Fusarium culmorum Produces NX-2 Toxin Simultaneously with Deoxynivalenol and 3-Acetyl-Deoxynivalenol or Nivalenol

Fusarium culmorum is a major pathogen of grain crops. Infected plants accumulate deoxynivalenol (DON), 3-acetyl-deoxynivalenol (3-ADON), or nivalenol (NIV), which are mycotoxins of the trichothecene B group. These toxins are also produced by F. graminearum species complex. New trichothecenes structurally similar to trichothecenes B but lacking the carbonyl group on C-8, designated NX toxins, were recently discovered in atypical isolates of F. graminearum from North America. Only these isolates and a few strains of a yet to be characterized Fusarium species from South Africa are known to produce NX-2 and other NX toxins. Here, we report that among 20 F. culmorum strains isolated from maize, wheat, and oat in Europe and Asia over a period of 70 years, 18 strains produced NX-2 simultaneously with 3-ADON and DON or NIV. Rice cultures of strains producing 3-ADON accumulated NX-2 in amounts corresponding to 2–8% of 3-ADON (1.2–36 mg/kg). A strain producing NIV accumulated NX-2 and NIV at comparable amounts (13.6 and 10.3 mg/kg, respectively). In F. graminearum, producers of NX-2 possess a special variant of cytochrome P450 monooxygenase encoded by TRI1 that is unable to oxidize C-8. In F. culmorum, producers and nonproducers of NX-2 possess identical TRI1; the reason for the production of NX-2 is unknown. Our results indicate that the production of NX-2 simultaneously with trichothecenes B is a common feature of F. culmorum.


Introduction
Fusarium head blight (FHB) is a cosmopolitan disease of small-grain cereals caused by several Fusarium species with a high economic impact [1]. Fusarium graminearum and F. culmorum belong to the predominant agents of FHB [2]. Infection of grain crops with Fusarium spp. causes yield losses and contamination of grains with toxic metabolites (mycotoxins), impairing food safety [3,4].
Here, we report that most isolates of F. culmorum collected in Europe and Asia produce NX-2 toxin simultaneously with DON and 3-ADON or NIV.

Mycotoxin Production in Rice Cultures
Our set of 20 F. culmorum strains consisted of 14 strains isolated from maize, wheat, and oat in Germany and 6 strains obtained from other laboratories and culture collections, which were isolated in different countries in a time span of 70 years. The analysis of rice culture extracts revealed an accumulation of NX-2 at concentrations larger than 1 mg/kg in 18 cultures (Table 1). All strains except one were primarily 3-ADON producers; one isolate was assigned to the NIV chemotype. The culture of this strain also accumulated fusarenon X, which is common in NIV producers. The concentration of 3-ADON in rice cultures of DON/ADON producers was high, exceeding 100 mg/kg in the cultures of 12 out of 20 strains. The concentrations of NX-2 and 3-ADON in the cultures of the strains of the 3-ADON chemotype were tightly correlated ( Figure 1). NX-2 in these cultures accumulated to levels corresponding to 2-8% of 3-ADON. In the culture of the only isolate of the NIV chemotype (isolate 59.6st), the concentrations of NIV and NX-2 were comparable ( Table 1).  (Table 1) is represented by a single data point (black dots). Isolate 59.6st, which was assigned to the NIV chemotype, and isolates 227.2 and IPP0999, in which the concentrations of NX-2 were below the limit of detection, were excluded.

Confirmation of the Structure of NX-2
Putative NX-2 accumulating in rice cultures of F. culmorum was originally identified by comparing its retention time in HPLC and MS/MS fragmentation with data obtained with purified NX-2 from Prof. Franz Berthiller (BOKU, Vienna, Austria). Because the production of NX-2 by F. culmorum has not been reported before and an isomer of NX-2 could possess the same retention time and generate product ions with the same m/z values, we purified putative NX-2 from rice cultures of F. culmorum 240.2sp ( Figure 2) to verify its  (Table 1) is represented by a single data point (black dots). Isolate 59.6st, which was assigned to the NIV chemotype, and isolates 227.2 and IPP0999, in which the concentrations of NX-2 were below the limit of detection, were excluded.

Confirmation of the Structure of NX-2
Putative NX-2 accumulating in rice cultures of F. culmorum was originally identified by comparing its retention time in HPLC and MS/MS fragmentation with data obtained with purified NX-2 from Prof. Franz Berthiller (BOKU, Vienna, Austria). Because the production of NX-2 by F. culmorum has not been reported before and an isomer of NX-2 could possess the same retention time and generate product ions with the same m/z values, we purified putative NX-2 from rice cultures of F. culmorum 240.2sp ( Figure 2) to verify its structure. Approximately 5 mg of pure metabolite was obtained from 480 g of dry rice culture (see Section 5.4 for details).

Figure 2.
Purification of NX-2 toxin from rice culture of F. culmorum 240.2sp. Three-week-old rice cultures were extracted with methanol/water/acetic acid and putative NX-2 was enriched by chro matography on a C18 cartridge (a), polar-modified C18 column (b), and Sephadex LH-20 (c). The purity of the metabolite was established by HPLC-MS in a full-scan mode (d). Red boxes mark frac tions collected for purification.
One-( 1 H and 13 C) and two-dimensional ( 1 H, 13 C-HSQC, 1 H, 13 C-HMBC, 1 H, 1 H-COSY NMR spectroscopic analysis was performed on the purified metabolite (Supplementary Figure S1-S5). The spectroscopic data were in accordance with published data for NX-2 (Supplementary Table S1). The structure of NX-2 is shown in Figure 3. One-( 1 H and 13 C) and two-dimensional ( 1 H, 13 C-HSQC, 1 H, 13 C-HMBC, 1 H, 1 H-COSY) NMR spectroscopic analysis was performed on the purified metabolite (Supplementary Figure S1-S5). The spectroscopic data were in accordance with published data for NX-2 (Supplementary Table S1). The structure of NX-2 is shown in Figure 3.

Species Assignment and Investigation of Polymorphisms in TRI1
The examination of conidia assigned all isolates analyzed for trichothecene production (Table 1) to Fusarium culmorum. Morphological characterization was complemented by the analysis of the melting curves of amplicons of taxonomically informative genes encoding translation elongation factor 1α (TEF-1α) and the second largest component of the RNA polymerases II (RPB2) [34]. Species identification was further strengthened by the analysis of the full-length sequence (1753 nt) of the TRI1 gene, which encodes cytochrome P450 monooxygenase catalyzing oxygenation of calonectrin on C-7 and C-8. The sequences of TRI1 obtained from isolates of F. culmorum used in this study (GeneBank accession numbers OM144918 to OM144937) were aligned with a set of reference sequences (Supplementary Table S3) and used for phylogenetic analysis using the maximum-likelihood method. Separation of F. culmorum from other Fusarium species was highly supported, indicating that TRI1 is taxonomically informative in Fusarium species producing trichothecenes.
The reason for selecting the TRI1 gene for the analysis was that the product of TRI1 catalyzes biosynthetic steps distinguishing NX toxins from trichothecenes B, and that polymorphisms in TRI1 differentiating F. graminearum strains producing NX toxins from nonproducers were identified [30]. As expected, all F. graminearum strains not producing NX toxins were separated from all strains producing NX toxins. In contrast, no polymorphisms separating F. culmorum strains producing NX toxins from nonproducers were found in the TRI1 gene ( Figure 4).
The amino acid sequences of the translation products of the TRI1 gene, designated Tri1, were identical for all isolates of F. culmorum in this study. Isolates of F. graminearum producing NX toxins differed from nonproducers in 14 amino acid residues within the heme-binding motif [35] (Table 2). In all F. culmorum strains used in this study, comprising 18 producers and 2 nonproducers of NX-2, these amino acid residues were identical, and they matched the corresponding residues in the strains of F. graminearum that did not produce NX toxins. Thus, the reason for the production of NX-2 by F. culmorum is not an NX-specific form of TRI1 found in NX-2 producers of F. graminearum. Table 2. Amino acid residues specific for the production of NX-2 in the translation product of TRI1 of F. graminearum and the corresponding residues in F. culmorum. Species-specific positions are highlighted in purple for F. culmorum and blue for F. graminearum. Positions reported to distinguish NX-2-producing strains of F. graminearum [35] are marked red. TRI1  3  5  8  33  35  100  115  134  210  252  254  256  346  361  373  418  430 450   Table 3) and reference sequences were subjected to maximum likelihood analysis [36] assuming the Tamura−Nei model [37]. Bootstrap values (1000 replications) are shown next to the nodes. The production of NX-2 was determined by HPLC-MS/MS; n.d. stands for no data available. Nucleotide sequences were deposited at NCBI with the accession nos. listed in Supplementary Table S3.

Production of NX-2 Is a Characteristic Feature of F. culmorum
So far, production of NX-2 has exclusively been reported in F. graminearum, which shares a high level of genomic similarity with F. culmorum, produces the same mycotoxins, and has the same host range [6,38]. The production of NX-2 was discovered in an atypical population of F. graminearum from the Midwest of the USA, named the Northland population, which did not produce any known trichothecene [33]. Producers of NX-2 were later found in this population [30], but only a small fraction of F. graminearum strains collected in the area produced NX toxins. A later study on F. graminearum in wheat in Ontario (Canada), however, reported that 80% of the investigated strains produced NX toxins [16]. A key finding of the current study is that NX-2 production is not limited to F. graminearum, and, contrary to the previous assumption [35], it is not endemic to northern USA and southern Canada.
The ability to produce NX-2 appears to be a common feature of F. culmorum. Among the 20 strains of F. culmorum isolated in Europe and Asia over a period of 70 years, 18 strains produced NX-2 (Table 1). We suppose that the production of NX-2 toxins in cultures of F. culmorum remained unnoticed because NX toxins were not monitored in routine surveys. No commercial standards for NX-2 were available at the time of writing.  Table 3) and reference sequences were subjected to maximum likelihood analysis [36] assuming the Tamura−Nei model [37]. Bootstrap values (1000 replications) are shown next to the nodes. The production of NX-2 was determined by HPLC-MS/MS; n.d. stands for no data available. Nucleotide sequences were deposited at NCBI with the accession nos. listed in Supplementary Table S3. 3. Discussion 3.1. Production of NX-2 Is a Characteristic Feature of F. culmorum So far, production of NX-2 has exclusively been reported in F. graminearum, which shares a high level of genomic similarity with F. culmorum, produces the same mycotoxins, and has the same host range [6,38]. The production of NX-2 was discovered in an atypical population of F. graminearum from the Midwest of the USA, named the Northland population, which did not produce any known trichothecene [33]. Producers of NX-2 were later found in this population [30], but only a small fraction of F. graminearum strains collected in the area produced NX toxins. A later study on F. graminearum in wheat in Ontario (Canada), however, reported that 80% of the investigated strains produced NX toxins [16]. A key finding of the current study is that NX-2 production is not limited to F. graminearum, and, contrary to the previous assumption [35], it is not endemic to northern USA and southern Canada.
The ability to produce NX-2 appears to be a common feature of F. culmorum. Among the 20 strains of F. culmorum isolated in Europe and Asia over a period of 70 years, 18 strains produced NX-2 (Table 1). We suppose that the production of NX-2 toxins in cultures of F. culmorum remained unnoticed because NX toxins were not monitored in routine surveys. No commercial standards for NX-2 were available at the time of writing. Because infection of grain crops with F. culmorum is widespread, we assume that contamination of grains and grain products with NX-2 might be common.

Both Chemotypes of F. culmorum Produce NX-2
Isolates of F. culmorum belong to the 3-ADON and NIV chemotypes; the 15-ADON chemotype is absent [26][27][28]39,40]. Most F. culmorum isolates studied in this work belonged to the 3-ADON chemotype and produced NX-2 toxin. The only strain of the NIV chemotype produced NX-2 toxin, too ( Table 1). Investigation of further isolates has yet to confirm that this generally holds for strains of the NIV chemotype. In F. graminearum, all producers of NX toxins identified in the first report from the USA belonged to the 3-ADON chemotype [30]. Thus, production of NX-2 by a strain of the NIV chemotype is another unique feature distinguishing NX-2 production in F. culmorum and F. graminearum. Another group studying F. graminearum strains from Canada confirmed this finding: NX-2 producers were rare, and all of them (including five isolates from Ontario, see below) were assigned to the 3-ADON chemotype [41]. The most recent study on F. graminearum from Ontario reported contradictory results, assigning most NX-2 producers to the 15-ADON chemotype [16]. Production of NX-2 was verified by chemical analysis in both studies. In the study of Kelly et al. [41], the 3-ADON/15-ADON chemotype was assessed by PCR-RFLP while Crippin at al. [16] used both PCR-RFLP and chemical analysis. Incongruencies between the chemotype prediction by PCR-RFLP and the results of chemical analysis, reported by several studies [14,16,29,42], cannot explain the contradictory chemotype assignments mentioned above because both studies used the same PCR assay (TRI3/TRI12), yet Crippin et al. [16] reported that all their NX-2 producers belonged to the 15-ADON chemotype while Kelly at al. [41] assigned all their NX-2 producers to the 3-ADON chemotype. Crippin at al. [16] also analyzed trichothecenes in cultures on three growth media by HPLC-MS/MS, using a special elution gradient for the separation of 3-ADON and 15-ADON (these mycotoxins often co-elute, and they cannot be reliably distinguished by MS fragmentation). They reported the concentrations of 15-ADON, but unfortunately not the concentrations of 3-ADON and/or DON.
The acetylation of C-3-OH of NX-2 ( Figure 3) is reminiscent of 3-ADON. Varga et al. [30] showed that when part of the translation product of TRI1 in a strain of the 15-ADON chemotype was replaced with a TRI1 segment specific for the production of NX-2, the recombinant strain accumulated NX-4 (a derivative acetylated on C-15-OH, reminiscent of 15-ADON) rather than NX-2. The replacement of NX-specific polypeptide in an NX-2 producer with a protein segment from a 15-ADON producer converted the NX-2-producing strain into a 3-ADON producer. Both results strongly support the hypothesis that NX-2 producers in F. graminearum developed from strains of the 3-ADON chemotype. The controversy between the assignment of chemotypes to NX-2 producers from Ontario in [41] and [16] remains unresolved.

Fusarium culmorum Produces NX-2 Simultaneously with DON and 3-ADON or NIV
A search for new trichothecenes in F. graminearum that facilitated the discovery of NX toxins was motivated by a failure to detect any trichothecene in a group of atypical strains of F. graminearum from the USA [30,33]. These strains produced NX toxins but no trichothecenes B. The recent study from Ontario cited above [16] reported that cultures of F. graminearum accumulating NX-2 simultaneously accumulated 15-ADON.
Cultures of NX-2-producing F. culmorum strains in our study accumulated 3-ADON (most strains) or NIV (a single strain). In cultures of the 3-ADON chemotype, NX-2 accumulated at amounts corresponding to 2-8% of 3-ADON, and the content of NX-2 and trichothecenes B was tightly correlated (Figure 1). In contrast, the concentrations of NX-2 to 15-ADON in F. graminearum cultures in the study of Crippin et al. [16] were not correlated; the ratio NX-2/15-ADON varied from 0.003 to 33. The only strain in our work producing NX-2 at an amount comparable to trichothecenes B was a strain of the NIV chemotype (Table 1). In rice cultures of this strain, NX-2 reached 75% of the concentration of NIV and 66% of total trichothecenes B. Future studies must clarify whether the relatively high production of NX-2 is a typical feature of strains of F. culmorum with the NIV chemotype. In F. graminearum, no strain of the NIV chemotype has so far been reported to produce NX-2.

NX-2 Production by F. culmorum Is not Caused by a Variant of TRI1
The reason for the production of NX toxins by certain strains of F. graminearum is that these strains harbor a special variant of the TRI1 gene [30]. The gene, which is only present in genomes of trichothecene-producing Fusarium species [43], encodes cytochrome P450 monooxygenase (calonectrin C7/C8 hydroxylase) Tri1, which catalyzes oxidation of C-7 and C-8 [44,45]. The difference between TRI1 of NX-2 producers and nonproducers allowed a PCR-RFLP assay for NX-2 producers [46]. We have not found such a polymorphism in the TRI1 gene of F. culmorum.
In F. graminearum, changes in Tri1 specific for NX toxins occurred in the heme-binding motif close to the C-terminus. Furthermore, Ramdass et al. [47] found a new potential glycosylation site in the enzyme of NX-2 producers. It was located at a large distance from the heme-binding site, but the authors suggested that glycosylation may modulate enzyme activity via protein folding. None of these changes occurred in the TRI1 gene of NX-2-producing F. culmorum. Some residues in the heme-binding segment differed from corresponding residues in F. graminearum (purple in Table 2), but these residues occurred in NX-2 producers and nonproducers.
Factors other than the amino acid sequence of Tri1 must suppress its activity towards C-8 of calonectrin in F. culmorum strains producing NX-2. The enzyme is located inside the endoplasmic reticulum, with two hydrophobic segments close to the ends of the protein crossing the membrane [47]. Other proteins and especially other cytochromes P450, which compete with Tri1 for the same NADPH-cytochrome P450 reductase, share this location. Cytochromes P450 are known to interact with each other and form heterooligomers, which modifies their activity [48]. The interaction of Tri1 with other proteins anchored in the membrane of endoplasmic reticulum may suppresses the oxidation of C-8 in NX-2-producing strains of F. culmorum.

NX-2 and the Aggressiveness of F. graminearum and F. culmorum
In F. graminearum, a shift from the 15-ADON to the 3-ADON chemotype observed in the last decades supports the hypothesis that the 3-ADON chemotype is more aggressive. For instance, an increase in the 3-ADON chemotype at locations where 15-ADON producers used to be predominant was reported in a study from North America [49]. Many studies of the population structure in F. graminearum elucidated the relationship between genotype, chemotype, and aggressiveness, aiming to explain how selection and gene flow shaped the populations (e.g., [17,20,35,41,46,[49][50][51][52]). Field studies documented the success of the 3-ADON chemotype of F. graminearum, but they could not address its cause, which requires a biochemical approach. According to a hypothesis from the lab of Gerhard Adam [24], acetylation of C-3-OH prevents detoxification of DON, which is a virulence factor, by plant UDP-glycosyltransferases. In NX-2, the hydroxyl on C-3 is also protected by acetylation, and Varga at al. [30] suggested that the production of NX-2 may benefit F. graminearum during the infection in the same way. Glucosylation of DON takes place in the cytoplasm while the target of DON is protein synthesis in rough endoplasmic reticulum, where the acetyl group would have to be removed. This is plausible because the endoplasmic reticulum is rich in hydrolases [53]. Varga et al. [30] also speculated that the lack of carbonyl on C-8 circumvents detoxification by glutathionylation. This hypothesis holds for F. cumorum, too.
Can the effect of NX-2 on the aggressiveness of F. graminearum or F. culmorum be proved? Field trials with natural isolates are unlikely to generate a conclusive proof. The same situation exists for the claim that the 3-ADON chemotype is more aggressive than the 15-ADON chemotype, which was supported by some studies [20][21][22][23] yet rejected by others [18,19,52]. Field isolates differ in many properties modulating aggressiveness. Some of them are likely linked to the chemotype. Isogenic strains differing only in the gene in question are required. Regarding the relative aggressiveness of the chemotypes 3-ADON and 15-ADON, strains with swapped TRI8 genes or TRI8 chimeras, constructed by Alexander et al. [54], could be used. Similarly, infection experiments with F. graminearum strains harboring TRI1 with swapped domains controlling the NX-2 production, constructed by Varga at al. [30], would reveal the effect of NX-2 production on aggressiveness.

Do further Fusarium Species Produce NX Toxins?
Comparison of the amino acid residues in the Tri1 sequence distinguishing NX-2producing isolates from nonproducers in F. graminearum (Table 2) showed that certain residues specific for the production of NX-2 are also present in Tri1 of other Fusarium species. For instance, Tri1 in a particular F. sambucinum strain shared three residues with NX-2-producing strains of F. graminearum, and it differed from Tri1 of strains that did not produce NX-2 in another three positions. We suggest that this species should be examined for NX-2 production. Fusarium cerealis and F. pseudograminearum harbor TRI1 genes very similar to TRI1 of F. culmorum (Figure 4), and the amino acid sequence in the heme-binding segment of their Tri1 protein is identical with the corresponding sequence in F. culmorum ( Table 2). Examination of these species for the production of NX toxins, too, appears worthwhile. A universal PCR-based assay for NX-2 production does not seem feasible.

Can F. culmorum Produce DON and NIV Simultaneously?
It is generally assumed that strains of F. culmorum and F. graminearum produce either NIV or DON but not both. We believe that this view is biased by the low sensitivity of analytical methods in the past, which only detected trichothecenes at high concentrations (e.g., [27]). Reports of the production of both trichothecenes by a single strain were largely overlooked. For instance, Foroud et al. [3] write in their recent review "NIV chemotypes do not produce DON". Experimental studies such as [20,55,56], the first two of which are cited in [3], clearly showed simultaneous production of DON and NIV by single strains of F. culmorum and F. graminearum.
DON and NIV have a common precursor. In NIV producers, the precursor is hydroxylated at C-4 by Tri13 [13]. We suggest that before the entire pool is hydroxylated, some precursor enters the path leading to DON, and thus all cultures producing NIV accumulate small amounts of DON and/or its acetylated derivatives. The presence of NIV in cultures of DON producers can be explained by a residual activity of TRI13 or the activities of hydroxylases with a relaxed substrate specificity. In line with this reasoning, cultures of the NIV-producing strain 59.6st in our work accumulated relatively large amounts of DON while cultures of most DON and ADON producers contained small amounts of NIV (Table 1). Similar results were reported in F. graminearum ( Table 6 in [56]).
We anticipate that with the widespread use of sensitive analytical methods, small amounts of NIV will often be found in cultures of Fusarium strains producing DON, and substantial amounts of DON will be found in cultures of all strains producing NIV.

Conclusions
Most strains of Fusarium culmorum produce NX-2 toxin simultaneously with deoxynivalenol, 3-acetyldeoxynivalenol, or nivalenol. Strains producing NX-2 do not possess a specific variant of the TRI1 gene known from NX-2-producing strains of F. graminearum.

Fungal Strains
The strains of F. culmorum are listed in Table 3. Strains isolated in the course of this study were obtained from maize grains, rachis, stalks, and oat grains according to Leslie and Summerell [57]. Briefly, samples were surface sterilized for 10 min with 0.1% silver nitrate or 3% sodium hypochlorite, rinsed, and placed on potato dextrose agar (PDA). Isolates were purified via single-spore cultures and grown on PDA for colony characteristics and on low-nutrient agar (SNA, [58]) under long-wave UV light for morphological characterization of spores. For long-term storage, fungal cultures were freeze-dried.

DNA Methods
Fungal DNA was extracted using a CTAB-based protocol [59] from 10 mg of lyophilized mycelium and dissolved in 50 µL of TE (10 mM Tris, 1 mM EDTA, pH 8.0). Segments of marker genes TEF-1α and RPB2 were amplified, and the PCR products subjected to highresolution melting curve (HRM) analysis as described previously [34]. The TRI1 gene was amplified for sequencing as four overlapping fragments (Table 4) Supplementary Table S2. PCR products were precipitated with isopropanol, washed with 80% ethanol, and sequenced at the facilities of Macrogen Europe (Macrogen Europe, Amsterdam, The Netherlands).
The sequences were quality-trimmed with Chromas version 2.6.6 (Technelysium Pty. Ltd., South Brisbane, Australia) and assembled to full-length gene sequences; the accession numbers are listed in Supplementary Table S3. Multiple sequence alignments were performed with ClustalW [60] in MEGA version 10.1.8 [36]. Phylogenetic relationships among TRI1 genes in Fusarium spp. were investigated using maximum-likelihood analysis using MEGA X under the assumption of Tamura's and Nei's substitution model [37]. CAATTCCAATCGCAGACAA [46] a Location of the PCR product within the nucleotide sequence of the gene TRI1 (1753 bp).

Mycotoxin Extraction and HPLC-MS/MS
Rice cultures were prepared in 50-mL Falcon tubes (Sarstedt, Nümbrecht, Germany) by autoclaving 3 g of dry polished rice with 5 mL of tap water. The tubes were inoculated with plugs of PDA (0.5 cm diameter) overgrown with 5-day-old mycelium. Rice medium incubated with agar plugs without mycelium served as a control. The cultures were incubated for 21 days at 25 • C in the dark. Fungal metabolites were extracted by shaking with 30 mL acetonitrile/water/acetic acid (84:15:1 (v/v/v)) overnight. Extracts were dried in a vacuum concentrator (Martin Christ, Osterode am Harz, Germany) and residues redissolved in methanol/water (20:80 (v/v)) as described previously [61]. Mycotoxin analysis was carried out using a 1290 Infinity II HPLC system (Agilent Technologies, Waldbronn, Germany) coupled with a 6460 triple quadrupole detector (Agilent Technologies, Waldbronn, Germany). The separation was performed on a Zorbax Eclipse Plus C18 column, 50 × 2.1 mm with 1.8 µm particle size (Agilent Technologies, Waldbronn, Germany). The column oven temperature was 40 • C. Mobile phase A was water with 0.1% formic acid (v/v), and phase B was methanol with 0.1% formic acid (v/v). The gradient was as follows: 0 to 0.2 min, 5% B; 0.2 to 8 min, 5% to 35% B; 8 to 8.5 min, 35% to 98% B; 8.5 to 12 min, 98% B; 12 to 12.5 min, 98% to 5% B; 12.5 to 16 min, 5% B. The calibration curve included 11 concentrations from 0.48 to 500 µg/L. A blank was analyzed after every 7th sample and a quality control standard after every 15th sample. The metabolites were detected in a multiple reaction monitoring (MRM) mode. The acquisition parameters and the limits of detection (LOD) and quantification (LOQ) are listed in Table 5. Limit of detection (LOD) and limit of quantification (LOQ) were calculated based on the standard deviation of the blank [62]. c m/z of product ions used for quantification are underlined.