Abstract
Alternaria molds produce a variety of chemically diverse secondary metabolites with potentially adverse effects on human health. However, data on occurrence in food and human exposure is inconsistent for some of these mycotoxins. Membrane filtration is a frequent step in many sample preparation procedures for LC-MS-based methods analyzing food contaminants. Yet, little is known about the possibility of adsorptive phenomena that might result in analyte losses. Thus, we treated a complex extract of Alternaria toxins with several types of syringe filters and unraveled the impact on its chemical composition by LC-MS/MS. We observed significant, and in some cases complete, losses of compounds due to filtration. Particularly, two key Alternaria toxins, alternariol (AOH) and its monomethyl ether (AME), were heavily affected. As a comparison with published food surveys indicating a correlation of the type of filtration used with lower incidence reports in food, our results point at a possible underestimation of AME in past exposure assessment. Also, perylene quinones were greatly affected by filtration, underlining the importance to take this into consideration during analytical method development. Furthermore, we applied the comet assay in HT-29 cells to elucidate the impact of filtration on the genotoxicity of the extract. We observed strong coincidences with the loss of epoxide-carrying metabolites and also an intriguing induction of oxidative DNA damage by yet toxicologically uncharacterized Alternaria toxins. In conclusion, we highlight potential issues with sample filtration and call for a critical re-evaluation of previous food occurrence data in the light of the results at hand.