Simultaneous Determination of Multiple Mycotoxins in Swine, Poultry and Dairy Feeds Using Ultra High Performance Liquid Chromatography-Tandem Mass Spectrometry.

A reliable, sensitive and accurate multiple mycotoxin method was developed for the simultaneous determination of 17 mycotoxins in swine, poultry and dairy feeds using stable isotope dilution (13C-ISTD) and (ultra)-high performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS). A simple QuEChERS-based method (quick, easy, cheap, effective, rugged and safe) was developed consisting of soaking with a solution of 1% formic acid followed by extraction with acetonitrile, clean-up with C18 sorbent and finally adding 13C-ISTD before the UHPLC-MS/MS analysis. The chromatographic condition was optimized for separation and detection of the 17 mycotoxins using gradient elution. The method's performance complied with the SANTE/11813/2017 standard and had mean recovery accuracies in the range 70%-120% and precision testing of % relative standard deviation (RSD) £ 20%. The limit of detection and limit of quantification values ranged from 0.25 to 40.0 ng/g and 0.5 to 100.0 ng/g, respectively. Finally, the method was applied to analyze feed samples, with the results showing that fumonisins, zearalenone, aflatoxin B1 and deoxynivalenol were the most prevalent mycotoxins contaminating the feed samples.


Introduction
Mycotoxins are toxic secondary metabolites produced by filamentous fungi, especially by the Aspergillus, Fusarium and Penicillium genera that contaminate agricultural commodities and animal feeds [1]. Nowadays, more than 300 mycotoxins with various types of toxicity (including immunosuppressive, hepatoxic, mutagenic, carcinogenic and estrogenic effects) in mammals have been identified in agricultural products and result in substantial adverse economic impacts [2,3]. The Food and Agriculture Organization of the United Nations estimated that approximately 25% of

LC-MS/MS Analysis
Amounts of 5 mM ammonium formate (AmF) and 0.2% formic acid (FA) in both Milli-Q water and methanol were used in ESI (+), whereas 5 mM AmAc in Milli-Q water and pure MeOH were used for ESI (−). In the first stage of our study, the two above-mentioned mobile phases were tested for analytes detection in ESI (+) and ESI (−).
For ESI (+), the 12 analytes including the isotopical mycotoxins were detected in ESI (+) especially for AFs that were at levels even more than 10 times that using an AmF solution with FA compared to the use of AmAc with 1% AA. The chromatogram for the FBs had good peak shapes and the highest sensitivity using AmF with 0.1% FA.
For ESI (−), the type B trichothecenes (except 15-AcDON and ZEA) were detected and had the highest sensitivity using the acetate buffer solution with acetic acid for the mobile phase [18,31]. Amounts of 5 mM AmF and 0.1% FA in both Milli-Q water and methanol solutions were used in ESI (+), whereas 5 mM AmF and 0.1% AA in both Milli-Q water and methanol solutions were used for ESI (−). Keeping the same ionic strength over the gradient was essential to improve the peak shape and retention time reproducibility. The developed method showed the better separation for acetylated deoxynivalenol (3-AcDON and 15AcDON) than those in different ionization mode [16]. The extract ion chromatograms (EIC) of the 17 mycotoxins including 15 isotopically internal standards in ESI (+) and ESI (−) are illustrated in Figures 1 and 2

LC-MS/MS Analysis
Amounts of 5 mM ammonium formate (AmF) and 0.2% formic acid (FA) in both Milli-Q water and methanol were used in ESI (+), whereas 5 mM AmAc in Milli-Q water and pure MeOH were used for ESI (−). In the first stage of our study, the two above-mentioned mobile phases were tested for analytes detection in ESI (+) and ESI (−).
For ESI (+), the 12 analytes including the isotopical mycotoxins were detected in ESI (+) especially for AFs that were at levels even more than 10 times that using an AmF solution with FA compared to the use of AmAc with 1% AA. The chromatogram for the FBs had good peak shapes and the highest sensitivity using AmF with 0.1% FA.
For ESI (−), the type B trichothecenes (except 15-AcDON and ZEA) were detected and had the highest sensitivity using the acetate buffer solution with acetic acid for the mobile phase [18,31]. Amounts of 5 mM AmF and 0.1% FA in both Milli-Q water and methanol solutions were used in ESI (+), whereas 5 mM AmF and 0.1% AA in both Milli-Q water and methanol solutions were used for ESI (−). Keeping the same ionic strength over the gradient was essential to improve the peak shape and retention time reproducibility. The developed method showed the better separation for acetylated deoxynivalenol (3-AcDON and 15AcDON) than those in different ionization mode [16]. The extract ion chromatograms (EIC) of the 17 mycotoxins including 15 isotopically internal standards in ESI (+) and ESI (−) are illustrated in Figures 1 and 2, respectively.

QuEChERS-Based Procedure
The optimized QuEChERS procedure was performed with slight modifications from Dzuman et al. (2014) [8]. Three different percentages of FA, consisting of 0.5%, 1.0% and 2.0% at soaking step were evaluated for use in the sample extraction. In addition, 1.0% FA showed the best extraction efficiency and provided satisfied recovery values that were better than those for 0.5% and 2.0% FA for all analytes ( Figure 3). In short, the test sample of 1.0 g and soaking solvent of 1% FA were added with stable isotope dilution (internal standard) after the sample had been cleaned up using dSPE. The performance of the optimized QuEChERS method was in line with previous publications [8,22]. The results for spiked swine, poultry and dairy feeds are shown in Figures

QuEChERS-Based Procedure
The optimized QuEChERS procedure was performed with slight modifications from Dzuman et al. (2014) [8]. Three different percentages of FA, consisting of 0.5%, 1.0% and 2.0% at soaking step were evaluated for use in the sample extraction. In addition, 1.0% FA showed the best extraction efficiency and provided satisfied recovery values that were better than those for 0.5% and 2.0% FA for all analytes ( Figure 3). In short, the test sample of 1.0 g and soaking solvent of 1% FA were added with stable isotope dilution (internal standard) after the sample had been cleaned up using dSPE. The performance of the optimized QuEChERS method was in line with previous publications [8,22]. The results for spiked swine, poultry and dairy feeds are shown in Figures 4

Method Validation
The results of linearity and sensitivity are reported in Table S1. The method produced good linearity over the relevant working range, with the r 2 value being greater than 0.995. The limit of detection (LOD) values in the matrices ranged from 0. 25 Figure 5. Extract Ion Chromatogram (EIC) of spiked swine feed samples in ESI (+); 10 ng/g for AFB1, AFB2, AFG1 and AFG2, with 80 ng/g for T-2, HT-2, DAS, NEO, and OTA, 400 ng/g for 15-AcDON, 300 ng/g for FB1 and 90 ng/g for FB2.

Method Validation
The results of linearity and sensitivity are reported in Table S1. The method produced good linearity over the relevant working range, with the r 2 value being greater than 0.995. The limit of detection (LOD) values in the matrices ranged from 0.25 ng/g for the aflatoxins in poultry feed matrices to 40.0 ng/g for DON in dairy feed matrices, respectively. The limit of quantification (LOQ) ranged from 0.5 ng/g for aflatoxins in poultry feed matrices to 100.0 ng/g for DON in dairy and poultry feed matrices (Table S1). The LOQ parameters showed the lower amount for some of trichothecenes mycotoxin such as 15-AcDON (50 ng/g), NIV (40 ng/g), FusX (40 ng/g), HT-2 [8 ng/g] and NEO (8 ng/g) in dairy feed matrix with the previous report [8] and for aflatoxin group; AFB1 (80 ng/g) AFB2, AFG1 and AFG2 (4 ng/g) [9]. The recovery and precision values were with the acceptable criteria, in the range 70%-125% and the %RSD values were less than 20% [32] for all 17 mycotoxins, as summarized in Tables S2-S4 for the swine, poultry and dairy feeds, respectively. The identification requirement of the relative ion ratio from sample extracts was lower than 30% for all 17 mycotoxins [33].

Matrix Effect Study
The study used %SSE to evaluate the matrix effects in the three types of feed matrices. If the suppression or enhancement was marginal, the %SSE would be very close to 100%; if there was strong suppression/enhancement, the %SSE would deviate from 100%. In the swine feed samples, the %SSE was in the range 82.5%-119.3%, except for DON which exhibited strong signal suppression with the %SSE less than 50% (47.4%). In the poultry feed samples, the %SSE was in the range 76%-115.4%, except for DON which exhibited strong signal suppression (%SSE 7.49%), with strong signal enhancement for NEO and DAS (%SSE 137% and 142%, respectively). In the dairy feed samples, the %SSE was in the range 89.3%-113.7%, except for DON and ZEA which produced the same results as for the poultry feed samples, namely strong signal suppression with %SSE 2.5% and 3.5%, respectively. Regarding signal enhancement, the %SSE was greater than 120% (123.8%) for DAS. The %SSE values of the three types of feed matrices are summarized in Figure 6. All results of the matrix effect, the quantification of mycotoxin using matrix matched-calibration as isotopically labeled as the internal standard are necessary.
Toxins 2020, 12, x FOR PEER REVIEW 10 of 18 matrices to 40.0 ng/g for DON in dairy feed matrices, respectively. The limit of quantification (LOQ) ranged from 0.5 ng/g for aflatoxins in poultry feed matrices to 100.0 ng/g for DON in dairy and poultry feed matrices (Table S1). The LOQ parameters showed the lower amount for some of trichothecenes mycotoxin such as 15-AcDON (50 ng/g), NIV (40 ng/g), FusX (40 ng/g), HT-2 [8 ng/g] and NEO (8 ng/g) in dairy feed matrix with the previous report [8] and for aflatoxin group; AFB1 (80 ng/g) AFB2, AFG1 and AFG2 (4 ng/g) [9]. The recovery and precision values were with the acceptable criteria, in the range 70%-125% and the %RSD values were less than 20% [32] for all 17 mycotoxins, as summarized in Table S2, S3 and S4 for the swine, poultry and dairy feeds, respectively. The identification requirement of the relative ion ratio from sample extracts was lower than 30% for all 17 mycotoxins [33].

Matrix Effect Study
The study used %SSE to evaluate the matrix effects in the three types of feed matrices. If the suppression or enhancement was marginal, the %SSE would be very close to 100%; if there was strong suppression/enhancement, the %SSE would deviate from 100%. In the swine feed samples, the %SSE was in the range 82.5%-119.3%, except for DON which exhibited strong signal suppression with the %SSE less than 50% (47.4%). In the poultry feed samples, the %SSE was in the range 76%-115.4%, except for DON which exhibited strong signal suppression (%SSE 7.49%), with strong signal enhancement for NEO and DAS (%SSE 137% and 142%, respectively). In the dairy feed samples, the %SSE was in the range 89.3%-113.7%, except for DON and ZEA which produced the same results as for the poultry feed samples, namely strong signal suppression with %SSE 2.5% and 3.5%, respectively. Regarding signal enhancement, the %SSE was greater than 120% (123.8%) for DAS. The %SSE values of the three types of feed matrices are summarized in Figure 6. All results of the matrix effect, the quantification of mycotoxin using matrix matched-calibration as isotopically labeled as the internal standard are necessary.

Occurrence of the Mycotoxins in Animal Feed
The developed method was applied to investigate the occurrence of the 17 toxins in 300 feed samples consisting of swine (n = 100), poultry (n = 100) and dairy feeds (n = 100). In the swine feed samples, more than 75% were contaminated with Fusarium mycotoxin, especially FB2 (77%), (FB1 85%) and ZEA (91%). The other mycotoxins contaminating the feed samples were: DON (43%), AFB1 (34%), NIV (18%), 15-AcDON (16%), AFB2 (13%), HT-2 (7%), AFG1 (4%), DAS (2%) and AFG2 (1%). However, T-2, NEO, FusX and OTA were not detectable in the swine feed samples. The results were consistent with the contaminants prevalent for Fusarium mycotoxin in a previous report [26,34]. In the poultry feed samples, there was a somewhat similar situation for AFB1 (77%) and ZEA (72%) in line with a previous publication [1] with a range of 0.27%-326.4 µg/kg for AFB1 and 1.3-235.8 µg/kg for ZEA, respectively. The FBs were the most contaminants in the poultry samples (90%) ranges of 16.0-2645.5 µg/kg for FB1 and 6.4-573.3 µg/kg for FB2. The trichothecene mycotoxins, especially for type B, consisting of DON, NIV and 15-AcDON were found in more than 25% of samples, but type A trichothecene, consisting of T-2 HT-2 and DAS were found at a lower level (7%) for positive samples. The dairy feed samples had the same prevalent contaminants as the swine feed samples but with lower concentrations of the Fusarium mycotoxins, especially for FB2 (45%), (FB1 62%) and ZEA (46%). However, AFG2, NEO, FusX and OTA were not detectable in the dairy feed samples. (Table 1 and Figure 7). The mycotoxin levels in the feed samples almost complied with the EU regulation with exceptions in the poultry and dairy feed samples. Specifically, in the poultry feed samples, the contamination of AFB1 was higher than the regulatory limit (20 ng/g) in three samples (range 24.4-32.64 ng/g), while contamination with AFB1 was higher than 5 ng/g for four samples (range 6.58-14.88 ng/g). For the extracted ion chromatogram (EIC) of some naturally contaminated samples were shown in Figures S1 and S2, respectively.

Conclusions
Our results demonstrated the successful development of a stable isotope dilution based on a QuEChERS sample preparation protocol and the LC-ESI-MS/MS method for the simultaneous determination of 17 mycotoxins in animal feed samples. This method was an excellent tool for the unambiguous identification of the 17 selected mycotoxins in the sampled swine, poultry and dairy feeds. The developed method was successfully validated according to the SANTE/11813/2017 standard and was applied to real feed samples. The results showed contamination by multiple mycotoxins with Fusarium mycotoxins such as FBs, ZEA, type B trichothecene (especially DON) and Aspergillus mycotoxins (especially AFB1) co-occurred most commonly in the animal feeds. The mycotoxin levels in feed samples almost complied with the EU regulations. However, further studies with a larger sample size are needed to confirm these data.

Preparation of Standards Solution
Combinations of the unlabeled standard mycotoxin stock solutions were prepared in methanol to provide a working standard solution at different concentrations: 2000 ng/mL for DON, 15-AcDON, 3-AcDON, NIV and FusX; 400 ng/mL for T-2, HT-2, DAS, NEO and OTA; 50 ng/mL for AFB1, AFB2, AFG1, AFG2 and ZEA; 1500 ng/mL for FB1; and 450 ng/mL for FB2. For the purpose of method validation for spiking experiments, working standard solutions were immediately prepared and stored in amber vials at −20 • C for one week.

Preparation of Isotopically Internal Standards
The concentration of isotopically internal standard (ISTD) working solutions were: 10.0 ng/mL for [ 13

Feed Samples
The 300 feed samples consisting of swine feed (n = 100), poultry feed (n = 100) and dairy feed (n = 100) were randomly collected from animal farms in different regions of Thailand. All samples were ground with rotor milling ZM200 (Retsh GmbH, Hann, Germany) into fine powder (0.50 mm) and stored at −20 • C before analysis.

QuEChERS-Based Procedure
The sample preparation protocol applying the QuEChERS-base procedure was developed based on Dzuman et al. [8] Briefly, 1 g of homogenized feed sample was weighed into a 50-mL polypropylene (PP) centrifugation tube, followed by the addition of 10 mL of 1% aqueous formic acid solution. The tube was closed and the sample was allowed to soak for 30 min. Then, 10 mL of acetonitrile was added into the soaked sample and shaken using a laboratory shaker (IKA Labortechnik; Staufen, Germany) for 30 min at 240 RPM. The phase partition was induced by the addition of 1 g NaCl and 4 g of MgSO 4 . The tube was immediately shaken for 30 s to prevent coagulation of the MgSO 4 and then centrifuged (Kubota; Tokyo, Japan) for 5 min at 10,000 RPM. A sample of 2 mL of the acetronitrile phase was placed into a 15 mL PP tube containing 0.1 g of C 18 silica sorbent and 0.3 g of MgSO 4 which were mixed and then centrifuged for 1 min. The purified extract was evaporated to dryness at 40 • C. The residue was reconstituted in 960 µL 20% MeOH and then 40 µL of [ 13 C]-ISTD working solution were added. The mixture was passed through a 0.22 µm nylon filter before being used in the LC-MS/MS analysis.

LC-MS/MS Analysis
The 17 target mycotoxins were analyzed using the UHPLC-MS/MS method. Chromatographic separation was developed according to [22]. The analysis used an ExionLC™ AD system (AB SCIEX; Toronto, ON, Canada) was equipped with an Accucore analytical column (100 × 2.1 mm i.d., 2.6 µm particle size; Thermo Scientific; San Jose, CA, USA) maintained at 25 • C. The mobile phase differed for the ESI (+) and ESI (−) analyses, with 5 mM ammonium formate and 0.1% formic acid (v/v) both in deionized water (A) and MeOH (B) being used in ESI (+), whereas 5 mM ammonium acetate and 0.1% acetic acid (v/v) both in deionized water (C) and MeOH (D) were used for ESI (−). The gradient elution was identical in both ESI (+) and (−), starting at 0% B/D followed by a linear change to 20% B/D in 4 min and subsequently using linear changes to 40% B/D in 5.5 min and 100% B/D in 10.5 min. Then, the column was washed for 2.5 min with 100% B/D followed by a reconditioning for 3 min using the initial composition of mobile phases. The flow rate was stable at 0.4 mL/min throughout the run and 3 µL of sample extract was injected into the LC-MS/MS system. The ExionLC™ AD system was coupled to a QTRAP 5500 tandem mass spectrometer (ABSCIEX; Toronto, ON, Canada), equipped with an electrospray (ESI) ion source operated in both positive and negative mode. The ESI (+) ion source parameters were: needle voltage 4500 V; curtain gas 30 psi; nebulizer (Gas1) and turbo gas (Gas2) 55 psi; and turbo gas temperature 500 • C. The ESI (−) source parameters were: needle voltage −4500 V; curtain gas 30 psi; nebulizer (Gas 1) and turbo gas (Gas2) 55 psi; and turbo gas temperature 500 • C. The declustering potential (DP), collision energy (CE) and collision cell exit potential (CXP) were optimized during infusion of individual analytes (10-200 ng/mL) using manual infusion. The MRM transitions of unlabeled mycotoxins and isotopically internal-standard-dependent parameters are summarized in Table 2. Table 2.
MS/MS parameters for determination of 17 mycotoxins and 15 isotope labeled internal standards.

Method Validation
The method performance characteristic parameters (linearity and ranges, accuracy, precision, LOD and LOQ) were determined for the samples of swine, poultry and dairy feeds. The analytes were quantified using an internal matrix-matched calibration standard with post spiking calibration curve for 15 mycotoxins excluding FusX and NEO (using external matrix-matched calibration standard) at the following levels: 1.0-40.0 ng/g (corresponding to 0.1-4.0 ng/mL) for AFB1, AFB2, AFG1, AFG2 and ZEA; 40-1600 ng/g (corresponding to 4-160 ng/mL) for DON, 15-AcDON, 3-AcDON, NIV and FusX; 8-320 ng/g (corresponding to 0.8-32 ng/mL) for T-2, HT-2, DAS, NEO and OTA; 30-1200 ng/g (corresponding to 3-120 ng/mL) for FB1; and 9-360 ng/g (corresponding to 0.9-36 ng/mL) for FB2. The accuracy and precision (repeatability, expressed as relative standard deviation (RSD) in %) were determined within-day by analyzing five replicates at three levels (low, intermediate, high). The inter-day precision was determined at the same level as the within-day precision on three different days. The LOQ values were estimated using the concentration of analytes which provided a signal-to-noise ratio (S/N) greater than 10 and the LOD values were defined as the minimum concentration of analytes which provided S/N values greater than 3.
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6651/12/4/253/s1, Figure S1: Extract Ion Chromatogram (EIC) of naturally contaminated of AFB1 at 326.4 ng/g in poultry feed sample; A: Quantifier, B: Qualifier and C: IS-AFB1. Figure S2. EIC of naturally contaminated of AFB1 at 14.88 ng/g in dairy feed sample, Table S1. Linearity ranges, limit of detection (LOD) and limit of quantification (LOQ) of the optimized UHPLC-MS/MS method for simultaneous determination of mycotoxins, Table S2. Accuracy and precision for mycotoxin determination in optimal LC-MS/MS conditions for swine feed samples, Table S3. Accuracy and precision for mycotoxin determination in optimal LC-MS/MS conditions for poultry feed samples, Table S4. Accuracy and precision for mycotoxin determination in optimal LC-MS/MS conditions for dairy feed samples.
Author Contributions: K.N.; methodology, formal analysis, validation, and writing-original draft preparation, S.P.; methodology and validation, M.G., Q.Z. and Z.Z.; study design and visualization, P.L. and A.F.L.; conceptualization, suggestion and supervision A.P.; conceptualization, funding acquisition, methodology, validation, supervision, project administration and writing-review and editing. All authors have read and agreed to the published version of the manuscript.
Funding: This work was funded by the Faculty of Veterinary Medicine, Kasetsart University, Bangkok, Thailand.

Conflicts of Interest:
The authors declare that there are no conflicts of interest.