Blast searches of the annotated genes of Lxx against publicly available microbial genomes revealed that the gene coding for the DNA replication/repair protein RecF (Accession: AE016822; locus tag: LXX_RS00020) was unique and conserved across Lxx genomes and thus used as a potential target for Lxx detection. 5 LAMP primer sets, named LXX20-1, LXX 20-2, LXX20-3, LXX20-4 and LXX20-5 were subsequently designed. Previously published gene encoding the UDP-N-acetylglucosamine 2-epimerase (Accession: AE016822; locus tag: LXX14026) was also used as a target for LAMP primer development (Taylor et al., 2003). LAMP primers were designed using the LAMP primer development software Primer Explorer (https://primerexplorer.jp/e/) and following the guidelines by Notomi et al. (Notomi et al., 2000). 16S rRNA primers previously reported by Liu et al. were also used in this study (Liu et al., 2013). The LAMP primers were also checked for self-priming using Thermo Fisher Scientific’s primer analyzer tool(https://www.thermofisher.com/au/en/home/brands/thermo-scientific/molecular-biology/molecular-biology-learning-center/molecular-biology-resource-library/thermo-scientific-web-tools/multiple-primer-analyzer.html). Details of all six primer sets are available in Table 1 .

15mL crude juice samples were collected at two sugar factories and processed immediately or stored at -20°C until processed later. Lxx DNA was purified from these samples using the dipstick DNA technique as described by Zou et al. and Mason et al. (Zou et al., 2017; Mason and Botella, 2020). Briefly, the exposed cellulose DNA binding zone of the dipstick was dipped in the crude juice, following which it was soaked for 2 minutes in 500μL extraction buffer (20mM Tris (pH 8), 25mM NaCl, 2.5mM EDTA, 0.5% Sodium Dodecyl Sulfate (w/v), 2% PVP (w/v)); dipped 10x times in 800μL wash buffer (10mM Tris (pH 8.8), and 8mM MgCl₂) and any bound DNA was eluted directly into the reaction tube containing 45μL LAMP reagents.

For total DNA extraction, frozen crude juice samples were thawed and allowed to sediment at room temperature for 30 minutes, followed by heat lysis of equal volumes of the sample in the presence of lysis buffer (4% SDS (w/v), 500mM NaCl, and 50 mM EDTA) at 65°C for 30 minutes. The total DNA was then purified using SPRI beads (AMPure XP, Beckman Coulter, Australia) according to manufacturer’s instructions and stored at -20°C until further use.

Unless stated, LAMP reaction mixture contained 0.8M betaine, 20mM Tris-HCl (pH 8.8), 50mM KCl, 10mM (NH₄)₂SO₄, 0.1% Tween^® 20, 8mM MgSO₄, 1.2mM dNTP (each), 0.32 U/mL Bst 2.0 warm start polymerase (New England Biolab), 0.2μM (each) F3 and B3 primers, 0.8μM (each) loop forward and reverse primer, 45 μL of reactions were incubated at 65°C for 60 minutes in our custom made hand held diagnostic device (Botella, 2022). When necessary, the amplified products were visualized using agarose gel electrophoresis, the raw values and real time curves were analyzed using Microsoft Excel. In case of fluorescent LAMP reactions, Syto9 dye was added at a final concentration of 1.25 μM and reactions incubated at 65°C for 60 minutes in a Bio-Rad CFX 96™ Real-Time System (Bio-Rad Laboratories, Australia).

The real-time quantitative PCR (qPCR) reactions were performed in a Bio-Rad CFX 96™ Real-Time System (Bio-Rad Laboratories, Australia) and were performed in 96-well plate, with three technical replicates for each sample. Each 25 μL reaction mixture contained 12.5 μL of GoTaq® qPCR master mix (2X) (Promega, Australia), 1 μM each of Lxx202FB/Lxx331R, 2 μL of total DNA template (extracted as described above), 5.5 μL ultra-pure water. The standard curves were generated using a serial dilution of the 130-bp purified amplicon of the Lxx 16S-23S rRNA intergenic transcribed spacer (ITS) region as described by Grisham et al.(Grisham et al., 2007). The no template control contained 2 μL of ultra-pure water. Thermal cycling consisted of denaturation at 95°C for 5 minutes, followed by 40 cycles of 95°C for 10 seconds, 60°C for 30 seconds and 72°C for 20 seconds. Post amplification final melting curve of 65°C to 97°C was also performed. The efficiency of primers was calculated according to Rasmussen et al.(Rasmussen, 2001). PCR amplification of the Lxx UDP-N-acetylglucosamine 2-epimerase and Lxx 16S-23S rRNA ITS gene was performed in 20 μL reaction volumes, with 10 μL GoTaq® green master mix (2X) (Promega, Australia), 0.5 μM each of S1000-2 F3/B3 primers and 2 μL of total DNA from crude juice or xylem extracted from infected plant. Cycling conditions were as following: 95°C for 2 minutes; 35 cycles of 95°C for 30 seconds, 60°C for 30 seconds, 72°C for 30 seconds; and a final extension at 72°C for 5 minutes. The PCR products were visualized by agarose gel electrophoresis.

The 230-bp amplicon of UDP-N-acetylglucosamine 2-epimerase gene (for LAMP reaction) and 130-bp amplicon of Lxx 16S-23S rRNA ITS gene (for qPCR) were purified using QIAquick PCR kit as per manufacturer’s instruction (Qiagen, Australia). For LAMP, 10⁶, 10⁵, 10⁴, 10³, 10², 10 and 1 fg of purified DNA was added to clean sugarcane juice sample, DNA was purified using the dipstick method as described in section 2.2 and eluted into LAMP reaction of 45 μL each. For qPCR, 4x10⁴, 4x10³, 4x10², 40 and 4 fg of DNA was directly added to each qPCR reaction tube of 25 μL volume, with three technical replicates for each DNA concentration. These concentrations of Lxx DNA were also used for generation of standard curves for qPCR. The thermal conditions and composition of the amplification reactions were as described in Sections 2.3 and 2.4.

GraphPad Prism (version 9.5.1) was used for all statistical analyses. For calculation of LAMP and qPCR diagnostic sensitivity, specificity, and area under the curve (AUC), the ROC curve analysis embedded in the software was used. When comparing the outcomes of two or methods, a chi-square test with 95% confidence interval was applied. One-way ANOVA followed by Dunn’s multiple comparison test was applied when comparing differences between different LAMP reaction composition parameters. Graphical rendering of data and schematics was done using Microsoft Office suite, GraphPad Prism and BioRender (BioRender.com).

Five LAMP primer sets, named LXX20-1, LXX20-2, LXX20-3, LXX20-4 and LXX20-5 were designed targeting the DNA replication/repair protein RecF (Accession: AE016822; locus tag: LXX_RS00020) ( Table 1 ). In addition, a LAMP primer set named S1000-2 was designed targeting a gene encoding the UDP-N-acetylglucosamine 2-epimerase (Accession: AE016822; locus tag: LXX14026)(Taylor et al., 2003). Finally, a primer set (called 16S RSD) published by Liu et al. was also evaluated (Liu et al., 2013). All seven LAMP primer sets were tested for their ability to yield amplification products using purified Lxx DNA and their susceptibility to produce amplification products by self-priming in the absence of Lxx DNA (i.e. no template controls or NTCs). In the absence of Lxx DNA ( Figure 1 , left side), primer sets 16S RSD, LXX20-1, LXX20-2, LXX20-3, LXX20-4 and LXX20-5 generated amplification bands of different intensities; whereas primer set S1000-2 failed to produce any detectable amplification. All 7 primer sets produced strong amplification in the presence of Lxx DNA ( Figure 1 , right side), evidenced by the characteristic LAMP ladder-like bands. Based on these results, the primer set S1000-2 was selected for further experiments.

The S1000-2 primers F3/B3 were used in PCR reactions and the sequence identity of the PCR amplicons confirmed using Sanger sequencing. The ability of the S1000-2 F3/B3 primers to detect different L. xyli subsp xyli strains was established by testing xylem extracts from plants from suspected RSD positive farms from 13 different growing locations across the Australian sugarcane industry ( Supplementary Table 1 ). The S1000-2F3/B3 primers successfully amplified a 230bp amplicon in 11 out of the 13 samples analyzed ( Supplementary Figure 1 ), and Sanger sequencing revealed strong sequence conservation among all isolates. The specificity of the S1000-2 primers towards Lxx UDP-N-acetylglucosamine 2-epimerase gene was ascertained by sequencing a 230bp amplicon in total DNA extracted from a Lxx infected crude juice sample ( Supplementary Figure 2 ). The sequencing results revealed high percentage of sequence identity to Lxx strain CTCB07 genome and low sequence homology to other prevalent environmental bacteria.

Sugar mill laboratory staff has limited time for sample processing and are not usually proficient in molecular biology techniques. We therefore decided to use the dipstick DNA purification method as it combines simplicity with speed (Zou et al., 2017; Mason and Botella, 2020). Crude mill juice samples contain diverse plant and non-plant materials, high concentrations of sugar, silt and soil which exert a strong inhibitory effect on DNA amplification (Henson and French, 1993; Lopes and Damann, 1993). Crude mill juice samples can also show large variability in the quantity of silt/soil present as can be seen in the different amounts of pellet observed after centrifugation of five independent samples, obtained from a region with high incidence of RSD ( Figure 2A ). In an initial approach, 500µL of crude juice samples were centrifuged for 5 mins at 16,000 x g to collect bacterial cells before discarding the supernatant and resuspending the pellet in the same volume of extraction buffer. DNA dipsticks were used to purify DNA from the solution and release it into LAMP reactions containing the S1000-2 primer set for Lxx detection. Using this method, three out of five samples analyzed produced positive LAMP amplification products 18-20 minutes after starting the reaction ( Figure 2B ). The negative amplification results obtained for samples 2A and 7A could be due either to the absence of Lxx bacteria in the juice or the presence of inhibitors in the amplification reaction. In an effort to reduce the amount of insoluble matter, 1.6 ml of crude mill juice samples 2A and 7A were allowed to settle for 60 mins before carefully collecting 500µL of sample from the top and subjecting it to the above-described treatment (i.e. centrifugation plus resuspension in extraction buffer) ( Figure 2C ). Dipstick purification of DNA from resuspended samples followed by LAMP amplification failed to detect the presence of Lxx in sample 2A, while the addition of the settlement step resulted in a positive amplification for sample 7A. Simultaneous treatments were also performed spiking the initial juice with xylem from an RSD-positive plant. In the case of sample 7A, spiking failed to produce an amplification product in the non-settled sample while settling produced a positive amplification indicating that the pre-settlement step is efficient in reducing inhibitors from the crude mill juice and improves the reproducibility of the LAMP assay. In contrast, the presence of spiked Lxx DNA was detected in the settled and non-settled treatments for sample 2A ( Figure 2D ) suggesting that the initial negative result ( Figure 2B ) was due to the absence of detectable Lxx DNA in the sample.

Consistent with the practical needs of a busy sugar mill with a high turnover rate of juice samples, we further investigated how to streamline the crude juice sampling process. For this purpose, we investigated whether it was necessary to centrifuge samples to collect bacterial cells or whether there was enough free Lxx DNA present in the crude juice to allow detection by LAMP amplification. We obtained 31 juice samples from a sugar mill (location undisclosed) during the 2020 crushing season and assayed them in our laboratory (ex-situ) using an initial settlement step of 60 mins followed by (a) centrifugation of samples plus resuspension (Spin) or (b) direct assay of the supernatant (No-spin). The RSD status of the samples was confirmed by purification of the Lxx DNA using dipsticks followed by conventional qPCR using primers previously developed for Lxx detection (Grisham et al., 2007). Both methods produced comparable results (results not shown), thus a preliminary trial at a mill was implemented during the 2021 crushing season to initiate technology transfer and to test the practical limitations of performing molecular analysis in a mill laboratory not designed for molecular biology procedures. Of the 56 juice samples assayed for RSD, 26 samples tested positive using the Spin method, whereas 31 samples tested positive using the No-Spin method, with no significant differences observed between the two methods, this experiment was repeated twice (n=2) ( Figure 3 ). As the two sampling methods did not significantly affect the RSD-LAMP results, we decided to adopt the gravitational sedimentation step for 30-60 minutes followed by dipstick purification. The LAMP reaction results were validated using the qPCR method with primers described by Grisham et al., and the template for qPCR was generated by using the total DNA extraction method as described in section 2.2.

To improve the robustness of the LAMP assay for Lxx detection and minimize the occurrence of non-specific false positive amplifications, we investigated the effect of betaine and primer concentrations. Previous work by our group and others have established that reducing betaine concentration in the reaction mix can significantly decrease the amplification time of the target DNA (Zou et al., 2020), hence a concentration of 0.5M betaine was used in the initial Lxx LAMP assays. One of the drawbacks of isothermal LAMP amplification is its propensity to produce non-specific amplification products due to primer dimerization in the absence of DNA template (Botella, 2022). Our experiments revealed that the use of 0.5M betaine occasionally produced self-amplification in no-template controls (NTCs) resulting in the production of false positives ( Supplementary Figure 3 ), therefore we performed a betaine concentration titration experiment. Addition of 0.8M betaine to the LAMP reaction produced no amplification in water control samples (n=6), whereas 0.5M betaine produced amplification in 1 out of 6 NTCs ( Figure 4 ). Excessive betaine (1M) as well as a lack of it resulted in unspecific amplification in 50% of the samples assayed.

We also investigated the effect of primer concentration on the specificity of the LAMP amplification detected by fluorescence and turbidity-based methods. The S1000-2 primer mix was tested at the concentrations recommended by the manufacturer of the Bst 2.0 warm start polymerase used for amplification (New England Biolabs) (1.6μM FIP/BIP, 0.2μMF3/B3, 0.8μM LoopF/LoopB) (named 1x) as well as 0.5x and 0.25x. Fluorescence monitoring of the reaction in a real time thermocycler revealed that when using 1x primer concentration unspecific amplification in NTCs was first observed at a reaction time of 36.6 mins and became increasingly frequent after 50 mins ( Figure 5A ). Reducing primer concentration to 0.5x produced 3 amplifications out of 21 NTCs at incubation times of 46 minutes and above, while the use of 0.25x primer concentration produced only one false positive at ~ 60 minutes. Reduced primer concentration also conveyed a small delay in amplification. Turbidity based detection of LAMP amplification is less sensitive than fluorescence and, in our experiments, resulted in a significant delay in the detection of amplification products when using 0.25X primer concentration (p=0.0177) ( Figure 5B ). Thus, our results indicate that the likelihood of false positive amplification in the RSD assay can be significantly decreased by using a 2-fold reduction of primer concentration over that recommended by the manufacturer.

We used the industry-wide accepted fluorescence-based qPCR Lxx detection method adapted from Grisham et al. (Grisham et al., 2007) to validate our RSD-LAMP assay which utilizes the dipstick method for DNA isolation and turbidity-based detection. RSD-LAMP assays were performed in samples containing serial dilutions of a fragment of the L. xyli subsp. xyli UDP-N-acetylglucosamine 2-epimerase gene (produced by PCR). LAMP produced turbidity detectable Lxx amplification with as little as 1fg of DNA within 55 minutes ( Figure 6A ), with Lxx DNA concentrations >1 fg showing amplification between 19-30 minutes. In comparison, qPCR showed reliable fluorescence-detectable amplification of 4fg of Lxx 16S-23S rRNA ITS ( Figure 6B ). These limits established for both LAMP and qPCR assays were applied as threshold cut-off for statistical analyses and evaluation of both diagnostic methods.

During the 2021 crushing season, a large-scale experiment was performed on-location at two different mills. A total of 619 juice samples from individual rakes were assayed for RSD (518 at Site A and 101 at Site B), using the optimized sampling and reaction conditions. In short, crude juice samples were left to settle for 45 minutes before using DNA dipsticks to purify the DNA and load LAMP reactions. Amplification was monitored by the appearance of turbidity using a purpose-made light reading device (Botella, 2022). Based on the limit of detection of 55 minutes for LAMP, samples showing amplification within 55 minutes in LAMP reactions were deemed as infected, whereas samples showing no amplification or late amplification (after 55 minutes) were classified as healthy. After LAMP analysis at the mills, samples were brought back to the laboratory, DNA was purified using heat and chemical lysis, followed by magnetic bead purification and analyzed by using 2μl of DNA template by qPCR. Samples containing more than 4fg of Lxx DNA were classified as Lxx-infected. These cut-off criteria were applied to all samples yielding dichotomous results for both diagnostic assays i.e. “Infected” or “Healthy”. The Receiver Operating Characteristic (ROC) curve is an effective method of evaluating the performance of diagnostic tests enabling the calculation of their specificity (false positive rate), sensitivity (true positive rate), and diagnostic predictive power (Mallett et al., 2012; Trevethan, 2017). The area under the curve (AUC) of a ROC is a measure of a diagnostic test’s predictive power, with higher AUC values indicating the discriminatory accuracy of a test. The RSD-LAMP and qPCR diagnostics tests yielded AUC of 1, thereby suggesting that both diagnostic tests have excellent discriminating power. When assessed independently using the ROC curve, the LAMP reaction time cut-off (LAMP Ct) of 55 minutes was found to be a good discriminator between infected and healthy crops, with a sensitivity of 94.67% and specificity of 100% ( Figures 7A, B ). Similarly, when using a threshold load >4fg of purified Lxx DNA from juice samples in the qPCR assay, the sensitivity and specificity were 98.23% and 100%. However, comparison of the pairwise sensitivity and specificity values across all Lxx DNA concentrations revealed that increasing the threshold above 4fg rapidly reduced the sensitivity of the qPCR assay. In contrast, if a Lxx DNA concentration cut-off of 2fg was used, the sensitivity improved to 100%, with specificity decreasing to 89%. As the aim of the assay is to detect all infected samples, a cut-off of 2fg of Lxx DNA load was chosen for any further analyses.

When the 2 diagnostic methods were compared to each other the newly developed RSD-LAMP assay showed a 77.05% correlation of with the Lxx qPCR assay. The RSD-LAMP assay indicated disease incidences of 42.08% and 6.93% in sites A and B respectively; while the qPCR-based detection method indicated 32.62% and 1.98% disease incidence at Site A and B, respectively ( Figure 8 ). Although the differences between the LAMP and the qPCR methods are not significant, the values provided by the -LAMP assay are closer to the known disease incidence in the two growing regions compared to the qPCR assay.