Specific Detection of Klebsiella variicola and K. oxytoca by Loop-Mediated Isothermal Amplification

Klebsiella spp. are opportunistic pathogens with clinical, veterinary and plant-associated isolates. A previous study showed that bacterial ooze from wetwood of severely declined ironwood trees in Guam contained Ralstonia solanacearum, Klebsiella variicola and K. oxytoca. In this study, Loop-Mediated Isothermal Amplification (LAMP) detected K. variicola and K. oxytoca specifically, using unique primer sets designed individually for each organism. Each LAMP detected its target specifically, while showing negative results for non-target bacteria and negative controls. LAMP detected Klebsiella in inoculated-ironwood stem tissues and bacterial ooze. Due to the presence of plant inhibitors, different sampling protocols were tested. Soaking plant tissue samples to allow diffusion of bacteria into solution, followed by boiling, provided optimum detection of Klebsiella directly from plant samples. False negatives obtained when using crushed plant samples were eliminated by including an enrichment step, involving plating and 12-h incubation. DGGE (denaturing gradient gel electrophoresis) and a colony-blot immunoassay using a Klebsiella-specific antibody also detected Klebsiella in inoculated ironwood. DGGE bands and antibody crossreactions from closely related enterobacters showed the potential for false positive results. The nature of LAMP makes it ideal for point-of-care testing, and when combined with the specificity of the LAMP primers developed in this study, demonstrates its potential as a routine field test for Klebsiella in ironwood in Guam, as well as clinical and veterinary diagnosis of Klebsiella infection. Additionally, the regions targeted for detection in this study have application across all forms of molecular-based diagnostics.


Introduction
Ironwood (Casuarina equisetifolia) is a common forest tree in the Pacific islands and is widely propagated in Australia, India and China for firewood, wood for construction, land reclamation and windbreaks. Ironwood is salt tolerant and grows on nutrient-poor sandy soils common on coastlines [1]. For the past twelve years, ironwood trees on Guam have been undergoing a slow decline and thousands have died throughout the island [2][3][4]. Ironwood decline is associated with thinning foliage, dieback of branches, and wetwood formation. Various causal agents, including the root and butt rot fungus Ganoderma spp., termites, tree-care practices, and most recently bacteria, have been implicated [5]. However, the disease etiology has not been clearly established [2,4,6,7]. A disease showing similar symptoms was first reported in Mauritius [8], and later in India [9]. The disease symptoms include defoliation, yellowing, wilt, dieback and finally death of the trees.
In a 2011 survey of ironwood decline in Guam, involvement of Ralstonia solanacearum (Rs), a bacterial pathogen that causes wilt in a large number of plant hosts including ironwood, was suspected when ooze from diseased trees produced positive reactions with Rsspecific immunostrips (Agdia, Inc.). Following the initial observation by Putnam in 2011 [10], numerous samples from symptomatic plants were sent to Hawai'i for isolation and identification. However, Rs could not be cultured, resulting in an additional survey in Guam in 2012. Three complimentary approaches were used to isolate and identify the pathogen: growth on modified Rs-semiselective medium (mSMSA), Rs-specific immunostrip assays, and confirmation with a Rs-specific loop-mediated isothermal amplification (LAMP) reaction [11,12]. The identity of isolates was determined in Hawai'i and the two predominant types of bacteria associated with ironwood decline were characterized [13,14]. Bacteriological tests, along with multiplex PCR, BOX-PCR, and phylogenetic analysis of the dnaA RIF marker [15], were performed to further characterize and evaluate the genetic diversity among the Rs isolates [13,14]. Non-Rs isolates were presumptively identified as Klebsiella spp. using bacteriological tests [13,14,16,17]. Most strains showed closest similarity to K. variicola following sequence analysis of the 16S rRNA gene, while one showed closest match to K. oxytoca. Strains identified as Rs were capable of wilting tomato seedlings within 7-10 days and ironwood seedlings within 10-30 days. Klebsiella strains did not produce symptoms in inoculated plants, but were re-isolated from stem tissues as far as 10 cm above the inoculation site, indicating possible colonization of the plant. Results indicated, for the first time, that Rs is involved in the decline of ironwood in Guam and Klebsiella spp. are associated with wetwood, a phenomenon associated with various bacteria and reported in other forest ecosystems [18,19].
Ironwood decline has not yet been observed in Hawai'i, or even Saipan, which is much closer to Guam, even though Rs is present throughout the Pacific. Molecular and immunodiagnostic tools for rapid pathogen identification are needed to obtain relevant data in future field studies to better understand ironwood decline and its potential to spread to other islands of the Pacific. Immunostrips (Agdia, Inc.) and LAMP technology using a hand-held SMART-DART device [11,12] are available for Rs detection, but similar tests are needed to determine the possible role of Klebsiella spp. in ironwood decline. Once developed and validated, fieldready diagnostic tools can be used to determine the association of these two bacteria within naturally infected symptomatic plants, and lack of association with healthy ones. These assays could also be used to screen nursery trees for latent and active infections, thereby providing tools to produce healthy nursery stocks for the replenishment of dead ironwood trees and reforestation in Guam. This study compares several diagnostic approaches, including a LAMP assay, for detecting Klebsiella spp. in ironwood, as well as mixed cultures with Ralstonia solanacearum and other bacterial genera.

Study design
DNA from Klebsiella spp. and Ralstonia solanacearum strains were tested individually, as well as in mixtures of two different genera or species. Individual and mixed DNA were further tested in the presence of plant tissue. Molecular tests of DNA included PCR followed by DGGE (denaturing gradient gel electrophoresis) analysis and LAMP. DNA from non-Klebsiella and non-Ralstonia bacteria were also tested with LAMP to determine specificity.
Cultures of Klebsiella spp. and Ralstonia solanacearum strains were used individually, as well as in mixtures of two different genera or species, and tested with a colony-blot assay, in the absence and presence of plant tissue. Cultures were also inoculated into plants individually and as mixtures. Samples were taken from plants and tested with LAMP, DGGE, and colony-blot assay.

Plant inoculation
Plant inoculations were performed according to a previously established protocol [13,14]. Briefly, bacterial suspensions at 10 8 CFU/ ml were prepared from 48 hr cultures of test strains from TZC plates. Seeds of C. equisetifolia trees were collected from the University of Hawaiʻi at Mānoa campus, germinated on damp filter paper, placed in the dark and then transplanted to community pots. Established seedlings were transplanted to 6.7 cm pots containing Sunshine Mix #4 (Sun Gro Horticulture, Agawam, MA). Roots of 15-week-old seedlings were wounded by drawing a sterile scalpel through the soil at four sides of the pot in the root zone. Then, 10 ml of the bacterial suspension were pipetted into the soil after the wounding process. Older ironwood seedlings, ~17 weeks old, were inoculated with 50 ml of bacterial suspension in the same manner. Two plants were co-inoculated with Ralstonia solanacearum and Klebsiella variicola, but one was an older, 10-month old ironwood (Plant 1) that was used to try to produce bacterial ooze. The ironwood seedlings were then grown under the same conditions as described previously [14].

Plant sampling
Plant samplings were performed according to a previously established protocol [14]. Briefly, stem sections of plants were surface sterilized in 10% Clorox (final concentration: 0.8% sodium hypochlorite) for 30 seconds and triple-rinsed in sterile de-ionized water. Stem sections were then macerated in ~200 µl of sterile deionized water and the subsequent suspension was streaked onto TZC. Following a 24 h enrichment phase on TZC, the identity of the re-isolated bacteria was presumptively determined by appearance, and then confirmed using the various tests described hereafter. Tests were also performed directly on the extracted plant samples [15][16][17][18][19]. Bacterial ooze from plant samples was also collected by soaking stem sections in water to allow bacteria to diffuse from the plant material. For co-inoculated plants, five sample sections were taken from Plant 1, while eight were taken from a younger one (Plant 2), with sections being numbered sequentially away from the inoculum site. Tests were performed directly on bacterial ooze, as well as isolated colonies following a 12-h and 24-h enrichment.

DNA extraction
A Chelex DNA extraction was performed on isolated bacteria, as well as plant samples. Briefly, 0.75-1.0 ml of 40% Chelex 100 resin (Bio-Rad, Hercules, CA) in 1X TE buffer (10 mM Tris HCl and 1 mM EDTA at pH 8) with 10% Triton X-100 (Sigma-Aldrich, St. Louis, MO) was added to each sample. Samples were mixed by pipeting vigorously and vortexing and then heated to 95°C for 10 minutes on a digital heat block. Samples were stored at 4°C. Various other DNA extraction protocols were used on plant samples to remove inhibitors [20][21][22][23][24], including the PowerPlant Pro DNA isolation Kit (MO BIO Laboratories Inc., Carlsbad, CA).

DGGE
DGGE primers (341F(-GC)/907RA) used in this study were reported previously to amplify the variable V3 region of the 16S rRNA gene [25][26][27]. PCR reactions for 341F(-GC)/907RA were performed in GMI-1000 --  . PCR reaction conditions were as follows: an initial denaturing at 94°C for 5 min, followed by 25 cycles of denaturing at 94°C for 30 s, annealing at 57°C for 1.5 min, and elongation at 72°C for 1 min, with a final elongation at 72°C for 10 min. PCR products were resolved using DGGE. PCR primer sequences are listed in (Table 2). DGGE analysis was performed with a DGGE-2000 system (C.B.S Scientific Co.), using a 0.75 mm-thick 6% polyacrylamide gel (ratio of acrylamide to bis-acrylamide, 37.5:1) with a linear denaturing-agent gradient of 35-65% (100% denaturant agent was defined as 7 M urea with 40% formamide) that was submerged in 0.5X TAE buffer (40 mM Tris, 40 mM acetic acid, 1 mM EDTA; pH 7.4) at 60°C. PCR samples were mixed with 6 µl of dye solution [0.1% bromphenol blue (w/v), 70% glycerol (v/v)] and applied to the gels. Electrophoresis conditions were 5 h 30 min at 75 V. Gels were stained for 30 min in 1X TAE buffer with ethidium bromide and visualized using the Foto/Analyst Express System (Fotodyne Inc., Hartland, WI). Products were excised from gels and resuspended in sterile water [28]. Products were reamplified using 341F/907RA primers (Table 2), under the same reaction conditions, and then sequenced for identification. Sequences were queried in the National Center for Biotechnology Information (NCBI) database using the Basic Alignment Search Tool (BLAST) [29].

DNA sequencing
PCR products were cleaned for sequencing using ExoSAP-IT (Affymetrix, Santa Clara, CA), according to the manufacturer's instructions. Cleaned PCR products were sequenced at the Greenwood Molecular Biology Facility at the University of Hawaiʻi sequencing facility, using each forward and reverse primer, according to specifications.

Colony-blot immunoassay
The colony blot immunoassay was performed according to methods established previously [30,31], with modifications. Briefly, 1 μl of bacterial suspension (OD A600 =0.1), or plant extract, were spotted onto 0.45 μm nitrocellulose membranes (NitroBind, Osmonics, Inc., Minnetonka, MN) and allowed to dry completely. Dried nitrocellulose membranes were blocked in 5% PBS-milk for 1 h and then boiled in PBS for 10 min to eliminate endogenous enzymatic activity. Membranes were then washed three to five times using 0.05% PBS-Tween 20.
Membranes were subsequently incubated with Klebsiella species antibody (73/28) (ThermoFisher Scientific, Waltham, MA) at a 1:2000 dilution for 30 min, followed by washing three times for five minutes each with 0.5% PBS-Tween 20. Membranes were then incubated with goat anti-mouse antibody conjugated to alkaline phosphatase (1:2000) (Southern Biotech, Birmingham, AL) for 30 min and washed three times for five minutes with 0.5% PBS-Tween 20. Membranes were developed with 1-Step NBT/BCIP (ThermoFisher Scientific, Waltham, MA) for 15-30 min, rinsed with distilled water, and air-dried. Colony immunoblots were analyzed under a dissecting microscope.

LAMP
LAMP primers developed in this study are shown in Table 2

Sensitivity assay
Bacterial DNA was isolated using the Wizard Genome DNA Purification Kit (Promega, Madison, WI). DNA was quantified using the NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA) and diluted to a starting concentration of 1 ng/μl, with subsequent 10-fold serial dilutions made to reach a final concentration of 1 fg/μl. LAMP reactions were performed in triplicate, using ddH 2 O as the negative control. . All characteristic bands were excised, re-amplified and sequenced. BLAST analysis of returned sequence data revealed that the characteristic Ralstonia and Klebsiella bands were indicative of each respective bacterium, thus demonstrating the utility of DGGE, and its potential ability to be used without sequencing by including appropriate controls. However, in the mixed sample, which contained other enteric bacteria (Figure 1, Lane 4), sequence data could not be obtained for the Klebsiella band. A strong band was produced for Ralstonia solanacearum, a weak band for Klebsiella, and no distinct bands for all other enteric bacteria. Similarities between bands of the enteric bacteria and Klebsiella could explain the lack of additional bands. Additionally, the Klebsiella band in the mixed plant sample was very faint (Lane 7). Inefficient recovery of DNA from the excised band is responsible for the inability to obtain sequence data.

Colony-blot assay
A Klebsiella-specific antibody was tested for specificity using pure bacterial suspensions in a colony-blot immunoassay format (Figure 2a

LAMP
Two sets of LAMP primers were designed to specifically detect K. oxytoca and K. variicola. Both LAMP reactions were highly specific, only detecting their respective Klebsiella species and producing no reaction with any of the other bacteria tested in this study (Table 1), as well as negative controls. Positive LAMP reactions were recorded visually via turbidity, using the specified primers and a LAMP reaction master mix (dNTP mix (1.2 mM), 10X ThermoPol Reaction Buffer (New England Biolabs, Ipswich, MA) (2 mM), betaine (1 M), MgSO 4 (4 mM), Bst DNA polymerase (New England Biolabs, Ipswich, MA) (8 U) and ddH 2 O) that was described previously [32,33]. Under the current protocol, positive reactions were visible in real-time and commonly occurred within 10-20 minutes.
The sensitivity of the Hcp1-LAMP reaction was determined and DNA in the range of 5 ng to 5 pg showed positive reactions in approximately 10 minutes, while DNA in the range of 500 fg to 5 fg showed positive reactions in approximately 15 minutes. However, at the 5 fg range of DNA, positive reactions were not always obtained due to the low amount of DNA in the starting sample. The sensitivity of the UGH-LAMP reaction was less than that of the Hcp1-LAMP, with a detection limit of ~5 pg of DNA.
LAMP was performed on plant sample extracts following a 12-h and 24-h enrichment phase on solid media. All samples from known inoculated plants were positive by LAMP, while negative control plants were negative by LAMP. Non-Klebsiella bacteria that were present in the plant samples were tested separately and were negative with LAMP. In order to make a single test for both K. variicola and K. oxytoca, the LAMP primers were multiplexed into a single reaction. The multiplexed test was able to detect both Klebsiella spp.; however the sensitivity of the combined assay was compromised, as known to occur during multiplexing [34,35].
LAMP was further tested directly on plant sample extracts from two co-inoculated plants. Extracts were streaked onto plates and the first two sections of Plant 1 showed growth of Klebsiella and Ralstonia, while only Ralstonia grew from sections 3-5. In Plant 2, Klebsiella grew only from the first five sections. Initial testing using turbidity revealed that plant inhibitors did affect the LAMP reaction to a certain degree (data not shown). For real-time LAMP detection, extracts were processed, using various methods, and tests revealed no significant difference in results between direct boiling, 1:1 sample and TE buffer with boil, Chelex extraction, or GES buffer, using buffers 1 and 2 and sample at a 0.5:0.5:1 ratio. All Klebsiella-negative samples, including negative and Ralstonia controls, were all negative by LAMP. The LAMP reaction did not react with boiled samples from Plant 2 at sections 1, 3 and 4, TEbuffered samples from Plant 2 at section 2, and GES-buffered samples from Plant 1 at section 2. All other samples were positive by LAMP, including all Chelex-extracted samples. Included in the LAMP test was one Klebsiella-inoculated plant sample that had DNA extracted using the PowerPlant Pro DNA isolation kit, which is explicitly designed to extract DNA from plants and remove plant inhibitors. This sample was positive by LAMP, and reactions were the most robust; however, this kit is high in cost and therefore not practical for field use/studies. The SMART-DART hand-held real-time fluorescence monitoring device performed equally to the in-lab real-time machine.

Discussion
The genus Klebsiella is comprised of opportunistic pathogens that are frequently isolated from humans and animals [36]. In humans, Klebsiella pneumoniae and K. oxytoca are particularly involved with nosocomial infections [36,37], such as septicemia, pneumonia and urinary tract infection. In animals, Klebsiella spp. are mostly associated with sepsis, infections of the urinary and respiratory tracts, and mastitis, with Klebsiella pneumoniae and K. oxytoca being isolated frequently from domestic livestock [38][39][40]. These disease syndromes can cause serious economic consequences in some cattle herds [41]. The ecological habitats of Klebsiella include surface water, sewage, soils and plants, as well as mucosal surfaces of mammals [42]. Recently, a new species of Klebsiella was proposed, K. variicola, having clinical and plant-associated isolates [43].
In Guam, the association found between ironwood decline and Ralstonia led to the discovery of an association between Klebsiella and the same disease [11,12]. An association was also found between wetwood symptoms observed in infected trees and Klebsiella spp., which were determined to be Klebsiella oxytoca and K. variicola [13,14]. The current study, which focused on evaluation and design of diagnostic tools, will enable a more extensive survey of Guam ironwood trees to further investigate the association between Klebsiella-and Ralstoniainfected ironwood trees. DGGE, an immunoassay and a molecularbased assay (LAMP) were all tested for their ability to identify Klebsiella in ironwood.
DGGE is a method used to discern bacterial populations in mixed environmental samples. However, the complexity of DGGE makes it less desirable as a routine test for field studies. Additionally, results generally require DNA sequencing of bands following gel separation, which is not readily available to researchers in the field. Here, we attempted to circumvent the need for sequencing. Knowing what specific bacteria to look for and by running appropriate controls, we thought characteristic bands could be found that would be representative of Ralstonia and Klebsiella. While we were able to demonstrate this, the inability of DGGE to separate bands from closely related Enterobacter species contained in a mixed sample showed the possibility for false positive results. This fact, in combination with the laboriousness of DGGE, resulted in not recommending it as a possible test to survey Guam ironwood for Klebsiella infection.
Immunostrips (Agdia, Inc.) are available for detection of Ralstonia, but there are none for detection of Klebsiella. We explored the possibility of making an immunoassay that would be later commissioned into immunostrips for rapid detection of Klebsiella. We decided to test a commercial Klebsiella antibody for its potential as a rapid detection system, using a simple colony-blot test. As with most immunoassays, we observed cross-reactions with closely related bacteria. The Klebsiella antibody was made using Klebsiella aerogenes. Klebsiella species are constantly being reclassified into other genera and/or species. Interestingly enough, a very strong cross reaction was observed with Enterobacter aerogenes, even stronger than the interaction with Klebsiella oxytoca and K. variicola. Since this Klebsiella antibody reacted with a closely related Enterobacter sp. associated with ironwoods, this immunoassay is not recommended for routine field testing of Klebsiella-infected ironwood in Guam.
The potential for false positive results using the immunoassay and the inability to clearly separate bands using DGGE led to the design of a specific molecular-based assay using loop-mediated amplification technology. The LAMP assay proved to be the most useful in this study. It was able to specifically detect Klebsiella, having no positive reactions with non-Klebsiella bacteria, and the two primer sets allowed differential detection of Klebsiella oxytoca and K. variicola. LAMP primers in this study were designed to amplify a hemolysin-coregultaed protein (Hcp1) family type VI secretion system (T6SS) effector (K. variicola specific) and an unsaturated glucuronyl hydrolase (UGH) (K. oxytoca specific). These regions were found by whole-genome comparisons. BLAST searches of these genes resulted in matches only to each designated Klebsiella species, demonstrating that these primer sets should be highly specific for each Klebsiella species to which they were designed. LAMP tests using non-target bacteria confirmed the specificity of these genes as targets. Testing a broader panel of nontarget bacteria would improve specificity claims; however, in silico data from BLAST searches combined with test results provided in this study, together, suggest a high specificity.
As with any molecular-based assay, plant inhibitors are always a concern. Once an effective assay is developed, sample processing remains the underlying factor. The LAMP assays presented here were affected by plant inhibitors to an extent. Plant samples produced some false negatives, but there were no false positive results and Chelex extraction was able to remove enough inhibitors to allow proper detection directly from woody tissues. Ultimately, to avoid any problems with plant inhibitors, it is recommended to simply enrich samples on solid media [17,44,45]. Klebsiella grows extremely fast and a simple 12-h incubation on media will eliminate all false results.
LAMP was chosen because it is the most widely researched isothermal nucleic acid method, which allows greater trouble-shooting during development [46]. LAMP is comparable to equivalent PCR-, immunoassay-or culture-based detection methods [47], with detection limits as low as five copies [48]. Moreover, its isothermal nature makes it suitable for field diagnostics.
Molecular methods such as PCR have been evaluated for rapid identification of Klebsiella in human clinical specimens [37,49]. However, Klebsiella species are difficult to identify and are often misclassified in clinical microbiology laboratories [16,50,51]. Phenotypic distinction between K. pneumoniae and K. oxytoca isolates based on existing biochemical tests are time consuming, laborious and not very reliable [52]. Current procedures, such as BIOLOG and API systems, often fail to differentiate between species of Klebsiella [50,53]. Differentiation of Klebsiella is a key component of clinical responses to infection. In Japan, in 2013, a patient with sepsis caused by Klebsiella variicola died due to misidentification as Klebsiella pneumoniae by a commonly used automated identification system [54]. The need for diagnostic tests that can properly identify Klebsiella at the species level has been nationally recognized, as the National Institute of Health and the National Institute of Allergy and Infectious Disease recently put out a request for proposal for an R01 grant that included Klebsiella as a target organism (Funding Opportunity Announcement Number RFA-AI-14-019).
Due to the isothermal nature of LAMP, it offers an easy and rapid assay for clinical medicine [89]. The LAMP reactions presented in this study not only specifically detect Klebsiella, but also distinguish K. oxytoca and K. variicola. Moreover, the target regions that these LAMP primers were designed to could become the new focus for developing Klebsiella-detection systems, having application to numerous molecular detection platforms. In conjunction with portable real-time fluorescence monitoring devices, such as the SMART-DART, these LAMP reactions could provide beneficial point-of-care diagnostics for clinical, veterinary and agricultural Klebsiella spp.