TechniquesOpen Access icon OPENOpen Access license

Isothermal Amplification and Lateral-Flow Assay for Detecting Crown-Gall-Causing Agrobacterium spp.

    Authors and Affiliations
    • Skylar L. Fuller
    • Elizabeth A. Savory
    • Alexandra J. Weisberg
    • Jessica Z. Buser
    • Michael I. Gordon
    • Melodie L. Putnam
    • Jeff H. Chang
    1. All authors: Department of Botany and Plant Pathology; first and seventh authors: Molecular and Cellular Biology Program; and seventh author: Center for Genomic Research and Biocomputing, Oregon State University, Corvallis 97331.

    Published Online:


    Agrobacterium is a genus of soilborne gram-negative bacteria. Members carrying oncogenic plasmids can cause crown gall disease, which has significant economic costs, especially for the orchard and nursery industries. Early and rapid detection of pathogenic Agrobacterium spp. is key to the management of crown gall disease. To this end, we designed oligonucleotide primers and probes to target virD2 for use in a molecular diagnostic tool that relies on isothermal amplification and lateral-flow-based detection. The oligonucleotide tools were tested in the assay and evaluated for detection limit and specificity in detecting alleles of virD2. One set of primers that successfully amplified virD2 when used with an isothermal recombinase was selected. Both tested probes had detection limits in picogram amounts of DNA. Probe 1 could detect all tested pathogenic isolates that represented most of the diversity of virD2. Finally, the coupling of lateral-flow detection to the use of these oligonucleotide primers in isothermal amplification helped to reduce the onerousness of the process, and alleviated reliance on specialized tools necessary for molecular diagnostics. The assay is an advancement for the rapid molecular detection of pathogenic Agrobacterium spp.

    Agrobacterium is a genus of bacteria that includes soilborne plant pathogens which can infect hundreds of different families of herbaceous and woody dicots, including ornamental trees, shrubs, fruit and nut trees, and grapevine (Otten et al. 2008). As the causative agent of crown gall disease, members of Agrobacterium cause malformations that manifest as tumors. These disfigurations in growth render some agricultural plants unsellable because of loss in aesthetic appeal. In young plants, infection by Agrobacterium spp. can compromise health due to decreased nutrient and water acquisition. In some species of plants, such as grapevine and rose, Agrobacterium infection can be systematic and persist without any overt expression of disease symptoms (Martí et al. 1999; Tarbah and Goodman 1987; Yakabe et al. 2012).

    In a commercial setting, the pathogen can be introduced into a production setting via damaged galls from diseased plants and disseminated by irrigation or improperly sanitized tools. At present, there are just a few commercially available protectants and no curatives (Ryder and Jones 1991). In total, pathogenic Agrobacterium spp. cause an estimated 40% reduction in yield in the orchard and nursery industries, with annual losses in the millions of dollars (Kennedy and Alcorn 1980; Moore et al. 2001). The best course of action includes early detection and eradication of the diseased plants (Burr and Otten 1999).

    Based on the more classical classification scheme, the Agrobacterium genus includes at least five recognized species that are distinguished based on phenotypic differences and host range. These include Agrobacterium tumefaciens, A. rhizogenes, A. vitis, A. rubi, and A. larrymoorei. However, regardless of phylogenetic classification, all pathogenic isolates require a tumor-inducing (Ti) plasmid to cause crown gall disease to plants (Christie and Gordon 2014). To infect plants, the bacteria require a wound, which releases attractants and provides ingression points. Perception of the attractant leads to the activation of vir genes on the plasmid, which execute a unique process of interkingdom gene transfer and genetic transformation of plants (Gelvin 2012). The vir-encoded protein products are necessary to process and translocate, via a type IV secretion system, T-DNA into plant cells. The T-DNA is a bordered segment of the plasmid that, upon transfer, leads to expression of genes that perturb plant hormone signaling and unregulated growth, manifested as crown galls, and cause the host to synthesize abundant levels of opines, nutrients for the infecting Agrobacterium spp.

    Early detection is a critical component in managing Agrobacterium-caused diseases. Culturing Agrobacterium spp. from diseased tissues is inadequate for diagnosis, because nonpathogenic variants are widespread. Accepted methods include polymerase chain reaction (PCR) amplification of loci unique to the Ti plasmid directly from extracts of symptomatic plant tissue, or verifying the virulence of isolated bacteria. The virD2 gene is the most conserved among pathogenic Agrobacterium spp. and was consequently advanced as a target for detection (Haas et al. 1995; Wang et al. 1990). VirD2 nicks the T-DNA from the plasmid and pilots the resulting T-strand into the host nucleus (Herrera-Estrella et al. 1988, 1990; Young and Nester 1988). Due to the diversity of plasmid sequences, multiple sets of primer sequences have been designed, each with limited specificity (Bini et al. 2008; Haas et al. 1995; Johnson et al. 2013). For example, the primer sequences reported by Haas et al. (1995), though designed to what was identified, at the time, as the most conserved portion of the virD2 gene, have limited use in detecting plasmids carried by A. vitis strains, commonly associated with infected grapevines (Bini et al. 2008).

    Though PCR is widely accepted and used by diagnostic labs, the method nevertheless requires expertise and specialized equipment and can be time consuming, depending on the number of samples. Loop-mediated isothermal amplification (LAMP) eliminates reliance on equipment for thermocycling or visualizing amplified products and can produce results in 30 min (Notomi et al. 2000). Nonetheless, LAMP requires the ability to maintain a constant high temperature and primers that anneal to six different regions of the target gene sequence, making it difficult to design LAMP for use against diverse alleles of a targeted gene.

    Recombinase polymerase amplification (RPA) is an isothermal method that works at a constant low temperature using stable reaction components that do not require refrigeration (Piepenburg et al. 2006). This method has been used for on-site detection of a variety of pathogens that affect human, animal, and plant health (Hansen et al. 2016; Lau et al. 2017; Londoño et al. 2016; Moore and Jaykus 2017; Wang et al. 2016). RPA relies on a recombinase protein to mediate primer invasion of the target DNA, and continuous amplification takes place via strand-displacement synthesis. Additionally, modified oligonucleotides used in RPA can be coupled with a modified probe to allow for visual detection on a disposable lateral-flow strip. The lateral-flow strip has two antibodies to immobilize and detect the amplicon, which is dually labeled with fluorescein isothiocyanate and biotin. Results can be visualized in as little as 30 min.

    Our goal was to develop a specific molecular detection assay to accelerate diagnostics of pathogenic Agrobacterium spp. primarily associated with hosts other than grapevine. We describe the application of RPA and a lateral-flow-based method for detecting pathogenic, crown-gall-causing Agrobacterium isolates. Primers and probes were designed to detect multiple alleles of virD2 and, thus, are broadly applicable for detecting most pathogenic Agrobacterium spp. As expected, these new oligonucleotide tools accurately distinguished between a plasmid-carrying and a plasmid-lacking strain and, furthermore, detected all tested variants of virD2. We additionally demonstrate that the detection limit of RPA is 3.5 pg of DNA extracted from cultured bacteria. Finally, we highlight the rapidity of the method. DNA extracted from gall tissues and cultured bacteria can be added directly to the RPA reaction and virD2 can be detected via the lateral-flow strip, dramatically reducing cost, time, and effort required for detection.


    Agrobacterium isolates.

    Agrobacterium spp. used in this study are described in Table 1. Bacteria were maintained on solid MGY (mannitol-glutamic acid-yeast) medium at 28°C or grown overnight at 28°C in MGY medium with shaking at 200 rpm (Keane et al. 1970).

    TABLE 1. Agrobacterium isolates used in this studya

    Induction of crown galls on Nicotiana benthamiana.

    Crown galls were induced using the method previously described (Anand et al. 2007). Overnight cultures were pelleted and bacterial cells were washed with sterile water. The bacteria were resuspended to a concentration of 109 CFU/ml in liquid Agrobacterium induction media (1/10 Murashige-Skoog basal salts; 2-N-morpholino-ethanesulfonic acid, pH 5.8, at 0.5 mg/ml; and 1% glucose) supplemented with acetosyringone (150 μg/ml) and placed at room temperature (24°C) for 14 to 16 h. The induced cultures were washed with sterile water and resuspended in 0.9% NaCl. Galls were induced on 4- to 6-week-old Nicotiana benthamiana by wounding the stems with a 22-gauge needle dipped in bacterial suspension. For a mock-inoculated control, plants were similarly wounded using a needle dipped in induction medium. Plants were maintained in a growth chamber with a daily cycle of 10 h of light and 14 h of darkness at 25°C.

    Extraction of genomic DNA.

    Agrobacterium DNA was extracted using the DNeasy Blood and Tissue Kit per the manufacturer’s instructions (Qiagen). Galls or mock-inoculated plant tissues were surface sterilized for 1 min in 95% ethanol, washed three times with sterile water, and cut finely with a sterile razor blade. Cut tissues were then transferred to a lysing matrix tube (Lysing Matrix A; MP Biomedicals) with 400 μl of cetyltrimethylammonium bromide buffer (100 mM Tris-Cl, pH 8.0; 20 mM EDTA, pH 8.0; 1.4 M NaCl; 2% [wt/vol] cetyltrimethylammonium bromide; and 1% polyvinyl pyrrolidone 40,000). The tube containing the plant tissue was homogenized at 6.0 m/s for 40 s on a Fast-Prep 24 (MP Biomedicals). The lysing matrix tube was then incubated in a heat block at 65°C for 10 min. The DNeasy Blood and Tissue Kit was used to extract DNA from the lysates (Qiagen).

    A NanoDrop ND-1000 spectrophotometer (Thermo Scientific) was used to determine the quality and quantity of the extracted genomic DNA. Genomic DNA was diluted in sterile deionized water to a concentration of 35 ng/μl and serially diluted 10-fold from 35 ng/μl to 3.5 fg/μl.

    Genome sequencing, assembly, and analyses.

    Total genomic DNA was used to prepare Nextera XT libraries, and the resulting multiplexed libraries were sequenced on an Illumina HiSeq 3000 to generate 150-mer paired-end sequencing reads (Center for Genome Research and Biocomputing [CGRB], Oregon State University). FastQC was used to assess sequencing reads for quality ( BBduk v.35.82, with the parameters “ktrim = rk = 23 mink = 9 hdist = 1 minlength = 100 tpe tbo”, was used to remove adapter sequences from the reads. SPAdes v. 3.1.1, with the parameters “–careful -k 21,33,55,77,99”, was used to correct errors and de novo assemble the reads ( (Bankevich et al. 2012). Blobtools was used to assess assemblies and, if necessary, guide elimination of contigs likely derived from contaminating bacteria (blobtools v0.9.19.6;​record/439101#.WXeqMITytph) (Kumar et al. 2013). Prokka was used to annotate the genome sequences (Seemann 2014).

    The virD2 sequence from A. tumefaciens C58 was used as a query in a BLASTN (version 2.5.0) search against the National Center for Biotechnology Information nr database, subset to Rhizobium/Agrobacterium (taxid:227290), using the autoMLSA package with relaxed settings (minimum coverage = 45%) (Davis et al. 2016). In total, 38 sequences were identified. The virD2 sequence was also extracted from the genome sequence of A. vitis 80/94. MAFFT (version 6.864b) with default settings was used to align 39 total virD2 sequences (Katoh and Standley 2013). The virD2 tree was generated using RAxML (version 8.1.17) and visualized in FigTree (version 1.4.3) (Stamatakis 2014). CLC Sequence Viewer (Qiagen) was used to align the virD2 sequences and to identify the most conserved sequence regions.

    PCR and RPA assays.

    Primers and probes were designed according to the instructions of the manufacturer (TwistDx Limited). Primer and probe sequences are provided in Table 2. The bottom-strand primer was modified to include a 5′ biotin tag and probes included a 5′ FAM (6-fluorescein amidite) group, a dSpacer, and a 3′ C3-spacer.

    TABLE 2. Oligonucleotides used in this study

    Standard PCR consisted of 0.4 mM each primer, 2.5 mM dNTP, 1× ThermoPol buffer, 0.5 U of Taq DNA Polymerase (New England Biolabs), and 1.0 μl of genomic DNA template (of varying concentration) in a final volume of 25 μl. Cycling parameters were a 10-min denaturation at 94°C; followed by 30 cycles of 30 s at 94°C, 30 s at 55°C, and 90 s at 72°C; followed by 10 min at 72°C. Products were resolved on a 2.0% 1× Tris-acetate EDTA (TAE) agarose gel, stained with ethidium bromide, and visualized under UV light.

    RPA was optimized following the manufacturer’s instructions (TwistAmp Basic; TwistDx Limited). Reactions consisted of 0.48 μM each primer, 29.5 μl of rehydration buffer, 12.2 μl of water, and 1.0 μl of genomic DNA (of varying concentration). A volume of 2.5 μl of 280 mM magnesium acetate (MgAc) was added to initiate the reaction, which was subsequently incubated at temperatures ranging from 20 to 40°C. Reaction times of 1, 10, 20, 30, and 40 min were tested. The RPA products were purified using the QIAquick PCR purification kit (Qiagen), resolved on a 2.0% 1× TAE agarose gel, stained with ethidium bromide, and visualized under UV light.

    Products from PCR and RPA were sequenced on an ABI 3730 DNA Analyzer (CGRB, Oregon State University). The PCR products were prepared using previously described methods (Creason et al. 2014). The RPA products were purified using the QIAquick PCR purification kit (Qiagen).

    For RPA coupled to lateral-flow detection, reactions consisted of 0.42 μM top-strand primer, 0.42 μM biotin-labeled bottom-strand primer, 0.12 μM probe, 29.5 μl of rehydration buffer, 12.2 μl of water, and 1.0 μl of genomic DNA (of varying concentration). Reactions were vortexed briefly and then added to a freeze-dried pellet provided by the manufacturer (TwistAmp nfo; TwistDx Limited). The reaction was initiated by addition of 2.5 μl of 280 mM MgAc. Reactions were incubated at 37°C for 30 min. The dual-labeled amplicon was visualized using a lateral-flow dipstick (Milenia Biotec GmbH). The product (1 μl) was diluted in 49 μl of 1.0× phosphate-buffered saline with Tween (PBST) and 10 μl was applied to the base of the dipstick. The lateral-flow strip was subsequently submerged in 100 μl of 1.0× PBST and incubated at room temperature until appearance of the positive control band, typically within 2 min.


    Design of oligonucleotides and selection of isolates for testing.

    The virD2 gene is the most conserved gene of oncogenic plasmids and was previously developed for use in discriminating pathogenic Agrobacterium spp., making it a good marker for molecular diagnostics (Haas et al. 1995). To improve upon the use of virD2 in discriminating more of the known pathogenic genotypes from nonpathogenic bacteria, we first generated a maximum-likelihood phylogenetic tree based on all publically available virD2 sequences (Fig. 1). This set of sequences was augmented by the inclusion of the virD2 sequence extracted from the genome sequence of A. vitis 80/94, determined herein. The 39 virD2 sequences from the Ti plasmids formed five distinct clades and those from the root-inducing (Ri) plasmids, which are traditionally associated with hairy root disease, formed separate clades. The virD2 sequence from A. vitis 80/94 grouped with other vitopine-catabolizing A. vitis strains of clade IV. We confirmed that the 39 virD2 sequences accurately represented the diversity of the gene by comparing them to another approximately 130 virD2 sequences (A. J. Weisberg and J. H. Chang, unpublished data).

    Fig. 1.

    Fig. 1. virD2 genes of oncogenic tumor-inducing (Ti) and root-inducing (Ri) plasmids of Agrobacterium spp. form distinct clades. Maximum-likelihood phylogenetic tree of virD2 sequences from diverse Agrobacterium spp. In total, 39 sequences were identified using BLASTN, aligned, and used as input for tree generation. Isolates are distinguished based on host type (woody, herbaceous, grapevine, or lab strain) and opine catabolism. Clades of Ti plasmids are labeled. Isolates used for testing are boxed.

    Download as PowerPoint

    As previously described, the virD2 genes have a conserved region in the 5′ end of the sequence that has been used as a target for amplification (Supplementary Figure S1) (Bini et al. 2008; Haas et al. 1995; Johnson et al. 2013). Because of the demonstrable utility of this region, we based our design of oligonucleotide reagents on the 5′ end of virD2. We targeted three sequences within the first 500 bp of the 5′ region and designed primer sequences to meet the requirements for use in RPA assays (Fig. 2A). The lateral-flow detection method requires an oligonucleotide probe that anneals internal to the amplification primers. To this end, we targeted two regions internal to the sequences for designing primers (Fig. 2A). These regions, however, were not as conserved in one clade of A. vitis Ti plasmids (Fig. 2B). Nonetheless, our focus is on isolates found on hosts other than grapevine. Therefore, we designed probe 1 to anneal to the 5′ end of the amplified region and target the same region as primer Haas_A (Haas et al. 1995). To be more inclusive, probe 1 included degenerate sites. Probe 2 was designed to anneal central to the amplified region and is a derivative of Haas_C (Haas et al. 1995).

    Fig. 2.

    Fig. 2. Oligonucleotide primers and probes were designed to the 5′ region of virD2. A, Schematic of the 5′ approximately 500 nucleotides of virD2 and relative positions of designed oligonucleotide primers. B, Alignment of virD2 sequences from tested Agrobacterium spp. with primer and probe sequences designed for use in recombinase polymerase amplification (RPA) coupled to lateral-flow detection. Shading indicates percent identity of the consensus. Inverted triangles indicate position of abasic sites in probes 1 and 2. C, Image of ethidium-bromide-stained gel with amplicons derived from the RPA basic assay (combinations indicated). D, Image of ethidium-bromide-stained gel with amplicons derived from standard polymerase chain reaction (PCR). For both RPA and standard PCR, in total, 35 ng of DNA from Agrobacterium tumefaciens C58 was used as a template (C58). Water and DNA from LMG215 (data not shown) were used as negative controls. Sizes of the amplified products are listed.

    Download as PowerPoint

    The phylogenetic tree was also used to guide the selection of a diverse set of isolates that represent each of the virD2 clades for use in testing and optimizing the RPA assays (Fig. 1). Isolates representing multiple Agrobacterium spp. as well as different plasmid-encoded opine metabolism types were chosen from our collection. Host type from which isolates were obtained (herbaceous, woody, and grapevine) was also used as a selection criteria. Although not a direct target of our assay design, one grapevine isolate was included to test whether the use of the primers and probes could be extended. We also included in subsequent tests LMG215, a nonpathogenic isolate collected from Humulus lupulus (the common hop). We generated a draft genome sequence of LMG215 and identified it as a member of the A. tumefaciens species. Based on molecular markers, LMG215 clusters most closely to reference isolate A. tumefaciens strain B6 and, thus, is a member of genomospecies 4 (data not shown) (Costechareyre et al. 2010). Moreover, the de novo assembled genome sequence does not include a Ti or Ri plasmid. The absence of an oncogenic plasmid was verified by the failure of any sequencing reads from LMG215 in aligning to any reference Ti or Ri plasmid sequence. LMG215 is used hereafter as a negative control in testing the specificity of molecular diagnostic tools.

    Testing primers and probes.

    Six combinations of primers were tested using the RPA basic kit and purified genomic DNA from A. tumefaciens C58 at 35 ng/μl as a template, and compared with their efficacy in standard PCR. In RPA, four of the tested combinations yielded products of expected sizes (Fig. 2C). Sequence analysis of the products showed them to be identical to the targeted sequences of virD2. In standard PCR, all six pairs successfully amplified a product of expected size and were identical in sequence to their targeted region of virD2 (Fig. 2D). The difference in results between RPA and PCR was not surprising and is consistent with the need to test primers in RPA basic, prior to advancing to use in RPA nfo. Additionally, regardless of the primer pair tested, no products were observed in reactions that lacked DNA or had genomic DNA from LMG215, the plasmid-lacking isolate (LMG215 data not shown). The primer pairs are specific to virD2-encoding and, thus, pathogenic Agrobacterium bacteria. Because primers Haas_A and VirD2_r1 best met recommendations of the manufacturer, all subsequent tests relied on the use of this pair of primers. VirD2_r1 was modified with the addition of a 5′ biotin tag for lateral-flow analysis and is subsequently referred to as VirD2_r1_Biotin.

    The efficacy of the probes was tested using the RPA nfo assay, and included primers Haas_A and VirD2_r1_Biotin and genomic DNA from A. tumefaciens C58 at 35 ng/μl. We optimized probe detection by testing combinations of duration of reaction (time) and temperature, which were varied but kept within manufacturer-recommended ranges. Temperature had a strong effect (Fig. 3). Reactions with either probe reliably amplified a product at 40°C in as little as 10 min, albeit with low signal intensity. In contrast, neither of the two probes yielded detectable products when the lowest temperature was used (20°C). Extending the time for amplification could mitigate some temperature effects. With incubation of 20 min at 37°C, both reactions yielded a prominent and visible product. Reactions that included probe 2 successfully amplified a product at a temperature as low as 30°C. Extending reaction times to 30 min allowed for reactions to work at a temperature as low as 25°C when using probe 2. In contrast, reactions with probe 1 were not successful at temperatures lower than 37°C, regardless of the duration of the reaction. Based on these comparisons, probe 2 appeared to be more robust than probe 1. In subsequent tests, a reaction condition of 37°C for 30 min was used.

    Fig. 3.

    Fig. 3. Time and temperature were optimized for using recombinase polymerase amplification (RPA) nfo to detect virD2 of Agrobacterium tumefaciens C58. Two probes were independently tested in RPA nfo using primers Haas_A and VirD2_r1_Biotin and 35 ng of genomic DNA from A. tumefaciens C58. Reactions were run at the indicated times (top) and temperatures (bottom) and assayed using lateral-flow strips. The “test” band indicates trapping of amplicons dually labeled with FAM (6-fluorescein amidite) and biotin, which are complexed with gold-labeled anti-FAM antibodies and captured via antibiotin antibodies. Gold particles that are not complexed with the amplicon are captured by antirabbit antibodies at the “control” line, which serves as a control for flow of the strip. Reactions with sterile water were included for each probe and temperature tested and assessed for 40 min. Results were repeated three times with similar results.

    Download as PowerPoint

    Testing the detection limit of RPA.

    We determined the detection limit of lateral-flow strip-based detection of RPA. Genomic DNA from A. tumefaciens C58 was serially 10-fold diluted and tested using primer sets Haas_A + VirD2_r1_Biotin with probe 1 or probe 2 (Fig. 4A). Results were compared with that of a standard end-point PCR of 35 cycles. RPA with either oligonucleotide set reliably detected a product from as little as 3.5 pg of input DNA. A product can be detected from as little as 3.5 fg of starting DNA but results were difficult to discern and more difficult to repeat (data not shown). In contrast, the use of end-point PCR could only yield a faintly detectable product if the amount of DNA was more than 35 pg; results were more reliable with 350 pg of input DNA. These data indicate that lateral-flow strip detection of the RPA assay, using either probe, has a detection limit 10 to 100× lower than that of traditional PCR methods at detecting virD2 from a genomic DNA template.

    Fig. 4.

    Fig. 4. Recombinase polymerase amplification (RPA) nfo probes have good limits of detection for virD2 of Agrobacterium tumefaciens C58. A, Genomic DNA was resuspended at 35 ng/μl and serially diluted 10-fold. A volume of 1.0 μl from each concentration was tested; the dilutions are indicated. The genomic DNA was also amplified using standard end-point polymerase chain reaction of 35 cycles. Products were visualized on a 1% Tris-acetate EDTA agarose gel. The size of the products is indicated. B, Bacteria were grown overnight, concentrated to 1 × 1010 CFU/ml, and diluted 10-fold; the number of CFU, quantified based on optical density at 600 nm, are indicated. For both, the two probes were independently tested in RPA nfo, with primers Haas_A and VirD2_r1_Biotin. Reactions were done for 30 min at 37°C and assayed using lateral-flow strips. The “test” band indicates trapping of amplicons dually labeled with FAM (6-fluorescein amidite) and biotin, which are complexed with gold-labeled anti-FAM antibodies and captured via antibiotin antibodies. Gold particles that are not complexed with the amplicon are captured by antirabbit antibodies at the “control” line, which serves as a control for flow of the strip. Results were repeated three times with similar results.

    Download as PowerPoint

    Using aliquots of DNA from the same starting stock of A. tumefaciens C58 DNA, the total assay time of the RPA method was compared with that required for traditional PCR detection. The RPA method coupled with lateral-flow detection could detect virD2 in as little as 30 min, whereas the traditional method, including thermocycling and resolving products via gel electrophoresis, took upwards of 3 h. This represents a substantial savings in time and cost for labor, especially if there is a large number of samples.

    We next tested whether RPA could be used to detect virD2 directly from liquid culture-grown bacterial cells of A. tumefaciens C58. Using serial dilutions of the bacterial suspension, we found that RPA was successful in yielding a product from as few as 10,000 untreated bacterial cells (Fig. 4B). We next asked whether the process could be expedited even further. Single colonies of bacteria were selected from an agar plate and either resuspended in 100 μl of water or added directly in the RPA reaction. For the former treatment, 1.0 μl of the resuspended bacteria was either immediately transferred to the RPA reaction or transferred after 30 min of boiling. A positive product was detected in each of the three cases (data not shown). The ability to confirm the presence of virD2 directly from a bacterial colony represents an additional savings in time and costs associated with labor.

    Testing the sensitivity of RPA.

    We tested the RPA nfo assay on a panel of Agrobacterium isolates that were selected based on criteria described above (Table 1). Both probes were tested on DNA extracted either from liquid culture-grown bacteria or galls artificially induced on N. benthamiana (Fig. 5). For assays using DNA from artificially induced galls, the negative controls were DNA from plants that were mock inoculated or inoculated with nonpathogenic isolate LMG215. Concentration of DNA from galls varied from 110 to 450 ng/μl, and 1 and 5 μl of each DNA sample was tested. Neither of the oligonucleotide sets yielded a detectable product in reactions that included DNA from isolate LMG215 or from mock-inoculated plants. These results confirm the specificity not only of the amplification primers but also of the oligonucleotide probes. RPA with probe 1 showed far greater applicability and yielded products from all the tested pathogenic isolates, either from bacterial or gall DNA extracts, including A. vitis 80/94 (Fig. 5). All gall DNA extracts worked, regardless of volume or starting concentration. Despite showing superior robustness in previous optimization experiments, probe 2 proved less suitable in this application, being able to detect only three of the six tested Agrobacterium spp., with results being consistent between the volumes tested (Fig. 5). Each of these three detected isolates were isolated from woody hosts, carry plasmids that belong to clade V, and encode for nopaline metabolism (Fig. 1). The difference in efficacy was not surprising, considering the design of the two probes (Table 2).

    Fig. 5.

    Fig. 5. Recombinase polymerase amplification (RPA) nfo probes vary in their ability to detect virD2 alleles of Agrobacterium spp. The two probes were independently tested in RPA nfo, using primers Haas_A and VirD2_r1_Biotin. In total, 35 ng of DNA extracted from culture-grown isolates of tumor-inducing (Ti) plasmid-carrying Agrobacterium spp. (C) or from infected galls (G) was tested. White arrows show site of inoculation of mock treatments. Circles show galls. DNA from the Ti plasmid-negative strain LMG215 and mock-inoculated plants (M) were included as negative controls. Reactions were done for 30 min at 37°C and assayed using lateral-flow strips. The “test” band indicates trapping of amplicons dually labeled with FAM (6-fluorescein amidite) and biotin, which are complexed with gold-labeled anti-FAM antibodies and captured via antibiotin antibodies. Gold particles that are not complexed with the amplicon are captured by antirabbit antibodies at the “control” line, which serves as a control for flow of the strip. Results were repeated twice with similar results.

    Download as PowerPoint


    Crown gall continues to be a significant problem for the nursery and fruit and tree nut industries. Here, we present a facile, rapid, and specific molecular diagnostic tool for detecting diverse phytopathogenic Agrobacterium spp.

    One of our goals was to develop a molecular diagnostic tool that is generalizable and sufficient for detecting most, if not all, of the Agrobacterium spp. and Ti plasmid variants commonly encountered on plants other than grapevine. Representative virD2 alleles, including approximately 130 generated in a separate study (A. J. Weisberg and J. H. Chang, unpublished data), were used to guide the design of oligonucleotide tools. Primers Haas_A and VirD2_r1 showed the greatest applicability and could detect all the isolates tested (Fig. 5). Based on in silico analyses, it is predicted that the use of these two primers could be extended to detecting Ri plasmids as well as isolates with Ti plasmids that belong to clade I, which are currently not available in our collection (Fig. 1). The design of the probes was more challenging. For probe 1, we introduced two degenerate sites, which likely contributed to increasing the ability to detect a wide range of Ti plasmids. However, based on the high number of polymorphic sites in clade I of the A. vitis Ti plasmid, these pathogenic variants may not be detectable with this probe. Probe 2 could detect from lower amounts of input DNA but had lower utility and could only detect plasmids of clade V.

    In addition to the applicability of this method for detecting most Agrobacterium spp., RPA offers advantages in being exceptionally rapid, with a good limit of detection. We could reliably detect virD2 from as little as 3.5 pg of DNA. This level is comparable with that of quantitative PCR, without the need for specialized equipment (Johnson et al. 2013). Moreover, the DNA extraction or culturing steps could be circumvented and bacteria or DNA from gall tissue could be added directly to reactions. As few as 10,000 cells are necessary. For tests using DNA from gall tissues, we did not attempt to normalize the amount used because of the challenges in determining the fraction that is derived from the bacteria. Nonetheless, despite the likelihood that only a small fraction of the DNA was derived from Agrobacterium cells, RPA coupled to lateral flow could adequately detect virD2. Using DNA from bacteria and in direct comparison with the traditionally used process for PCR-based diagnostics, we estimate that the method described in this study saved 2 h and 40 min, providing a savings in time, effort, and cost.

    However, there were some challenges in applying RPA and additional difficulties that have yet to be overcome. The design of oligonucleotide tools is restricted by the need to target three conserved regions that are within an amplifiable distance and, thus, knowledge of the genetic variation of the targeted sequence is required. Interestingly, we found the that buffer that was provided for use with the lateral-flow strips was unusable, because it consistently led to false positives when negative control RPA reactions were applied (data not shown). Using 1× PBST instead of the provided buffer solved this issue. Specific to Agrobacterium spp., there are documented cases in which plants showing symptoms of crown gall fail to yield culturable pathogenic Agrobacterium bacteria (Bélanger et al. 1995). It is possible that commonly used semiselective media are insufficient for supporting growth of all taxa of Agrobacterium. It is also likely that galls are devoid of or occupied by few plasmid-carrying bacteria, because opportunistic bacteria may have subsequently colonized, outcompeted, and displaced founder pathogenic genotypes. Finally, despite the depth of our sequencing effort, we can never exclude the possibility that yet-to-be discovered Ti plasmids with more divergent virD2 sequences exist and evade currently used detection methods.

    In summary, we analyzed a large dataset of diverse virD2 sequences to design oligonucleotides for use in a coupled RPA and lateral-flow assay to detect pathogenic Agrobacterium spp. associated with plants other than grapevine. Assay conditions were optimized and we used isolates that represent most of the genetic diversity to demonstrate the developed tools to be efficacious with all parameters examined. The RPA method shows promise for use in the diagnostic laboratory setting and represents a significant step toward providing onsite detection methods that can be used by nonexperts and completed without specialized equipment.


    We thank H. Lederhos for her assistance and the members of the CGRB for their assistance with whole-genome and DNA sequencing. This work was supported by the United States Department of Agriculture (USDA) National Institute of Food and Agriculture (NIFA) award 2014-51181-22384 to J. H. Chang and M. L. Putnam. E. A. Savory was supported by USDA-NIFA postdoctoral fellowship 2013-67012-21139. A. J. Weisberg is supported by USDA-NIFA postdoctoral fellowship 2017-67012-26126. Finally, we thank the Department of Botany and Plant Pathology for its support of S. L. Fuller and the computing infrastructure. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.