Moving from qPCR to Chip Digital PCR Assays for Tracking of some Fusarium species causing Fusarium Head Blight in Cereals

Fusarium Head Blight (FHB) is one of the major diseases affecting small-grain cereals, worldwide spread and responsible for severe yield and quality losses annually. Diagnostic tools, able to track Fusarium species even in the early stages of infection, can contribute to mycotoxins’ risk control. Among DNA-based technologies for Fusarium detection, qPCR (single and multiplex assays) is currently the most applied method. However, pathogen diagnostics is now enforced by digital PCR (dPCR), a breakthrough technology that provides ultrasensitive and absolute nucleic acid quantification. In our work, a panel of chip digital PCR assays was developed to quantify Fusarium graminearum, F.culmorum, F. sporotrichioides, F. poae and F. avenaceum. The primers/probes combinations were evaluated on pure fungal samples with cdPCR technique, in comparison with the qPCR approach. Moreover, the cdPCR assays were applied to quantify Fusarium in durum wheat and oat samples, naturally contaminated or spiked with fungal DNA. For a better evaluation of infection level in plants, duplex assays were developed, able to co-amplify both plant and fungal DNA. To the best of our knowledge, this is the first study directed to the application of digital PCR to Fusarium diagnosis in plants.


Introduction
Fusarium Head Blight (FHB) is one of the major diseases affecting small-grain cereals, it is worldwide spread and responsible for severe yield and quality losses annually.
Several fungal species, mainly of the Fusarium genus, have been identified as the etiological agents of such a disease. In European cultivation environments, FHB occurs, mainly, because of Fusarium graminearum and Fusarium culmorum presence, but also Fusarium poae, Fusarium pseudograminearum, Fusarium avenaceum, Fusarium sporotrichioides and Fusarium langsethiae. Most of the species associated with FHB, in advantageous environmental conditions, invade the ear of the cereals and produce toxic secondary metabolites-mycotoxins-that contaminate the grain. FHB, therefore, compromises not only the yield but also the grain safety and quality due to the accumulation of mycotoxins in infected kernels. Depending on species and chemotypes, Fusarium can produce A and B trichothecenes: type A trichothecenes include highly toxic mycotoxins, such as T-2 and HT-2, meanwhile type B trichothecenes include, among others, deoxynivalenol (DON), nivalenol (NIV) and acetyl-NIV. Moreover, fumonisins, zearalenone, beauvericin and enniatin B can accumulate in cereal grains after bibliographic search, focusing on DNA-based methods and aiming to conduct Fusarium diagnostics in small-grain cereals, was published in 2009-2019 [9]. By applying the appropriate filters, 50 publications have been selected and analyzed to derive information, among others, on the molecular technology used. The obtained results are schematically shown in Figure 1, from which it can be inferred that the qPCR (single and multiplex assays) is the most widespread method, followed by multiplex PCR, LAMP-based protocols and metabarcoding.
Strains were stored on potato dextrose agar (PDA, Liofilchem, Teramo, Italy) at 4 • C until use. Fungal DNA was extracted from lyophilized mycelium previously grown on PDA medium, according to the procedure described by Al-Samarrai and Schmid [16]. DNA concentrations were determined using Qubit ® fluorimeter (Life Technologies™, Invitrogen, Monza, Italy)
The plants were grown in the experimental fields of Research Centre for Genomics and Bioinformatics, in the 2015 and 2016 seasons, without any fungicide treatment. The 23 wheats and oats were grown in 3-m 2 plots, in triplicate. At maturity, the plots were harvested, and 20 gr of grains were sampled from each of the three plots and bulked. The 60-gr bulked sample was then milled into a fine powder using an analytical mill (IKA Universal mill M20, IKA-Werke GmbH, Staufen, Germany) and stored at 4 • C until analysis.
Plant genomic DNA was extracted in triplicate from 100 mg samples from the 60-gr bulked milled grains using DNeasy Plant Mini Kit (Qiagen Italia, Milano, Italy) according to the manufacturer's instructions. DNA concentrations were determined using Qubit ® fluorimeter (Life Technologies™, Invitrogen, Monza, Italy). The DNA extracted were analyzed with qPCR to evaluate the presence of Fusarium. A subset of these samples was analyzed with cdPCR for Fusarium quantification. Moreover, a second subset of plant DNA samples was spiked with fungal DNA. For the preparation of such samples, batches of 20 ng plant DNA were added with 250, 100, 10, and 1 pg of fungal DNA. Table 1 reports the primers and probes sequences. Primer Express 3.0.1 Software (Life Technologies™, Invitrogen, Monza, Italy) was used to design F. spo, F. gram/culm and Avena dig assays. Multiple Primer Analyzer (Thermo Fisher Scientific, Monza, Italy) was used to verify the absence of self-complementarity and primer dimer formation.  For the determination of reaction efficiencies, standard curves were generated by plotting the Ct (Cycle Threshold) values versus the log10 amount of pure DNA of the different Fusarium (10-fold dilution series).  Table 2. End-point fluorescence data were collected in QuantStudioTM 3D Digital PCR Instrument and files generated were analyzed using cloud-based platform QuantStudioTM 3D AnalysisSuite dPCR software, version 3.1.6. Each sample was analyzed in duplicate.

Fungal Samples
The fungal DNA stocks were initially quantified with Qubit and the same dilutions of F. sporotrichioides, F. graminearum, F. culmorum, F. poae and F. avenaceum DNA were amplified with both qPCR and cdPCR techniques. A dynamic range of 0.5-0.0005 ng of fungal DNA was considered in all the assays. The same primers/probe combinations and the same amplification conditions were applied in both techniques. Table 3 reports the qPCR assays' efficiencies, as determined with qPCR. Table 3. The R2 coefficients and amplification efficiencies of the four assays targeting Fusarium species were calculated in qPCR with the standard curve approach, using six calibration points with three PCR replicates each and the formula E = 10 −1 /slope.  Figure 2 reports examples of the amplification patterns obtained after cdPCR analysis of some fungal DNA samples.  Figure 2 reports examples of the amplification patterns obtained after cdPCR analysis of some fungal DNA samples. Linearity between DNA dilution factors and copies/uL (cdPCR determined) has been found for all the assays (Figure 3) as well as high correlation levels between theoretical and cdPCR measured copies/µL for F. sporotrichioides (R 2 = 0.987), F. graminearum (R 2 = 0,999), F. poae (R 2 = 0.999) and F. avenaceum (R 2 = 0.999). Linearity between DNA dilution factors and copies/uL (cdPCR determined) has been found for all the assays (Figure 3) as well as high correlation levels between theoretical and cdPCR measured copies/μL for F. sporotrichioides (R 2 = 0.987), F. graminearum (R 2 = 0,999), F. poae (R 2 = 0.999) and F. avenaceum (R 2 = 0.999). The Limit of Detection (LOD) and the sensitivity in cdPCR, for the four assays tracking Fusarium, were calculated with QuantStudioTM 3D AnalysisSuite dPCR software, as reported in Table 4. Table 4. Limit of detection expressed as copies/μL and sensitivity of the four cdPCR assays.

Target
Limit of Detection Sensitivity The Limit of Detection (LOD) and the sensitivity in cdPCR, for the four assays tracking Fusarium, were calculated with QuantStudioTM 3D AnalysisSuite dPCR software, as reported in Table 4. Precision refers to the ability of distinguish between two measurements with a certain confidence. The AnalysisSuite TM Software calculates precision as the ratio of the maximum deviation of the confidence interval to the mean value, therefore it expresses the tightness of the confidence interval: the lower the precision, the tighter the confidence interval. In Figure 4, the precisions and the corresponding quantification values obtained amplifying fungal DNA dilutions with the four assays are reported. The highest DNA dilutions move rare target samples below the lower limit of detection and outside the supported dynamic range.

Plant Samples
cdPCR assays were validated on cereal samples naturally infected with mycotoxigenic fungi or spiked with fungal DNA. As previous step to cdPCR analyses, a set of durum wheat and oats samples were analyzed with the qPCR assays described in Table 2 to track fungal species. The rationale behind this preliminary step was to individuate samples naturally contaminated and samples free of fungal contamination and therefore suitable for the preparation of artificially contaminated ones. Three classes were found: i) samples in which no fungal species has been detected; ii) samples contaminated with one Fusarium species; iii) samples contaminated with two or more Fusarium species.
Starting from the data, the following two subsets of samples were further analyzed with cdPCR.
• Naturally contaminated samples, belonging to classes ii) and iii); • Synthetic samples created by spiking plant DNA (extracted from samples found not contaminated) with fungal DNA at different concentrations.
The fungus quantifications in the two subsets of samples obtained with the qPCR and cdPCR assays are reported in Table 5.

Plant Samples
cdPCR assays were validated on cereal samples naturally infected with mycotoxigenic fungi or spiked with fungal DNA. As previous step to cdPCR analyses, a set of durum wheat and oats samples were analyzed with the qPCR assays described in Table 2 to track fungal species. The rationale behind this preliminary step was to individuate samples naturally contaminated and samples free of fungal contamination and therefore suitable for the preparation of artificially contaminated ones. Three classes were found: i) samples in which no fungal species has been detected; ii) samples contaminated with one Fusarium species; iii) samples contaminated with two or more Fusarium species.
Starting from the data, the following two subsets of samples were further analyzed with cdPCR.
• Naturally contaminated samples, belonging to classes ii) and iii); • Synthetic samples created by spiking plant DNA (extracted from samples found not contaminated) with fungal DNA at different concentrations.
The fungus quantifications in the two subsets of samples obtained with the qPCR and cdPCR assays are reported in Table 5. The contaminated plant samples belonged to Triticum durum and Avena sativa species; therefore, a further objective of our study was to develop a duplex assay, able to co-amplify both plant and fungal DNA. The rationale behind the duplex assay's development is to evaluate the impact of relevant quantity of plant DNA on the functioning of primers and probes of the fungal PCR assays. Moreover, the ratio between the quantity of Fusarium and plant DNA can be informative about the infection level in a sample. Two genic targets for durum wheat and oat were selected from the literature or newly developed. The Grano CO2 assay, designed on TaHd1 gene sequence and developed by Morcia et al. [18] was used to track Triticum genus. A new assay (Avena dig assay) was designed on actin1 gene sequence to track Avena genus. Grano CO2 and Avena dig assays amplification efficiencies, evaluated in qPCR, have values of 99.6 and 111%, respectively.
The compatibility of the tests to trace the fungal and plant species in cdPCR was evaluated comparing the precision values of the simplex vs. duplex assays. Correlation values ranged from a minimum of 0.97 to 0.99; therefore, the assays are fully compatible and can be organized in duplex mode. The contaminated plant samples belonged to Triticum durum and Avena sativa species; therefore, a further objective of our study was to develop a duplex assay, able to co-amplify both plant and fungal DNA. The rationale behind the duplex assay's development is to evaluate the impact of relevant quantity of plant DNA on the functioning of primers and probes of the fungal PCR assays. Moreover, the ratio between the quantity of Fusarium and plant DNA can be informative about the infection level in a sample. Two genic targets for durum wheat and oat were selected from the literature or newly developed. The Grano CO2 assay, designed on TaHd1 gene sequence and developed by Morcia et al. [18] was used to track Triticum genus. A new assay (Avena dig assay) was designed on actin1 gene sequence to track Avena genus. Grano CO2 and Avena dig assays amplification efficiencies, evaluated in qPCR, have values of 99.6 and 111%, respectively.
The compatibility of the tests to trace the fungal and plant species in cdPCR was evaluated comparing the precision values of the simplex vs. duplex assays. Correlation values ranged from a minimum of 0.97 to 0.99; therefore, the assays are fully compatible and can be organized in duplex mode. Figure 5 reports, as examples, cdPCR plots of oat DNA spiked with F. avenaceum DNA dilutions.

Discussion
In this work we propose four cdPCR assays for detection and quantification of mycotoxigenic Fusarium, etiological agents of Fusarium Head Blight in small-grain cereals. The assays were organized as duplex assay to simultaneously quantify the fungus and the plant species. The logic behind the development of molecular tools for Fusarium diagnosis rely on the possibility to increase fungal control in plants. The fungal DNA can be tracked in the plant during the initial phase of infection, when visible symptoms are absent. Such early diagnosis can mitigate mycotoxin contamination problems in the harvested grains thanks to appropriate fungicidal treatments applied in the right temporal window as well as segregation of highly infected field sectors. We focused on Fusarium species worldwide spread in cereal cultivation areas: F. graminearum and F.culmorum, which are widely recognized as the most important DON producers in small-grain cereals [19], F. poae which shows a NIV chemotype although not all isolates produced NIV in vivo [20], F. sporotrichioides, a T-2 and HT-2 toxins producer which is frequently isolated in some temperate regions of Europe [21] and F. avenaceum, an enniatin and beauverin producer [22].
As already stated, several molecular assays have been recently developed for Fusarium diagnosis but, to the best of our knowledge, none based on a digital PCR has been proposed until now.
Our assays fill this gap, giving the chance to identify and quantify the presence of mycotoxigenic Fusarium in small-grain cereal samples with digital PCR technology. Such new assays can be now practically used in Fusarium diagnosis.
As reported by other authors [23,24], dPCR has several advantages in a comparison with qPCR.
 As the most advantageous feature, dPCR relies on absolute quantification of the target operated by dPCR; on the contrary, "results generated from qPCR were relative to calibration curve and were not the actual number of copies in a sample itself" [23].  Secondarily, the high sample partitioning ensures accurate results even at very low target copy numbers as well as detection of rare targets even in a high background of non-target DNA [24],

Discussion
In this work we propose four cdPCR assays for detection and quantification of mycotoxigenic Fusarium, etiological agents of Fusarium Head Blight in small-grain cereals. The assays were organized as duplex assay to simultaneously quantify the fungus and the plant species. The logic behind the development of molecular tools for Fusarium diagnosis rely on the possibility to increase fungal control in plants. The fungal DNA can be tracked in the plant during the initial phase of infection, when visible symptoms are absent. Such early diagnosis can mitigate mycotoxin contamination problems in the harvested grains thanks to appropriate fungicidal treatments applied in the right temporal window as well as segregation of highly infected field sectors. We focused on Fusarium species worldwide spread in cereal cultivation areas: F. graminearum and F.culmorum, which are widely recognized as the most important DON producers in small-grain cereals [19], F. poae which shows a NIV chemotype although not all isolates produced NIV in vivo [20], F. sporotrichioides, a T-2 and HT-2 toxins producer which is frequently isolated in some temperate regions of Europe [21] and F. avenaceum, an enniatin and beauverin producer [22].
As already stated, several molecular assays have been recently developed for Fusarium diagnosis but, to the best of our knowledge, none based on a digital PCR has been proposed until now.
Our assays fill this gap, giving the chance to identify and quantify the presence of mycotoxigenic Fusarium in small-grain cereal samples with digital PCR technology. Such new assays can be now practically used in Fusarium diagnosis.
As reported by other authors [23,24], dPCR has several advantages in a comparison with qPCR.
As the most advantageous feature, dPCR relies on absolute quantification of the target operated by dPCR; on the contrary, "results generated from qPCR were relative to calibration curve and were not the actual number of copies in a sample itself " [23]. Secondarily, the high sample partitioning ensures accurate results even at very low target copy numbers as well as detection of rare targets even in a high background of non-target DNA [24], Lastly, dPCR is less sensitive to contaminants eventually present in the samples; complex biomolecules such as humic acid can, in fact, significantly inhibit qPCR reactions, but dPCR can overcome this lack thanks to its endpoint quantification [23].
Our cdPCR assays have a LOD ranging from 2 to 13 copies/µL; this level of sensitivity is suitable to Fusarium diagnosis purposes in field, for FHB control, for fungicide treatments optimization and breeding purposes. The main disadvantages we encountered, compared with qPCR, are related to the expenses of the analysis, amount of sample analyzed in a certain time. Controversial are the opinions on the commercial cost of dPCR vs qPCR assays. The processivity can be improved by multiplexing, as suggested by Demeke and Dobnik [24]; it is also related to the different instruments available on the market.
Our specific experience highlighted the necessity of specific laboratory skills for both qPCR and cdPCR as well as similar supporting instruments in the laboratory.
In conclusion, our position on the topic is that dPCR has the potential to replace qPCR in some diagnostic fields, e.g., for Genetically Modified Organisms detection [24]. With regard to microbiological routine diagnostics, the two technologies can be considered complementary, and therefore advantageously used in combination. qPCR technology, as previously stated, requires a reference standard curve for quantification, although standardized reference materials of plant pathogens are generally unavailable. Digital PCR, on the contrary, gives an absolute quantification of the molecular target as output and can, therefore, be proposed for characterization of the calibrators needed for standard curves in qPCR analyses. As suggested by other authors [25], dPCR can hypothetically be exploited for the production of calibrators.