Untargeted LC-MS based 13C labelling provides a full mass balance of deoxy- nivalenol and its degradation products formed during baking of crackers, bis- cuits and bread

Deoxynivalenol (DON) is considered to be one of the most important contaminants in cereals and food commodities produced thereof. So far it is not clear i) to which extent DON is degraded during baking and ii) if a degradation results in reduced toxicity. We have elucidated the fate of DON during baking of crackers, biscuits and bread, which were produced from fortified dough and processed under pilot plant conditions. Untargeted stable isotope assisted liquid chromatography (LC) high resolution mass spectrometry was used to determine all extractable degradation products. Targeted LC – tandem mass spectrometry based quantification revealed that DON was partially degraded to isoDON (1.3 – 3.9 %), norDON B (0.2 – 0.9 %) and norDON C (0.3 – 1.2 %). A DON degradation of 6 % (crackers), 5 % (biscuits) and 2 % (bread), respectively, was observed. In vitro translation experiments indicate that isoDON is less toxic than DON.

Deoxynivalenol (DON) is a mycotoxin that is produced following infestation of plants with Fusarium species. It is most prevalent in cereal commodities such as wheat, barley, oat and corn. Due to the potential health risk associated with DON (e.g. nausea, vomiting, diarrhea, abdominal pain), regulatory bodies such as the European Commission, imposed maximum levels for DON in grain commodities (European Commission, 2006; Van Egmond, Schothorst, & Jonker, 2007). Raw materials are tolerated to have a higher contamination level than the finished food product, as food processing is thought to lower the contamination level. This reduction can potentially be achieved by a) physical processing of the grains (e.g. cleaning, sorting, dehulling, milling), b) dilution of the contaminated flour with non- contaminated ingredients (e.g. water, fats, sugar), and c) degradation (e.g. due to thermal treatment).Whereas physical processing was shown to reduce DON levels in flour by removing DON-enriched by- products, the effect of the baking process on the DON content during the production of bakery wares seems controversial (Karlovsky et al., 2016; Knutsen et al., 2017; Q. Wu, Ku, Humpf, & Klímová, 2017). The reported change of DON content due to baking ranged from an increase of 45 % to a decrease of 50 %, which is shown in Fig. 1 (Bergamini et al., 2010; De Angelis, Monaci, & Pascale, 2013; Generotti et al., 2015; Kostelanska et al., 2011; Kushiro, 2008; Lancova et al., 2008; Scudamore, Hazel, Patel, & Scriven, 2009; Suman, Manzitti, & Catellani, 2012; Valle-Algarra et al., 2009; Vidal, Morales, Sanchis, Ramos, & Marín, 2014; Vidal, Sanchis, Ramos, & Marín, 2015; Voss & Snook, 2010; L. Wu & Wang, 2015).Figure 1: Deoxynivalenol (DON) change during baking as reported in literature.

In our view, this high variation in the reported degradation rates might be partly explained by different ways to calculate the DON degradation rate. Errors may arise from changes in moisture content (MC) or the dilution with non-contaminated ingredients used for preparing the dough which were not accounted for. Contamination with masked forms might even lead to a DON increase. Lack of consideration of recovery factors and matrix effects can also lead to non-accurate results.Although the degradation of DON during baking has been investigated in many studies, the knowledge of the fate of DON is still incomplete. So far, the DON degradation products isoDON, deepoxy-DON (DOM- 1), norDONs A, B, C, D, F and DON lactone have been described in baked food samples (Bretz, Beyer, Cramer, Knecht, & Humpf, 2006; Greenhalgh et al., 1984b; Kostelanska et al., 2011; Vidal et al., 2015; Vidal, Sanchis, Ramos, & Marín, 2017; L. Wu & Wang, 2015). Fig. 2 depicts the chemical structures of themost important DON degradation products, which have already been quantified in processed food samples. Until now, it is not clear whether further degradation products and/or matrix entrapped or bound forms are formed. For a comprehensive assessment of the health risks of eating bakery products produced from DON contaminated flour, the occurrence and the toxicity of all degradation products hasand its formed degradation products by subsequently employing targeted LC tandem MS (MS/MS) analysis. This study presents the first untargeted tracer fate baking study of DON with the aim to identify and quantify all extractable degradation products. Thus, it significantly differs from previously published papers, which estimated the effect of the baking process on DON solely on the observed decrease of the DON concentration. An LC-MS/MS based method was developed and validated for the analyzed matrices and provides the most accurate quantitative results on DON degradation to date. Moreover, the potency of isoDON to inhibit protein synthesis, which is strongly indicative for the overall toxicity of isoDON, is reported for the first time.

2.Materials and Methods
2.1.Chemicals and reagents
Acetonitrile (ACN, gradient grade) was purchased from VWR International GmbH (Vienna, Austria). Acetic and formic acid (both LC–MS gradient grade) were obtained from Sigma Aldrich (Vienna, Austria). LC-MS grade methanol was purchased from Roth (Karlsruhe, Germany). In all experiments, ultra-pure water (purified by a Purelab Ultra system ELGA LabWater, Celle, Germany) was used. Solid DON and U- [13C15]-DON as well as liquid calibrant solutions of DOM-1, DON-3-Glc, DON and U-[13C15]-DON were supplied by Romer Labs GmbH (Tulln, Austria).

2.2.Synthesis of reference standards of the DON degradation products isoDON (Schwartz-Zimmermann et al., 2017) and norDONs A, B and C (Bretz et al., 2006) were synthesized according to published procedures. The structure of the compounds was confirmed by 1H and 13C NMR (see supplementary material). The purity of the compounds was assessed by HPLC-UV (see supplementary material).

2.3.Preparation of bakery products
To determine the concentrations of modified forms of DON, the flour used for the baking trials was first screened by an LC-MS/MS based multi-mycotoxin assay (Malachová, Sulyok, Beltrán, Berthiller, & Krska, 2014).The weight of the ingredients and the dough (see below) was measured with a PE 3000 scale (Mettler Toledo, Greifensee, Switzlerland). The dough was prepared by mixing the ingredients with a 3 L Dito Sama plenary mixer (Electrolux, Stockholm, Sweden) . Fermentation steps were performed in a leavening cell (FermaLievita Alaska, Bologna, Italy). Baking was carried out in a pilot-scale oven (Tagliavini, Parma, Italy). The MC was measured by weighing the dough after heating 5 g of sample at 105 °C for 6 h.For each commodity (crackers, biscuits and bread), three treatment groups were prepared for the dough: a) naturally contaminated (i.e. unfortified), b) fortified with DON (1000 µg) and c) fortified with DON (1000 µg) and U-[13C15]-DON (1000 µg). For the fortified treatment groups, DON and U-[13C15]-DON were dissolved in the water used for preparing the dough to ensure homogenous distribution.Crackers: The dough for the crackers was prepared in two steps.

First, the preparation of the first dough (sponge) and secondly, the preparation of final dough from the sponge. The sponge was prepared by mixing flour, water, salt, and yeast. It was fermented at 26 °C for 19 h at 85 % relative humidity (RH). The final dough was prepared by adding flour, water and malt extract to the sponge. The dough was fermented at 24 °C for 3.5 h at 90 % RH. Crackers were shaped (approximately 10 g each) from the final dough (1058 g, 26 % MC), baked at 250 °C for 5 min and dried at 100 °C for 30 min. The crackers (1026 g) consisted of 97 wt% flour, yeast and malt extract and 3 wt% water.Biscuits: First, the following ingredients were mixed in two separate bowls: a) butter, sugar, milk powder, sunflower oil and b) water, honey, eggs, sodium bicarbonate, ammonium bicarbonate. Secondly, the content of the two bowls was combined and mixed thoroughly. Thirdly, flour was added and the resulting dough was mixed thoroughly. The final dough (1000 g, 14 % MC) was used to form the individual biscuits (approximately 11 g each). The biscuits were baked at 180 °C for 8 min and dried at 100 °C for 10 min. The biscuits (910 g) consisted of 56 wt% flour, 40 wt% fat and sugar and 4 wt% water.Bread: The dough for the bread was prepared by mixing the ingredients (flour, salt, sugar, yeast, vinegar, oil and water) thoroughly. 2 different loafs of 450g each (40% MC) were shaped, fermented at 38 °C for 1 h at 90 % RH and finally baked at 200 °C for 15 min. The bread (4000 g) consisted of 65 wt% flour, yeast, oil and salt and 35 wt% water.

2.4 Untargeted tracer fate study of DON
An untargeted metabolomics workflow, which was first used to detect metabolites of DON in wheat (Kluger et al., 2013), was utilized to study the fate of DON during baking. In short, the dough was treated with a mixture of non-labelled DON and the labelled U-[13C15]-DON tracer. The finished food commodities were analyzed by LC-HR-MS. Signals originating from the tracer were extracted from the raw data by the software MetExtract II (Bueschl et al., 2017).

2.4.1 Sample preparation
2.00 g aliquots of flour or the ground final products were extracted with 8.00 mL of MeOH:H2O (1:1, v:v) or ACN:H2O (84:16, v:v), respectively. The ACN:H2O (84:16, v:v) extract was reconstituted in MeOH:H2O (1:1, v:v) after drying down. Subsequently, the extracts were centrifuged at 10 °C and 15000 rpm for 10 min and 600 µL of the supernatant was transferred into an HPLC vial. Sample work-up was done in duplicate.

2.4.2 LC-HR-MS measurement
The samples were analyzed on an ultra high performance liquid chromatography (UHPLC) Vanquish system coupled to a QExactive HF Orbitrap (both ThermoFisher), which was equipped with a heated electrospray ionization (HESI) source. Five µL of sample extract were injected for chromatographic separation on an XBridge C18 column (150 x 2.1 µm ID, 3.5 µm, Waters), which was protected by a pre- column and maintained at 25 °C. At a constant flow rate of 250 µL/min, a linear gradient program with water containing 0.1 % formic acid as eluent A and MeOH containing 0.1 % formic acid as eluent B was used: An isocratic step at 5 % solvent B for 2 min was followed by a linear gradient to 95 % solvent B until 32 min. The column was washed at 100 % solvent B for 5 min and re-equilibrated at 5 % B for 8 min. The ESI interface was operated in fast polarity-switching mode, using 55 psi sheath gas, 5 psi auxiliary gas, a spray voltage of 3.0/3.5 kV respectively for the positive and negative ionization mode, S-lens 55 eV, capillary temperature 320 °C and auxiliary gas temperature 350 °C. FT-Orbitrap was operated in profile mode (scan range, m/z 100-1500) with a resolving power of 120,000 FWHM (at m/z 200) and automatic gain control setting of 3×106 with a maximum injection time of 200 ms. Mass accuracy was calibrated on a weekly basis or on demand using Pierce LTQ Velos ESI positive and Pierce negative ion calibration solution (ThermoFisher) in 1:1 mixture with manual addition of appropriate levels of butyrate. The system was operated with TUNE 2.8 SP1 software and the direct control plugin for Chromeleon 7.2 SR4. Data processing was performed with XCalibur (ThermoFisher).

2.4.3 Automated data processing
Data-processing consisted of the following steps: i) Each MS scan was inspected for pairs of mass peaks, M (non-labelled degradation product) and M’ (labelled degradation product) with an m/z difference proportional to the number of tracer derived carbon atoms (max. allowed m/z deviation ± 3 ppm). Additionally, the intensity ratio of the M and M’ peaks was tested to be approximately 1 (± 60 %). ii) Subsequently, the extracted ion chromatograms (XICs) were calculated for the detected mass peaks M and M’ and tested for co-eluting chromatographic peak pairs (max. allowed m/z deviation ± 5 ppm). Any such pair was considered a putative DON-degradation product. iii) Finally, all detected compound ions were convoluted by correlating their retention times and peak shapes. Correctly assigned degradation products were listed with a 12C monoisotopic m/z for M, a retention time (tR) and the number of tracer derived carbon atoms. The annotation of the degradation products was carried out by the comparison with reference standards.

2.5 Quantification of DON and its degradation products
The degradation of DON and the increase of its degradation products due to baking were verified for three food commodities. UHPLC-MS/MS was used in combination with a) stable isotope dilution analysis (SIDA) or b) matrix-matched calibration and correction for the extraction recovery (RE) to ensure accurate results.

2.5.1 LC-MS/MS measurement
UHPLC-MS/MS analysis was performed on an Agilent 1290 series UHPLC system coupled to a 6500+ QTrap mass spectrometer equipped with an Ion-Drive Turbo V ESI source (both SCIEX, Foster City, CA, USA). Chromatographic separation was performed on a Kinetex C18 column (150 x 2.1 mm, 2.6 µm) protected by a SecurityGuard ULTRA pre-column of the same stationary phase (both Phenomenex, Aschaffenburg, Germany). Mobile phases consisted of ultra-pure H2O containing 0.1 % of acetic acid (eluent A) and ACN containing 0.1 % of acetic acid (eluent B). Chromatographic separation was achieved at a flow rate of 250 µL/min using linear gradient elution: 0–1 min: 5 % B, 1-12 min: 5-35 % B, 12.1–14 min: 95% B, 14.1-17 min: 95 % B. The injection volume was 4 µL.Tandem mass spectrometric detection was performed in multiple reaction monitoring mode in negative ESI polarity for DON and isoDON and positive ESI polarity for norDONs A, B and C. The ESI settings were: temperature 400 °C, ion spray voltage -4500 V, curtain gas 35 psi, ion source gas 1 60 psi, ion source gas 2 40 psi and collision gas (N2) high. The following transitions were monitored for 100 ms each (declustering potential (DP), collision energy (CE) and cell exit potential (CXP) are given in brackets): DON-3-glucoside: quantifier: m/z 517>427 (DP –50 V, CE –30 V, CXP –11 V), qualifier: m/z 517>59 (DP – 50 V, CE –85 V, CXP –7 V); DON and isoDON: quantifier: m/z 355>265 (DP –50 V, CE –24 V, CXP –13 V), qualifier: m/z 355>59 (DP –50 V, CE –36 V, CXP –9 V); [13C15]-DON: quantifier: m/z 370>279 (DP –50 V, CE –24 V, CXP –13 V), qualifier: m/z 370>59 (DP –50 V, CE –36 V, CXP –9 V); DOM-1: quantifier: m/z 339>59 (DP -50 V, CE -40 V, CXP -9 V), qualifier: m/z 339>249 (DP -50 V, CE -17 V, CXP -15 V); norDON A, C:
quantifier: m/z 267>231 (DP 36 V, CE 11 V, CXP 12 V), qualifier: m/z 267>203 (DP 36 V, CE 14 V, CXP 15 V); norDON B: quantifier: m/z 267>231 (DP 36 V, CE 11 V, CXP 12 V), qualifier: m/z 267>219 (DP 36 V, CE 15 V, CXP 12 V). Representative chromatograms and method validation parameter can be found in the supplementary material. Analyst software version 1.6.3 (Sciex, Foster City, CA, USA) was used for instrument control and peak integration. Further data evaluation was carried out in Microsoft Excel 2013.

2.5.2 Quantification of DON and its degradation products by targeted LC-MS/MS analysis
The concentration of DON and its degradation products was quantified in fortified (1000 µg DON) and unfortified bakery products. Ground and homogenized samples (5.00 ± 0.01 g) were weighed into 50 mL polypropylene tubes and extracted with the fourfold volume (20.0 mL) of extraction solvent (ACN:H2O, 84:16, v:v) for 90 min on a GFL 3017 rotary shaker (GFL, Burgwedel, Germany). The supernatant was transferred into a microinsert HPLC vial. 200 µL of the sample extract were dried down under a gentle stream of pressurized air, reconstituted in 100 µL of H2O and centrifuged at 10 °C and 4000 rpm for 10 min. This way, for instance, a DON concentration of 1000 µg/kg corresponds to 500 µg/L in the injected solution (conversion factor of 2). 4 µL aliquots of the sample extract were injected together with 0.2 µL internal standard working solution (U-[13C15]-DON, 500 µg/L) into the HPLC system. The concentration values of the analytes and their standard deviations were calculated from tenfold work-up of the bakery products and triplicate injection of each sample extract.
Linear calibration curves were obtained for each analyte by plotting i) the relative response of the analyte to the internal standard (SIDA) or ii) analyte signal (neat solvent calibration) versus the analyte concentration. All results were reported in µg of the analyte per kg of the food commodity.DON and isoDON were quantified against the U-[13C15]-DON internal standard using SIDA. The concentration range encompassed 20 – 800 µg/L (corresponding to 40 – 1600 µg/kg) for DON and 2 – 40 µg/L (corresponding to 4 – 80 µg/kg) for isoDON using eight calibration levels.For DON-3-glucoside, DOM-1, norDONs A, B and C, quantitation was carried out by neat solvent calibration. The concentration range encompassed 0.5 – 10 µg/L (corresponding to 1 – 20 µg/kg) using eight calibration levels.

2.5.3 Recovery evaluation and correction
Recovery factors were determined for DON and its degradation products for each matrix. Recovery experiments were performed by spiking blank samples (0.50 g) with the appropriate volume of spiking solution prior to extraction in quintuplicate to obtain a concentration of 1000 µg/kg (DON), 50 µg/kg (isoDON) and 20 µg/kg (DON-3-glucoside, DOM-1, norDONs A, B and C), respectively. Spiked samples were stored uncapped overnight at room temperature to allow solvent evaporation and to achieve equilibrium between analytes and matrix. The samples were extracted with 2.00 mL of extraction solvent and for further sample preparation the above protocol was followed. The concentration of the analytes and their standard deviation was calculated from five replicates of spiked samples. Each sample extract was injected in triplicate. The found concentration was then compared to the theoretical value. All results were corrected by the respective recovery factor (see supplementary material).

2.5.4 Complete recovery study
To study whether the spiked DON was completely recovered, two subsequent extractions were performed on the finished bakery products. After the first extraction, the samples were centrifuged, the supernatant was removed and the extract was analyzed. With the remaining solids, a second extraction was performed and the extract analyzed. The concentration of DON was determined in both extracts and the RE after two extractions was calculated (see supplementary material).

2.5.5 Estimation of LOD and LOQ
For DON and isoDON, no blank bakery products were available. Therefore, the approach to estimate the limit of detection (LOD) and limit of quantification (LOQ) based on the S/N level of the analytes (LOD: S/N 3, LOQ: S/N: 10) in the neat solvent calibration curve and the applied recovery factor (see supplementary material) was used consistently for all analytes.

2.6 In vitro translation assays
The toxicity of isoDON on eukaryotic ribosomes was determined using coupled in vitro transcription/translation [TnT® T7 Coupled Rabbit Reticulocyte Lysate System and TnT® T7 Coupled Wheat Germ Extract System both from Promega (Madison, WI, USA)]. The assays were essentially performed as described (Varga et al., 2015) with the exception that DON and isoDON were dissolved in water and the reticulocyte assays were stopped after 20 minutes. At least three assays with three independent dilutions were performed for each substance and each translation system. A dose response curve was fitted to the individual measurement points using the drc package (Ritz, Baty, Streibig, & Gerhard, 2015) in R version 3.3.1 (2016-06-21).

To elucidate the fate of DON during baking an untargeted tracer fate study was carried out to determine all formed extractable degradation products. This degradation study was carried out for three commodities (differing in ingredients, fermentation agent, fermentation time and received heat load during baking) to cover a wide variety of possible degradation and conversion products that could be formed under different processing conditions.

3.1 Untargeted tracer fate study of DON
The bakery products were prepared from dough that was spiked with non-labelled DON and 13C-labelled DON tracers. The unique isotope pattern and perfect co-elution of the non-labelled and 13C-labeled compounds were utilized by the software tool MetExtract II (Bueschl et al., 2017) to filter all DON degradation products from the high-resolution mass spectrum of the extract of the bakery products. Thereby, any modification of the tracer that occurred during baking (e.g. conjugation or degradation) could be detected in an untargeted manner without prior knowledge of the respective degradation products.In addition to the intact tracer DON, three degradation products were found in the bakery products, independent of the commodity. Degradant 1 (C15) had the same molecular weight and a similar retention behavior as DON. Degradant 2 and degradant 3 (both C14), lost a fragment with m/z 30 which was ascribed to a loss of CH2O. For the annotation of the degradation products, in-house synthesized reference standards of isoDON (C15) and norDONs (C14), which had already been described as DON degradation products in food matrices (Bretz et al., 2006; Greenhalgh et al., 1984a), were used. Degradant 1, 2 and 3 were confirmed to be isoDON, norDON B and norDON C, respectively. norDON A and DOM-1 were not detected. No fermentation specific conversion products, e.g. DON-3-glucoside (Kostelanska et al., 2009), were detected. The LC-HR-MS spectra of a selected biscuit sample before and after analysis with the used analytical and data processing workflow are depicted in Fig. 3.standards.

3.2 Targeted quantification of DON and its degradation products
The DON degradation products isoDON, norDON B and norDON C, which were detected in the bakery products within the untargeted tracer fate study, were subsequently quantified by means of a validated LC-MS/MS method. Moreover, the concentrations of the DON conversion (DON-3-glucoside) and degradation products (norDON A, DOM-1) which were described in the literature were monitored. Fig. 4 summarizes the concentration of the analytes in the naturally contaminated (nat. cont.) flour and bakery products and in the fortified (fort.) bakery products.Figure 4: Deoxynivalenol (DON), DON-3-glucoside, isoDON, deepoxy-DON (DOM-1), norDONs A, B and C concentration of the flour and bakery products, which were produced under pilot plant conditions. The concentration of the analytes was given as the mean ± standard deviation (calculated from tenfold work-up and triplicate analysis of each extract) per kg flour or finished food product. Naturally contaminated (Nat. cont.) samples were produced from flour without the addition of DON. For the fortified samples (Fort.), 1000 µg DON were added to water used for preparing the dough resulting in a spiking concentration of 1000 µg DON/kg crackers, 1090 µg DON/kg biscuits and 1050 µg DON/kg bread. n.d.: not detectable (< LOD)The flour which was used to produce the bakery products, was naturally contaminated with DON-3- glucoside, DON and interestingly also with isoDON, norDONs B and C. norDON A and DOM-1 were not detected. An increase of DON-3-glucoside beyond the concentration resulting from natural contamination was not observed. 3.2.1 Mass balance To determine the extent of the DON degradation, a mass balance of DON and its extractable degradation products was calculated. Furthermore, an incomplete mass balance would indicate that matrix entrapped/bound forms of DON were formed during baking.The degradation of the DON which was added to the dough and the formation of the degradation products was calculated according to:𝐶ℎ𝑎𝑛𝑔𝑒 𝑑𝑢𝑒 𝑡𝑜 𝑏𝑎𝑘𝑖𝑛𝑔 = µ𝑔/𝑘𝑔𝑎𝑛𝑎𝑙𝑦𝑡𝑒,𝑓𝑜𝑟𝑡. − µ𝑔/𝑘𝑔𝑎𝑛𝑎𝑙𝑦𝑡𝑒,𝑛𝑎𝑡.𝑐𝑜𝑛𝑡. µ𝑔/𝑘𝑔𝐷0𝑁,𝑠𝑝𝑖𝑘𝑒𝑑 𝑡𝑜 𝑑𝑜𝑢gℎ The DON content was found to slightly decrease due to baking in all three commodities. 93 ± 2 %, 97 ± 4 % and 100 ± 7 % of the spiked DON could be recovered in baked crackers, biscuits and bread, respectively. This corresponds to a DON decrease of 7 ± 2 % (cracker), 3 ± 4 % (biscuits) and 0 ± 7 % (bread). The decrease of DON was accompanied by the corresponding increase of the degradation products (Fig. 5).Figure 5: Deoxynivalenol (DON) degradation products (isoDON, norDON B and norDON C) that were formed during the production of crackers, biscuits and bread from fortified dough under pilot plant conditions in percent of the spiked DON concentration. Change of the concentration of the extractable DON degradation products. Error bars represent the standard deviation which was calculated from ten replicates.In all three commodities, the main derivative of DON during baking was isoDON, which accounted for 64 % (crackers), 81 % (biscuits) and 70 % (bread) of all found degradation products. norDON B and norDON C were quantified to be minor degradation products.Quantitative analysis of DON was carried out by correcting the results for the RE of the analytes, which was in the range of 80 - 90 %. To study whether the incomplete recovery was due to the formation of non-extractable forms of DON, higher recoveries were aimed for. After two successive extractions, the recoveries of DON increased to 95 % in all three commodities. The observed incomplete recovery could arise from an analytical error or the formation of matrix bound/entrapped forms of DON. Although we could not fully exclude the formation of matrix entrapped/bound forms of DON during baking, the results suggest that their potential formation is negligible. concentrations (IC50) were estimated from dose response curves and their error bands (corresponding to a confidence level of 95 %), which were fitted to the individual data points (x). 4.Discussion 4.1 Identification of DON degradation products in complex food matrices DON was previously found to be partly converted to isoDON, DON-lactone, HM-DON-lactone, noDONs A to F under alkaline conditions (Bretz et al., 2006; Greenhalgh et al., 1984a; Grove, 1985; Schwartz- Zimmermann et al., 2017; Young, 1986; Young, Blackwell, & ApSimon, 1986) and to DOM-1 under acidic conditions (Mishra, Dixit, Dwivedi, & Pandey, 2014). In food matrices, five of the mentioned degradation products, isoDON, DOM-1, norDONs A, B and C, have been confirmed so far by the comparison with reference standards (Bretz et al., 2006; Greenhalgh et al., 1984a; Vidal et al., 2015, 2017).In further studies, accurate mass, fragmentation pattern and elution behavior were used to tentatively identify DON degradation products. Kostelanska et. al identified norDON A, B, C, D and DON-lactone as degradation products of DON in roasted wheat and baked bread samples (Kostelanska et al., 2011). Wu and Wang observed norDON A, B, C, F in bread crust, while these compounds were not found in the bread crumb (L. Wu & Wang, 2015).In our study, we found that isoDON, norDON B and norDON C were formed from DON during the production of crackers, bread and biscuits. DOM-1 and norDON A were not detected. Upon comparing the retention behavior of our reference standards under the same conditions as used in the original publication (Bretz et al., 2006), we concluded that isoDON had been wrongly annotated as norDON A in the previous study (see supplementary material). As DOM-1 was found to be present as natural contamination in the flour and did not increase significantly during baking, we envisage that the formerly reported increase of DOM-1 during baking (Vidal et al., 2015, 2017) had not resulted from degradation of DON. Especially in complex food matrices, matrix derived signals might lead to the false positive identification of DON degradation products, even if UHPLC in combination with highly selective detection modes such as MS/MS or HR-MS are used. This might have been the case for norDON A, D and DON lactone (Kostelanska et al., 2011) and norDON A, F (L. Wu & Wang, 2015). Similar fragmentation patterns may have led to the wrong annotation of isoDON as norDON A in (Bretz et al., 2006). Furthermore, matrix related signals corresponding to the degradation products can lead to wrong analyte identification. Therefore, the lack of reference standards might lead to false positive identification and/or wrong annotation of DON degradation products. Our results support the only two studies which used reference standards for identification and reported isoDON (Greenhalgh et al., 1984a) and norDON B, C (Bretz et al., 2006) as DON degradation products in processed food samples. 4.2 Toxicity of the found degradation products Most of the studies on the behavior of DON during baking focused on the change of the DON content under the assumption that the degradation of DON results in detoxification. However, this assumption had not been verified so far. In our work, for the first time, the full spectrum of DON degradation products that are formed during baking was elucidated. NorDON B and C were shown to be considerably less cytotoxic to human kidney cells than DON (Bretz et al., 2006). This might be either due to a reduced uptake into the cell and/or a reduced toxicity at the molecular target, the ribosome. IsoDON was shown to inhibit protein translation in plant and mammalian ribosomes only at a 94-fold (wheat ribosomes) and 60-fold (rabbit ribosomes) higher concentration than DON. As trichothecenes are potent inhibitors of eukaryotic protein synthesis (Pestka, 2010), these results suggest that isoDON is likely to be less toxic than DON. The cytotoxicity of isoDON is currently investigated and will be published elsewhere. To determine whether the DON degradation is accompanied by a detoxification, more information on the uptake, metabolization and distribution of the formed degradation products is needed. 4.3 DON degradation Studies which use a reliable analytical methodology and realistic baking conditions often report a decrease of the DON concentration during baking ranging from 0 – 20 %. This indicates that DON is mostly stable during baking, which is in accordance with its high thermal stability. From an analytical point of view, a small change of the analyte concentration is difficult to quantify, since the measurement uncertainty could be in the same range as the observed change in concentration due to degradation. Under the conditions chosen in this paper, a DON reduction of 0 – 7 % was observed. However, the standard deviation of the obtained concentration values was in a similar range of 2 - 7 %. Therefore, the degradation of DON could not be accurately quantified by evaluating the decrease of the DON concentration. Hence, we quantified the DON decrease by measuring the increase of its degradation products (assuming the formation of matrix entrapped/bound forms was negligible). The sum of the degradation products, which we consider to be equivalent to the DON degradation, was 6.0 ± 0.3 % (crackers), 4.8 ± 0.2 % (biscuits) and 1.7 ± 0.4 % (bread), respectively. Based on our results, we conclude that the DON degradation can be quantified more accurately based on the increase of the degradation products. This can be illustrated by the following example: Assuming an initial DON concentration of 1000 µg/kg and an analytical precision of 5 %, a degradation of 6 % would result in a remaining DON concentration of 940 ± 47 µg/kg and a concentration of 60 ± 3 µg/kg of the formed degradation products, respectively. Thus, in the first case the calculation of the DON degradation would be based on a DON decrease of 6 ± 5 %. Whereas in the second case, the DON degradation would be based on an increase of its degradation products measured with much better precision, i.e. 6.0 ± 0.3 %. As isoDON was found to be the main degradation product, monitoring only the isoDON increase might be seen as a realistic measure for DON degradation in the future. In our view, this simplification would reduce the cost of analysis (less reference standards and shorter run times) while at the same time providing an accurate value for the DON degradation. 4.4 Fate of DON during baking The same three degradation products with a similar ratio of occurrence were found in the three different commodities. Therefore, we conclude that the formation mechanism is largely independent of the recipe used and the baking conditions.Under the chosen conditions (i.e. bakery wares produced from dough that was spiked with DON and processed under pilot plant conditions), DON was degraded by 6 % in crackers, 5 % in biscuits and 2 % in bread. As pilot plant conditions mimic, but are not identical to the conditions employed during the industrial production of bakery wares (i.e. processing of dough produced from naturally contaminated flour on industrial lines), more experiments are needed in order to study the extent of the DON degradation that can be achieved during the industrial production of bakery wares.Based on the chemical nature of the DON derivatives (e.g. no conjugation to organic molecules), we hypothesize that it is mainly the impact of heat which facilitates DON degradation during baking. The heat impact depends on the a) oven temperature, b) baking time, c) surface to volume ratio, and d) water content of the food commodity. Baking time and temperature were already found to be the most important parameters to influence the degradation of DON (of up to 20 %) during the production of whole-wheat crackers (Suman et al., 2012). In bread, a commodity which does not receive a high heat load (low ratio of surface to volume, high water content), DON degradation was considerably lower than in crackers (Bergamini et al., 2010). In addition to the heat impact, the addition of raising agents such as NaHCO3, which increase the pH value, was shown to facilitate DON degradation during the production of biscuits (Generotti et al., 2017). 5.Conclusion A stable isotope assisted untargeted fate study was carried out to identify all degradation products of DON that are formed during baking. DON was found to partially degrade to three degradation products: isoDON, norDON B and norDON C. The quantification of the parent mycotoxin and all its degradation products revealed that isoDON was the main degradation product with norDON B and norDON C formed to a lower extent. No fermentation specific conversion products (e.g. DON-3-glucoside) were detected. In vitro protein translation inhibition data suggest that the formation of the degradation products could result in a reduced toxicity. In order to determine whether DON degradation truly results in detoxification, information on the uptake, metabolization and distribution of the degradation products is needed. DON degradation during baking could not be accurately quantified by evaluating the decrease of the DON concentration alone, as the precision (2 – 7 %) was in a similar range as the degradation due to baking. It is likely that by not considering the uncertainty of the measurement results, DON degradation was previously over- or underestimated. To quantify the DON degradation during baking with better precision (0.2 – 0.4 %), the sum of the degradation products was evaluated here.The fate of DON was studied in bakery wares, which were produced from fortified dough and processed under pilot plant conditions. DON was found to degrade by 6 % in crackers, 5 % in biscuits and 2 % in bread. As the surface to volume ratio of crackers and biscuits is high compared to bread, it is not surprising to find a higher DON degradation. Pilot plant conditions mimic but are not identical to B02 conditions which are employed during real industrial processing. Therefore, more experiments with bakery products manufactured from naturally contaminated flour and processed on industrial lines are within industrial bread-making technology.