Plasma sample preparation

Blank plasma was obtained by the collection of heparinised blood from broiler chickens (Ross 308) and pigs (Landrace). The animals were fasted twelve hours prior to blood collection. Plasma was obtained by centrifugation (2851 x g, 10 min, 4 °C) of the blood. The blank plasma was pooled, homogenised and stored at ≤ – 15 °C until the moment of use for the preparation of matrix-matched calibrators and quality control samples.

Calibrator and quality control samples. To 250 µL of blank plasma, 5 µL of a 1 µg/mL internal standard working solution (13C15-DON) and appropriate volumes of the standard mixture working solutions (1 µg/mL and 100 ng/mL) were added to obtain calibrator samples with mycotoxin concentrations of 1, 2, 5, 10, 20, 50, 100 and 200 ng/mL. After vortex mixing, acetonitrile was added up to a volume of 1 mL to precipitate plasma proteins. The samples were vortex mixed again, followed by a centrifugation step (10 min at 8517 x g, 4 °C). The

86 supernatant was transferred to a new tube and evaporated to dryness under nitrogen at 80 °C. The sample was then redissolved in 200 µL of UPLC-grade water, micro-filtrated and analysed by means of LC-MS/MS.

Incurred samples. To 250 µL of plasma, 5 µL of a 1 µg/mL internal standard working solution were added. After vortex mixing, the samples were subjected to the same sample preparation procedure as the calibrator samples.

Blank samples. After the addition of 750 µL of acetonitrile to 250 µL of blank plasma, the samples were extracted in the same way as the calibrator samples.

2.4 LC-MS/MS analysis

For chicken plasma all compounds were eluted with a gradient of UPLC-grade water + 0.1% glacial acetic acid (mobile phase A) and UPLC-grade methanol + 0.1% glacial acetic acid (mobile phase B) at a flow rate of 300 µL/min (Figure 2 C). The gradient started at 5% B for one minute, in six seconds the gradient increased to 20% B and this was maintained up to 5 minutes. From 5 minutes to 5.1 minutes the % B was augmented to 50 and was held up to 8 minutes from where it increased to 95% in six seconds which was maintained for one minute. Afterwards, the gradient was restored to its initial conditions. Column and autosampler temperatures were set to respectively 60 and 5 °C, injection volume (partial loop) was fixed at 10 µL. For pig plasma identical parameters were used, the sole difference was the use of 0.3% glacial acetic acid instead of 0.1%.

The mass spectrometers were operated in the multiple reaction monitoring (MRM) mode with two ion transitions for each target analyte. Instrumental and compound specific parameters were optimised by the direct infusion of either 1 µg/mL (TSQ Quantum Ultra) or 10 ng/mL (Xevo TQ-S) standard solutionsin methanol/ultra-pure water (50/50; v/v) + 0.1% acetic acid at a flow rate of 10 µL/min. For the TSQ, the instrumental mass spectrometry parameters were set as follows: vaporizer temperature 50 °C, capillary temperature 350 °C and ion sweep gas pressure 2.0. The other parameters varied for ESI- and ESI+ respectively, spray voltage 3500 V (-) and 5000 V (+), sheath gas pressure 40 arbitrary units (au) (-) and 49 au (+), auxiliary gas pressure 10 au (-) and 25 au (+), source CID 5 (-) and -5 (+). Compound specific MS parameters, together with precursor and product ions used for quantification

87 and qualification, are given in Table 1. All compounds were detected in negative electrospray ionisation mode (ESI-) as [M+Hac-H]- adducts, except for 15ADON which exhibited better sensitivity when measured in ESI+, measured as the protonated precursor ion ([M+H]+). For the Xevo TQ-S, the desolvation gas flow rate was fixed to 1000 L/h with a temperature of 600 °C, the cone gas flow rate was set at 150 L/h, the source temperature was set at 150 °C and the capillary voltage was optimised at 3.5 kV. Dwell times of 44 - 52 ms/transition were chosen. In Table 1, compound specific MS parameters such as cone voltage and collision energy are mentioned.

Table 1. Compound specific MRM ion transitions and MS-parameters; Rt= retention time; IS= internal standard;

a _{quantifier ion; for chicken plasma tube lens offset is mentioned, for pig plasma cone voltage.} Measured form/adduct Precursor ion (m/z) Product ion (m/z)

Rt (min) Tube lens offset /cone voltage Collision Energy ESI modus Chicken DON [M+Hac-H]- _{355.2 265.1}a _{3.50 75 15 -} 295.1 3.50 75 10 - DOM-1 [M+Hac-H]- 339.1 249.1a 4.75 85 15 - 59.10 4.75 85 35 - 3ADON [M+Hac-H]- 397.1 306.8a 7.05 95 15 - 337.1 7.05 85 10 - 15ADON [M+H]+ _{339.2 321.2}a _{7.00 80 15 +} 136.9 7.00 80 20 + 13

C15-DON (IS) [M+Hac-H]- 370.2 279.1a 3.50 75 15 -

310.1 3.50 75 10 - Pig DON [M+H]+ _{297.1 249.1}a _{3.50 20 9 +} 203.4 3.50 20 14 + DOM-1 [M+H]+ _{281.1 215.1}a _{4.75 20 12 +} 137.0 4.75 20 16 + 3ADON [M+H]+ 339.0 231.1a 7.05 30 10 + 213.2 7.05 30 14 + 15ADON [M+H]+ _{339.0 231.1}a _{7.00 30 10 +} 213.2 7.00 30 14 + 13_C 15-DON (IS) [M+H]+ 312.0 245.2a 3.50 20 10 + 263.0 3.50 20 10 + 2.5 Validation

Given the unavailability of reference materials, validation was performed on spiked blank plasma samples. Both recommendations as defined by the European Community (Commission Decision 2002; Heitzman 1994) and the Veterinary International Conference on Harmonisation (VICH 2009) served as validation guidelines. The developed method was single laboratory validated.

Linearity of the response of the compounds was assessed by means of three matrix-matched calibration curves consisting of seven calibration points in the range of 1-200 ng/mL. The

88 correlation coefficients (r) and goodness-of-fit coefficients (gof) were determined, limits were set to ≥0.99 and ≤20%, respectively.

Within-day accuracy & precision were determined by analyzing six samples at a low concentration level (LOQ of the compounds) and at a high concentration level (100 ng/mL). Values for the relative standard deviation (RSD) could not exceed 2/3 of the RSDmax, calculated according to the Horwitz equation (given below). The acceptance criteria for accuracy were: −30% to +10% and −20% to +10% for concentrations between 1 and 10 ng/mL, and ≥10 ng/mL, respectively. Between-day accuracy & precision were assessed by analyzing the low and high concentration levels in threefold on three consecutive days (n=3x3). The acceptance criteria for accuracy were identical to the values given above and RSD values could not exceed the RSDmax. The formula to determine RSDmax are given below.

Within-day precision: RSDmax = 2(1−0.5logConc) × 2/3

Between-day precision: RSDmax = 2(1−0.5logConc)

The LOQ was calculated as the lowest concentration for which the method had acceptable results with regards to accuracy and precision. It was determined by spiking six plasma samples at 1 or 2 ng/mL. The LOQ was also established as the lowest point of the calibration curve. The LOD was calculated using the samples spiked at the LOQ level (n=6) corresponding to the lowest concentration that could be determined with a signal-to-noise (S/N) ratio of 3.

Carry-over was evaluated by analysing a mixture of mobile phase A and B (50/50; v/v) directly after the highest calibrator (200 ng/mL).

The specificity, the capability of the method to distinguish signals of the analytes from any other substances or interferences, was determined on six blank plasma samples. For an acceptable specificity the S/N ratio of possible interfering peaks with similar retention times in these samples could not exceed the S/N ratio of the analyte(s)’ LOD.

Recovery and matrix effects. Two types of matrix-matched calibration curves (on pooled blank plasma derived from six different animals) were prepared, one by spiking the blank

89 calibrator samples before and one after extraction. A third calibration curve was prepared in standard solution. All curves consisted of seven calibration points in the range of 1-200 ng/mL. The slopes of these calibration curves (external calibration, without IS) were compared to calculate the apparent recovery (RA = 100 × slope spiked before extraction/slope standard solution), the matrix effect denoted as signal suppression/enhancement (SSE = 100 × slope spiked after extraction/slope standard solution) and the recovery of the extraction step (RE = 100 × slope spiked before extraction/slope spiked after extraction). Regarding SSE, values ≤ 1 indicate ion suppression due to matrix effect, values ≥ 1 are caused by ion enhancement (Matuszewski et al. 2003).

To test applicability of the method on plasma from other animals, the validation as described above was also executed on pig plasma.