Genome-informed diagnostics for specific and rapid detection of Pectobacterium species using recombinase polymerase amplification coupled with a lateral flow device

Pectobacterium species cause serious bacterial soft rot diseases worldwide on economically important fruit and vegetable crops including tomato and potato. Accurate and simple methods are essential for rapid pathogen identification and timely management of the diseases. Recombinase polymerase amplification (RPA) combined with a lateral flow device (LFD) was developed for specific detection of Pectobacterium sp. directly from infected plant materials with no need for DNA isolation. The specificity of RPA-LFD was tested with 26 Pectobacterium sp. strains and 12 non-Pectobacterium species and no false positive or false negative outcomes were observed. RPA primers and probe for host control were also developed to detect the host genome for enhanced reliability and accuracy of the developed assay. The detection limit of 10 fg was obtained with both sensitivity and spiked sensitivity assays. No inhibitory effects were observed on the RPA assay when targets (pathogen and host) were directly detected from infected potato and tomato sap. The developed RPA assay has numerous applications from routine diagnostics at point-of-care, biosecurity, surveillance and disease management to epidemiological studies. In addition, this tool can also be used to discover reservoir hosts for Pectobacterium species.


Results
Target selection in-silico validation. Whole genome sequences of Pectobacterium and other genera downloaded from NCBI GenBank genome database and were explored for unique and conserved target selection of Pectobacterium. The P. carotovorum genome accession number NC_012798 was used as a reference for whole genome alignment and analysis. TyrR family transcription regulator gene (tyrR) was selected and targeted for RPA primers and probe design. Designed primers and probe were blasted against the NCBI GenBank database for in-silico validation. No similar sequences occurred with other genera including Dickeya. The primers and probe location in the genome was completed and visualized in ring image (Fig. 1). The ring image output shows a comparison of a reference genome sequence of P. carotovorum colored with dark blue in the center circle (inner most) with other closely and distantly related bacterial genomes, P. carotovorum subsp. carotovorum (NC_18525), P. carotovorum subsp. brasiliensis (NZ_CP020350) P. atrosepticum (NZ_CP009125), P. wasabiae (NC_013421), D. zeae (NZ_CP006929), D. dadantii (NC_014500), D. solani (NZ_CP009460), E. amylovora (NC_013961),  (Fig. 1). The orthologous average nucleotide tool predicted overall similarity among all tested bacterial genera. Within species of Pectobacterium, 88.5-90.8% similarity was calculated whereas the closely related genus Dickeya showed 74.9-76.2% similarity with Pectobacterium sp. (Fig. 2). A low nucleotide similarity was observed with E. amylovora, X. vesicatoria and C. michiganensis subsp. michiganensis (Fig. 2). Plasmid sequences were not included in these analyses.
Identity confirmation and phylogenetic relationships. All strains included in inclusivity and exclusivity panels were confirmed by amplifying partial dnaA gene regions; obtained sequences were blasted against NCBI GenBank nucleotide and genome databases. Sequences were also aligned, and a phylogenetic tree was generated (Fig. 3). Separate clusters were formed which further cross confirmed the BLASTn outcome. Two major clusters of Pectobacterium sp. and Dickeya sp. were formed. These two major clusters further sub-clustered according to their species or subspecies (Fig. 3). The C. michiganenis subsp. michiganensis and C. michiganensis subsp. nebraskensis showed the maximum dissimilarity with the other strains included in the analyses. Details of identity percentages among the strains are depicted in Fig. 4. Sensitivity assays. Serially diluted (1 ng to 1 fg) genomic DNA of P. carotovorum subsp. carotovorum (A5280) was used to perform the sensitivity assay. The spiked assay was performed by adding 1 µl (59 ng) of healthy tomato host DNA to each reaction to confirm the inhibitory effect of host genomic DNA on RPA assay's detection limit. Both sensitivity and spiked sensitivity assays detected P. carotovorum subsp. carotovorum genomic DNA up to 10 fg (Fig. 6). The outcome sensitivity detection limit was not affected when 1 µl (59 ng) of tomato  host DNA was added to each reaction (Fig. 6). Non-template control was included to eliminate the possibility of getting a false positive due to cross contamination.

Specificity assays.
Host control. RPA primers and probe were also designed targeting genomic DNA of hosts (tomato, potato, pepper and eggplant) to enhance the reliability and accuracy of the developed assays. These RPA primers and probe detected the genomic DNA of tomato fruit, tomato leaves, potato tuber, fruit of pepper and eggplant, and were used as host control in separate reaction to monitor each reaction for reliable sample/DNA prep and reagent acceptability. Each run included a positive control and a non-template control. The host control primers and probe reacted with the tested Solanum hosts and were detected on lateral flow strips.
Detection of Pectobacterium sp. in infected host tissue. The specificity and detection capabilities of the developed RPA assay was evaluated with infected tomato fruits and potato tubers. All tomatoes were inoculated with five different strains of P. carotovorum and three potato tubers were inoculated with two of the P. carotovorum strains. Three other potatoes were individually inoculated with a strain of D. chrysanthemi or D. solani. Consensus sequences of dnaA gene were aligned and used to calculate the phylogenetic relationships among the strains using NJ tree building method. Consensus tree was generated using Bootstrap resampling method with 1000 replicates. Details of the strains are given in Table 1. Positive results were obtained only from tissues of tomato and potato infected with P. carotovorum whereas no test line was observed with potato tissues infected with either of the Dickeya species (Fig. 7A). These results confirmed that the developed RPA assay can accurately detect the Pectobacterium sp. from infected plant materials.
Each infected sample was also tested with host RPA primers and probe as a host control to confirm the accuracy of DNA isolation. All showed positive results with host DNA isolated from infected tissues. No amplification occurred on the non-template control (Fig. 7B).
Rapid sample prep with TE buffer for RPA detection. The rapid sample prep using TE buffer was evaluated with infected tomato fruits and potato tubers. Three tomato fruits and three potato tubers were artificially inoculated with P. carotovorum subsp. carotovorum and samples were prepared for RPA detection using TE buffer. Positive detection was obtained from infected tissues of tomato and potato with P. carotovorum subsp. carotovorum whereas no test line was observed in the non-template control (Fig. 8). Each infected sample was further tested with host RPA primers and probe as a host control to confirm the accuracy and the capacity of the sample prep; all showed positive results. No amplification occurred on the negative control (Fig. 8). These results confirmed that RPA-LFD assay can directly be used for specific detection with no need of DNA isolation. Total assay from sample prep to LFD based detection can be performed in less than 35 minutes without the need of lab equipment (Fig. 8).

Discussion
We developed and validated a recombinase polymerase amplification (RPA) assay to detect and discriminate Pectobacterium sp. from other closely related genera including Dickeya sp. and non related bacterial genera such as Clavibacter, Xanthomonas, and Ralstonia. We also developed a rapid sample prep procedure for RPA detection using TE buffer which reduced the sample prep time to less than one minute. LAMP methods for detection of the closely related genera Pectobacterium and Dickeya have already been developed by Yasuhara-Bell and coworkers 16 , but LAMP assays have the disadvantage of requiring a high temperature for amplification (65 °C) and an expensive instrument. RPA is a comparatively new isothermal technique that can detect target DNA in 15-30 min at 37-42 °C with no requirement for sophisticated and/or expensive instruments 20 . In addition, RPA displays greater resistance to inhibitors and can be used with samples of greater complexity as compared to other isothermal methods 21,22 . RPA is less affected by inhibitors than the well known LAMP method (M. Arif, unpublished information). Twist Dx RPA reagents are available in lyophilized form; the entire kit can be easily transported to the field without the need for cold storage. Results are easily visualized and  interpretation does not require specific training or instrumentation. Ouyang and coworkers 23 developed a Razor Ex-based sensitive field detection method for the plant pathogenic bacterium, Xylella fastidiosa subsp. pauca but its use required an expensive Razor Ex instrument and training for operation. Reagent cost per reaction was high for this instrument and limited the use of this method for routine in-field diagnostics 24 . Due to its high sensitivity, robustness and accuracy, RPA methods using a lateral flow device are an excellent option for both in lab and out of lab detection of Pectobacterium. Accurate and rapid detection and identification of disease-causing bacteria are one of the most significant prerequisite aspects of disease control 25 . The successful application of RPA assay for detection of Pectobacterium sp. using a primer set and probe targeting TyrR family transcriptional regulator gene has been demonstrated in this study (Fig. 1). Specificity was confirmed using results from a wide range of strains in an inclusivity panel and strains from other both closely-and distantly-related plant pathogenic bacterial species. No false positive or false negative outcomes were observed (Fig. 5, Table 1). Moreover, no cross reactivity was observed when potato tuber tissues were infected with Dickeya chrysanthemi and D. solani, whereas tomato and potato tissues infected with P. carotovorum were all positive confirming the presence of Pectobacterium sp. (Fig. 7). In this study, the application of RPA-LFD assay for detection Pectobacterium from infected tomato fruit and potato tuber samples was achieved without extraction of DNA; the sample prep method required about 12 mins with no need of lab equipment (Fig. 8). These results have shown that the test is highly reliable for detecting Pectobacterium in the field. The short reaction time at 37 °C has several advantages, one of which is that samples can be warmed in hand without the need for sophisticated equipment. Secondly, because plant inhibitors have no effect on the amplification process, the test procedure does not require DNA extraction. Diagnostic methods should be sufficiently sensitive to detect a low amount of the target pathogen in an infected sample without producing false negative results. The developed RPA assay is sensitive, detecting 10 fg of pure Pectobacterium DNA and 10 fg Pectobacterium DNA mixed with host DNA. The detection limit was the same for Pectobacterium DNA as well as bacterial DNA mixed with host DNA indicating that host DNA had no adverse impact on the assay (Fig. 6). The inclusion of a host control in the diagnostic assay targeting the host genome enhances the reliability of the outcome. The developed RPA assay also included host specific RPA primers and probe to detect the genome of potato, tomato, pepper and eggplant. The newly-developed RPA assay is ready to use for in-field and/or storage applications for accurate and robust detection of Pectobacterium sp. in tomato and potato.

Conclusion
We have established a novel lateral flow strip-based RPA assay for Pectobacterium detection that provides unique advantages with respect to speed, specificity, sensitivity, ease of visual detection and simplicity of equipment required. The developed RPA assay does not require pure DNA isolation and a sample can be prepared for RPA in less than a minute. RPA primers and probe were developed to discriminate between host and pathogen, enhancing the reliability of the assay. All these features indicate that the RPA assay can be a useful tool for laboratory and field applications for specific detection of Pectobacterium.

Materials and Methods
Source of bacterial strains and DNA isolation. The bacterial strains used in both inclusivity and exclusivity panels are listed in Table 1. Strains were stored in −80 °C and re-grown on TZC medium (10 g/L peptone, 5 g/L glucose 17 g/L agar, and 0.001% 2, 3, 5-triphenyle-tetrazolium chloride added after autoclaving); plates were incubated at 26 °C ± 2 °C. The Wizard Genomic DNA Purification Kit (Promega, Madison, WI) was used to extract the genomic DNA from all strains according to the manufacturer's protocols. The concentration of extracted DNA was measured using a NanoDrop TM 2000C (Thermo Fisher Scientific Inc, Worcester, MA). Extracted DNA was stored in a freezer at −20 °C. For host samples, healthy peppers, eggplants, tomatoes, and potato were grown in the greenhouse. Fruits and tubers were surface sterilized with 10% sodium hypochlorite for 1 min and washed three times with sterilized water. The Wizard Genomic DNA Purification Kit was used to extract the genomic DNA from all fruit samples following the manufacturer's protocols.
Endpoint PCR and dnaA sequencing. The dnaA gene region of bacterial strains was amplified using primer sets designed for Clavibacter, Dickeya, Pectobacterium, Ralstonia, and Xanthomonas (Supplemental Table 1) 26 . A 20 μl PCR reaction contained 10 μl of GoTaq Green Master Mix (Promega), 0.5 μl of 10 mM of each forward and reverse primer, 1 μl template DNA, and 8 μl Ultra-Pure DNase/RNase-Free distilled water (Thermo Fisher Scientific). PCR was performed with an initial denaturation step at 95 °C for 5 minutes, followed by 35 cycles of 94 °C for 20 seconds, 60 °C for 20 seconds and 72 °C for 20 seconds, followed by a three mins extension at 72 °C and hold at 12 °C. Agarose gel (1.5%) electrophoresis was used to separate all PCR amplicons, stained with 0.4 μg/ml ethidium bromide, and bands were visualized under a UV illuminator. PCR was performed in a T100 Thermal cycler (Bio-Rad, Hercules, CA). Amplified PCR products were treated using ExoSAP-1T (Affymetrix Inc., Santa Clara, CA). A 5 μl of post-PCR reaction and 2 μl ExoSAP-IT reagents were combined and incubated at 37 °C for 15 min following by an incubation of 80 °C for 15 min. Each treated template was sequenced for both sense-and antisense-strands using corresponding genus-specific primers. Sequencing was performed at the Advanced Studies in Genomics, Proteomics and Bioinformatics facility (ASGPB), University of Hawaii at Manoa, Honolulu, HI.
Sequence analyses. The sequences of partial dnaA gene regions of tested strains were manually edited in order to generate error free consensus sequences for all the strains. Manually edited error free consensus sequences were compared against the NCBI GenBank nucleotide database using BLASTn algorithm. Consensus sequences were aligned and used to reveal the phylogenetic relationships among the strains using NJ tree building method. The consensus tree was generated using the Bootstrap resampling method with 1000 replicates 27 . Geneious 10.2.3 was used for editing, alignment, and generation of phylogenetic trees. The Sequence Demarcation Tool v1.2 was used to calculate percent pairwise similarity among the strains.
Pectobacterium genus-specific RPA primers and probe design. The genomes of P. carotovorum subsp. carotovorum, P. carotovorum subsp. odoriferum, P. atrosepticum, P. wasabiae, D. zeae, D. dadantii, D. solani, E. amylovora, C. michiganensis subsp. michiganensis, C. michiganensis subsp. nebraskensis, X. vesicatoria and R. solanacearum were retrieved from NCBI GenBank genome database. Genomes were aligned using Mauve (2.4.0); and P. carotovorum subsp. carotovorum genome NC_012917 was used as a reference genome. Generated locally Collinear Blocks (LCBs) were analyzed to hunt for unique regions for Pectobacterium sp. Regions within the tyrR family transcriptional regulator gene were selected to design RPA primers and probe to specifically detect all Pectobacterium sp. RPA primers and probes were designed manually and checked for thermodynamic characteristics following the parameters described by Arif and Ochoa-Corona 28    using the BLAST Ring Image Generator (BRIGS) 29 ; ncbi-blast 2.6.0+ database was used to compare and generate BRIG image. Average Nucleotide Identity (ANI) was calculated using Orthologous Average Nucleotide Identity Tool (OAT) 30 . The reverse primer PCRT-RPAR1 was labeled with biotin at 5′ position, and similarly, probe was labeled with FAM ( Table 2). The Internal Transcribed Spacer (ITS) region was used to design the primers and probe to target host genomes of tomato and potato; primers and probe information are provided in Table 2

RPA assay.
A 50 μl reaction was performed with the TwistAmp Exo ® kit following the manufacturer's protocols. Each reaction contained 29.5 μl of rehydration buffer, 0.6 μl of (10 μM) of probe, 2.1 μl (10 μM) of each forward and biotin-labeled reverse primer, 11.2 μl nuclease-free water, 2 μl of DNA template or plant sap, 2.5 μl of magnesium acetate to activate the RPA reaction. RPA reactions were performed at a constant temperature of 37 °C for 30 min. Each RPA run was conducted with positive control and a non-template control (NTC; water). After amplification, 2 μl of RPA product was added to a mix of 400 μl of nuclease-free water and 100 μl of buffer (Milenia Biotec). A Lateral Flow Device was vertically inserted into the dilution mix and left for 5 min.

Bacterial-infected fruit materials and DNA extraction (host control). Individual tomato fruit
was mechanically inoculated in a biosafety hood with five bacterial strains of Pectobacterium sp., A5280, A5278, A5368, A2688, and A3048 using 10 µl of 10 8 (CFU/ml) of each bacterial suspension. Individual potato tubers were inoculated with the Pectobacterium sp. strains A5278, A5368 and Dickeya sp. strains A5415, A5581, and A5582. All inoculated fruits and tubers were incubated for 24 h at 28 °C. Afterward, the DNA was extracted using the Wizard Genomic DNA Purification Kit following the manufacturer's protocols. Each DNA sample was checked using RPA for Pectobacterium and host DNA.  RPA sensitivity and spiked assays. Both, sensitivity and spiked assays were performed following the procedure mentioned by Larrea-Sarmiento et al. 31 and Dobhal et al. 32 . DNA of P. carotovorum subsp. carotovorum strain A5280 was 10-fold serially diluted from 1 ng to 1 fg and used for sensitivity assays. The spiked assay was performed by adding 1 µl of host DNA (98 ng/µl) in each 10-fold serially diluted sensitivity reaction. NTC was included in each assay to confirm the reliability of the assay.
RPA assay test with infected fruit samples without DNA prep. Individual tomato fruits and potato tubers were surface sterilized by immersing in a beaker containing 10% sodium hypochlorite for 3 min followed by three successive washes with sterilized water. Tomato fruits were mechanically inoculated in a biosafety hood (BSC-II) with P. carotovorum (strain A5278) using 10 µl of 10 8 (CFU/ml) of bacterial suspension. Potato tubers were cut into thin slices using a sterilized knife. Cut potato slices were placed inside the Petri dishes containing filter paper with 1-2 ml of sterile water to make a moist chamber. Bacterial colonies from P. carotovorum (strain A5278) were stab inoculated into the middle of potato slice using a sterile tooth pick. Tomato fruits and tubers were incubated for 24 h at 28 °C. The target P. carotovorum, tomato, and potato were prepared using 20-50 mg infected fruit tissues and 400 µl (TE) buffer. Samples were macerated using an Eppendorf tube and pestle for 2 min. Tubes containing macerated tissues were left 10 min at room temperature to settle down the debris. Afterward, 2 μl supernatant was used in RPA reactions.