Multiplex real-time PCR for the detection of Clavibacter michiganensis subsp. michiganensis, Pseudomonas syringae pv. tomato and pathogenic Xanthomonas species on tomato plants

A multiplex real-time PCR method based on fluorescent TaqMan® probes was developed for the simultaneous detection of the tomato pathogenic bacteria Clavibacter michiganensis subsp. michiganensis, Pseudomonas syringae pv. tomato and bacterial spot-causing xanthomonads. The specificity of the multiplex assay was validated on 44 bacterial strains, including 32 target pathogen strains as well as closely related species and nontarget tomato pathogenic bacteria. The designed multiplex real-time PCR showed high sensitivity when positive amplification was observed for one pg of bacterial DNA in the cases of Clavibacter michiganensis subsp. michiganensis and Pseudomonas syringae pv. tomato bacteria and 100 pg for bacterial spot-causing xanthomonads. The reliability of the developed multiplex real-time PCR assay for in planta detection was verified by recognition of the target pathogens in 18 tomato plants artificially inoculated by each of the target bacteria and tomato samples from production greenhouses.


Bacterial strains and plant samples
Bacterial cultures (listed in S1 and S2 Tables) were obtained from the National Collection of Plant Pathogenic Bacteria (NCPPB, London, UK), National Collection of Agricultural and Industrial Microorganisms (NCAIM, Budapest, Hungary), Collection of University of Warwick (HRIW, Wellesbourne, UK) and Crop Research Institute (CRI, Prague, Czech Republic). In total, 44 pathogenic bacterial strains with different geographical origins were tested. The set of samples included 32 strains of target bacteria (six strains of Cmm, eight strains of Pst, 18 strains of BSX), 10 closely related species and two nontarget bacterial pathogens of tomato. The isolates were grown on Mueller-Hinton Agar (HiMedia, Mumbai, India) at 25˚C (genus Pseudomonas) and 28˚C (genera Clavibacter and Xanthomonas) for 2-3 days prior to DNA extraction. Positive controls for in planta detection were obtained from artificially inoculated tomato seedlings (Solanum lycopersicum, cv. Mandat) grown in isolated greenhouse conditions. Reference strains of Cmm (NCPPB 2979), Pst (NCPPB 1106), Xe (NCPPB 2968), Xv (NCPPB 422), Xg (NCPPB 881) and Xp (NCPPB 4321) grown overnight in Mueller-Hinton Broth (HiMedia, Mumbai, India), and bacterial suspensions of approximately 10 8 CFU.ml -1 were prepared in 0.9% sterile physiological saline solution. For each pathogen, 10 one-month-old seedlings were artificially inoculated. The bacterial suspension of Cmm (0.01 ml) was injected into the tomato stems using a sterile needle and syringe [34], and the suspension of Pst and each of the BSX pathogens were sprayed on the leaf surface using a hand atomizer (BOSCH PFS 55, Bosch, Germany) [35]. After the development of symptoms, three plants inoculated with each bacterium were randomly selected; symptomatic parts were used for DNA extraction and pathogen detection. As a negative control, the genomic DNA from healthy plants of seven tomato cultivars (Pedro, Palava, Sonet, Mandat, Curranto, Charmant and Gallant) was used. Seeds and seedlings of all tomato cultivars were provided by the company Moravoseed CZ a. s.

DNA extraction
Total genomic DNA of bacterial cultures and DNA of tested plants was extracted with the NucleoSpin Tissue kit (Macherey-Nagel, Düren, Germany) according to the manufacturer´s instructions. The DNA of plant samples was isolated from approximately 100 mg of homogenized plant tissue. The DNA concentration of the samples was estimated with a Modulus™ Single Tube Multimode Reader (Turner BioSystems, CA, USA) and adjusted to a final concentration of 4-5 ng.μl -1 for bacterial cultures and 50 ng.μl -1 for the plant samples. A concentration of 50 ng.μl -1 was used for nontarget bacterial cultures to increase the probability of their detection in case of nonspecific amplification and to prevent possible false positive results.

Primer and probe design
Primers and probes (Table 1) were designed using Primer-BLAST [36] and CLC Main Workbench 6.5 (CLC Bio, Aarhus, Denmark) software. The detection system for Cmm used the previously published TaqMan1 probe CMM-TP targeting the region of 16-23S rRNA [29] that allows the differentiation of Cmm from other Clavibacter michiganensis subspecies. Newly designed Cmm primers were constructed to be fully compatible with the Cmm probe. For the detection of Pst, the target region of RNA polymerase sigma factor hrpL (hypersensitivity and response pathogenicity) was used based on its involvement in pathogenicity. The hrp gene cluster is considered essential for symptom formation in host plants and for hypersensitive responses in nonhosts. The GTP-binding membrane protein lepA (hypothetical protein for elongation factor 4) was used as the target for BSX. LepA promotes the back translocation of tRNAs on the ribosome during the elongation cycle and has a high conservation in certain bacteria [37,38,39], thus allowing highly specific detection. All primers were tested for individual specificity by in silico analysis using a Blastn search (GeneBank/NCBI, BLAST 2.2.31+) in which nonredundant collection (nr/nt) and highly similar sequences (megablast) settings were used. The self-complementarity, primer-dimer estimation and melting temperatures of each oligonucleotide were evaluated using Multiple Primer Analyzer (https://www.thermofisher. com/cz/en/home/brands/thermo-scientific/molecular-biology/molecular-biology-learningcenter/molecular-biology-resource-library/thermo-scientific-web-tools/multiple-primeranalyzer.html) and OligoAnalyzer Tool (https://www.idtdna.com/pages/tools/oligoanalyzer). For multiplexing, the following combinations of fluorophores and quenchers were used: FAM-BHQ1 (Pst), HEX-BHQ1 (Cmm) and Cy5-BHQ2 (BSX).

Optimization of simplex real-time PCR assays
Detection systems were first optimized as simplex assays using DNA from isolates from the NCPPB collection (S1 and S2 Tables) and seven healthy cultivars of tomato. All samples were tested in triplicate on the Roche LightCycler 480 Instrument II (Roche Diagnostics, Basel, Switzerland). The reactions consisted of 10 μl of LightCycler 480 Probes Master (Roche Diagnostics Corp., Basel, Switzerland), 1.2 μl of each of the forward and the reverse primer (10 μM), 0.24 μl of probe (10 μM) and 2 μl of DNA template. The final reaction volume of 20 μl was achieved by the addition of nuclease-free water (Ambion, TFS, Waltham, USA). Annealing temperatures of 60, 62 and 65˚C were compared to optimize the thermal profile. The best combination of specificity and reaction efficiency was obtained with an annealing temperature of 65˚C. The thermal profile of the reaction was set to 95˚C for 10 min followed by 40 cycles of 95˚C for 10 s, 65˚C for 30 s and 72˚C for 1 s. A plate read was set to occur during the 72˚C cycle. The thresholds were set automatically according to manually selected noise bands within the assays for each TaqMan1 probe.

Evaluation of assay sensitivity
The sensitivity of the simplex assays was evaluated on serial dilutions of DNA isolated from cultures of Cmm (NCPPB 2979), Pst (NCPPB 1109) and Xe (NCPPB 2689) reference strains. Only Xe was chosen from the BSX complex due to its prevalence in diseased tomatoes. DNA samples were diluted to approximately 50 ng.μl -1 with nuclease-free water, and then 10-fold dilution series (5×10 1 -5×10 −4 ng.μl -1 ) were prepared. All samples were tested in triplicate.  [29] https://doi.org/10.1371/journal.pone.0227559.t001 TaqMan1 real-time PCR assays were performed on the Roche LightCycler 480 Instrument II (Roche Diagnostics Corp., Basel, Switzerland) using the mastermix LightCycler 480 Probes Master (Roche Diagnostics Corp., Basel, Switzerland) and the thermal profile as described above for the simplex assay optimization. The efficiency of each reaction was determined by a standard curve calculated with qPCR cycler software (LightCycler1 480 Software, Version 1.5.1.62). For the absolute quantification, the fit points method analysis module was used for all the analyses.

Specificity of the multiplex real-time PCR assay and sensitivity analysis
Once the simplex reaction conditions were optimized, the three assay reactions were multiplexed into one. The triplex assay was performed on the Roche LightCycler 480 Instrument II using 96 multiwell plates. The reaction mixes were comprised of 10 μl of LightCycler 480 Probes Master (Roche Diagnostics Corp., Basel, Switzerland), 1.2 μl of each of the forward and reverse primers (10 μM), 0.24 μl of each probe (10 μM) and 2 μl of DNA template. The final reaction volume of 20 μl was achieved by the addition of nuclease-free water (Ambion, TFS, Waltham, USA). The thermal profile of the reaction was set to 95˚C for 10 min followed by 40 cycles of 95˚C for 10 s, 65˚C for 30 s and 72˚C for 1 s. A plate read was set to occur during the 72˚C cycle. All samples were tested in triplicate. The multiplex assay was verified on 44 isolates, which are listed in the S1 and S2 Tables. The sensitivity of the assay was evaluated using the same serial dilutions of target bacterial DNAs as in simplex sensitivity analyses.

Mixed template experiment
Template DNA from Cmm (NCPPB 2979), Pst (NCPPB 1109) and Xe (NCPPB 2689) were mixed together in a 1:1:1 ratio (50 ng.μl -1 starting concentrations), and the 10-fold dilution series in the range of 5×10 1 -5×10 −4 ng.μl -1 were tested in simplex and triplex real-time PCR assays. Three replicates of each dilution were used. Briefly, leaves suspected for pathogen presence were surface sterilized with 0.5% sodium hypochlorite and four pieces with approx. sizes of 1 cm 2 were placed into 1 ml of physiological solution (0.9% NaCl) for 15 min. For plating on media, 50 μl of the extract was used.

In silico analysis of primers and probe specificity
Oligonucleotide primers were first tested for formation of self-dimers and primer-dimers. According to the Multiple Primer Analyzer tool, self-dimers do not form. The possible primer-dimers are presented in S1 Fig. The specificity of designed primers was tested by in silico analysis using a Blastn search (GenBank/NCBI). Designed sequences were compared with sequences available in the GenBank database using a threshold of 95% similarity. The organisms possibly amplified by the designed oligonucleotides are presented in S3 Table. The in silico analyses showed that the combination of oligonucleotides for Cmm and Pst detected only the target organisms. The primers and probe used for BSX detection showed the possible amplification of five other bacteria, Xanthomonas axonopodis pv. commiphorae, Xanthomonas campestris pv. arecae, Xanthomonas campestris pv. musacearum, Xanthomonas fuscans subsp. fuscans and Xanthomonas vasicola pv. vasculorum. However, these pathogens have not been described on tomato plants.
Designed sequences were also compared with partial or complete genome sequences of target bacteria available in GenBank. The primers and probe designed for Cmm detection had

Specificity and sensitivity of simplex real-time PCR
Detection systems were tested on DNA from the isolates listed in S1 and S2 Tables and seven healthy cultivars of tomato. At the annealing temperature of 65˚C, only strains X. axonopodis pv. phaseoli and X. hortorum pv. carotae showed nontarget amplification in the assay for BSX detection (S4 Table).
The sensitivity of the simplex reaction was evaluated on serial dilutions of genomic DNA purified from Cmm (NCPPB 2979), Pst (NCPPB 1106) and Xe (NCPPB 2689) ( Table 2). The

Specificity of multiplex real-time PCR assay
The multiplex assay was tested on the bacterial isolates listed in S1 and S2 Tables. The specificity of the designed primers and probes was tested on 32 target bacteria, 10  The assay for Pst identification (FAM channel) successfully amplified seven out of eight Pst isolates, although detectable fluorescence was not recorded for Pst strain NCPPB 878. The sequence amplified from NCPPB 878 with hrpL primers in conventional PCR showed variability at the hybridization site of the TaqMan1 probe in two positions where a nucleotide substitution of T instead of C was present in the Pst reference sequence (NC_004578) (Fig 1). Nonspecific amplification was not observed for the other bacteria.
Primers and the probe designed for the detection of the lepA gene of BSX pathogens (Cy5 channel) proved reliable for identification of all BSX members, X. euvesicatoria, X. vesicatoria, X. gardneri and X. perforans, as well as for the X. axonopodis pv. vesicatoria strains. Nonspecific amplification was observed only in the case of X. axonopodis pv. phaseoli and X. hortorum pv. carotae. The reactions with the other bacteria showed negative results.
No amplification occurred from all samples of plant genomic DNA extracted from healthy tomato plants.

Sensitivity of the designed multiplex real-time PCR assay
The sensitivity of the multiplex real-time PCR exhibited consistent amplification of the 16S rRNA and hrpL genes from 1 pg of genomic DNA in the reaction ( Table 4). The sensitivity of detection was 100 times lower for the of lepA gene compared to the 16S rRNA and hrpL genes. The reaction efficiency reached values of 97.8% for Cmm, 95.5% for Pst and 99.3% for BSX detection.

Mixed template experiment and in planta detection
The DNA of Cmm, Pst and Xe mixed at a ratio of 1:1:1 was diluted in a 10-fold dilution series and tested by the real-time PCR assay. Positive results were obtained up to the 1000-fold dilution in both simplex and multiplex assays (Figs 2 and 3), which represented the amount of 1 pg of genomic DNA in the reaction. In the case of the 1000-fold dilution, the Cmm (HEX) and Pst (FAM) detection exhibited mean Ct values of approximately 30 in both the simplex and multiplex reactions. The Ct values for Xe detection were higher than those obtained for Cmm The in planta detection of the designed assay was verified on artificially inoculated onemonth-old tomato plants. The target pathogens were successfully detected by multiplex realtime PCR assay using 16-23S rRNA, hrpL and lepA genes on appropriate channels in all tested samples (Table 5).

Testing the designed multiplex assay on greenhouse samples
Leaf samples of four tomato cultivars randomly collected from plants in production greenhouses were tested for the presence of pathogenic Cmm, Pst and BSX by the designed multiplex real-time PCR assay. None of the 20 samples showed positive detection of the target organisms. The presence of pathogens was also tested by cultivation of leaf extracts on selective media, but characteristic colonies were not observed.

Discussion
This study provides the first report of a multiplex TaqMan1 real-time PCR assay that simultaneously targets the Cmm, Pst and BSX complex. For the detection of these pathogens, conventional multiplex PCR was optimized by Ö zdemir [1] using previously published systems for Cmm [33], Pst [40 and Xav [41] detection. However, this system was tested only on three cultures, Cmm ICMP 2550 (NCPPB 2979), Pst ICMP 2844 (NCPPB 1106) and Xav ICMP 9592,  which were recently classified as X. euvesicatoria [15]. Additionally, the specificity analyses were confirmed neither on a broader spectrum of target and nontarget isolates nor on plant samples. Because Cmm and BSX are quarantine organisms, the use of real-time PCR methods is more accurate for providing information about the health of tested material. Real-time PCR assays using TaqMan1 probes for Cmm detection were previously published [29,42,43]. Available detection systems use the target sequence of the internal transcribed spacer (ITS) located between the 16S and 23S rDNA, as in this study. This region is generally used to avoid possible false-negative results that can be reported when plasmidborne genes encoding pathogenicity (pat-1 and celA) are targeted [44,45]. The specificity of the designed assay for Cmm was confirmed by the specific detection of C. michiganensis and   (Table 3). Nevertheless, based on the nonspecific amplification of Pectobacterium carotovorum subsp. carotovorum and Pst isolate CRI 211, the cycle cutoff value of Ct 37 is recommended. The application of a cut-off value in qPCR protocols is frequently used by diagnostic laboratories to prevent false positive results [46]. The qPCR system targeting putative two-component system sensor kinases [47] for Cmm identification that was recommended by EPPO [48] also applies the limit of Ct 35. This protocol was not used in this study because of the possible hybridization of the primers RZ_ptssk 10/RZ_ptssk 11 with Pst isolates, as a difference in two nucleotides between both primers and the Pst reference sequence was found by in silico analysis.
The designed system for the identification of Pst pathogens targets the hrpL gene from the hrp cluster, which was reported as suitable for PCR-specific detection [25,26]. The hrpL gene encodes the alternative sigma factor required for the expression of the hrp gene cluster in the Pseudomonas syringae group [49,50], which is essential for symptom development in host plants and the hypersensitive response in nonhosts [51,52]. The specificity of the reaction was proven for seven of eight tested isolates of Pst. Nonspecific amplification was not observed. In the case of strain NCPPB 878, lack of detection was caused by nucleotide substitutions at the probe hybridization site. However, the nucleotide sequence of the target locus of the hrpL gene showed 97% identity with Pseudomonas syringae pathovars aptata, panici, pisi and syringae but only 88% identity with the Pst reference sequence (GenBank Acc. No. NC_004578) when the Blastn search (BLAST 2.2.31+) was applied. Additionally, the NCPPB collection does not declare the authenticity of isolate 878, although the origin from S. lycopersicum and its pathogenicity was confirmed by NCPPB.
The published real-time PCR protocols for BSX pathogens are focused mainly on the distinction of the involved species [32,53]. Since the worldwide distribution of X. perforans [54] and X. gardneri [55] was reported, the presence of all BSX species should be tested. In the detection of the BSX group, the lepA gene was successfully amplified from all four species and natural isolates from the CRI collection. Nonspecific amplification was obtained only for the isolates X. axonopodis pv. phaseoli and X. hortorum pv. carotae. Based on the in silico analyses of the designed oligonucleotides, the amplification of X. a. pv. phaseoli should be prevented by incompatibility with the BSX probe. Nevertheless, none of these pathogens were reported on tomato plants; thus, a false positive result is not expected.
In sum, the designed assay proved high specificity in both simplex and multiplex reactions. The specificity of the multiplex assay was not influenced by the involvement of the primers and probes used except for Pectobacterium carotovorum subsp. carotovorum and Xanthomonas axonopodis pv. phaseoli, which were detected by the assay designed for BSX. The reliable results were also obtained when target DNAs were mixed and 1000 times diluted to the final amount of approx. 0.1 ng of genomic/total DNA per reaction. The verification of plant samples confirmed the usability of the designed detection assay, which allows the recognition of quarantine organisms. The negative result for samples from commercial greenhouses confirmed the high level of phytosanitary practice that is required for the ecological approach to be applied by concerned companies. The verification of the designed assay on seeds was not performed based on the quarantine character of pathogenic Cmm and BSX and strict sanitary controls that created the unavailability of infected seeds for this study. Nevertheless, this multiplex real-time PCR based on three TaqMan1 probes was shown to be a useful tool for the quick, specific and sensitive detection of the quarantine and economically important tomato bacterial pathogens Clavibacter michiganensis subsp. michiganensis, Pseudomonas syringae pv. tomato and bacterial spot-causing xanthomonads in both bacterial cultures and tomato plants.
Supporting information S1 Fig. Results for primer dimer detection.