How different microfilters affect the recovery of eleven EU-regulated mycotoxins

Certified analytical standards of the following mycotoxins were obtained from Romer Labs Biopure (Romer Labs, Tulln, Austria): AFB1, AFB2, AFG1, and AFG2 (2.0 g/mL for AFB1/AFG1, 0.5 g/mL for AFB2/AFG2), DON (100 g/mL), FB1 and FB2 (50 g/mL), ZEA (100 g/mL), T-2 (100 g/mL), HT-2 (100 g/mL), and OTA (10 g/mL). LC-MS-grade acetonitrile (ACN) and methanol (MeOH) were obtained from J. T. Baker (J. T. Baker, Deventer, The Netherlands). LC-MS-grade formic acid (FA) and LC-MS ammonium formate (AFNH₄) were supplied by Sigma-Aldrich (Sigma-Aldrich, Louis, MO, USA). Water (H₂O) was purified to the ultrapure grade with a Niro VV (Nirosta d.o.o., Osijek, Croatia) or a Purelab flex system (ELGA LabWater, Woodridge, IL, USA). Syringe filters (13 mm in diameter with a 0.22 µm membrane pore size) were obtained from the following suppliers: nylon (NY) from Kemolab d.o.o. (Veliko Polje, Croatia), polytetrafluoroethylene (PTFE) from Kefo d.o.o. (Sisak, Croatia), while polyethersulphone (PES), mixed cellulose ester (MCE), and cellulose acetate (CA) were obtained from Obrnuta faza d.o.o. (Pazin, Croatia).

The multi-mycotoxin standard solution was prepared by mixing the eleven mycotoxin standards with ACN, and then by diluting this mixture one hundred times with a solvent mixture of ACN/H₂O/FA (49.5/49.5/1, v/v/v) as shown in Table 1 to obtain analyte concentrations corresponding to the middle calibration point of the previously developed and validated multi-mycotoxin method by Kovač et al. (1). This solvent mixture composition for standard dilution matches the ratio of solvent components in real dilute-and-shoot without the matrix.

The multi-mycotoxin standard solution was microfiltered in such a way that the first three filtrate drops (filtered solution 1) and the following filtrate (filtered solution 2) were collected separately into two different vials and injected (in triplicate) into a UHPLC-MS/MS system. The unfiltered multi-mycotoxin solution was injected into the system as is (Figure 2). The whole procedure was performed in duplicate. To process the obtained data on peak areas for each compound, we took averages of repeated measurements and calculated the relative recovery (expressed with a percentage) for each mycotoxin and filter membrane material by comparing the average area of the filtered solution (1 or 2) with unfiltered standard solution according to Equation 1, as follows:

Figure 2

For instrumental analysis we used a UHPLC (Acquity H-Class, Waters, Milford, MA, USA) equipped with a quaternary pump system, coupled with a triple quadruple mass spectrometer (XEVO TQ-S micro, Milford, MA, USA) using a ZSpray electrospray interface (ESI). For chromatographic separation of the analytes we used a reverse phase column XBridge Peptide BEH C18 (150×3.0, 2.5 µm) (Waters, Milford, MA, USA) maintained at 50 °C, which we achieved with gradient elution of eluent A (aqueous solution of 5 mmol/L AFNH₄+0.02 % FA) and eluent B (MeOH) at a constant flow rate of 0.6 mL/min. The elution started with 75 % of eluent A in gradient, with a hold time of 0.5 min. Then we switched to 10 % of eluent A for 6 min, followed by equilibration to initial elution conditions after 2 min, giving a total run time of 13 min. The sample injection volume was set to 10 µL. Analytes were identified with the ESI-MS/MS working in the multiple-reaction monitoring (MRM) mode in positive and negative polarity, with two MS/MS transitions acquired per analyte (Table 2). The MS and MS/MS parameters used for analysis in both positive and negative ESI were as follows: capillary voltage 3.5 kV (+) and 2.5 kV (–), source temperature 150 °C, desolvation temperature 550 °C, desolvation gas flow 1000 L/h, and cone gas flow 50 L/h (both nitrogen). Argon gas was used for collision at a pressure of about 0.4 Pa in the collision cell. The obtained UHPLC-MS/MS chromatogram of the eleven mycotoxins in the unfiltered solution is shown in Figure 3.

Figure 3

Statistical data analysis was performed using Statistica 13.3. software (TIBCO Software Inc., Palo Alto, CA, USA). Significant difference was set to p<0.05. Before we ran statistical tests, we tested the data for normality of distribution using the Shapiro-Wilk test and for the homogeneity of variance using Levene’s test. All variables that had normal distribution and homogenous variance (AFB1, AFG2, ZEA) were then tested with the t-test, and analysis of variance (ANOVA), while other variables were tested with the Mann-Whitney U test and Kruskal-Wallis H test. In the box and whisker graphs medians are shown as central points, interquartile ranges (IQR) as boxes, and ranges as whiskers, since most data did not have normal distribution.