Development of a real-time PCR method for the specific detection of the novel pear pathogen Erwinia uzenensis

Erwinia uzenensis is a plant-pathogenic bacterium, recently described in Japan, which infects pear trees, causing the ‘bacterial black shoot disease of European pear’ (BBSDP). Like other Erwinia pear pathogens, E. uzenensis causes damp, black lesions on young shoots resembling those of E. amylovora, but not blossom blight, fruitlet blight or wilting of the shoot tip. The distribution of E. uzenensis seems restricted to the country where it was reported up to now, but it may spread to other countries and affect new hosts, as is the current situation with E. piriflorinigrans and E. pyrifoliae. Fast and accurate detection systems for this new pathogen are needed to study its biology and to identify it on pear or other hosts. We report here the development of a specific and sensitive detection protocol based on a real-time PCR with a TaqMan probe for E. uzenensis, and its evaluation. In sensitivity assays, the detection threshold of this protocol was 101 cfu ml-1 on pure bacterial cultures and 102–103 cfu ml-1 on spiked plant material. The specificity of the protocol was evaluated against E. uzenensis and 46 strains of pear-associated Erwinia species different to E. uzenensis. No cross-reaction with the non-target bacterial species or the loss of sensitivity were observed. This specific and sensitive diagnostic tool may reveal a wider distribution and host range of E. uzenensis initially considered restricted to a region and will expand our knowledge of the life cycle and environmental preferences of this pathogen.


Introduction
Erwinia uzenensis is a Gram-negative bacterium, non-spore-forming, facultatively anaerobic, pathogenic on European pear (Pyrus communis), causing damp, black lesions on young shoots, which occasionally extend from the shoot through the petioles to the main vein of leaves [1]. The disease has been named 'bacterial black shoot disease of European pear' (BBSDP) and has been recorded only in Japan and only in pear trees [1,2]. The symptoms of BBSDP generally a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 cycle of E. uzenensis and its interaction with the other pear-associated bacteria in the natural environment.

Bacterial strains
Bacterial strains employed in this study are listed in Table 1. The strains of E. uzenensis (LMG 25842, LMG 25844 and LMG 25845), obtained from the Belgian Coordinated Collection of Micro-Organisms (BCCM TM ), as well as all the other Erwinia strains employed in specificity and sensitivity assays, were grown on King's B medium [8] at 26˚C for 48 h. Escherichia coli strain JM109 used for cloning experiments was incubated at 37˚C on LB [9] broth or agar plates.

Sequencing of the E. uzenensis ribosomal operon
Sequence determination of the ribosomal operon of E. uzenensis was performed with PCR amplicons obtained by applying the 16SrRNA gene universal primer FGPS6 and the 23SrRNA gene reverse primer FGPL132' [10] (Table 2) on the three E. uzenensis strains employed in this study (Table 1). Briefly, 5 μl of bacterial suspension of 10 6 -10 7 cfu ml -1 were added in a 50 μl reaction mixture volume containing 1 U Taq DNA polymerase (Biotools B&M Labs S.A., Madrid, Spain), 0.1 μM of each primer, 0.2 mM of each of four dNTPs and 3 mM MgCl 2 . The mixture was subjected to an initial denaturation step of 3 min at 94˚C, followed by 40 cycles of 35 s at 94˚C, 1 min at 55˚C and 2 min at 72˚C. The amplification products obtained were visualized by electrophoresis on 1% (w/v) agarose gel in 0.5× TAE buffer (40 mM Tris base, 40 mM acetic acid, 1 mM EDTA) and ethidium bromide staining. A 1-kb DNA molecular weight ladder (Invitrogen, USA) was used in gel electrophoresis.
For all three strains, the FGPS6/FGPL132' PCR resulted in two bands migrating closely together in the gel (described as 'H' and 'L' bands in the Results section); for each strain, the two bands were excised together from the gel using the UltraClean 15 DNA purification kit (MoBio Laboratories, USA), ligated with pGEM-T-Easy vector (Promega, USA) and transformed into E. coli JM109 chemically competent cells, following the instructions of vector's supplier. Colony PCR with the universal primers M13F and M13R (Promega, USA) ( Table 2) identified E. coli clones with ligated inserts corresponding to the 'H' or 'L' band. From the obtained E. coli clones, recombinant plasmid DNA was extracted using the REAL miniprep turbo kit (REAL, Spain). Sequencing of the ligated inserts was performed by IBMCP (Univ. Pol. de Valencia-CSIC, Spain) using the universal primers M13F and M13R (Promega, USA) that anneal to either side of the vector's cloning site. The primers 1389R [11] and R16-1 [12] ( Table 2), which anneal internally to the inserts, were also used for sequencing, to achieve coverage of the whole length of the cloned inserts. All primers used in this study are listed in Table 2 and their position at the rRNA operon is shown in Fig 1. The sequences obtained were assembled using the software Vector NTI 8 (Life Technologies, USA) and the consensus sequences of the inserts corresponding to the 'H' or 'L' band for the three E. uzenensis strains under study were obtained. These sequences were then subjected to BLASTn analysis against the GenBank non-redundant nucleotide database (NCBI).

Structural annotation of sequenced ribosomal region
The structural organization of the sequenced region of the rRNA operon of each of the three E. uzenensis strains was inferred by BLASTn analysis against the GenBank non-redundant nucleotide database (NCBI), as well as by applying the software RNAmmer [13] and tRNA scan-SE [14]. RNAmmer predicts rRNA genes (5s, 16s, and 23s), while the tRNAscan-SE predicts transfer RNA genes (tRNAs) in the 16-23S rRNA internal transcribed spacer (ITS). They were applied using default values, after setting the organism type to 'bacteria'.

Design of real-time PCR primers and probe
In order to find a sufficiently variable area to design primers and probe to discriminate E. uzenensis from the other Erwinia species, the consensus sequence of the 'H' bands along the ITS region of the three E. uzenensis strains was compared using software CLUSTALW [15] with to anneal to the region of the ITS that showed the highest variability between the strains analyzed, in order to achieve the best specificity (Figs 2 and 3). A TaqMan probe (UzeP) was also designed inside the selected region. This probe includes the ZEN Internal Quencher (Integrated DNA Technologies, Inc., USA) that enables less background and increased signal (better sensitivity) in the PCR assays (Table 2).

Real-time PCR conditions
The real-time PCR assays were carried out in a StepOne thermocycler (Applied Biosystems, ThermoFisher, Waltham, MA, USA). The template used was a culture suspension of 10 8 cfu ml -1 , and dilution series from 10 5 to 10 1 cfu ml -1 of the E. uzenensis LMG25844 strain ( Table 1). The amplification was performed in a 25 μl reaction volume containing 5 μl template (sample), 12.5 μl 2x reaction mix (Quantimix Easy Probes kit, Biotools B&M Labs S.A., Madrid, Spain), 0.2 μM of each primer (UzeF-UzeR) and 0.1 μM of TaqMan probe (UzeP). The amplification conditions included an initial denaturation step at 95˚C for 10 min, followed by 40 cycles, each one consisting of 15 sec at 95˚C and 30 sec at 60˚C.

Specificity and sensitivity assays
The specificity of the primers and probe designed in this work to detect E. uzenensis was first evaluated by BLASTn search of their sequences against GenBank database. Additionally, specificity assays using the developed real-time PCR protocol were performed on bacterial suspensions of 38 strains of E. amylovora, 2 strains of E. piriflorinigrans, 2 strains of E. pyrifoliae and 4 strains of non-pathogenic Erwinia species, all listed in Table 1. The analyses were performed twice with all the listed bacterial strains. The analysis was extended to include real-time PCR detection protocols previously developed for other pear pathogenic Erwinia species, in order to check possible cross-reactions with E. uzenensis. Therefore, the three E. uzenensis strains were analyzed with real-time PCR protocols developed for E. amylovora by Gottsberger et al. [16] and Pirc et al. [17], for E. pyrifoliae by Wensing et al. [18], and for E. piriflorinigrans by Barbé et al. [3], using 10 6 cfu ml -1 bacterial suspensions. Additionally, sensitivity and specificity assays of the developed primers and probe were performed in bacterial mixtures consisting of E. uzenensis and one of the other three pear pathogenic Erwinia species used in the specificity assays. Thus, there were three bacterial mixtures assayed, namely E. uzenensis:E. amylovora, E. uzenensis:E. pyrifoliae, and E. uzenensis:E. piriflorinigrans. The initial bacterial suspensions used for the preparation of the mixtures had a concentration of 10 6 cfu ml -1 , and were mixed in ratios 1:1, 1:2, 2:1 v/v. Same ratios for E. uzenensis:water were used as control. A second series of assays was conducted on similar mixtures prepared using initial bacterial suspensions with a concentration of 10 5 cfu ml -1 . All bacterial mixtures prepared were subjected to real-time PCR following the protocol described above. The experiments were performed in triplicate.
Sensitivity assays of the developed real-time PCR were performed on serial dilutions of bacterial suspensions and also on spiked samples of pear material. Briefly, a suspension of E. uzenensis LMG25844 strain was adjusted to 10 8 cfu ml -1 at 600 nm (OD 600 0.2), and serial dilutions from 10 5 to 10 1 cfu ml -1 were prepared. A plate count on King's medium B was used to confirm the bacterial concentration of each suspension. For spiked samples, 1 g of healthy pear tissue (twigs, flowers, buds, leaves, fruits) was slightly crushed with 50 ml of antioxidant maceration buffer, as described in the protocol of the European and Mediterranean Plant Protection Organization (EPPO) for E. amylovora [19] and then spiked with the serially diluted suspensions of E. uzenensis described above. In specific, 1 ml of the spiked sample consisted of 990 μl of macerated plant tissue in PBS and 10 μl of the bacterial suspension to reach a final concentration of 10 5 to 10 1 cfu ml -1 .
The serially diluted bacterial suspensions were used directly in the real-time PCR reactions, while for the spiked samples, the DNA used was extracted following the protocol of Llop et al. [20]. In both cases, 5 μl of suspension or DNA were used per 25 μl reaction. The bacterial suspensions were prepared in 10 mM PBS in all the analyses. The negative control used included DNA extracted from healthy plant material macerated in PBS similarly to the spiked samples, while the no template control included only PBS.
The bacterial dilutions prepared for the sensitivity assays were employed as real-time PCR DNA standards to generate a standard curve of the fluorescence signal in relation to the amount of template. The threshold cycle values (Ct values) were plotted versus the known amount of bacteria.

Cloning and sequencing of the E. uzenensis ribosomal operon
Gel electrophoresis of the PCR products generated with primers FGPS6 and FGPL132' on the three E. uzenensis strains revealed the presence of two bands of ca. 2000 bp and 2250 bp for all of them. These bands were marked as 'H' or 'L', according to their higher or lower size, respectively (Fig 4). The presence of two bands in the PCR reactions suggests that the three isolates possess multiple rRNA operons with intragenomic ITS variation, which generally stems primarily from differences in tRNA gene composition, as already observed in other Erwinia species [21].
The two bands were cloned and sequenced for each of the three strains to deduce the information needed for the subsequent selection of the appropriate region for primers-probe design. Comparison of the derived sequences by the BLASTn algorithm (NCBI) confirmed that 'L' and 'H' bands correspond to sequences that are highly homologous to the 16S-ITS-23S rRNA region of other Erwinia strains, for example they are 98% identical over 99% of their length to that of E. pyrifoliae (GenBank Accession no: Furthermore, using the RNAmmer and the tRNAscan-SE software [13,14], the structural annotation of the E. uzenensis sequences obtained in this work was predicted with regard to the location of non-coding RNA sequences. For all three strains, a 16S rRNA gene was predicted to span a region of 1530 bp in the sequences corresponding to both the 'L' and 'H' bands. This 16S rRNA gene was �99% identical among the three E. uzenensis strains. However, in all three strains, the 'L' and 'H' bands differed in the organization pattern and size of their ITS region. In the case of 'H' bands, two tRNA molecule sequences were predicted in the ITS region: one tRNA Ile at location 1611 bp to 1684 bp, and one tRNA Ala at location 1841 bp to 1913 bp. In the case of 'L' bands, only one tRNA molecule sequence was predicted in the ITS region: a tRNA Glu at location 1608 bp to 1680 bp. Finally, in all three strains and in both the 'H' and 'L' bands, the sequences included 128 bp of the 5' end of the 23S rRNA gene (Fig 2).

Real-time PCR development and determination of its specificity and sensitivity
The comparative analyses of the ITS consensus sequence of the E. uzenensis strains, as determined in the 'H' bands, with that of strains of related Erwinia species (E. amylovora and E. pyrifoliae) resulted in determining a 129 bp region in the ITS, suitable to design a pair of primers (UzeF, UzeR) and probe (UzeP) (Fig 3; Table 2). The developed real-time PCR protocol employing these primers and probe performed well when using a dilution series of culture suspension (10 5 to 10 1 cfu ml -1 ) of the E. uzenensis LMG25844 strain. Cycle threshold (Ct) values were linearly correlated with the quantity of the target sample and permitted the construction of the standard curve (Fig 5). The correlation coefficient (R 2 ) was 0.996, with a slope of -3,4 and efficiency of 96,6%.
Different assays were performed to check both specificity and sensitivity of the developed real-time PCR protocol. Specificity was firstly analyzed in silico by BLASTn searches of the sequences of the primers and probe designed in this work. These analyses showed that whereas primer UzeR hybridized with several locations within the E. amylovora ITS (100% identity in 19 out of the 20 nts), primer UzeF and probe UzeP did not give enough matches against sequences of the known, closely related pear pathogens. This supports the reliability of the PCR developed by in silico analyses. Additionally, when the developed real-time PCR protocol was applied on all pathogenic and non-pathogenic Erwinia strains included in Table 1, only the E. uzenensis strains provided a positive result Furthermore, E. uzenensis strains did not provide positive results with any of the real-time PCR protocols developed to detect the other pathogenic Erwinia species. The sensitivity of the developed PCR using the bacterial suspensions of E. uzenensis reached the level of 10 1 cfu ml -1 , and with the spiked pear samples, the level of 10 2 −10 4 cfu ml -1 depending on the plant material analyzed ( Table 3). The sensitivity of the real-time PCR was not affected in the presence of other Erwinia species, with the same Ct values as the E. uzenensis:water control, even in ratios of 1:2 (data not shown). The real-time PCR was positive only when E. uzenensis was included in the mix, and negative when E. amylovora, E. pyrifoliae or E. piriflorinigrans were tested alone.

Discussion
Some pear pathogenic Erwinia species constitute a serious threat for the cultivation of this economically-important crop, as well as for other rosaceous fruit trees (apple, quince, loquat, etc.) and ornamental species of economic value. E. amylovora is the most destructive species of this genus, considering its large host range [19] and the severity of symptoms caused [7], resulting in significant crop losses [7]. However, other pathogenic Erwinia species may contribute to significant losses on these crops, as well. This is supported by certain features of these Erwinia pathogens: a) the presence of plasmids genetically similar to the ubiquitous, virulence-related plasmid pEA29 of E. amylovora, such as the plasmid pEJ30 of Erwinia sp. strains causing bacterial shoot blight on pear in Japan [22] and the plasmid pEPIR37 of E. piriflorinigrans causing pear blossom necrosis, which has been proven to increase symptom development when introduced in a E. amylovora strain cured of pEA29 [23]; b) their possibly underestimated occurrence, due to the lack of reliable diagnostic methods, as suggested by the recent records showing the occurrence of E. pyrifoliae and E. piriflorinigrans in new geographic areas and new hosts [3,4], (Smits, T. personal communication); c) the presence of transferable plasmids in these pathogens [24] which suggests that interchange of genetic information through horizontal gene transfer events is possible, and lastly d) the occurrence of mixed infections on host plants, as shown for E. amylovora and E. piriflorinigrans which were isolated jointly from loquat plants outbreaks in Spain [3] indicating a possible synergistic effect in enhancing symptom development. These features substantiate the need for specific and sensitive detection protocols for all pathogenic Erwinia species, that can provide: a) reliable identification of the different pathogens affecting a host, especially when the possible causal agents are so closely related genetically and their disease symptoms can be misidentified due to their similarity; b) further insight into the epidemiological behavior of each species, alone, or in combination of two or more of them, as well as the real host range of each species. The need for such detection protocol is obvious for E. uzenensis, a newly-described pear pathogenic Erwinia species for which not much information is available about its geographical distribution, host range, and life cycle. To address this knowledge gap, a new molecular detection method, based on a realtime PCR protocol using a specific probe, has been developed in this work, showing reliability in specificity and sensitivity, even with mixed related bacteria in the same sample.
Because little genetic information is available on E. uzenensis, its ribosomal operon was sequenced and the ITS region was selected to design primers and probe for real-time PCR with high specificity. Three E. uzenensis strains were used in the study, as the number of such strains available in the bacterial collections worldwide is very limited. The comparisons performed on the E. uzenensis ITS consensus sequence against the respective region of the other Erwinia pathogenic species allowed the selection of a region sufficiently variable to design a pair of primers and a TaqMan probe with good specificity against closely related pear-associated Erwinia species. The developed real-time PCR protocol was evaluated using bacterial cell suspensions and plant material artificially spiked with E. uzenensis. Due to the very restricted geographical distribution of this pathogen (recorded only in Japan), plant tissue naturally infected by this pathogen was not available to the authors for further evaluation of the protocol. Nevertheless, the results obtained showed that the developed real-time PCR protocol is specific in not detecting any of the other closely related Erwinia species tested and shows a good sensitivity level in detecting E. uzenensis in cell suspensions and in spiked plant material. The high sensitivity obtained could be explained by the presence of DNA from dead cells that would increase the template available for amplification by PCR not reflected in the counting by plating, as reported in previous works [25]. The differences in the sensitivity level obtained with the various types of plant material used could be due to the dissimilar efficiency of the DNA extraction protocol to remove the inhibitors present in variable amounts in fruit, leaf, bark and stem tissues. Similarly to the detection methods recently developed for other newlydescribed pathogenic Erwinia species [3,4], this protocol would provide an important tool for diagnostic and epidemiological studies, expanding our knowledge on this new species. As the symptoms caused by E. uzenensis on pear can be misidentified as being caused by E. amylovora or other Erwinia species, it is not recommended to make a diagnosis on the basis of symptoms observation. Furthermore, the identification of E. uzenensis is confusing because it shares phenotypic characteristics with other pathogenic and non-pathogenic Erwinia species. The real-time PCR protocol presented here accurately identifies E. uzenensis in spiked plant material and could be used in surveys investigating the current distribution of this pathogen and in epidemiological studies that would provide information on its life cycle in the plant tissues.