Reference strains for assay development were selected from a previous molecular characterization of blotch pathogens in which whole genomes were sequenced from 30 blotch-associated Pseudomonas (van der Wolf et al., 2016). Reference strains for “P. gingeri,” include LMG 5327^T, LMG 5328, IPO 3754, IPO 3777, IPO 3776, IPO 3769, IPO 3757, IPO 2767, and IPO 3756. Reference strains for P. tolaasii include LMG 2342^T, ATCC 51310, and ATCC 51309. Reference strains for the unknown brown-blotch causing Pseudomonas include LMG 2343 (formerly identified as P. tolaasii) and IPO 3765. Reference strains for each of the pathogens originate from symptomatic mushroom tissue and have been shown to demonstrate pathogenic behavior on fresh mushroom caps.

Target regions against “P. gingeri” and the unknown Pseudomonas sp. were selected from Multi-Locus Sequence Alignment (MLSA) on whole genome sequences of Pseudomonas isolated from symptomatic mushrooms (van der Wolf et al., 2016; unpublished results). Reads from reference strains of each pathogen were aligned, and the consensus sequence was divided into 500 bp sequence blocks. Blocks that mapped to the pathogens of interest, but did not map to strains from any other outgroups, and had no single nucleotide polymorphisms (SNPs) were selected as the target regions. The sequence blocks were checked for specificity with NCBI-BLAST. Target regions against P. tolaasii were selected on the “tolaasin” gene fragment (accession numbers AY291584, AY228241, U16024, and AF291753) from LMG 2342^T and other isolates. Target regions were checked for homology to other bacterial species, in particular to taxonomically related pseudomonads via NCBI-BLAST. Assay development was done for six target regions that were specific to target Pseudomonas sp.

For each pathogen, two amplicons, with length between 80 and 120 bp, were designed on different loci within their respective target regions. TaqMan^TM primers and probes were designed using the PrimerQuest web-tool (Integrated DNA Technologies), or Primer Express (Applied Biosystems) with the oligo size between 18 and 25 bp (Table 1). The amplicons are described in Supplementary Table S1. All amplicons were subjected to BLAST searches against the Pseudomonas database¹ to check for degeneracy. The specificity of the primers and probes was also checked via NCBI-BLAST. In total, six assays were developed against “P. gingeri” (Pg2, Pg6), P. tolaasii (Pt1, Pt2), and the unknown Pseudomonas sp. (Pu4, Pu10). The assays were multiplexed to also detect Xanthomonas campestris pv campestris (Xcc) (Köhl et al., 2011) and the Pseudomonas genus (Pp) (Lloyd-Jones et al., 2005).

The detection of bacterial DNA in varied samples was performed using TaqMan^TM probe technology, on a QuantStudio^TM 12K Flex from Applied Biosystems (Thermo Fisher Scientific, United States). Per reaction, the PCR mix included 0.25X ROX dye II, 1X TaKaRa^TM Premix Taq^TM DNA Polymerase (Takara Bio), 300 nM forward and reverse primers from all three targets, and 100 nM of FAM, VIC and ATTO labeled probes with NFQ-MGB quencher for target pathogens, X. campestris pv campestris and generic Pseudomonas, respectively. The total volume was made up with DNAse and RNase free water to 12.5 μl per reaction. Activation of Taq-polymerase was done at 95^°C for 2 min The qPCR amplification conditions consisted of denaturation for 15 s at 95^°C and extension for 60 s at 60^°C, for 40 cycles. A Threshold cycle (C_T) value lower than 35 was considered a positive reaction. qPCR reactions were performed with 1 ng of target DNA, in duplicates.

For determining the limits of detection, several types of positive controls were tested. The first positive controls consisted of gBlocks^® (Integrated DNA Technologies, United States). These synthetic oligonucleotides contain concatenated sequences of all amplicons to be tested within a multiplexed assay. gBlocks^® were designed per pathogen, for both target amplicons along with X. campestris pv campestris and generic Pseudomonas amplicons interspaced with some extra flanking nucleotides from the original gene sequence. The gBlock sequences are described in Supplementary Table S1. A serial dilution of the gBlocks^® from 10⁸ to 0 copies of target DNA was used to plot the calibration curves. Controls for water samples consisted of sterile water spiked with a pathogen suspension (in Ringers solution) to obtain a 10-fold serial dilution from 10⁸ to 0 cfu/ml. Controls for casing soil samples consisted of pure soil DNA, extracted from soil samples spiked with serially diluted pathogen suspension from 10⁸ cfu/g to a final concentration of 0 cfu/g of soil. All controls were spiked with three pathogens together. The negative control was DNAse and RNase free water.

Bacterial isolates were grown on King’s B or Trypic Soy agar for 24 h at 25^°C before DNA extraction. The entire panel of bacterial strains are described in Supplementary Table S2. Total DNA was extracted from 200 mg of bacterial colonies scrapped from a plate for each isolate using Wizard Magnetic DNA Purification System for Food (Promega, United States) according to the manufacturer’s protocol, including the DNase-free RNAse treatment. All the DNA for the testing panel was quantified fluorometrically using a Quant-iT PicoGreen dsDNA assay Kit (Thermo Fisher Scientific, United States) on the Infinite M200 PRO microplate reader (Tecon, Switzerland).

Assays were also evaluated by testing a variety of environmental samples. Bacterial suspensions were prepared from pure cultures in Ringers solution, and serially diluted from an upper concentration of 10⁸ cfu/ml (OD₂₈₀ = 0.1). Biopsies from symptomatic tissue were weighed, homogenized in 1 ml of Ringers solution, and the extract was filtered using BIOREBA bags (Bioreba, Switzerland). For water samples, 100 ml of tap-water was centrifuged at 9,000 g for 15 min at 4^°C. The supernatant was discarded, and the pellet re-suspended in 1 ml of sterile DNAse and RNAse free water. The bacterial suspensions, tissue extracts and water samples were centrifuged at 11,200 g, then heated to 100^°C for 10 min, and 1 μl of the supernatant was used for qPCR.

For casing soil samples received from mushroom farms, 250 mg of sampled material was processed for a semi-automated DNA isolation. The PowerMag Soil DNA Isolation kit from MoBio Technologies (Qiagen) was used to perform a total DNA extraction, eluted into 100 μl of Tris-EDTA buffer, using a Kingfisher Flex (Thermo Fisher Scientific) according to manufacturer’s protocol. 1 μl of the resultant DNA (concentration as extracted) was used for qPCR.

For diagnostic tests, 10⁵ cfu of X. campestris pv. campestris in 0.01 M phosphate buffer saline was added to water, mushrooms and soil samples before DNA extraction, to allow for relative quantification, by controlling for loss of bacterial DNA during sample processing.

Data analyses was performed on RStudio with R version 3.4.0 (R Core Team, 2013). Threshold cycle (C_T) values from TaqMan^TM qPCR were imported from the QuantStudio^TM 12K Flex Real-Time PCR Software (Thermo Fisher Scientific, United States). When the C_T values were plotted against the copies of template DNA, amplicon efficiency was calculated from the slope of the calibration curve, as 10^–1/slope−1 (Kubista et al., 2006). Pathogen populations in the samples were quantified relative to known concentrations of X. campestris pv. campestris by the 2^–ΔΔCT method (Livak and Schmittgen, 2001), as a fold change ratio.

In this study, six TaqMan^TM qPCR assays were designed, to detect two different loci for each of the three blotch pathogens (Figure 1). Each target-pathogen assay was multiplexed with two additional assays that quantify Xanthomonas campestris pv. campestris (Xcc) (Köhl et al., 2011) as an extraction control, and the genus Pseudomonas (Lloyd-Jones et al., 2005) for the total Pseudomonad population (Pp). In total, six triplex assays were developed; two for “P. gingeri,” (i) Pg2-Pp-Xcc (ii) Pg6-Pp-Xcc; two for P. tolaasii, (i) Pt1-Pp-Xcc and (ii) Pt2-Pp-Xcc; and two for the unknown Pseudomonas sp., formerly classified as P. tolaasii, (i) Pu4-Pp-Xcc (ii) Pu10-Pp-Xcc.

Assay Pt1 and Pt2 amplify two fragments of the tolaasin gene (AF291753, AY291584). Assay Pg2 amplify the LysR transcriptional regulator gene fragment (PNQ94452) whereas assay Pg6 amplifies a hypothetical protein fragment (PNQ88072). Assays Pu4 and Pu10 amplify two unknown gene fragments. The Xcc assays also amplifies a hypothetical protein fragment (QCX70676). The Pp assay amplifies the 16S gene fragment (MK294319). In the reference genome sequences all amplicon targets are single copy. The sequences of primers and probes for the triplex assays are described in Table 1. The amplicons sequences are described in Supplementary Table S1. In silico homology of amplicons to taxonomically-related Pseudomonas sp. using NCBI-BLAST, is described in Supplementary Table S2.

The multiplexed assays for all target pathogens were tested against positive controls comprising of gBlocks^®, in replicates of three. The TaqMan^TM assays were evaluated according to the Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines (Bustin et al., 2009). The calibration curves showed good linearity, symbolized by the high correlation coefficients (R² > 0.95). The amplification efficiencies (E) estimated from the slope of the curve, varied within 91.6–113%. The standard curves are illustrated in Figure 2. Delta Rn values (ΔRn) varied between 2.1 and 3.0 for the assays, suggesting a good magnitude of the amplification signal. The Xcc and Pp control assays both showed high precision, indicated by the low variation observed in qPCR results when tested in replicates of six, within and between experiments. Their repeatability (intra-assay variation) and reproducibility (inter-assay variation) is observed from the confidence intervals in Figure 3.

A panel to test the diagnostic specificity, inclusivity and sensitivity was assembled from international (LMG, CFBP, ATCC, and NCPPB) and local culture collections. This panel included 18 bacterial blotch pathogens, 18 mushroom-associated Pseudomonas, 28 plant pathogenic Pseudomonas, and 41 other plant pathogenic bacteria (34 species from 14 genera). The assays were tested on 1 ng of total DNA from pure bacterial cultures. The true positive (TP), true negative (TN), false negative (FN), and false positive (FP) results are summarized in Tables 2, 3.

Diagnostic specificity of an assay is a measure of the negative samples that are correctly identified. It is described as TN/(TN + FP). The assays tested negative against the majority of other Pseudomonas strains and an assortment of plant-pathogenic bacteria. Very few cross reactions were observed with non-target strains, resulting in an overall specificity of 0.99, 0.99, 1, 1, 0.99, and 1 for Pg2, Pg6, Pt1, Pt2, Pu4, and Pu10, respectively. All non-specific amplifications were characterized by a very late detection of target DNA in one or both replicates, implying a weak positive reaction, where 35 > C_T > 40. Therefore C_T < 35 was considered an unambiguous positive result.

Diagnostic sensitivity of an assay is a measure of the positive samples that are correctly identified. It can be described as TP/(TP + FN). The six target assays, Pg2, Pg6, Put1, Pt2, Pu4, and Pu10, all tested positive for each of their respective reference strains, characterized by an early detection of DNA, where C_T < 25. No false negatives were detected, implying a high diagnostic sensitivity of 1. The mean C_T values from diagnostic specificity and sensitivity tests are described in Supplementary Table S3.

The detection thresholds of all assays increase across target DNA (gBlocks^®), spiked water and soil samples, indicated by the mean C_T values (Figure 4). The limit of detection (LOD), is described as the lowest concentration at which 95% of the positive samples are consistently detected, characterized by C_T < 35. When tested against serially diluted gBlocks^®, assays Pg6, Pt1, Pt2, Pu4, Pu10, and Xcc had a LOD of 10 copies of target DNA. Assays Pg2 and Pp had a higher LOD of 100 copies of target DNA. For water samples, the LOD for assays Pt1, Pu4, and Pu10 was 10³ cfu/ml. Assays Pg2, P6, and Pt2 had a higher LOD of 10⁴ cfu/ml. For casing soil samples, assays Pt1 and Pt2 had the lowest LOD of 10³ cfu/g. Assays Pg2, Pg6, Pu4, and Pu10, had a higher LOD of 10⁴ cfu/g. The control assays, Xcc and Pp, gave consistent positive reactions at C_T∼28 and C_T∼30 for X. campestris pv campestris and the total Pseudomonas population in the soil, respectively.

Water, soil and symptomatic tissue from mushroom farms were tested to apply the newly designed assays. For primary detection, assays Pt1, Pg2, and Pu4 were employed, and the results were verified with the secondary assays, namely, Pt2, Pg6, and Pu4. Over the course of 10 independent batches in which samples were processed, diseased mushrooms, casing soil, peat, and tap water was sampled from mushroom farms across Netherlands and Belgium, and stored at -20^°C until tested.

Diseased mushrooms were received from 50 sources varying in mushroom farm, soil, compost, cultivar, environmental conditions, and harvest cycle. Out of 95 symptomatic cap tissue tested, 78% of the samples were strongly positive (C_T < 25) for “P. gingeri” with assays Pg2 and Pg6, 42% of mushrooms tested positive for the unknown Pseudomonas species with the Pu10 and Pu4 assays, and 20% of the mushroom tested positive for P. tolaasii with Pt1 and Pt2 assays. Only 9.4% of the symptomatic samples tested negative against all assays. Several samples also showed a secondary infection with another pathogen, when multiple assays tested positive on the mushroom tissue. Immense variability exists between the pathogen densities on biopsies taken from mushroom caps, ranging from 4.8 × 10² cfu to 7.9 × 10⁶ cfu. Pathogen density also varied with the severity of the disease symptoms.

Peat, before being mixed into casing soils in the mushroom farms, was received from 9 geographical sources. Out of 14 samples of peat tested, only 2 were positive for “P. gingeri” with assays Pg2 and Pg6, whereas 6 samples tested positive for the unknown Pseudomonas species with assays Pu4 and Pu10. These samples were not tested with assays Pt1 and Pt2 for P. tolaasii. Pathogen populations were consistently detected in the range from 1.2 × 10⁵ to 4.2 × 10⁵ cfu per gram. Co-occurrence of the pathogens was observed in only one peat sample, when it tested positive for multiple assays. Seven peat samples tested negative against all assays. Tap water from all samples tested unanimously negative for the blotch pathogens, characterized by C_T > 40.

All assays demonstrated high diagnostic specificity when tested against the bacterial panel, although occasionally weak false positives were detected, as observed in Supplementary Table S3. Multiple negative controls were also tested within the same setup, such as (a) no template control (b) sterile water control, (c) DNA extracted from a uninoculated plate of media. Contamination from positive controls during the qPCR, in reagents, or during DNA extraction is thus ruled out. In such samples, when DNA from pure bacterial cultures is used as template, due to very high DNA concentrations, template DNA can sometimes align with the fluorescent probe despite limited homology, leading to weak false positive reactions. Environmental samples, such as casing soil and water, contain much lower levels of pathogens, with the exception of symptomatic mushroom tissue. Hence these false positives are unlikely to occur when the diagnostic assays are applied during routine screening of raw materials used in mushroom farms. Furthermore the use of a C_T threshold value, determined from the assay validation data, allows interpretation of true positive and negative samples.

For diagnostic purposes, precise quantification of the very low density samples by qPCR is critical. All six target assays showed good linearity across the dilution range for their calibration curves, even at lower densities. The calibration curves of all assays showed similar slopes across the gBlocks^®, water and soil samples, indicating that the qPCR efficiency is uninfluenced by sample type. The mean C_T values for the assays, at a given concentration of pathogen, increased from gBlocks^® and water to soils samples. This is potentially due to DNA loss during the extraction process, and can be addressed by improving DNA isolation methods for soil samples, which in turn can also reduce the presence of qPCR inhibitors such as humic and fulvic acids.

These assays can be successfully applied to identify and quantify pathogens on mushrooms, casing soils and peat sources at much lower densities than previously possible. Former thresholds of P. tolaasii density on mushroom caps were in the range from 7.7 × 10⁴ to 2 × 10¹⁰ cells per cap (Nair and Fahy, 1972; Nair and Bradley, 1980; Preece and Wong, 1982; Olivier et al., 1997). In the casing soil, pathogen densities were earlier reported in the magnitude of 10⁷–10⁹ cfu/g of soil (Nair and Fahy, 1972). For both caps and soils, 100-fold lower pathogen density thresholds were detected with the newly developed TaqMan^TM assays. This increased sensitivity, makes it now possible to screen raw materials used in mushroom cultivation, such as spawn, peat, compost and casing soil for presence of bacterial blotch pathogens.

Previously, the use of DNA-based molecular detection methods for routine diagnostics has been demonstrated for agricultural pathogens (Ward et al., 2004; Ophel-Keller et al., 2008; Bonants and te Witt, 2017). However, several considerations and limitations exist in the use of these assays to determine and provide insights into the “health status” of mushroom farms. The relationship between pathogen populations in the casing soil, and a disease outbreak is rather volatile. Blotch outbreaks depend on a variety of other factors also, which include environmental conditions, growing practices, cultivar type, source of compost and type of casing soil (Godfrey, 2003). The direct translation of pathogen population densities retrieved from these assays into prediction of an economic risk requires good knowledge and cautious interpretation of other disease indicators.

Previously known diagnostic methods based on colony morphology and growth characteristics (Wong and Preece, 1979), microscopy (Preece and Wong, 1982; Soler-Rivas et al., 1999) and biochemical features (Goor et al., 1986), were time consuming and prone to false positive or -negative results. Conventional PCR assays (Kwon et al., 2000; Lee et al., 2002) require the use of gel electrophoresis for post amplification analyses, which make the assays less suitable for high throughput screening. Taqman^TM-qPCR assays designed here, provide the opportunity for quantitative high-throughput detection of blotch pathogens at high precision, sensitivity and specificity in multiple environmental samples. Additionally, the assays ensure repeatability and accuracy in the diagnostic measurements.

A major limitation of these assays is that they only offer information on abundance of the pathogens, not its viability (Vincelli and Tisserat, 2008). The active fraction of pathogens cannot be calculated, as DNA-based assays cannot differentiate between dead and living cells (Wolffs et al., 2005). To quantify the fraction of viable cells, agents such as propidium monoazide may be used, that only penetrate dead cells with a compromised cell membrane and bind to DNA, after which the complexes can be to selectively removed prior to the TaqMan assay (Fittipaldi et al., 2012). The presence of viable cells by TaqMan assays may also be determined by enrichment of target cells in or on selective growth media (Schaad and Frederick, 2002) or reverse-transcriptase PCR to amplify mRNA instead (Kobayashi et al., 2009). However, these two methods do not quantify the fraction of viable cells accurately.

These assays present the first opportunity to quantify exact pathogen populations in environmental samples, and consequently study the microbial ecology and population dynamics of the blotch pathogens. They can be used in field-trials for testing the efficacy of biocontrol agents and disinfectants in mushroom beds. Based on population dynamics during the mushroom cropping cycle, they can be used to improve the dose and time of application. They can also be used to study the survival rates of these pathogens in the post-harvest supply chain, when introduced from the raw-materials, under varying storage conditions. Despite being limited to three different pathogens, these tools generate many fundamental insights for disease prevention, control and management.

For use of these assays as a monitoring system, an up-scaled sampling strategy needs to be explored for an industrial analysis. The number of samples that are representative of a batch, and the frequency of sampling that takes into account the effect of storage, seasonal variation, and bulk quantities, remain largely undetermined. Additionally, the use of an agent for removal of DNA from dead cells, could allow quantification of only the viable pathogen populations. Detailed information on the survival rates and population dynamics of the pathogen is essential to place quantitative information about pathogen densities into context. Finally, field trials are also required to understand the variation in the inoculum threshold of the soil for disease outbreaks, based on the type of casing soil, compost, cultivar, watering method and environmental conditions during production, etc.