The development and deployment of a field-based loop mediated isothermal amplification assay for virulent Dichelobacter nodosus detection on Australian sheep

Dichelobacter nododus is the causative agent of footrot, a major disease of sheep that creates welfare concerns and large economic loss. The virulence of D. nododus depends on the presence of extracellular proteases, AprV2 and AprB2, which differ by one amino acid. Strains possessing AprV2 can cause clinically virulent disease, while AprB2 may cause clinically benign disease. Current methods for detecting D. nodosus are difficult, laborious and time consuming. New techniques capable of rapidly detecting and typing D. nodosus are needed to aid control programs. Molecular methods, like real-time polymerase chain reaction (rtPCR) can detect aprV2 and aprB2, however, this assay is not field-deployable and cannot support local decision-making during an outbreak. Here we present a field-based molecular assay for detecting aprV2, using loop mediated isothermal amplification (LAMP). The aprV2 LAMP (VDN LAMP) assay was optimised to reliably detect aprV2 from laboratory purified genomic (gDNA) of virulent D. nodosus down to 5x10-3 ng μL-1, with time to positive (Tp) ≤ 16 minutes, while aprB2 was unreliably detected at 5 ng μL-1 from 16–20 minutes. The use of field collected samples that were rtPCR positive for aprB2 resulted in no amplification, while aprV2 positive field samples by VDN LAMP assay are defined as having Tps’ of < 20 minutes and melting temperature between 88.0–88.9°C. When compared to rtPCR, the VDN LAMP was shown to have a diagnostic specificity of 100% and sensitivity of 83.33%. As proof of concept, the VDN LAMP was taken on farm, with all processing occurring in-field. The on farm VDN LAMP successfully detected 91.67% aprV2 positive samples, no aprB2 positive samples (n = 9) or D. nodosus negative (n = 23) samples, with a kappa agreement of ‘almost perfect’ to rtPCR. This highlights the potential of the assay to inform local treatment decisions for management.


Introduction
Footrot is a bacterial disease of sheep, first recorded 200 years ago [1]. Its primary aetiological agent is Dichelobacter nodosus, an aerotolerant anaerobe capable of digesting the hoof material, causing painful lesions and underrun in the sheep foot [2]. Research into footrot and diagnostics in Australia has been conducted since the 1940's, progressing from histopathology [3], culturing [4], and now to molecular methods, such as real time polymerase chain reaction (rtPCR), for the detection and strain typing of D. nodosus [5]. Footrot has been traditionally classed into two forms, benign and virulent, based on the infecting D. nodosus strains' ability to digest the hoof substrate and the clinical presentation of the disease [6]. An association has been shown between severe lesions and the presence of an extracellular protease, AprV2, which is responsible for the overall elastase activity and development of footrot [7]. Conversely, the benign D. nodosus protease equivalent, AprB2, has been shown to cause mild lesions [7]. Genomic sequencing of over 100 strains of D. nodosus has shown two distinct clades, based on the aprV2/aprB2 protease genes [8], differentiated by dinucleotide polymorphism (TA/CG) at position 661-662. This change results in a single amino acid change (Tyr92-Arg), which affects the thermostability and elastase activity of the protease molecules [9,10].
Diagnosis of clinical benign or virulent forms of footrot helps to identify the cost-benefit to enacting on-farm control or eradication programs, based on production losses and cost of the program [11]. The clinical expression of footrot is reliant on the presence of an infecting D. nodosus strain, animal genetics, and environmental conditions. Wet and warm conditions, above 10˚C, have been reported to be ideal for D. nodosus to cause disease [1]. An infected sheep may show no or little sign of clinical disease if it has a natural resistance, or if the weather is not conducive to the bacteria expressing virulence factors and subsequently causing lesions [11]. In this way, infection can spread undetected, and may not cause disease under suboptimal conditions, with the virulence of the infecting strain potentially masked. The presence of a virulent strain does not guarantee the development of virulent footrot [12,13], but detection may be useful as an indication of clinical disease risk. In addition, footrot is a highly contagious disease, with spread of the bacteria occurring easily through pasture and yards, and the presence of asymptomatic carriers commonplace [14].
Laboratory based diagnostic tools, such as the gelatin gel, elastase and the intA polymerase chain reaction (PCR) do not always align with the clinical presentation and severity of disease and use of these assays is the cause of much debate [15][16][17]. A more informative approach is to use a laboratory-based footrot diagnostic test that can detect infection and identify the virulence potential of the infecting strain early in the disease. A recently published real-time PCR able to detect and differentiate aprV2 and aprB2 provides this information [5].
An extension of this diagnostic capability is to test on farm for infection, with a molecular assay that can be performed in a shearing shed or paddock, alongside other every day management practices. Fast identification of infection and strain typing of D. nodosus has the potential to improve biosecurity practises, reduce disease spread, and implement timely and cost-effective control methods. Loop mediated isothermal amplification (LAMP) is a nucleic acid amplification method that has been reported to be robust, simple to use and appropriate for application in the field [18]. LAMP reactions consist of four main primers, a forward and backward inner primer (FIP and BIP), and inner (F3) and outer primer (B3). These four primers are designed to recognise 6 distinct sequence regions on the target, with FIP and BIP containing complimentary sequences to form loop structures, and an additional two primers, loop forward (LF) and loop back (LB), can be designed to an additional 2 sequence regions, to form loop structures and amplification starting points, increasing reaction speed [19]. LAMP utilises a unique polymerase with strand displacement activity and as such does not require thermocycling, and lacks proof-reading ability during amplification [20,21]. LAMP is highly specific for detecting a target sequence within a species or gene, however can be challenging for SNP detection. Commercially available master mixes contain intercalating fluorescent dyes that allow the real-time monitoring of amplification. These properties combine to offer a fast, robust, specific and sensitive assay that uses simple, portable machinery, in field. We report the development of a LAMP that detects aprV2 in D. nodosus (VDN LAMP), the assay performance in comparison to a previously published rtPCR and the ability to identify virulent D. nodosus infections on farm.

Control strains
Genomic DNA (gDNA) was extracted from cultured cells of D. nodosus strain A198 (aprV2) (AC: 6466) and strain C305 (aprB2) (AC: 6465) using PrepMan Ultra Sample Preparation (Life Technologies) as per manufacturer's instructions. Genomic DNA (gDNA) extracted from these D. nodosus strains were used as controls throughout this study and were provided by DAFWA Diagnostics and Laboratory Services (Department of Agriculture and Food, South Perth, Western Australia). DNA was quantitated at 260 nm using a Nanodrop (ThermoFisher).

Sample collection
Samples for rtPCR and VDN LAMP use were obtained from the Victorian Government Veterinary Diagnostic Laboratory [22]. Samples were submitted by Victorian District Veterinary Officers, Animal Health Officers, and private veterinary practitioners during routine disease investigations, from December 2014 to July 2015.
For rtPCR, the interdigital skin or active margin of a lesion was swabbed with sterile cotton swabs (CLASSIQSwabs) and placed into 600 μL of phosphate buffered saline (PBS) (8.1 mM Na 2 HPO 4 3-, 137 mM NaCl, 1.4 mM KH 2 HPO 4 and 2.6 mM KCl) containing 20 mM ethylenediaminetetraacetic acid (EDTA), pH 8. Using the modified Egerton foot scoring system [6] ( Table 1) foot scores of the sampled feet were taken and recorded. Samples were stored and transported at 4˚C before use.
Animal experimental procedures for collection of field-based VDN LAMP samples were approved by the La Trobe University Animal Ethics Committee (ethics approval AEC17-21). VDN LAMP field swabs were collected as above from August-September 2017 into 500 μL alkaline-polyethylene glycol 200 (Sigma Aldrich), pH 13 [23]. Template was prepared by directly diluting the sample using a 10 μL inoculation loop into 990 μL H 2 O (MilliQ). Method of collection and foot scores were recorded as described. For both rtPCR and field-based VDN LAMP throughout, the highest scored foot (singular) was sampled per sheep.

DNA extraction
DNA for the specificity panel was extracted using the Power-Soil DNA isolation kit (MOBIO), as per manufacturer's instructions, with the following modifications at the elution stage: 30 μl of nuclease free water, pre-heated to 30˚C, was added to the white filter membrane, followed by an incubation period of five minutes at room temperature. Nucleic acid for rtPCR and lab based VDN LAMP was extracted and purified from collected samples using the MagMax Viral RNA extraction kit (Thermo Fisher Scientific), and Kingfisher-96 magnetic particle handling system (Thermo Fisher Scientific), as per manufacturer's instructions.

D. nodosus strain typing using rtPCR
The presence of aprV2 and/or aprB2 in samples was identified using primers, probes and cycling conditions as described by Stäuble et al. [5]. The AgPath-ID One Step RT-PCR Kit (Ambion) was used as master mix according to manufacturer's instructions, with primers and probes synthesised and supplied by Applied Biosystems. Reactions were carried out in 25 μL volumes and analysed using the 7500 Fast Real-Time PCR System (Life Technologies), with a set threshold of 0.05. Results are reported as cycling threshold (Ct) values, the point at which the sample signal exceeds the threshold of 0.05. , LF and LB were tested and a set of 5 primers consisting of a FIP, BIP, F3, B3 and LF were chosen. The 2bp SNP site (bold) overlaps at the 5' end of the F1 complementary region contained in FIP and the 5' end of the B1 region found in BIP, to facilitate discrimination between the aprV2 and aprB2. All primers were assessed for initial specificity using a BLAST search with GenBank.

Primer
Primer Sequence 5'-3' Sequence Region 5' -3' both the Genie II and Genie III (Optigene) real-time fluorometer, with results reported as time to positive (Tp) (minutes.seconds) and anneal derivative melting temperature (Tm) (˚C), given when the sample florescence crossed the pre-set threshold of the Genie machines. Primer concentration ratios of 1:10, 1:8, 1:6 and 1:4 (F3/B3:FIP/BIP) were assessed and 1:8 chosen to optimise the discrimination of aprV2 fluorescent signals from aprB2 fluorescent signals based on Tp. Temperature gradients from 60˚C-67˚C were performed and assessed for speed and discrimination. LAMP assays were run with a pre-heating step of 40˚C for 1 minute, followed by 20 minutes at 65˚C before an annealing step from 94˚C to 84˚C, at a rate of 0.5˚C/second, with these conditions used throughout.

Analytical performance of VDN D. nodosus LAMP
The analytical sensitivity of the VDN LAMP was determined using a standard curve of gDNA from virulent (aprV2-positive) and benign (aprB2-positive) D. nodosus as the template with final concentrations of 5, 0.5, 0.05, 0.025 and 0.005 ng μL -1 . Samples were run in triplicate across both the Genie III real-time fluorometer and the Genie II real-time fluorometer (n = 3, each dilution). Analytical sensitivity of the assay was determined in duplicate on the Genie II real-time fluorometer and repeated singularly on the Genie III real-time fluorometer. It was necessary to run samples in this manner due to machine processing limitations as Genie II can only run 8 samples at one time while Genie III can run 16 samples. A panel consisting of gDNA from 12 different bacterial species (Table 3) was used to determine specificity, kindly supplied by Dr Ashley Franks, La Trobe University. Additionally, samples processed with the MagMax Viral RNA Extraction kit, negative for D. nodosus (n = 46), positive for aprB2 (n = 7) and field prepared negative samples (n = 23) and aprB2 positive samples (n = 9), were also used to determine specificity of the VDN LAMP.
Reproducibility of the VDN LAMP was performed using sheep samples (n = 21) extracted and purified by the MagMax Viral RNA Extraction kit. These samples were run in triplicate over 3 different days to assess intra-assay variation, while triplicate aprV2 gDNA controls across sensitivity concentrations were used to calculate inter-assay variation.

Comparison of aprV2/aprB2 rtPCR to VDN LAMP
One hundred and forty-three samples were obtained retrospectively from the Victorian Government Veterinary Diagnostic Laboratory. Sample extraction for rtPCR was as described and rtPCR was performed as previously stated. Samples with a Ct value under 35 are considered positive for D. nodosus. VDN LAMP was performed as described above on the same nucleic acid extractions for comparison between methods to assess agreement.

Field performance
The VDN LAMP assay as developed in the laboratory was tested in-field as proof of concept, using the alkaline-PEG sampling method outlined above. Foot swabs were collected from August-September 2017 into 500 μL alkaline-PEG, diluted 1:100 into MilliQ H 2 O, and 5 μL of this dilution used directly as template in-situ on farm. Samples (n = 57) were collected from sheep on 3 different farms (Farm 1: Shire of Strathbogie, Farm 2: Shire of Southern Grampians and Farm 3: Shire of Hepburn), covering the full range of foot scores from 0-5, based on the Egerton foot scoring system (Table 1). Samples typically contained hair, faeces, soil and plant material. Biologically duplicate samples were also collected for screening using the aprV2/ aprB2 rtPCR as described above to compare results.

Statistics
Statistics were performed using Microsoft Excel 2016 and GraphPad Prism 6. Co-efficient of variation was calculated to indicate repeatability using The level of agreement between rtPCR and VDN LAMP was evaluated using Cohen's kappa statistic from Fleiss, Levin [24]

Design of LAMP primers and assay optimisation
Initial assessment of the VDN LAMP primer sets was undertaken using high quality gDNA from two previously characterised D. nodosus isolates, A198, as the aprV2 control, and C305, as the aprB2 control. Various primer set combinations and concentrations were assessed for performance ( Table 4). The primer combination in bold was chosen due to providing the fastest time to signal detection and the best discriminatory power between aprV2 and aprB2, based on Tps', using 0.5 ng μL -1 genomic DNA as template, and was subsequently used throughout. The recommended temperature for OptiGene GspSSD 2.0 Isothermal Mastermix (ISO-DR004) of 65˚C was found to give the fastest time to positive, with changes in temperature (60-67˚C) not affecting specificity of the reactions, only the time of amplification.

Analytical performance of VDN LAMP
To assess VDN LAMP sensitivity to ensure adequate detection and discrimination of aprV2 from aprB2, a serial dilution of gDNA for both aprV2 and aprB2 D. nodosus isolates was used.
Using diluted gDNA at final concentrations of 5, 0.5, 0.05, 0.025 and 0.005 ng μL -1 , the chosen VDN LAMP primer set amplified aprV2-positive gDNA within the range of 5-0.005 ng μL -1 reliably between a Tp range of 9.45-15.53 minutes (Fig 1), and with a Tm range of 88.00˚C-88.60˚C. Based on these results, a sample that gave a Tp of 16 minutes was considered positive for the presence of aprV2. Very low concentrations of virulent aprV2-positive gDNA (0.0005 ng μL -1 ) showed inconsistent amplification after 16 minutes. Genomic DNA from aprB2-positive strain was amplified within a range of 5-0.5 ng μL -1 (Fig 1), with times of 16.58-18.2 minutes (Table 5) and Tm's of 88.20˚C and 88.50˚C. Based on these initial observations, samples that gave a Tp range from 16-20 minutes were subsequently designated as 'uncertain', as both very low concentrations of aprV2-positive gDNA and very high concentrations of aprB2-positive gDNA gave results within this range. Amplification of only aprV2 or aprB2 was confirmed by a single Tm peak, seen within the range of 88.0˚C-88.9˚C, and was required to be counted as VDN LAMP positive. Samples that gave only a Tp, only a Tm or no Tp and Tm were considered VDN LAMP negative. A run was considered valid if the aprV2 gDNA control amplified with a Tp and Tm within the stated range, and the negative control had no amplification curve and anneal derivative melting temperature.
As the VDN LAMP assay's intended use is in the field where a variety of environmental bacteria will be present, specificity of the assay was tested. Genomic DNA from range of environmental bacteria listed in Table 3 was tested and no bacteria returned a positive result (S2 Fig), indicating the VDN LAMP assay is specific for D. nodosus. Additional specificity testing was performed on field collected foot swabs that were shown to be D. nodosus negative by rtPCR. None of the 46 D. nodosus negative samples with nucleic acid isolated and purified showed any positive amplification further showing that VDN LAMP is specific for D. nodosus (S1 Table).
To investigate repeatability of the assay, both control strains and a subset of field samples were used. Field samples with nucleic acid isolation and purification were used. The sample subset was chosen to provide a variety of Ct ranges and aprV2/aprB2 rtPCR designations. Inter-assay co-efficient of variation (CV) for aprV2 control Tps' ranged from 0-6.06%, most to least concentrated (Table 5). AprB2 controls gave an inter-assay CV range from 2.22-0.95% (Tp) ( Table 5). Using the field sample subset, intra-assay CV for individual samples ranged    (Table 7). For the 53 samples rtPCR aprB2 positive or D. nodosus negative, 0/53 were VDN LAMP positive (Table 7). This resulted in the definition of a VDN LAMP positive sample being any sample that gave a Tp 20 minutes and a corresponding single anneal derivative peak with a Tm of 88.0˚C-88.9˚C. It is recommended that samples with Tps' above 16 minutes are confirmed via rtPCR.
VDN LAMP decreased in sensitivity with an increase in the samples Ct value from aprV2/ aprB2 rtPCR (Table 8)

Field performance
To trial the field application of the VDN LAMP, 57 samples were collected over 3 volunteer farms. For a sample to be considered positive in the field, both a time to positive (Tp) and a single peak melting temperature (Tm), as previously defined, was required. A run was considered valid as defined previously. Foot scores, VDN LAMP, rtPCR results, and individual farm agreements between methods are summarised (Table 9).  discounted as rtPCR positive for aprV2 after the Ct cut off of 35 is applied, with VDN LAMP then correctly identifying 8/9 samples as aprV2 positive A total of 3 samples were both VDN LAMP and rtPCR negative for aprV2 and aprB2. Farm 3 had the highest proportion of severe lesions as shown by foot score. The pilot field study gave an overall VDN LAMP DSe of 91.67% (22/24), and DSp of 100% with no aprB2 positive (n = 9), or D. nodosus negative (n = 24) samples giving a positive VDN LAMP result (Table 10). These results suggest the VDN LAMP assay can identify virulent D. nodosus infections in the field.

Discussion
Footrot is a contagious disease that is easily spread throughout the farm. Traditionally, visual inspections are used to make a clinical diagnosis, with laboratory support occasionally used. Laboratory methods previously available for a diagnosis are time consuming and require   specialist growth media and expertise. The improved ability to detect virulent D. nodosus prior to the appearance of severe clinical symptoms can help inform control methods on-farm and reduce the severity of disease if treated early. LAMP assays have been developed for several livestock diseases [25][26][27] and are reported to offer a field suitable, easily performed molecular diagnostic assay for detection of infection. While many livestock LAMP assays have been reported, very few have progressed to field testing and demonstrated on-farm use [28,29] with most in-field LAMP testing occurring in plant diagnostics [30,31]. Here, we present a LAMP assay developed in the laboratory and applied in-field. The VDN LAMP assay with 5 primers that is presented in this paper is based on 7 regions of the aprV2 gene sequence found in D. nodosus, only one of which is differing between aprV2 and aprB2. As the primers have multiple shared sequence between virulent and benign D. nodosus, the FIP and BIP primers were used for their discriminatory power. This strategy has been used for SNP detection previously [32]. To help with differentiation, the FIP and BIP primers were designed to overlap at the SNP site, on the 5' end of each which was found to be most effective. The use of 5 primers in a LAMP assay has also been reported previously, to detect Ralstonia solanacearum [33]. Traditional methods of improving oligo binding stringency, like temperature adjustments, were seen here to adversely affect the amplification of the target. Due to this, the 5 primer set was chosen as it offered the best differentiation between aprB2 and aprV2 on high quality gDNA when based on Tp and also amplified a wide concentration range of aprV2 DNA consistently within the 20 minute run time. The aprV2 gene was chosen as the target due to the potential for severe lesion development, and the cost of control programs once disease progresses through to virulent clinical signs.
The original aim of the assay was to differentiate samples in the field with aprV2 present from those with benign D. nodsous (aprB2) infections. Although the VDN LAMP assay amplified aprB2 gDNA, it was a noticeably slower reaction than aprV2, only amplifying after 16 minutes and only when at high concentrations. Amplification of aprB2 from gDNA was 2 orders of magnitudes less than that of aprV2, with no concentration below 0.5 ngμL -1 showing positive results, in comparison to 0.005 ngμL -1 of virulent D. nodosus gDNA. It is recommended that samples amplifying after 16 minutes have the presence of aprV2 confirmed via rtPCR, based on gDNA laboratory results. It was noted that when used on field sample template with nucleic acid isolated and purified, samples rtPCR positive for a range of aprB2 Ct values were not VDN LAMP positive. It's thought the addition of remaining inhibitors in the extraction to the already poor amplification of aprB2 is responsible for hindering the reaction and the cessation of aprB2 amplification. No rtPCR aprB2 positive samples, both nucleic acid isolated and purified or field prepared, showed a positive VDN LAMP result. This continued observation supports the VDN LAMP only amplifying samples which have aprV2 present. For both the nucleic acid isolated and purified samples and the field processed data, diagnostic specificity of the VDN LAMP assay was 100%.
The sensitivity of VDN LAMP was less than that of rtPCR, which has been reported previously [34]. VDN LAMP DSe varied from 97.5-23.08%, depending on the samples Ct value, and therefore the number of bacteria collected on the swab ( Table 8). The diagnostic sensitivity observed indicates the in-field application of VDN LAMP use may lie at the flock level, with this being appropriate as typically sheep management decisions are applied at this level. It has been shown that the highest bacterial load occurs in the early stages of disease [35,36], and this is the concentration range of bacteria where VDN LAMP is most accurate. This suggests VDN LAMP use would be most beneficial at the early stages of clinical disease, to identify if D. nodosus aprV2 is present and provide information that can be acted upon before severe lesions are present. As a proof of concept for on farm use, the VDN LAMP was deployed on 3 farms with different foot scoring ranges. In all cases, samples were collected and processed on farm with minimal equipment. Reagents were transported in an cold box; a pipette was used to deliver reagents and all buffers and dilutions were carried out in microfuge tubes. A VDN LAMP run was considered valid if the positive and no template H 2 O control performed as expected, and samples were considered positive if a Tp before 20 minutes and a Tm between 88.0-88.9˚C was seen, with samples amplifying after 16 minutes confirmed via rtPCR as having aprV2 present. No runs failed and the average time on farm for collecting and processing 14 samples was 1.5 hours. Samples positive for aprV2 were detected on all 3 farms. In the case of farm 1, there was 1 instance of co-infection with virulent and benign D. nodosus as confirmed by rtPCR that was identified as aprV2 positive by the VDN LAMP, while 9 samples rtPCR positive for only aprB2 were VDN LAMP negative. The VDN LAMP missed 1 aprV2 positive sample on farm 2, with this sample giving a positive result that was later discounted as the Tm (87.86˚C) did not meet the quality control requirements of a positive sample. Farm 3, which had been recently diagnosed by a private veterinarian as having virulent footrot, had 8/9 samples rtPCR aprV2 positive samples correctly identified by the VDN LAMP. One aprV2 rtPCR positive sample was missed by the VDN LAMP. The remaining 4/13 samples from farm 3 were aprV2 negative on both the VDN LAMP and rtPCR.
The in-field VDN LAMP assay was able to correctly identify 55/57 (96.49%) samples when compared to rtPCR, without requiring any type of sample nucleic acid isolation or purification. The LAMP platform has been shown to perform when large amounts of inhibitors are present, and in non-sterile or laboratory conditions [37]. The sample substrate commonly found in the pilot study of 57 field samples included soil, faeces, plant material and wool. The field trial of VND LAMP has demonstrated on-farm results and has differentiated aprV2 from aprB2 with DSe of 91.67% and DSp of 100%, while being capable of being performed easily and consistently on-farm with minimal equipment. The development of this assay will facilitate changes in approach to footrot control by providing a fast and accurate diagnosis of infection on-farm.

Conclusion
We have developed a highly specific and sensitive LAMP assay for the detection of virulent D. nodosus, which was able to correctly identify the presence of the aprV2 gene in samples collected on swabs in both laboratory and field settings. In the future we envisage the use of the VDN LAMP assay on farm, with continuing improvement to the sample preparation methods and the use of lyophilized reagents.