf PCR fragment copies/reaction). The serial dilution of P7 'Ca P. solani' tuf PCR fragment alone (positive control) and DNA from healthy roots and leaves (negative control) were included. Moreover, the DNA from infected roots were testing by qPCR-HRM at different concentrations (5 ng, 50 ng, 500 ng/reaction).
The DNA concentration for the nested-qPCR-HRM analysis was selected by testing 1, 1/10, 1/100 and 1/200 dilutions of the PCR products from the first amplification. To determine the optimal PCR cycle number in the first-step of PCR 37 before the nested-qPCR-HRM analysis, several trials were carried out. The PCR programme was stopped every 5 cycles (from 10-35 cycles) to test the 10-fold serial dilutions of the P7 sample calibrator (1 to 1 × 10 −3 ng/μL) and the S-y5/4 sample that was positive to 'Ca. P. solani' (1 × 10 −1 to 1 × 10 −2 ng/μL). This experiment was carried out in duplicate and was repeated twice. For nested qPCR-HRM assays, the HRM reproducibility was estimated using the PCR template of 10-fold serial dilutions (from 1 × 10 −5 to 1 × 10 −10 ng/μL) of PCR fragments generated from the calibrators.
The qPCR-HRM and nested qPCR-HRM trials were carried out in duplicate over three independent experiments. Reproducibility of the results was checked by interassay analysis, and the Cq mean was calculated for all of the standard dilutions. The variability are expressed as SD and CV%.
Detection and characterisation of 'Ca. p. solani' on grapevine roots. For qPCR-HRM assays, 5 μL (1 ng/μL) DNA template was used for all of the experiments. For the nested-qPCR-HRM, the DNA extracted from root test samples in the first step was amplified using the fTuf1/rTuf1 primer set, using conventional PCR 38 . For the PCR mix, 10 ng DNA was included in each 20 µL PCR reaction, with 1 mM of each primer, 10 μL 2x EconoTaq Plus Green Master Mix (Lucigen; Tema Ricerca S.r.l., Castenaso, Bologna, Italy). The products from the first amplification were diluted 1/200 in ultrapure water, and 5 μL was used as the DNA template in the nested-qPCR-HMR assays. Finally, all qPCR-HRM and nested-qPCR-HRM amplifications were carried out in a total volume of 14 μL, which in addition to the DNA template described above, contained 7 μL SsoFast EvaGreen Supermix (Bio-Rad Laboratories, Hercules, CA, USA), and 1 μL of the designed primers (1 mM each). The reactions were subjected to the following conditions: initial denaturation step for 3 min at 98 °C, followed by 40 cycles of 20 s denaturation at 98 °C, and 40 s annealing-elongation at 60.5 °C. The final step included the melting curve analyses (0.2 °C step increments; 10 s hold before each acquisition), which were analysed from 70 °C to 95 °C. The quantification of the samples in the qPCR-HRM was performed according to the standard curve previously described. The qPCR-HRM and nested-qPCR-HRM amplifications were both performed using the CFX real-time PCR detection system, and analysed using the 'High-Resolution Melting analysis software' (Bio-Rad Laboratories). This software automatically clusters the samples according to their melting profiles and assigns confidence scores to each of the samples. The confidence level threshold for a sample to be included in a cluster was 99.0%. As controls, all of the root samples were subjected to qPCR-HRM and nested-qPCR-HRM, and the performances were compared with the data obtained by applying conventional nested PCR 38 , and RT-PCR using TaqMan fluorogenic exonuclease 21 . The conventional PCR was performed in three independent experiments, and all the qPCR-HRM and nested-qPCR-HRM trials were assessed in duplicate over three independent experiments.