A Method for Multiple Mycotoxin Analysis in Wines by Solid Phase Extraction and Multifunctional Cartridge Purification, and Ultra-High-Performance Liquid Chromatography Coupled to Tandem Mass Spectrometry

An analytical method using two solid phase extractions and ultra-high-performance liquid chromatography coupled to tandem mass spectrometry (UHPLC-MS/MS) was developed for the identification and quantification of 14 mycotoxins (patulin, deoxynivalenol, aflatoxins B1, B2, G1, G2, M1, T-2 toxin, HT-2 toxin, zearalenone, fumonisins B1, B2, B3, and ochratoxin A) in domestic and imported wines. Mycotoxins were purified with an Oasis HLB cartridge, followed by a MultiSepTM #229 Ochra. As a result, sufficient removal of the pigments and highly polar matrices from the red wines was achieved. UHPLC conditions were optimized, and 14 mycotoxins were separated in a total of 13 min. Determinations performed using this method produced high correlation coefficients for the 14 mycotoxins (R > 0.990) and recovery rates ranging from 76 to 105% with good repeatability (relative standard deviation RSD < 12%). Twenty-seven samples of domestic and imported wines were analyzed using this method. Although ochratoxin A (OTA) and fumonisins (FMs) were detected in several samples, the FM levels were less than limits of quantification (LOQs) (1 μg/L), and even the largest of the OTA levels was below the EU regulatory level (2 μg/L). These results suggest that the health risk posed to consumers from the wines available in Japan is relatively low.

Especially concerening aflatoxins, AFB1 has only been detected from crops, but in late years, the fungi producing AFs have changed under the influence of global warming, and the detection of AFB2 and G2 have increased. Therefore each country has strengthened regulation of AFs. These facts suggest that wine can be contaminated with mycotoxins for which the levels have not been legally controlled; hence it is essential to establish procedures for simultaneously determining multiple mycotoxins in wine. The MHLW also stipulated that the maximum level of total aflatoxins (AFs) is 10 μg/kg in all foods, and this has been applied in Japan since October 1, 2011 [23]. However, it will be time-consuming to separately determine mycotoxin levels in wine products.
While the mycotoxin regulations are being strengthened worldwide, we developed an accurate and sensitive method for the analysis of multiple mycotoxins in beer-based drinks, and this method shortened the time required for the assay [24]. In order to expand this multiple mycotoxin analysis, we considered to applying this method to foods, food-materials, and other alcoholic drinks. In this study, one of the alcoholic drinks, wine, was selected to use as analytical samples to include many matrices. This study thus aimed to develop a method for the high-throughput analysis of multiple mycotoxins in wine. The method was then applied to quantitatively determine mycotoxins in domestic and imported products. The fourteen mycotoxins that are attracting global attention due to significant toxicities were selected for the target analytes ( Figure 1).

Samples
Fourteen and thirteen samples of red and white wines, respectively, were randomly obtained from local supermarkets in Japan in 2010. All samples were then stored in a refrigerator until analysis.

Sample Preparation
A 5-mL sample of wine and 25 mL of 10 mM ammonium acetate aqueous solution were added into a 50-mL polypropylene centrifuge tube and were then mixed. The mixture was applied to an Oasis HLB cartridge previously conditioned with 5 mL of acetonitrile and 5 mL of 10 mM ammonium acetate aqueous solution. The cartridge was washed with 5 mL of 10 mM ammonium acetate aqueous solution. The mycotoxins retained in the cartridge were eluted with 5 mL of 10 mM ammonium acetate aqueous solution/acetonitrile (1/1 v/v) and then with 5 mL of acetonitrile. The eluate was mixed and evaporated to dryness at 40 °C under a nitrogen stream. The dried sample was dissolved with 1 mL of water. Subsequently 60 μL of formic acid and 5 mL of acetonitrile were added to the sample, and they were mixed together. The mixture was applied to a MultiSep #229 Ochra cartridge. Four mL of the purified eluate was evaporated to dryness at 40 °C under a nitrogen stream and was dissolved with 500 μL of 10 mM ammonium acetate aqueous solution/acetonitrile (85/15 v/v). Each sample was filtrated through a 0.20 μm PTFE filter immediately before UHPLC-MS/MS analysis.

UHPLC-MS/MS Analysis
An UHPLC-MS/MS analysis was performed using an ACQUITY UPLC TM system coupled to a Quattro Premier TM XE tandem quadrupole mass spectrometer (Waters, Manchester, UK). MassLynx 4.1 software equipped with QuanLynx software (Waters) was used to control the instruments and to process the data. An ACQUITY UPLC consisting of a binary pump, an auto-sampler, and a column heater was used. Chromatographic separation was carried out on an ACQUITY UPLC BEH C18 (1.7 μm, 2.1 × 100 mm; Waters). Solvent A was water, and solvent B was 2% acetic acid and 0.1mM ammonium acetate in methanol. In the previous study [24], it was clear that 2 gradient conditions were needed in multiple mycotoxin analysis because the carry-over of FMs and OTA were observed. The two gradient profiles were performed as follows: 0-1 min (5% B); 8 min (80% B), and 8.01-10 min (5% B) for the mycotoxins except for FMB1, FMB2, FMB3, and OTA: 0 min (55% B), 5 min (80% B), and 5.01-7 min (55% B) for FMB1, FMB2, FMB3, and OTA. The flow rate was set at 0.3 mL/min and the column temperature was 40°C. An auto-sampler was used to inject 5 μL.

Performance Evaluation
Because no authorized guideline for the analysis of multiple mycotoxins has been issued, we consulted "about the total aflatoxins analysis" provided by the Japanese Ministry of Health, Labour and Welfare in 2011 [25]. The performance of our developed method was evaluated on wine samples spiked with target mycotoxins. Prior to the spiking, we analyzed the samples and confirmed that no mycotoxins were naturally present. The coefficient of linearity was determined using the samples spiked with each mycotoxin at the following levels: 5, 10, 20, 50, and 100 μg/L for PAT, NIV, DON, and ZON; 0.2, 0.5, 1, 2, and 5 μg/L for AFB1, B2, G1, G2, M1, and OTA; and 1, 2, 4, 10, and 20 μg/L for HT-2, T-2, FMB1, B2, and B3. Recovery and repeatability (relative standard deviation (%RSD)) experiments involved 5 replicate measurements that were carried out on the same day using the samples spiked with each mycotoxin at the following levels: 20 μg/L for PAT, NIV, DON, and ZON; 1 μg/L for AFB1, B2, G1, G2, M1, and OTA; 4 μg/L for HT-2 and T-2; and 5 μg/L for FMB1, B2, and B3.

Optimization of Clean-up and Analytical Conditions
The method that we previously reported for the analysis of beer-based drinks with the QuEChERS extraction [24], was applied first to red wine samples, and spiked tests were performed to optimize the procedure for extraction and clean-up of 15 mycotoxins, including NIV. In the UHPLC conditions, the column length and gradient profile have been changed from those in the previous study because it is assumed that matrix effects are more likely to occur when the mycotoxins are analyzed in wine than in beer-based drinks. The pigments were not sufficiently removed from the prepared samples, and injection of these samples may result in the lower analytical precision and UHPLC-MS/MS system contamination. Therefore, to sufficiently remove the pigments, the MultiSep #229 Ochra multifunctional cartridge was used instead of the C18 cartridge. Multifunctional cartridges have been widely adopted for the analysis of mycotoxins. They feature multiple solid supports for reverse-phase, normal-phase, cation exchange, anion exchange, or graphite carbon black, each of which can be used depending on the characteristic of target analytes; this allows for efficient removal of matrix components. Our previous study shows that several target mycotoxins were adsorbed on a primary-secondary amine and graphite carbon black [24]. A MultiSep #229 Ochra is designed to minimize adsorption of the target mycotoxins; hence, the clean-up procedure was examined using it. Pigments were removed from the red wines with this cartridge. However, on the chromatograms (Figure 2(2)), matrix peaks were observed proximally to PATs, and the PAT peaks were not as sharp as those on the standard chromatograms. No peaks were identified for NIVs. In either case, PAT and NIV peaks were affected by the presence of matrix components. Considering the retention time, it is assumed that these components were highly polar. Although this pretreatment combined extraction by acetonitorile and clean-up using the MultiSep #229 Ochra, it appeared difficult for this method to remove the highly polar matrices. Therefore, to remove such matrices, the sample was extracted and purified using the OasisHLB, rather than the QuEChERS methodology. An OasisHLB is a solid phase extraction cartridge that contains styrene-divinylbenzene co-polymer. It holds low to intermediate-polar substances with six-membered rings and separates highly polar substances. Eventually, nearly all of the pigments were removed from the wine samples that were purified with the multifunctional cartridge after being extracted and purified with the OasisHLB. No peaks for the highly polar matrices were observed, and the peak shapes for PAT and NIV were improved (Figure 2(3)). Using this clean-up method, pigments and highly polar matrices were removed from the wine samples, and good chromatograms were obtained. Figure 3 (1 and 2) shows UHPLC-MS/MS chromatograms of the purified red wine samples spiked with mycotoxins.

Performance Evaluation
Results of the evaluation are summarized in Table 2. Linearity of the calibration curves for the wine samples spiked with each mycotoxin were >0.990. Recovery rates ranged from 76 to 105% with the repeatability (as %RSD) ranging from 3.4 to 11.8% except for NIV. Recovery for NIV was 43%, which affected the quantification performance. It is assumed that the highly polar NIV was hardly held by the OasisHLB and was eluted with the matrices. Limits of quantification (LOQs) for the mycotoxins were the lowest concentration values shown on the calibration curves: 5 μg/L for PAT, DON, and ZON; 0.2 μg/L for AFB1, B2, G1, G2, M1, and OTA; and 1 μg/L for HT-2, T-2, FMB1, B2, and B3.

Analysis of Commercially Available Samples
Twenty-seven domestic and imported wines available in Japan were analyzed using our developed method. Results are summarized in Table 3. In none of the samples tested were any of the target mycotoxins detected, except for FMs and OTA. FMs and/or OTA were detected in 6 of the red wine samples. All of the FMs detected were at values less than LOQ (1 μg/L). The maximum OTA value was 0.20 μg/L but this is less than its maximum level established by the EU (2 μg/L). This result indicates that the health risk posed to consumers from red wine is relatively low. However, it will be necessary to keep monitoring them in conjunction with other mycotoxins in the future because the co-occurrence of different FMs in 1 sample and the co-occurrence of FMs and OTA in 3 samples were observed.
Moreover, FMB1 and OTA were detected in 2 and 1 samples of white wines, respectively; the co-occurrence of FMs and/or OTA was not observed in all white wine samples. This result clearly indicates that white wine samples are less subject to mycotoxin contamination in comparison with red wines.
As for FMs and OTA detected in wine samples, no regulatory levels have been set in Japan. Therefore, it is necessary to control risk of the contamination of mycotoxins and to closely watch the domestic and foreign trends.