Additives: Vitamin mix (2%), Mineral mix (0.59%), Monosodiumphosphate-24% P (2%), Yttrium oxide (0.01%), Carophyll Pink-10% (0.05%).

#### *2.2. Exposure of Zebrafish and Salmon to Plant-Derived Aquafeeds*

The zebrafish and salmon included in the feeding experiments showed an overall normal growth performance (data not shown). Observable differences in growth rate between diet groups in on-growing salmon in the order of SPC15 > SPC30 > WG15 ≈ FM > WG30 were small but proportional to the feed intake by the same groups. The zebrafish study also included an exposure to PPC15 and PPC30 feed compositions, resulting in a slight growth reduction that had previously also been described for rainbow trout [48]. We considered, however, that the small weight gain differences observed in the present study would not significantly affect the analysis of potentially transmitted contaminants in fish muscle.

#### *2.3. Characteristics of Targeted Analytes in Method*

The mycotoxins and phytoestrogens included in the multi-analyte LC-HRMS/MS method had considerable differences in their molecular weights and structures (Table S1). Furthermore, there were sizeable differences in compound solubilities, e.g., between the hydrophilic DON, DON-3G, 3-ADON, 15-ADON and NIV and the lipophilic enniatins. These differences, as reflected by the logP (Table S1), became obvious in the order of retention on the reversed-phase LC-column (Figure 1). Molecular structure and logP were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Retention times differed with up to 30 min under the optimised chromatographic conditions of the ammonium acetate/MeOH gradient, while peak widths were small demonstrating good signal resolution. MeOH proved to be the best eluent for combining the different analytes in one LC method. Previous studies have shown that MeOH improves peak shape and sensitivity in the analysis of trichothecenes [32,33,35,37–39] and the same solvent has been used for phytoestrogen chromatography [45].

**Figure 1.** Chromatograms of targeted analysis of 100 μg/L in solvent of the 25 mycotoxins and phytoestrogens included in the multi-analyte LC-HRMS/MS method.

During method development, all compounds were analysed in positive and negative ESI mode for the determination of the highest peak intensities and best target ions, which included proton, ammonium, sodium and acetate adducts (Table 2). The HRMS/MS parameters were adjusted accordingly so that each compound was measured in targeted analysis under optimal conditions.



#### *2.4. Optimisation of Sample Preparation*

Appropriate sampling and sample extraction are prerequisites for the reliability of analytical methods [39–41]. Several studies describing sampling strategies for the mitigation of uneven contaminant distribution in different matrices have been published [34]. Sampling plans should aim at achieving pragmatic fit-for-purpose results, providing homogeneity while limiting sample sizes and numbers. In the present experiment, potential distributional heterogeneity was not an issue in the preparation of zebrafish samples since the whole carcasses of three fish were ground and extracted together. In contrast, the salmon fillets were of considerable size and could not be processed in total. Consequently, we attempted to obtain representative samples by punching out tissue at different places in fillet and combining aliquots after grinding (Figure 2a). Additional tissue punches were gathered for proteomic and immunological analyses that were foreseen for subsequent studies (Figure 2b). The composite diets had already a high degree of homogeneity due to the production process. We assumed therefore that the targeted analytes were evenly distributed in samples taken from a few places in the storage bags and ground together.

**Figure 2.** (**a**,**b**) Sampling scheme for homogenous sampling of representative aliquots from a salmon fillet. C: samples used for the chemical analyses in the present study. P and I: samples used for proteomic and immunological analyses in the same project.

Matrix effects impairing analytical method performance can be managed by using clean-up procedures, sample extract dilution, precipitation, filtering, matrix-assisted standard calibration curves and stable-isotope labelled ISTD [34,39]. Clean-up during sample preparation may include passing the extract through immunoaffinity columns or solid-phase extraction (SPE) cartridges, which can be filled with a variety of adsorbents. In the present study, we have not applied clean-up methods during sample preparation to avoid the potential loss of target analytes from surface adhesion. Additionally, the different molecular properties of the 25 compounds would optimally require the use of specific SPE materials. We have therefore attempted to develop a generally applicable sample preparation method by diluting the homogenised material with eight- to tenfold excess of adjusted solvent and using a one-step extraction procedure with subsequent submicron filtering.

Extraction conditions were optimised in a number of preliminary trials by determining recovery rates from spiked matrices with different acidic MeCN/water solvent compositions and, additionally, with a two-step MeCN/water approach [36,40]. However, the two-step extraction produced multiple aqueous and organic layers in the extract, making separation difficult and decreasing analyte recovery. The overall best results for the extraction of the target analytes from feed and fish were achieved with acidic MeCN/water (70:30) (Figure 3), similar to what has been described for other multi-mycotoxin

methods [36,37]. This solvent was also suitable for the phytoestrogens that have been extracted with MeOH/water in previous studies [44,45].

**Figure 3.** Recovery rates from spiked fish and feed matrices for the mycotoxins and phytoestrogens included in the multi-analyte LC-HRMS/MS method using optimised extraction solvent.

#### *2.5. Performance of the Multi-Analyte LC-HRMS*/*MS Method*

The performance characteristics of the new LC-HRMS/MS method for 25 mycotoxins and phytoestrogens were determined with regard to international standardised guidelines [49,50]. The specificity of the method for the selected analytes was excellent due to the high mass accuracy in full scan mode and targeted fragmentation (dd-MS2) (Figure 1; Table 2). The total run time was slightly increased in comparison to other multi-mycotoxin methods [32,33,35,37–39,41], leading to good chromatographic separation of the analytes. The high resolution of the analysis allowed us to resolve between isomers such as 3-ADON and 15-ADON, which previously has been sometimes a challenge [41].

The 25 analytes were detected with different sensitivities in fish and feed matrices differed considerably between the 25 analytes. The salmon matrix-assisted standard calibration curves showed high sensitivities for the enniatins, ZEN and the phytoestrogens daidzein and genistein, whereas the curve slopes were less steep for the trichothecenes, OTA, ergot alkaloids and remaining phytoestrogens. Interestingly, this order was not identical for solvent, zebrafish and feed matrices, comparable to results reported for other multi-mycotoxin methods that achieved different analyte sensitivities in matrices such as fruit, yoghurt, soya, hazelnut, pepper, wheat, maize, oat, rice, pasta and bread [33,35–38]. The effect of the signal enhancement or suppression by a specific matrix type can be illustrated by the connected SSE% value. Matrix impact is considered as insignificant for SSE 80-120%, while lower values indicate significant signal decrease and higher values signal increase [32,33,35,37–40]. In the present study, SSE varied from 67% to 115% for control fish feed, 58% to 173% for salmon, and 89% to 181% for zebrafish, with ENN A showing the highest signals in the feed and fish matrices (Table 3). Considering all analytes, the feed matrix generally suppressed signals, whereas the fish matrix caused signal enhancement.

Linearity of the standard calibration curves in different matrices was achieved for all analytes in the range 1.0 to 200 μg/L, with the exception of NIV, OTA, DON-3G and 15-ADON that were linear in the range 5.0 to 200 μg/L. The correlation coefficients (Table 2) were *R*<sup>2</sup> > 0.98 for all calibration curves, irrespectively of whether or not stable-isotope labelled ISTD, similar analogue-ISTD or no ISTD were

included. Considering the eight times or 10 times sample dilution during matrix extraction, the linear ranges corresponded to 8.0 (40)–1600 μg/kg for feed and salmon and 10 (50)–2000 μg/kg for zebrafish.

The limits of detection (LOD) and quantification (LOQ) in solvent, fish feed, salmon and zebrafish matrices are presented for the undiluted commodities (Table 3). The LOD ranged in solvent from 1 μg/L for ENN A1, B, B1 and genistin to 19 μg/L for NIV, in fish feed from 6 μg/kg for 15-ADON to 85 μg/kg for ENN A, in salmon from 21 μg/kg for glycitein to 144 μg/kg for NIV, and in zebrafish from 8.0 μg/kg for ergonovine and α-ergocryptine to 176 μg/kg for DON-3G. The corresponding LOQ were, as per the definition, 3.3 times higher (Table 3). The values were similar to data shown for comparable multi-mycotoxin methods. LOD ranging from 5.4 to 24 μg/kg for DON, 36 to 50 μg/kg for 15-ADON, 2.8 to 50 μg/kg for NIV, 0.2 to 47 μg/kg for ZEN, 1.0 to 18 μg/kg for T-2, and 0.7 to 12 μg/kg were reported in a number of different matrices [32,35–38]. In contrast, two methods that had been specially developed for the analysis of phytoestrogens in food products had established group LODs of, respectively, 250 μg/kg [44] and 15 μg/kg [45].

The precision of our multi-analyte LC-HRMS/MS method was demonstrated on the one hand by good day-to-day congruency of the solvent and matrix-assisted standard calibration curves. The coefficients of variation (% CV) for all data points in six independent experiments were generally less than 20% in solvent and less than 25% in feed, salmon and zebrafish matrices (data not shown), which was well within the guidance criteria [49]. On the other hand, precision was also assessed by intra-day and inter-day analysis of spiked quality control samples. The total within-laboratory precision was in the range of 1% for ZEN and ENN A to 17% for NIV in the feed matrix and 1% for ergonovine to 41% for NIV in the salmon matrix (Table 3). The precision data were comparable to values reported for other multi-mycotoxin methods in a variety of matrices [32,35,37,38,41]. Published precision data for phytoestrogen analysis in food commodities are scarce. When control samples were analysed using standard calibration in solvent, intra-day and inter-day% CV in the range of 1–13% were reached for a number of analytes [45].

Recovery rates in fish feed ranged from 19% to 161% for all mycotoxins and phytoestrogens in the newly developed method, with the exception of DON-3G, NIV, ergosine, ergotamine, ergocornine and α-ergocryptine that were retrieved less efficiently, and ENN A and ENN A1 that showed enhanced recoveries (Table 3). In the salmon matrix, the analytes were recovered with 69–127% except for a reduced performance for NIV and enhancement for genistein. In the zebrafish matrix, recovery rates of 41–98% were reached, except in DON-3G and NIV, which showed reduced values. The recovery rates established in the present study were similar to those determined with comparable methods ranging from 50% to 150% for a number of mycotoxins [32,35–41]. For phytoestrogens, recoveries between 89% and 107% in spiked solvent have been reported [45]. However, in different food matrices the rates were widely varying and in part very low, which is in strong contrast to our new LC-HRMS/MS method, showing remarkably low interference for phytoestrogen analysis in the three matrices considered (Table 3). Spiking experiments are widely used for the determination of recoveries in the validation of analytical methods, although they only can emulate naturally-contaminated samples to a certain extent. Preferably, the accuracy should be verified with a certified reference material, but this is currently not available for all target analytes and selected matrices of the LC-HRMS/MS method.



#### *2.6. Mycotoxins and Phytoestrogens in Fish Feed, Zebrafish and Salmon Tissues*

The in-house-validated multi-analyte LC-HRMS/MS method was used for the analysis of the customised fish feed and dietary exposed salmon and zebrafish. The feed analysis did not detect any of the targeted mycotoxins, with the exception of ENN B that was found in concentrations close to LOD in WG30 (data not shown). Norwegian aquafeeds ingredients contain generally only low amount of mycotoxins [13,23]. The highest mean contents were found in wheat (DON: 94 μg/kg; T-2+HT-2: 28 μg/kg) and maize (ZEN: 246 μg/kg), which was in compliance with the recommended maximum levels [24,25,51]. Considering that in the present study, the feed contained a maximum of 42% wheat-derived components (WG30) (Table 1), we did not expect sizable levels in the five diets. In contrast, survey data for finished feeds from Central Europe and Asia contained on average 165 μg DON/kg, 188 μg ZEN/kg and 2 μg OTA/kg [11]. Interestingly, our finding of ENN B in WG30 diets is in line with the relatively high prevalence of enniatins in cereals in Northern Europe. ENNs have shown considerable toxicity in in vitro studies and in mice [52]. Carry-over of ENN B and B1 from poultry feed into eggs has been demonstrated [22], but maximum levels for animal feed have not been established yet.

In view of the low mycotoxin content (<LOQ) in the customised feeds in the present study, we consequently did not detect any of the targeted analytes above the respective LOQ in salmon or zebrafish tissues. There were, however, traces of ENN B in several of the WG30-exposed salmon at concentrations close to the LOD, suggesting the carry-over potential of enniatins. A relatively high occurrence of ENNs, especially ENN B, in fish muscle and livers has been previously reported [22,53] and correlates with our data. Transfer of mycotoxins such as DON, T-2 and OTA from low-level contaminated wheat gluten-containing feed into fish fillets has also been demonstrated [13]. In contrast, when salmon was fed with diets containing 2 and 6 mg DON/kg or 0.8 and 2.4 mg OTA/kg for eight weeks, up to 19 μg DON/kg was measured in the muscle, whereas up to 5 μg OTA/kg was detectable in the fish livers [46]. Human exposure following high consumption of salmon fillets with the highest DON concentrations was estimated to amount to only 2% of the established tolerable daily intake (TDI) [46,54]. Consequently, our results in the present study show that the use of plant-based fish feed containing mycotoxins below the recommended maximum levels results in negligible health risks for consumers.

The phytoestrogen analysis of the diets included in the salmon and zebrafish feeding experiments showed dose-dependent levels of all targeted analytes in the soy protein containing feeds (data not shown). Mean concentrations ranged in SPC15 from 21 μg glycitein/kg to 786 μg daidzin/kg and in SPC30 from 40 μg glycitein/kg to 1356 μg daidzin/kg. Glucosidated forms occurred in higher concentrations than the corresponding aglycons, whereas an increase of the free form had been previously observed in extruded protein preparations [30]. In PPC15 and PPC30, 26 and 54 μg glycitein/kg were detected, respectively, confirming results from a screening study on fruits and vegetables [45]. Phytoestrogen levels in food and feed are not regulated so far, and the health risks or benefits of dietary exposure in humans and animals are still under discussion [28,44]. Still, considerable oestrogenic and thyrogenic activities have been determined in vitro in commercial Spanish fish feeds [18], and further evaluation is required. A survey of the phytoestrogen content in food products of animal origin detected the highest concentrations in soy-containing milk products and farmed salmon contained up to 40 μg/kg [47].

In the present experiment, we did not find phytoestrogen concentrations above LOQ in dietary exposed zebrafish or salmon, not even in the respective SPC30 groups. Information on the uptake of isoflavones in fish is not available, but considerable differences in bioavailabilities and biotransformation are reported for warm-blooded vertebrate species [55]. We have recently studied the metabolism of isoflavones in salmon liver microsomes (article in progress) and characterised the major metabolites. Chromatographic peaks corresponding to the retention times and m/z of these metabolites were, however, absent in the muscle of the dietary exposed fish suggesting an efficient detoxification mechanism and excretion of isoflavones without accumulation in the edible parts of fish. Equol,

an intestinal metabolite of daidzein, has not been studied in our experiment. Previous studies have suggested that isoflavone metabolisation by the intestinal microbiome varies considerably between producers and non-producers of equol [55]. When gibel carp (*Carassius auratus gibelio*) were exposed to 40–400 mg daidzein/kg in feed, the unchanged compound was recovered with 128 and 261 μg/kg in the fish muscle [56]. In contrast, equol was not found in any of the samples suggesting that fish could lack the necessary gut bacteria. Considering that the highest daidzein level in our experiments was with 0.2 mg/kg in SPC30, about 200-fold smaller than the lowest feed concentration in the gibel carp study, and considering the LOQ of the LC-HRMS/MS method in the fish matrix, the non-detectability of the targeted isoflavones in the salmon fillets was conclusive. However, we intend to investigate the metabolic fate of important isoflavones in fish in depth in a follow-up study.

#### **3. Conclusions**

The increasing use of vegetable ingredients in aquafeeds has motivated risk evaluations for mycotoxin exposure of farmed fish, which has resulted in the establishment of recommended maximum levels. Furthermore, the potential consequences of the presence of bioactive compounds such as isoflavones in plant-based feed should be monitored. We have therefore developed and validated a 25-in-1 LC-HRMS/MS method that is suitable for the survey of compliance to feed regulations and for the detection of undesirable compounds in fish fillets. The new method has excellent specificity for all analytes, while there are some differences in sensitivity due to the great diversity of molecular structures. The LOD and LOQ in fish feed, zebrafish and salmon matrices are sufficient to ensure that mycotoxin and phytoestrogen levels are below concentrations that might cause negative health effects. The accuracy of the method, described by precision and recovery of the included analytes, is satisfactory, confirming its applicability for screening and surveillance purposes. The applicability range is limited at present, however, due to the exclusion of aflatoxins. They will be added during the planned extension of the multi-analyte method. In zebrafish and salmon exposed to customised feed containing up to 30% wheat gluten, soy or pea protein concentrate, carry-over of mycotoxins or phytoestrogens could not be detected, confirming that fillets from fish fed commercial plant-based diets are safe for consumption.

#### **4. Materials and Methods**

#### *4.1. Chemicals*

LC-MS grade acetonitrile (MeCN), methanol (MeOH) and water (Optima, LC/MS grade,) were provided by Fisher Scientific (Loughborough, Leics., UK), and ethanol (EtOH) was obtained from VWR International (Lutterworth, Leics., UK). Acetic acid (CH3COOH) (>99.8%), formic acid (HCOOH) (>98%) and ammonium acetate (CH3COONH4) (>98%) were purchased from Merck KGaA (Darmstadt, Germany).

The mycotoxins deoxynivalenol (DON), 3-actetyl-deoxynivalenol (3-ADON), nivalenol (NIV), T-2 toxin (T-2), HT-2 toxin (HT-2), zearalenone (ZEN), deoxynivalenol-3-glucoside (DON-3G), 15-acetyl-deoxynivalenol (15-ADON), ochratoxin A (OTA), ergosine, α-ergocryptine and ergocristine as well as the stable isotope-labelled analogues U-[13C-15]-NIV, U-[13C-15]-DON, U-[13C-21]-DON-3G, U-[13C-17]-3ADON, U-[13C-17]-15ADON, U-[13C-22]-HT-2, U-[13C-24]-T-2, U-[13C-20]-OTA, U-[13C-18]-ZEN were provided by Romer labs (Tulln, Austria) as solutions in MeCN, ranging from 10 to 100 mg/L. Intermediate standard solutions at 10 mg/L were prepared for DON-3G and 15-ADON by dilution of stock solutions with MeCN. The enniatins A, A1, B, and B1 (ENN A, A1, B, B1), ergonovine, ergotamine, ergocornine, methysergide maleate salt (MetErg) and bromocriptine mesylate (BromCri) were provided as solids by Sigma-Aldrich (St. Louis, MO, USA). Stock solutions in MeOH or MeCN were prepared for ergot alkaloids in the range of 100 to 500 mg/L, and for enniatins with 200 mg/L in MeOH. A combined intermediate standard solution with 10 mg/L was prepared for both enniatins and ergot alkaloids by combining appropriate aliquots of

stock standard solutions, evaporating the mixture with a gentle stream of nitrogen and re-dissolving in MeCN/water (50:50). Finally, a combined standard solution containing all mycotoxins (Set A) was prepared by combining aliquots of stock or intermediate standard solutions, evaporating the solvent and re-dissolving in the appropriate volume MeCN/water (50:50) to obtain final concentrations of about 200 μg/L (200.0–200.12 μg/L, depending on the stock solution provided by the manufacturer).

The phytoestrogens daidzin, genistin, glycitin, daidzein, genistein, and glycitein were bought in crystalline form from Sigma-Aldrich (St. Louis, MO, USA), and stock solutions were prepared in MeOH or DMSO (glycitein) ranging from 500 to 1000 mg/L. Individual intermediate standard solutions at a concentration of 5 mg/L were prepared by dilution with MeOH. A combined standard solution (Set B; 200 μg/L) containing all phytoestrogens was prepared by further dilution in MeCN/water (50:50). The finished Set A and Set B solutions were stable at −20 ◦C for several months and used for the preparation of standard calibration curves.

Additionally, a 25-in-1 multi-analyte mixture was prepared and used in spiking experiments. All analytes were combined with regard to the concentrations of their respective stocks or intermediate standard solutions so that a final concentration of 25 μg/L per analyte was reached after spiking into feed, zebrafish and salmon samples. The multi-analyte mixture was evaporated and re-dissolved in MeCN/water (50:50). It was stable at −20 ◦C for about a month.

A combined internal standard (ISTD) solution for 15 mycotoxins, containing stable isotope-labelled analogues and the ergot homologues MetErg and BromCri, was prepared in MeCN/water (50:50) to reach final concentrations of 251 μg/L U-[13C-18]-ZEN, 500 μg/L U-[13C-22]-HT-2, 443 μg/L U-[13C-22]-T-2, 506 μg/L U-[13C-15]-DON, 502 μg/L U-[13C-17]-3ADON, 500 μg/L U-[13C-17]-15ADON, 500 μg/L U-[13C-20]-OTA, 530 μg/L U-[13C-15]-NIV, 530 μg/L U-[13C-21]-DON-3G, 624 μg/L BromCri and 500 μg/L MetErg. The different concentrations were chosen with regard to the respective measurement sensitivities in the developed multi-analyte LC-HRMS/MS method. The ISTD solution was stored at −20 ◦C, adjusted to room temperature (RT) and mixed thoroughly prior to use. It was added in a ratio of 1:5 to the study samples.

#### *4.2. Preparation of Fish Diets*

Diets with definite amounts of wheat gluten, soy protein concentrate or pea protein concentrate were produced at Nofima Feed Technology Centre, Fyllingsdalen, Norway. The diets were based on fishmeal (FM) as main protein source, which was replaced by 15% or 30% plant proteins. All diets contained 12% wheat that was required for binding in the extrusion process, in addition to minor inclusion of wheat as carrier for some of the additives used (Table 1). In total seven diets were produced: (1) control feed (FM), (2) 15% soy protein concentrate (SPC15), (3) 30% soy protein concentrate (SPC30), (4) 15% wheat gluten (WG15), (5) 30% wheat gluten (WG30), (6) 15% pea protein concentrate (PPC15), and (7) 30% pea protein concentrate (PPC30). The ingredients used for the preparation of diets included FM Norsildmel AS (Bergen, Norway), SPC from Agilia A/S (Videbæk, Denmark), PPC from AM Nutrition AS (Stavanger, Norway) and WG from Tereos Syral (Marckolsheim, France). All diets had an inclusion of 4% fish oil at extrusion. The feed were produced on a pilot scale twin-screw, co-rotating Wenger TX 52 extruder (Wenger, Sabetha, KS., USA) with a die of 2.5 mm diameter. After extrusion, the diets were dried for 40–70 min in a carousel dryer (Paul Klöckner, Verfahrenstechnik GmbH, Hachenburg, Germany) at 65 ◦C to a water content of 7–8%. The salmon diets 1 to 5 were, in addition, oil-coated with 16% fish oil after extrusion by vacuum-coating (Dinnissen, Sevenum, Netherlands) to meet the standard dietary inclusion of oil for the fish size studied. The salmon feed had a pellet size of 3.5 mm, while the zebrafish feed were ground and sieved to a pellet size of 0.6–0.8 mm.

#### *4.3. Feeding Studies in Zebrafish and On-Growing Salmon*

#### 4.3.1. Zebrafish

Four-month-old zebrafish (*Danio rerio*) (AB strain) with a mean weight of 0.214 g were distributed into 28 tanks (*n* = 16) and were maintained in a flow-through system with 20 % water exchange per hour (ZebTEC Stand-Alone Toxicology Rack, Techniplast, London, UK) under daily-monitored standard husbandry conditions, including a stable temperature of 28 ± 0.5 ◦C, pH 7.5, water conductivity of 1500 μS/cm and photoperiod of 12 h light:12 h dark at the Faculty of Biosciences and Aquaculture, Nord University, Bodø, Norway. The feeding study included 336 fish that were distributed into the system's 3.5-litre tanks according to the seven experimental diets. Four replicate groups per diet, each consisting of 12 fish (six per gender) in one tank (and an additional four fish to compensate for potential losses during the study period), were hand-fed twice daily with a total feed amount equal to 2.5% of their body weight over a period of 46 days. The feeding behaviour and health and welfare of the fish were regularly controlled. At the end of the study, the fish were not fed for 24 h prior to sampling. They were separated by gender and euthanised individually by transfer into a tank containing a lethal dose of 200 mg/L tricaine methanesulfonate (MS222) (Sigma-Aldrich, St. Louis, MO, USA), buffered with an equal amount of sodium bicarbonate. The liver, spleen and intestines were carefully dissected under a light microscope and immediately frozen in liquid nitrogen along with the rest of the carcass. All samples were stored at −80 ◦C for further analyses.

The zebrafish feeding study was conducted in compliance with the guidelines provided by the Norwegian Animal Research Authority (FOTS ID 12581, 27 July 2017) and approved by the Nord University (Norway) ethics committee.

#### 4.3.2. Salmon

One-year-old post-smolt Atlantic salmon (*Salmo salar*; salmo breed strain) with a mean weight of 223 g were randomly distributed into 15 experimental tanks (1 m3; *n* = 32) filled with seawater at the Nofima Research Station, Sunndalsøra, Norway. The oil-coated diets 1–5 were given to randomised triplicate tanks by automatic disc feeders. Excess feed was collected once daily for calculation of feed intake. The water temperature was maintained at an average of 10.6 (±0.6) ◦C. The oxygen level at the tank outlets was higher than 90% at study start and about 80% at the study's end. The water flow in each tank was set to 20 L/min.

The feeding was conducted for nine weeks. At the start of the experiment, 15 fish were sampled, and the muscle, liver and intestine were collected. After five weeks, muscle was sampled from one fish from each tank of the FM, SPC30 and WG30 groups. At the termination of the study, five fish from each tank were collected and weighed. The sampled fish were anaesthetised with 60–80 mg/L MS222, transferred and euthanised with a double dose (120–160 mg/L) MS222. Blood was drawn from the caudal vein using 2.5-mL vacutainers (VACUETTE® 2.5 mL Z serum separator clot activator; Greiner Bio-One, Kremsmünster, Austria) and centrifuged at 2500× *g* for 15 min at 4 ◦C (Allegra 6R Centrifuge, Beckman, Indianapolis, IN, USA), and sera were stored at −20 ◦C. The livers and intestines of the fish were removed, and tissue samples were frozen with liquid nitrogen and stored at −80 ◦C. Fillets were stored at −20 ◦C. The remaining fish in each tank were weighed in bulk, and their mean weight was calculated, including the sampled fish.

The salmon feeding study was performed in compliance with the national regulations for the use of animals in experiments [57]. The experiment was classified as not requiring a specific license [58] as none of the planned experimental treatments were expected to cause any distress or discomfort for the fish.

#### *4.4. Extraction of Fish Feed, Zebrafish and Salmon Samples*

#### 4.4.1. Fish Feed

Fish feed pellets were homogenised with a grinding mill (Retsch, Haan, Germany), and 2.5 g were weighed into 50-mL polypropylene tubes. After the addition of 20 mL extraction solvent, the samples were vortexed for 1 min, extracted on a horizontal shaker (Edmund Bühler, Tübingen, Germany) with 200 min−<sup>1</sup> at room temperature (RT) for 30 min, and centrifuged with 2000<sup>×</sup> *g* for 10 min at 4 ◦C (Beckman Coulter, Brea, CT, USA). The supernatants were transferred into fresh 50-mL tubes and let to settle overnight (ON) at 4 ◦C. Subsequently, 0.5 mL of the supernatants were centrifuged for 1 min at 20,000× *g* through 0.22 μm nylon filters (Costar Spin-X; Corning, Inc., Corning, NY, USA) and 40 μL of the filtrates were transferred into LCMS vials. Finally, 10 μL ISTD solution were added to each vial. Samples were store refrigerated until analysis by LC-HRMS/MS.

The composition of the extraction solvent was optimised during method development in spiking experiments. Multi-analyte mixture (50 μL) was added to 2.5 g ground feed, which was then kept under a laminar hood for 30 min, allowing the solvent to evaporate. Extractions were performed either in one step with 20 mL acidic (0.1% formic acid (FA)) MeCN/water mixtures of different compositions (50:50; 60:40; 70:30; or 80:20) or in two steps with acidic MeCN/water (I: 80:20; II: 20:80). Based on the best recovery rates for mycotoxins and phytoestrogens, MeCN/water (70:30; 0.1% FA) was selected for all further experiments.

#### 4.4.2. Zebrafish

Three frozen, gutted zebrafish, for each replicate and diet, were thawed and, after separation of the heads, ground to a fine powder with pestle and mortar in liquid nitrogen. The powdered tissue (0.1 g) was weighed and extracted with 1 mL extraction solvent (MeCN/water 70:30; 0.1% FA). The mixture was homogenised by ultra-sonication (Branson, Danbury, CT, USA) for 10 min at 30 ◦C, centrifuged at 4000× *g* for 10 min at 4 ◦C (Thermo Scientific, Waltham, MA, USA), and the supernatant was transferred into fresh 5-mL tubes. An aliquot (0.5 mL) was filtered as described before, and 40 μL of the filtrates were transferred into LCMS vials, mixed with 10 μL of the ISTD solution, and analysed by LC-HRMS/MS.

The recoveries of mycotoxins and phytoestrogens from the zebrafish matrix was investigated during method development by different acidic MeCN/water extraction solvents in spiking experiments with multi-analyte mixture.

#### 4.4.3. Salmon

The salmon fillets were half-thawed. Tissue pieces of equal size were sampled from four different areas using a steel puncher (0.5 cm in diameter) (Figure 2). The tissue samples were ground with a pestle and mortar, combined, and 2.5 g were transferred into a 50-mL tube, extracted with 20 mL extraction solvent (MeCN/water 70:30; 0.1% FA) and thoroughly homogenised for 40 s by ultra-turrax (Janke & Kunkel, IKA-Werke, Staufen, Germany). To avoid cross-contamination, the ultra-turrax was washed with water for 20 s between samples from the same fish tank and with water and MeOH for 40 s between samples from different tanks. The samples were vortexed for 30 s and extracted using a horizontal shaker (Edmund Bühler) with 200 min−<sup>1</sup> at RT for 1 h. Subsequently, they were centrifuged with 2000× *g* for 10 min at 4 ◦C (Beckman Coulter), and the supernatants were transferred into fresh 50-mL tubes and let to settle overnight at 4 ◦C. Subsequently, 0.5-mL aliquots were filtered as described before, and 40 μL of the filtrates were transferred into LCMS vials, mixed with 10 μL of the ISTD solution, and analysed by LC-HRMS/MS. The recovery of mycotoxins and phytoestrogens from the salmon matrix was investigated as described for zebrafish.

#### *4.5. Preparation of Matrix-Assisted Standard Calibration Curves*

Calibration curves in solvent were prepared by evaporating 200 μL Set A solution with nitrogen and re-dissolving with 200 μL Set B, resulting in a standard solution with 200 μg/L for all 25 analytes included in this study. The standard solution was serially diluted with MeCN/water (50:50) to produce calibrants with 200, 100, 50, 10, 5 and 1 μg/L. For the preparation of the matrix-assisted standard calibration curves, 40 μL aliquots of the calibrants were transferred into LCMS vials and 10 μL ISTD solution was added. They were evaporated with nitrogen at 40 ◦C and re-dissolved in the same volume of blank matrix extract that had been prepared either from control feed or from zebrafish or salmon in the respective FM-control groups by pooling equal volumes of replicates. The calibration standards were transferred into LCMS vials and analysed by LC-HRMS/MS.

#### *4.6. Development of the Multi-Analyte Liquid Chromatography High-Resolution Mass Spectrometry (LC-HRMS*/*MS) Method*

Multi-analyte analysis was performed on a Q-Exactive™ Hybrid Quadrupole-Orbitrap HRMS/MS equipped with a heated electrospray ion source (HESI-II) and coupled to a Vanquish UHPLC system (Thermo Scientific). The instrument setup was similar to that described in a previous study [36]; however, there were several modifications and different analytes were included. The HESI-II interface was operated at 300 ◦C, alternatively in positive and negative mode during one run. The parameters were adjusted as follows: spray voltage 3.2 and 2.5 kV (positive and negative mode, respectively), capillary temperature 280 ◦C, sheath gas flow rate 35 L/min, auxiliary gas flow rate 10 L/min, and S-lens RF level 55.

The Q-Exactive HRMS/MS was operated in full scan (FS) mode with the inclusion of targeted fragmentation (data-dependent MS/MS: dd-MS2). For full scans, the mass ranges were set to *m*/*z* 90–900 and 200–900 in negative and positive mode, respectively. FS data were acquired at a mass resolution of 70,000 full width half-maximum (FWHM) at *m*/*z* 200, while mass resolution was set to 17,500 FWHM at *m*/*z* 200 during dd-MS2. The automated gain control (AGC) target was set to 5 <sup>×</sup> 105 ions for a maximum injection time (IT) of 250 ms in the FS mode, whereas for dd-MS2 mode the AGC target was <sup>1</sup> <sup>×</sup> 105 and the IT was 100 ms. The inclusion list for the targeted analysis contained the *<sup>m</sup>*/*z*, retention times (RT), and normalised collision energies (NCE) (Table 2). NCE values were determined by direct infusion of standard solutions in the mobile phase (MeCN/water (50:50), containing 5 mM ammonium acetate and 0.1% acetic acid) by using a syringe pump at a flow rate of 5 μL/min. The quadrupole mass filter was operated with an isolation window of *m*/*z* 3. External mass calibration of the Q-Exactive HRMS/MS was performed every three days over the mass range *m*/*z* 90–2000, in accordance with the manufacturer's instructions. The identification of the 25 mycotoxins and phytoestrogens included in the multi-analyte method was supported by the determination of specific retention times, fragmentation patterns and accurate masses, which were obtained using a mass accuracy window of ±5 ppm with respect to the theoretical accurate masses (Table S1). Chromatographic separation was achieved at 30 ◦C on a 150 × 2.1 mm Kinetex reversed-phase F5 column (2.6 μm, 100Å; Phenomenex, Torrance, CA, USA) with a 0.5 μm × 0.004" ID, HPLC KrudKatcher Ultra Column In-Line filter. The flow rate of the mobile phase was 0.25 mL/min, and the injection volume was 1 μL. Eluent A was water and eluent B was MeOH (both containing 5 mM ammonium acetate and 0.1% acetic acid). Since the solubility of ammonium acetate in MeOH is limited, it was first dissolved in 25 mL water before MeOH was added. The total run time was 43 min, and gradient elution was employed starting at 3% B for 1 min, linearly increasing to 15% B in 15 min, to 79% B in 10 min, and finally, to 100% B in 13 min. After washing the column for 2 min with 100% B, the mobile phase was returned to the initial conditions and the column was eluted isocratically for 2.5 min. The column was regularly washed with 70% methanol to prevent cross-contamination. Calibration standards and samples were analysed in randomised order and intercepted with blank solvent samples to minimise analytical bias from sample positions and to reduce sample-to-sample carry-over.

#### *4.7. Validation of the Multi-Analyte LC-HRMS*/*MS Method*

The method was validated with regard to the guidelines established by the International Organization for Standardization [49,50]. The analytical selectivity was determined by the combination of LC retention time and high-resolution mass detection including dd-MS2 product ion qualifying of the different analytes. Measured peak areas were used for quantification. Sensitivity for the different analytes was expressed, by the slope of the respective six-point standard calibration curves (mean of three to four independent experiments) that were calculated by linear regression analysis in both solvent (MeCN 50:50) and the different matrices. The linear range was defined as the concentration interval, in which the regression coefficient *R*<sup>2</sup> was <sup>≥</sup> 0.96. Although internal standard calibrations were used for 15 of the analytes for the compensation of matrix interferences, potential suppression and enhancement (SSE%) of signals from the co-eluting matrix were estimated for all analytes as the ratio of the slope of the matrix-assisted standard calibration curve to the calibration curve in MeCN/water (50:50). If SSE values were above or below 100%, signal enhancement or suppression by the matrix could be assumed.

Considering the negligible noise in the extracted high-resolution mass chromatograms, the limits of detection (LOD) and limits of quantification (LOQ) of the 25 analytes were calculated based on the standard deviation of the y-intercept of the respective calibration curves and their corresponding slopes (m) as *LOD* = <sup>3</sup> <sup>×</sup> *SD <sup>m</sup>* , *LOQ* <sup>=</sup> <sup>10</sup> <sup>×</sup> *SD <sup>m</sup>* [59]. The accuracy of the method was assessed by determining recovery by spiking experiments and precision in terms of total within laboratory precision (RSiR) considering intra- and interday variabilities together [60]. Furthermore, coefficients of variation (% CV) were determined for all concentration points in the solvent and matrix-assisted standard calibration curves. Recovery rates were calculated for all analytes as the mean of three to four experiments at a spiking level of 25 μg/L. In a few cases, where the matrix-assisted standard curves in feed or fish matrices did not pass through the origin but showed a positive signal on the ordinate due to background noise, this was corrected by virtually moving the curve with parallel shift on the abscissa. The corresponding concentration difference was added to the spike concentration used in the recovery experiments according to Recovery(spike corrected) = (measured concentration − blank)/(spiked concentration + concentration difference to origin).

Measured results for fish feed and fish study samples were converted from concentrations (μg/L) into content in the respective matrix (μg/kg) by using the factors 0.1 for zebrafish and 0.125 for salmon and feed.

#### *4.8. Data Analysis*

The Q-Exactive was calibrated using Xcalibur software, version 2.2 (Thermo Scientific). The molecular formulas and exact masses of the target analytes were calculated using the built-in Qualbrowser of the Xcalibur 2.2 software, which was also applied for signal quantification. Microsoft Excel (Version 2016, Microsoft Corporation, Redmond, WA, USA) was used for basic statistics (e.g., calculation of mean, minimum and maximum values, regression and relative standard deviation).

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6651/11/4/222/s1, Table S1: Molecular characteristics of target analytes.

**Author Contributions:** Conceptualization, C.K.F. and L.I.; methodology, L.I.; software, A.J. and L.I.; validation, A.J.; formal analysis, A.J.; investigation, A.J.; resources, A.S.B., G.M.B. and L.I.; data curation, A.J.; writing—original draft preparation, A.J.; writing—review and editing, C.K.F., G.M.B., A.S.B., J.F. and L.I.; supervision, C.K.F., L.I and J.M.O.F.; project administration, C.K.F; funding acquisition, C.K.F. and J.M.O.F.

**Funding:** This work was supported by the Norwegian Research Council (RCN) as part of the projects SAFEFISH (RCN 254822), EPIGREEN (RCN 267944) and Amritha Johny's PhD scholarship grant.

**Acknowledgments:** We express our sincere gratitude to Anusha K. S. Dhanasiri at the Faculty of Biosciences and Aquaculture, Nord University, Bodø, Norway, for her substantial help with the zebrafish exposure study. We also would like to thank the technical staff at Nofima's Research Station for Sustainable Aquaculture, Sunndalsøra, Norway, for their valuable assistance in the salmon exposure study. Furthermore, we are thankful to Silvio Uhlig

in the Chemistry Section at the Norwegian Veterinary Institute, Oslo, Norway, for supporting the instrumental analyses performed in this study.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the result.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Perspective* **Aflatoxin Binders in Foods for Human Consumption—Can This be Promoted Safely and Ethically?**

#### **Sara Ahlberg 1,2,\*, Delia Randolph 1, Sheila Okoth <sup>3</sup> and Johanna Lindahl 4,5,6**


Received: 2 June 2019; Accepted: 12 July 2019; Published: 14 July 2019

**Abstract:** Aflatoxins continue to be a food safety problem globally, especially in developing regions. A significant amount of effort and resources have been invested in an attempt to control aflatoxins. However, these efforts have not substantially decreased the prevalence nor the dietary exposure to aflatoxins in developing countries. One approach to aflatoxin control is the use of binding agents in foods, and lactic acid bacteria (LAB) have been studied extensively for this purpose. However, when assessing the results comprehensively and reviewing the practicality and ethics of use, risks are evident, and concerns arise. In conclusion, our review suggests that there are too many issues with using LAB for aflatoxin binding for it to be safely promoted. Arguably, using binders in human food might even worsen food safety in the longer term.

**Keywords:** Aflatoxins; binding; food safety; biocontrol; food discipline

**Key Contribution:** Aflatoxin control by binders in human foods as a food safety measure raises concerns and risks not previously discussed. These issues have to be taken into consideration in research planning targeting improved food safety.

#### **1. Aflatoxins in Developing Country Food Chains with a Special Focus on Kenya**

Mycotoxins, including the important fumonisins, trichothecene toxins, zearalenone, and especially aflatoxins, have caused great concern in African and especially Kenyan markets over the last four decades. These mycotoxins are widespread, contaminating cereals, potatoes, bananas, cotton, and other plants. Additional mycotoxins, such as ochratoxins and patulin, are found in coffee, apples, and citrus fruits [1].

Aflatoxins are an important group of mycotoxins because there is strong evidence of their severe health impacts, causing liver cancer, especially among hepatitis B–positive people [2–4]. Extended exposure is implicated in immunodeficiency, immunosuppression, stunting, kwashiorkor, and interference with the metabolism of micronutrients in children [4]. High prevalence of aflatoxins in staples and consequently chronic exposure is common in regions where control and monitoring systems are poor and regulations are not enforced. Many studies find aflatoxins are present in high

levels in both feed and food chains in Africa, exposing consumers to aflatoxins, especially through staple foods [5].

Aflatoxins are produced by toxin-producing fungi *Aspergillus*, but fungal growth does not necessarily entail toxin production. Naturally occurring, there are non-toxic and toxic strains that produce aflatoxins at different levels [6]. Fungal growth and aflatoxin production are driven by climatic conditions. Any pre-harvest contamination of maize with *Aspergillus* fungi can lead to the accumulation of considerable aflatoxin levels when post-harvest conditions are adverse. However, post-harvest preventive measures against fungal contamination are more common than pre-harvest measures [7].

Acute aflatoxicosis is caused by consumption of large amounts of aflatoxins. This has occurred repeatedly in Kenya and other countries resulting in outbreaks with hundreds of human and thousands of animal deaths in the worst cases [8–10]. These widely reported cases have led to increased public concern and stimulated research efforts, policy changes, and investments into the research of suitable and effective mitigation interventions, and increased awareness of safety measures. However, these efforts have not been shown to decrease either the prevalence nor the dietary exposure to aflatoxins [9].

Kenya, a hot-spot of aflatoxins, has frequent, high, and not consistently improving prevalence of aflatoxins in staples and animal feeds. Aflatoxin studies report high proportions of cereals and feeds contaminated to some extent, and many samples exceed the allowable limits [8,11–14]. Likewise, fumonisins are found in almost all crops, often in co-occurrence with aflatoxins [8,15–22]. In consequence of the crop and feed contamination, almost all cattle milk is contaminated with aflatoxins [8,11–13,22–24].

Compared with other common foodborne hazards, aflatoxins are unusual because they can be formed only as a result of fungal infestation, usually at the farm level. This is exacerbated and spread by poor storage conditions. Once the aflatoxins are introduced, products are contaminated, and, if not removed from the chain at the control point when detected, they move further along the food chain and through processing. Heat treatments used in food production cannot eliminate the formed aflatoxins. Aflatoxins and other mycotoxins are invisible and can be detected only with modern analytical methods. However, if visible *Aspergillus* mould is present, this is an indicator of risk. The lack of control and monitoring in developing regions enables the supply of contaminated crops to reach the consumers.

Exposure to aflatoxins can be assessed through blood samples detecting albumin adducts or through detection of metabolites in milk or urine. Surveys report a wide range of exposure levels, from nondetectable to very high. Aflatoxin levels reported from Kenya during the 2010 outbreak were the highest ever reported (even up to 1200 pm/mg albumin) [5,10,25–27]. An indirect assessment of human exposure is the contamination level in food products.

Poverty is associated with poor availability and quality of foods, and this is also associated with aflatoxin exposure levels. Higher aflatoxin exposure levels were associated with the lowest socio-economic conditions in a study in Kenya, although all the women sampled were exposed [28]. In Africa, many small-holder farmers are women, who farm mostly for household consumption and informal markets and lack resources to avoid aflatoxin exposure.

Many mitigation methods have been suggested, from farm- to consumer-level interventions. Wild and Gong [29] have listed reasons for failures in aflatoxin control strategies. This list, which is relevant still a decade later, includes


• Aflatoxins are a multidisciplinary problem of agriculture, public health, and economics.

Staple foods in Africa are the most contaminated with aflatoxins and other mycotoxins. Promotion of healthy diets and diversification of food sources in the diet, e.g., increased diversity of legumes and vegetables, could be one significant way to decrease the levels of exposure. However, most people in Africa cannot afford diverse diets. Nonetheless, diversification of nutrient sources should be promoted, not only from the contamination exposure point of view, but also from agricultural and environmental diversity and nutritional perspective. Focus on staples and fungus-resistant maize can further decrease the promotion of diversity in diets and in agriculture, promoting further monocropping leading to decreased biodiversity levels, which are declining globally in alarming levels.

#### **2. Binding of Aflatoxins as a Biocontrol Method**

Novel approaches and new intervention methods focusing strongly on finding solutions to aflatoxin contamination have been called for. Risk mitigation and food safety improving measures have attracted funding resources, leaving other issues and problems, including other mycotoxins, behind. For example, aflatoxin research has benefited from a level of donor support disproportionate to the health burden it causes. According to the World Health Organisation and World Bank, aflatoxins are a relatively minor contributor to the overall health burden of foodborne disease [30,31], but the WHO report only includes the burden from hepatocellular carcinomas.

A specific approach to aflatoxin control is the use of aflatoxin-binding agents in foods. The principle is as follows: aflatoxins, which have contaminated foods, can be bound to an agent to mitigate the aflatoxin-induced health risks after consumption. Binders include bacteria cells, yeasts, proteins, and clays; the latter have been especially analysed for use in animal feeds. The hypothesis is that the binding agent and the bound toxin would pass through the gastrointestinal tract without, or at least with less, uptake and thus less damage caused by the toxins. Binding with lactic acid bacteria (LAB) is discussed below. Some other organic binding agents analysed have been yeasts [32,33].

Evidence of the binding ability of aflatoxins with LAB cells has been shown through a number of studies in laboratory conditions, some with 100% binding efficiency [32,34,35]. Binding is speculated to be an instant phenomenon [32,36–39], but also binding levels have been observed to increase over time [32,36,37,39–43].

Contrary to observed instant binding, some studies have reported no immediate binding at all [43,44]. Govaris [45] also noted several contradictions among the studies since the 1980s. Conflicting results have also been reported from storage studies. While Ahlberg [46] observed both increased binding over time and release of aflatoxins back to the matrix during 21-day trial, Barukˇci´c [47] and Govaris [45] reported binding levels to remain the same even for 21 days. Sokoutifar [44] recorded large amounts of aflatoxins bound to LAB strains up to 30 days at 21 and 37 ◦C. In practice, however, such high temperature conditions cannot be attained due to food integrity and safety risks.

While some authors reported increasing binding efficiency of LAB with increased aflatoxin concentration, others have reported decreasing effects or no difference or even both [34,41,45,46]. Binding has been shown to be dependent also on the concentration of the LAB cells [34,35,48].

Viability of bacteria strains has been considered a significant factor in binding. However, both viable and non-viable LAB strains have performed better in binding over the other in different studies, and no difference between the two has also been found [36,37,42,48–50]. These results have not brought clarity to the binding mechanisms, whether the binding effect is due to physical binding or influenced by the components produced by the bacteria.

Other factors affecting binding efficiency have been reported to occur in different food matrices such as milk or yoghurt, possibly explained by the compounds in the matrix [40,45], lower pH [45], or even higher pH [37,48]. In conclusion, external conditions seem to strongly affect the binding ability of aflatoxins by LAB.

One factor to consider in the binding analysis is the stability of the bound complex. Even simple washing can release 20–70% of the initially categorized bound aflatoxins back to detectable forms [32,34,36,37,39,51,52]. The stability of the formed bond is an important factor to assess the suitability of binding agents in food systems to reduce the harmful effects of aflatoxins.

As LAB are commonly used in dairy fermentation, the binding efficiencies of milk components have been studied. The milk protein casein is often speculated to be a binding agent in milk, the cheese making process being an indicator for this phenomenon due to the separation of whey and casein fractions. Aflatoxin binding has been concluded both to increase and decrease during cheese making [53]. In some of the binding studies, the controls without LAB cells show very low binding and reduction in aflatoxin shares (2–5%) compared with the binders [43,44]. These findings do not support the binding of milk components or casein to aflatoxins to be anything significant.

One of the first studies in binding concluded LAB removed as much as 80% of aflatoxins during cooking [17], which probably resulted a flourishing of interest in this research sector. Scientific evidence shows good potential in binding methods if certain criteria for evaluation are selected. However, when considering binding from wider perspective, serious concerns and problems arise, which have not been discussed or critically reviewed within these applications.

#### **3. Challenges with Interpreting the Results of Binding Aflatoxins with LAB**

Binding mechanisms and efficiency factors for LAB are not clearly understood and are considered still speculative in publications on binding. There seems to be no predictable factor affecting the binding efficiency and stability, resulting in the unpredictability and uncontrollability of the binding process. Optimal conditions for controlled and predictable binding have not been found. One factor can enhance binding shares in one study, but the same factor decreases the binding shares in another study. For example, the level of aflatoxin concentration is speculated to be one major factor in binding efficiency. It is especially important to bear this in mind because, as aflatoxins are contaminants, the levels and prevalence are unpredictable and vary significantly between batches, commodities, regions, and seasons. The approach to increase the safety of foods with aflatoxin binding with LAB cannot depend on the uncontrollable contamination level.

The binding analyses follow fairly simple procedures. Binders and LAB are mixed and possibly incubated in a liquid media (milk, broth, PBS, etc.) with aflatoxins. The mixture is then centrifuged, and the pellet is considered containing the bound aflatoxins attached to the LAB, as the free, unbound aflatoxins are considered remaining in the supernatant, the liquid media. It is possible that in this method the aflatoxins can be "trapped": physically pulled down by the other components of the binding analysis matrix to the pellet during centrifugation. This is even more likely when fermentation is taking place: LAB produce exopolysaccharides, high in molecular weight and large in structure constructing extracellular polymeric substances (EPSs) with proteins. These are partly responsible for the thickening of the product during fermentation. As any high molecular component will be pelletized during centrifugation, so are the fermenting products, which then can easily trap the aflatoxins and further falsely be detected as "bound".

For food safety purposes, both the binding efficiency and the stability of the formed bond are relevant. A weak formed bond releasing the aflatoxin would not have mitigation potential, despite the initial binding efficiency. If the binding phenomenon is only temporary, the suitability as a food safety method will not be relevant due to the uncontrollable conditions and risks induced. Several studies have reported how different levels of aflatoxins are released from formed aflatoxin and LAB complex under different conditions [32,34,36,37,39,51,52].

One major flaw in aflatoxin binding studies is the over-optimistic rhetoric used in the studies and conclusions. A number of studies observed binding in laboratory conditions with limited replications yet concluded it to be a suitable method of improving food safety. These conclusions contradict standard approaches to food safety measures, guidance, and regulations development, which would not support use of additives on the basis of inconclusive evidence. The phrase "aflatoxins could be removed" is often used in aflatoxin-binding studies, but in practice, the aflatoxins are still present in the food at the original levels, whether bound or not.

The analysis of binding of aflatoxins by LAB raises a question about the suitability of the methods. Aflatoxin contamination methods for screening contamination levels from foods uses the same analyses as the binding methods. These results of aflatoxin screening in different studies can sometimes show even higher aflatoxin contamination levels for the fermented food and milk products, which are incompatible with bound aflatoxins [23]. To further speculate, in principle, if the binding of aflatoxins to LAB, to milk components, or other food components occurs, all the analysed levels of aflatoxins from foods would be higher in reality than the given results indicate. Alternatively, it could be implied that the analysis methods for food contamination levels are not appropriate for the binding trials.

#### **4. Big Picture—Safe Food for All**

Promotion of aflatoxin binding at the consumer level of the food chain can signal to the producers and operators that the production of unsafe foods is acceptable as the problem could be solved later on. Such new principles can be extremely difficult to reverse later on, especially in poorly regulated markets. The awareness, knowledge, and practice of safety measures about mycotoxins and aflatoxins among farmers [7], producers, and consumers is limited [12,24,54], and promoting a method with uncertainties could easily create new misunderstandings and misperceptions of the causalities behind the contamination patterns and induced health risks.

Development of binders has taken a highly technological approach with little consideration to ethical, political, consumer acceptability, or legal implications. Yet using binders raises serious concerns and questions about risk, trade-offs, and entitlements that have not been discussed, let alone addressed. Without thoroughly understanding these aspects, it is likely that even if LAB is found to be technologically effective, it may not be adopted, or that if adopted, it could have unintended negative consequences.

Today, poor consumers patronising informal market chains cannot enjoy the same fundamental right to safe food as the wealthy consumers in formal markets in high income countries. In developing countries, market regulations, although inadequately implemented, mainly cover the formal markets, leaving informal markets unregulated [23,55]. In Kenya, among branded products sold in formal markets, lowest priced maize was 25% less likely to meet regulatory requirements for aflatoxins than the highest priced products [56]. Some indicators show that the situation might be worse in informal markets, but no systematic comparison has been done between the maize products sold in formal and informal markets in Kenya [56]. Aflatoxin exposure from milk among low-income consumers in urban Nairobi is higher than among mid-income consumers due to the higher aflatoxin levels in products sold in low-income areas and the higher milk consumption [23].

One effect of promoting fermentation with LAB to reduce aflatoxins in the informal sector could be the development of double standards in the food safety and food production systems. In principle, promoting different standards and procedures in different markets will create problems later in the upgrading and formalisation of traditional markets.

While aflatoxins are present in large parts of the world, high exposure levels in humans are mainly a problem in developing regions, and worst among poor purchasers. These people have often less access to information, and their understanding about the options, alternatives, and the relationship between actions at the beginning of the production chain and the consumption level may be lacking. Consumers in informal markets have limited access to the regulated markets without full market structure change.

Promotion of the use of aflatoxin binders in foods could potentially create new layers of problems. These have not received attention because the solution has been developed from a perspective of scientific functionality. Aflatoxins are by far the most studied mycotoxins [57], and when other mycotoxins start to gain more publicity, aflatoxin binding may appear inadequate as a solution. The role of social sciences should be promoted to create collaboration and multidisciplinary academic knowledge to develop new and suitable ways to work against aflatoxins and increase food safety [9].

Notably aflatoxin binding research has approached the issue from a one-component "silver bullet" solution instead of focusing on comprehensive food safety solutions at the farm and value chain level mitigating all the mycotoxins. Other mycotoxins are prevalent and occur together with aflatoxins. The binding solution is a rather simplified solution for a complex problem formed due to several factors and enchased by insufficient practices.

#### **5. Ethical Assessment to Improve Food Safety with Binders in Human Foods**

"Humans have a right to food free from mycotoxins that could cause significant health risk".

Declaration by the United Nations Environment Programme (UNEP) and the World Health Organization (WHO) International Programme on Chemical Safety (IPCS) [58]

An aflatoxin binding approach to foods inevitably requires testing and efficient analytical methods at consumer level. This is challenging for a number of reasons. First of all, promoting aflatoxin binding at the consumer level assumes people will deliberately be exposed to aflatoxins from contaminated foods and food products to assess the effectiveness, accepting something many citizens feel to be unacceptable. Second, aflatoxins are more of a problem in poorly regulated countries than in developed regions, especially among the poorest consumers.

Aflatoxins are one of the most regulated contaminants with allowable legal limits in commodities [59,60]. The role of the European Union in trading has pushed EU limits to be followed and adopted in regions with limited resources, creating a situation where limits are strict but resources are scarce to implement, monitor, and control the set limits. Also, the Codex Alimentarius has recommended limits for mycotoxins and aflatoxins, which can be adopted to national legislation [61]. The limits, whether reasonable and realistic from economic and trade perspective, are set to harmonize the safe food production systems to ensure the safety of the products.

The application of binders in human foods is in conflict with the principle of developed food safety regulations, set allowable limits, and regulation implementation and compliance by the operators. Even an emergency application can add new problems to the fragile and developing food safety systems. Officially approving a binder application in foods would be politically ambiguous. It is highly unlikely that developed regions, with strong regulatory systems, would allow aflatoxin binders at the consumer level as it is strongly against the principles of the current food regulatory systems. Implementing such binders legally only in developing regions with poor regulatory systems would raise concerns in terms of promoted double standards.

Clay supplements and LAB have been tested in human trials aimed to be used during an emergency aflatoxins outbreak situations [62,63]. In already poorly regulated regions, it would be challenging to keep the promotion of daily food safety measures and good practices separated from promoting a temporary solution or a quick-fix. It raises the concern at what threshold level would such an emergency outbreak be announced for the binding application in human foods to be "legal" or allowed. For example, in Kenya, many foods continuously contain aflatoxins above the allowable limits. Would high aflatoxin prevalence above legal limits permit the usage of binders in a specific time and region? Instead of supplementing binders to people in an aflatoxin emergency situation to encourage them to eat the contaminated, potentially high-risk maize, it would be more ethical to provide the replacement of safe maize for consumption.

#### **6. Who Can Choose What to Eat?**

A food safety method to control harmful contaminants should be robust, reliable, and functional in all conditions the contaminants are present. Also, to be suitable for the purpose, the methods needs to be available, feasible, understandable, and acceptable to the end users.

As the European Union is the largest economy in the world and a major trading partner for many countries, the EU legislation and standards are relevant globally. The EU has strict standards to ensure food safety and comprehensive regulation of practices to ensure the safety and quality of the products. It should be highlighted that set standards and limits alone cannot create food

safety, but the comprehensive food industry system from farm practices, through processing to consumers, all controlled and monitored by relevant institutions, can create a chain of controlled and traceable practices. This element of comprehensive approach is lacking in poorly regulated and in informal markets.

Where the legislation is set and executed throughout the practices in food production to protect consumers, unsafe products rarely enter the market chains and can be recalled if necessary. The binding application idea is also related to the food security status, but can the science and research community promote it in regions where people who have no institutional food safety protection, allowing them to consume foods that in regulated regions would be categorised unsafe and not fit for consumption? These are fundamental issues that should be discussed before any binding applications are taken to further testing. Figure 1 illustrates the separation between informal and formal markets and the most likely binder application channel and consequences.

**Figure 1.** The most likely application chain for binder method applied in informal markets focusing strongly to the consumer actually taking the risk. Implementation of binding method in formal markets would be highly unlikely as the approach conflicts strongly against the regulatory allowable limits set to the aflatoxins. Informal and formal markets currently are not equal and should be merged into formal markets to enable the same food safety standards, economic growth, and new value chains in one coherent food production system.

Consumers and end-users have very little influence on aflatoxin levels. Would consumers accept contaminated foods and milk for consumption with binding methods compared with better management at the farm- and supply chain–level to prevent the contamination altogether? Would poor and less-informed consumers be more approving toward the binding methods than informed, knowledgeable consumers who have more resources to understand the production chains and the consequences of the practices?

Judging from past trends, it is unlikely that the food safety standards and measures will be lightened. Consumers are increasingly conscious, information is ever more readily available, and consumers are demanding safer, high-quality foods produced sustainably, ethically, and fairly. Enabling and promoting the development of different food standards and measures in informal market sectors or poorly regulated regions is a very questionable approach to food safety, and the acceptability of binding applications should be brought to wider discussions from laboratories and the research community.

One of the most important questions in the binding applications should be, would you take it?

#### **7. Suggestions for Way Forward**

Using LAB to bind aflatoxins in foods may pose greater short- and long-term risks than benefits. Most important aspects are related to regulations, acceptability, and the creation of double standards when harmonized systems and merged markets are needed. Use of binding agents in foods contradicts all the existing principles and regulations set to ensure food safety. If such a method is promoted, the efforts to combat the aflatoxin problem at farm level and throughout the value chain, to eliminate and reduce the contaminants, could be compromised.

Aflatoxin control is not simple and needs a comprehensive approach covering food safety and economic development to address overall good farming and food production practices. Currently, food safety promotion through binders is discussed as an isolated factor, a magic bullet, to solve the problem. Over-reliance on technological solutions and inadequate attention to legal, ethical, political, and behavioural aspects of technologies as well as unintended consequences reduces the likelihood that agricultural innovations will have beneficial health and development outcomes. Now is the time to start addressing these neglected and important aspects of aflatoxin control.

Aflatoxin problems are prevalent especially in staples, and promoting diverse diets could reduce the exposure, especially from maize. Basically, all measures come with a cost, but creating new systems to promote increasing diversity in diets would directly contribute to diversity in crops in farming, creating resilience against climate change and unpredictable conditions. Promoting new value chains for staples and for a larger variety of plant and animal source foods can create new income sources for farmers while contributing to improved diets and decreased aflatoxin intake, directly contributing to a decreased public health burden from unsafe foods and unhealthy diets. When people become richer, they naturally diversify their diets, and aflatoxin exposure reduces. So, the promotion of development through economic and agricultural policy may be an indirect way of ending the scourge of aflatoxin [64]. Other public health approaches such as hepatitis B vaccination also have potential. Finally, the authorities' role to ensure the food safety in poorly regulated regions covering both informal and formal markets, but also promoting the merge of the two, should be strengthened significantly.

In final conclusion, there are too many issues with the aflatoxin binding methodology and results for it to be promoted. This review also highlights that binders for humans may be counter-productive for food safety.

**Author Contributions:** Conceptualization, S.A.; D.R.; S.O. and J.L.; Writing—original draft preparation, S.A.; Writing—review and editing, S.A.; D.R.; S.O. and J.L; Visualization, S.A.

**Funding:** This perspective received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Fumonisins: Impact on Agriculture, Food, and**

## **Human Health and their Management Strategies**

**Madhu Kamle 1, Dipendra K. Mahato 2, Sheetal Devi 3, Kyung Eun Lee 4, Sang G. Kang 4,5,\* and Pradeep Kumar 1,\***


Received: 4 May 2019; Accepted: 4 June 2019; Published: 7 June 2019

**Abstract:** The fumonisins producing fungi, *Fusarium* spp., are ubiquitous in nature and contaminate several food matrices that pose detrimental health hazards on humans as well as on animals. This has necessitated profound research for the control and management of the toxins to guarantee better health of consumers. This review highlights the chemistry and biosynthesis process of the fumonisins, their occurrence, effect on agriculture and food, along with their associated health issues. In addition, the focus has been put on the detection and management of fumonisins to ensure safe and healthy food. The main focus of the review is to provide insights to the readers regarding their health-associated food consumption and possible outbreaks. Furthermore, the consumers' knowledge and an attempt will ensure food safety and security and the farmers' knowledge for healthy agricultural practices, processing, and management, important to reduce the mycotoxin outbreaks due to fumonisins.

**Keywords:** Fumonisins; *Fusarium* spp.; food contamination; health issues; secondary metabolites

**Key Contribution:** This review gives insight into the occurrence of fumonisins, their outbreak and effects on human health, and agriculture and food along with their management strategies.

#### **1. Introduction**

Fumonisins are secondary metabolites produced in cereals by pathogenic fungi, namely *Fusarium verticillioides, Fusarium proliferatum*, and related species [1]. Moreover, *Aspergillus nigri* also produces fumonisins in the crop plants of peanut, maize, and grape [2–6]. The maize and maize-based products are most commonly infected with fumonisins besides their presence in several other grains (rice, wheat, barley, maize, rye, oat, and millet) and grain products (tortillas, corn flask, chips) [7,8] which have major influences on health. More than 15 fumonisin homologues have been known and characterized as fumonisin A, B, C, and P [9,10]. Further among fumonisin B, FB1, FB2, and FB3 are most abundant with FB1 being the most toxic form that can co-exists with other forms of fumonisin, i.e., FB2 and FB3 [11]. These (FB1, FB2, and FB3) forms are the main food contaminants. FB1 consists of a diester with propane-1,2,3-tricarboxylic acid (TCA) and 2-amino-12,16-dime thyl-3,5,10,14,15-pentahydroxyleicosane where hydroxyl (OH-) groups at the C-14 and C-15 positions involved with the carboxyl groups (-COOH) of TCA to form an ester. On the other hand, FB2 and FB3 are actually the C-5 and C-10 dehydroxy analogues of FB1 [12].

The toxins are linked with several health issues like cancer of the esophagus as evident from different regions of the world. Fumonisins are a very sensitive issue all around the world, which occur in Europe (51%) and Asia (85%) [13]. The occurrence of fumonisins with other related toxins in feed and food is reported in various countries like Argentina [14], Brazil [15], China [16], Italy [17], Portugal [18], Spain [19], Tanzania [20], and Thailand [21]. They are also reported to have toxic effects on the liver and nephron in all the tested animals [22]. In addition, FB1 is implicated with the incidences of hepatocarcinoma, stimulation and suppression of the immune system, defects in the neural-tube, nephrotoxicity, as well as other ailments. It is prominent as a promoter of hepatocarcinoma [23] where its synergistic interactions with aflatoxin B1 (AFB1) has been exhibited in animal models (rainbow trout and rats) for two stages, i.e., initiation and promotion of cancer [24–26]. The international agency for Research on Cancer (IARC) characterized FB1 as a group 2B possible carcinogen for human. Besides this, it can cause toxicity in several animals like rats, mice, and rabbits [27]. Further, a temporary maximum tolerable daily intake for fumonisins has been set as 2 μg/kg bw/day based on the lack of any observed adverse effects for nephrotoxicity in male rats by the joint Food and Agriculture Organization (FAO) and World Health Organization (WHO) [28].

#### **2. Major Source of Fumonisin**

Fumonisins are mainly produced by *F. verticillioides* and *F. proliferatum* and other *Fusarium* spp. The genus *Fusarium,* belonging to the family Nectriaceae, can be found as saprophytes in soil and plants worldwide [29]. *Fusarium* spp. colonize to the rhizospheres of plants and then subsequently enter into the plant system. Furthermore, *F. verticillioides* and *F. proliferatum* are known to be the most common pathogens of maize (*Zea mays*) [30]. Not only the crops, but also many popular ornamental plants (e.g., aster begonia, carnation, chrysanthemum, gladiolus, etc.) are frequently attacked by different *Fusarium* species, viz., *F. oxysporum, F. foetens, F. hostae*, and *F. redolens* at various stages of production [31].

*Fusarium*, on the other hand, infects orchids in both pathogenic and non-pathogenic forms. The non-pathogenic forms are either decomposers [32] or in mutual relation where they help in the germination of seeds and the color development of seedlings [33]. The non-pathogenic forms also help to mitigate the infection of *Fusarium* wilt on various crops [34]. Soils responsible for suppressing *Fusarium* wilt are found to be dominant in the *Fusarium* spp. like *F. oxysporum* and *F. solani* which are of agricultural importance [35,36]. The *Fusarium* species infect maize and produce fumonisins mainly at the pre-harvesting stage. Furthermore, fumonisin production has been observed during the post-harvest period; however, under adverse conditions of storage [37]. Dietary exposure of fumonisins can lead to several harmful outcomes in both farm and experimental laboratory animals. For example, these toxins are responsible for leukoencephalomalacia in horses [38], pulmonary edema syndrome in pigs [39], hepatotoxicity and nephrotoxicity in rats [40], and apoptosis in many other types of cells [41].

#### **3. Chemistry and Biosynthesis of Fumonisin**

Fumonisins (FBs) consist of two methyls (–CH3), one amine (-NH2), one to four hydroxyl (-OH- ), and two tricarboxylic ester groups located at different positions along with the linear polyketide-derived backbone. The biosynthesis step comprises the addition of two molecules of tricarballylic esters and one alanine-derived amine to a C-18 polyketide backbone [42]. FBs structural identity has been established, which are similar to sphingosine and are an integral part of cell signaling, growth, and communication [43]. It was believed that fumonisin formation could be controlled by disrupting the biosynthesis of sphingolipids [44]. The biosynthesis process of the toxin has been initiated to illustrate these cellular mechanisms and to design modified analogs [45,46]; however, to date, single total synthesis has been achieved by Pereira et al. [47]. There are intra-specific differences in the biosynthesis of fumonosins depending on the environmental conditions, e.g., temperature, the wavelength of light, humidity, and media composition for both the *Fusarium* spp. *F. verticillioides* and *F. proliferatum* [48]. Even the responses of strains were found to be different when the plant extracts were added from common hosts of *F. proliferatum* [49].

#### **4. Genes Responsible for Fumonisin Production**

Exploring the biosynthesis of trichothecene and fumonisin has revealed the gene cluster fumonisin biosynthetic gene (FUM in *Fusarium* and *Aspergillus*) which is responsible for the production of fumonisins, two transport proteins, and a transcription factor [50]. The expression of these genes is co-regulated and related to the FUM genes expression as well; however, it is influenced by ecological conditions [51,52]. The production of fumonisin is dependent on FUM1 which further expresses an enzyme complex known as polyketide synthase that catalyzes the initial step for fumonisin biosynthesis [53]. Furthermore, a positive correlation has been identified between the proportion of FUM1 transcripts being estimated by real-time RT-PCR and the proportion of fumonisins biosynthesized by the *F. verticillioides* and *F. proliferatum* species [54]. FUM19 lies at a distance of 35 kb downstream of the FUM1 gene that expresses an ATP-binding cassette responsible for exporting extracellular fumonisins [51]. Further, the expression of an aminotransferase by FUM8 functions to maintain the biologically active and mature FB1 molecule [55].

*A. niger* genome has a *Fusarium* FUM cluster homologue consisting of eleven homologues of the *Fusarium* genes namely fum1 (polyketide synthase), fum3, fum6, and fum15 (hydroxylase), fum7 (dehydrogenase), fum8 (aminotransferase), fum10 (acyl-CoA synthase), fum13 (carbonyl reductase), fum14 (condensation-domain protein), fum19 (ABC transporter), and fum21 (transcription factor) genes [56,57]. The FUM cluster in the *A. niger*, also known to have a dehydrogenase gene (sdr1), which is of a short-chain length, is absent in the *Fusarium* FUM cluster and its role in the process of fumonisin biosynthesis is unknown [56,57]. Further, the *Fusarium* FUM2 gene is also absent from the *A. niger* FUM cluster which causes hydroxylation at the C-10 backbone position of fumonisin [58]. The absence of a FUM2 homologue in the *A. niger* cluster has been seen to be consistent with other studies as well revealing that *A. niger* produces fumonisins (FB2, FB4, and FB6) only when it lacks a hydroxyl at C-10 [59–61]. In addition to these, genes like FUG1 and FST1 have been also confirmed to have an important role in fumonisin biosynthesis in *F. verticillioides* besides their role in maize kernel colonization [62,63]. Furthermore, Niehaus et al. [64] have identified 21 polyketide synthase (PKS) in the genome of the *F. proliferatum* where PKS3 and PKS11 are predicted to be linked with the biosynthesis of fumonisin.

#### **5. Occurrence in Food**

The contamination of foods by fumonisin depends on agroclimatic conditions (Table 1). The most commonly infected groups in food are the cereals (rice, wheat, barley, maize, rye, oat, and millet). The FB1 has been reported to contaminate numerous food products like asparagus and garlic [65], barley foods [66], beers [67], dried figs [68], and milk [69]. Maize (*Zea mays* L.) and maize-based products are one of the most commonly infected foods by FB1 [70]. Maize is used for manufacturing several products like tortillas and tortilla chips, corn flakes and corn starch, popcorn, grits, flour, and oils. However, the contamination by FB1 and FB2 is decreased by 59% during the manufacturing of tortilla chips from maize flour, while 60% for flour and 50% for grits and snack products due to the heat treatment by extrusion [71]. Further, several other products like cornflakes [72], the Portuguese maize bread [73], tea (black and herbal), along with some medicinal plants [74] have also been contaminated by fumonisins.



#### **6. E**ff**ects on Agriculture and Food**

Annually 25% of harvested crops are contaminated by mycotoxins, causing huge economic losses to agricultural and industrial commodities. These mycotoxins are stable in nature and do not eliminate during food processing, cooking, baking, roasting, and pasteurization. The meagre agricultural, as well as post-harvest practices like inappropriate drying techniques, handling procedure, packaging materials and methods, and storage and transport conditions, are responsible for the increased risk of fungal growth and fumonisin contamination [106]. Cao et al. [107] investigated the accumulation of fumonisins at different kernel developmental stages as well as during the drying of the kernel of hybrid varieties of white maize. They observed *Fusarium* (especially *F. verticillioides*) to be the most prevalent

genus for growth and contamination as compared to *Aspergillus* and *Penicillium*. The lower humidity of kernels favoured damage by insects along with fungal growth and accumulation of fumonisins [107]. The occurrence of fumonisins have been reported in edible plants like onion, garlic, asparagus, and pea seed [108,109]; in other cereals, mainly in wheat [84,110] as well as in crops like sorghum, beans (white, adzuki, mung), barley, soybean, asparagus spears, and figs [111,112]. Besides this, fumonisins have been found to impact the performance of aquatic animals like the Nile tilapia fingerlings and juveniles [113]. Fumonisins affected the hepatic expression of growth hormone receptor (GHR) and insulin like growth factor 1 (IGF-1) in these species, which is an indication that other aquatic animals and plants could also be affected by fumonisins posing a serious threat to food safety and security.

Fumonisins are an important class of mycotoxins produced by *F. proliferatum* and *F. verticillioides* along with others such as *F. napiforme, F. oxysporum, F. dlamini, F. nygamai* and *F. anthophilum* that are widely distributed, having potential health hazards to humans and animals [9]. These toxins are widely distributed in crops like corn, rice, sorghum, barley, and coffee. The exact causes of ear rot and kernel rot diseases is not well known but may be due to changes in weather such as dry weather followed by warm wet weather during flowering. The damage caused by the insect at the time of maturity allows strains present in nature to enter the ear and kernels. Rain before harvest may intensify the contamination of fumonisins in corn. Sometimes there are substantial amounts of fumonisins present in the non-symptomatic kernels of corn [114]. Yoshizawa et al. [103] reported the occurrence of fumonisins and aflatoxins in eighteen samples of corn in Thailand and found FB1 and FB2 and isolated *F. moniliforme* and *F. proliferatum* from the corn grit samples. Studies carried out in the USA reported the presence of FB1 and moniliformin in 34% of corn samples and 53% of corn-based food products, respectively [115]. A study in Brazil was conducted (during 2007–2010) to detect fumonisins in corn-based food products and reported that FB1 and FB2 were present in 82% and 51% of the examined products, respectively [116]. Contaminations by FB1 and FB2 observed in poultry broiler and feed fatting calves in South Korea [117]. Abdallah et al. [118] found the co-occurrence of FB2 and ochratoxin A and B in the date palm. In Brazil, it was reported that the production of fumonisins by *F. verticillioides* is found in both symptomatic and asymptomatic grains [119].

Furthermore, a survey was conducted in Japan for aflatoxin, ochratoxin A, and fumonisins contamination using HPLC and LC-MS. Results revealed that peanut butter is contaminated by aflatoxin, while orchratoxin A infection in oatmeal, rye, buckwheat flour, green coffee beans, roasted coffee beans beers, wheat flour, and wine. However, fumonisins were observed in popcorn, frozen corn, corn flasks, and corn grits [120]. Noonim et al. [59] analyzed the aflatoxin and fumonisin contaminations in different samples of Thai dried coffee, and it was noted that no *Fusarium* spp. were observed; however, *A. niger* was present in the coffee beans and produced fumonisins along with aflatoxins. A variable range of acetyldeoxynivalenol, deoxynivalenol, neosolaniol, fumonisin B1, and ochratoxin A contaminations were observed in Spanish coffee, and this variation was due to different methods of coffee brewing [121].

#### **7. Mechanism of Toxicity and Health E**ff**ects of Fumonisins**

#### *7.1. Mechanism of Toxicity*

FB1 predominates in 70% of the total FBs naturally occurring in infected food and feed samples [122]. FB1 express both acute and chronic symptoms in infected animals. FB1, though being an initiator of cancer, is non-genotoxic [123]. The major organs affected are liver and kidney; however, the severity of infection depends upon the strain and species [124]. The intestine, on the other hand, is a possible target for fumonisin toxicity [125]. FBs contamination has raised higher concern because of their interference with sphingolipid metabolism that ultimately leads to serious health concerns. Fumonisins are also linked to esophagal cancer and defects of the neural tube in humans [126]. Further, FB1 is the major causative agent for porcine pulmonary edema (PPE) [39], the toxicity of the liver and nephron in rodents [127], as well as cancers of the liver and esophagus in humans [128].

Franceschi et al. [129] studied the relationship between maize consumption and the risk of cancer of the upper digestive tract in the Pordenone Province in the north-eastern part of Italy. The population of this province has a high incidence of these neoplasms and shows particularly elevated levels of alcohol and tobacco use, in addition to high maize consumption. They observed that there were highly significant associations with frequent intake of maize emerging for oral cancer, pharyngeal cancer, and esophageal cancer. Dragan et al. [130] showed that the FB1 caused renal carcinomas in male rats and liver cancer in female mice. FB1 also induces apoptosis in many kidney cell lines, primary cell cultures, and also in vivo in rat liver and kidney [130,131]. Sun et al. [132] reported high contamination of FB1 in the food of the Huaian and Fusui city of China and suggested that FB1 may have a contributing role in human esophageal- and hepatocarcinogenesis. Further, Alizadeh et al. [133] studied 66 samples of both corn and rice from the Golestan province of Iran and observed high levels of FB1 contamination in both corn (223.66 μg/g) and rice (21.59 μg/g). They found a significant relationship between FB1 contamination in rice and the risk of esophageal cancer. Besides this, FB1 was found to be toxic to other cell lines. For example, FB1 triggers dose-dependent apoptosis and necrosis in esophageal carcinoma (SNO) cell lines in humans. Similarly, FB1 inhibited the activity of ceramide (CER) synthase, which is responsible for the acylation of sphinganine (Sa) and the recycling of sphingosine (So). This leads to an increment in the intracellular cytotoxic Sa-compound. Therefore, the variation of Sa/So proportions in urine and blood samples may denote the exposure of FBs in several animals; however, this has not been accurately validated [134].

#### *7.2. Health E*ff*ects of Fumonisin*

Equine leucoencephalomalacia first reported in 1891 is now revealed to be caused by consuming fumonisin-contaminated maize [135]. Further, the consumption of maize culture material infected by *F. verticillioides* [136] is responsible for the occurrence of porcine pulmonary edema (PPE) [30]. Since then, the outbreaks of PPE in the USA have been identified because of fumonisin infection. Further intake of fumonisin-affected diets by pregnant women causes neural tube defects in the developing fetus [126,137]. Sadler et al. [138] reported that FB1 has the potential to inhibit embryonic sphingolipid synthesis, produce embryotoxicity, and block folate transport and has been associated with increased prevalence of cancer and neural tube defects. On the other hand, Missmer et al. [126] reported the prevalence of neural tube defects (NTDs) doubled between 1990–1991 in Mexican–American women because they consume large amounts of corn in the form of tortillas, due to which they may be exposed to high levels of fumonisin. Fumonisin exposure increases the risk of NTDs and a dose above the threshold level may cause fetal death. Similarly, the exposure of fumonisin and its effect on esophageal and liver cancer is rare [132,139]. While no direct evidence of fumonisin hazard is found, its prolonged exposure may lead to cancer and birth defects in humans [140]. Moreover, the co-contamination of foods by fumonisin and aflatoxin has imposed risks of occurrences of outbreaks in southwest Nigeria [140], and the rural areas of Malawi in sub-Saharan Africa [141].

Besides this, the contamination of breast milk by fumonisins has been reported in several studies [142–144]. Recent studies have revealed the relationship between exposure to FBs and growth impairment in children [145–147]. According to Shirima et al. [146], fumonisin exposure negatively impacted child growth among children in Tanzania, which was confirmed based on urinary biomarker levels of fumonisin (UFB1). On the other hand, aflatoxin exposure had no significant impact on child growth. Furthermore, breastfeeding and weaning practices were considered to be associated with growth impairment in children due to exposure to FB1 [147]. The fumonisin carry-over has been observed in cow's milk as well [69]. Therefore, the incidence of fumonisin in human breast milk and its consumption by infants cannot be ignored, as the milk is a crucial part of infants' nutrition [148].

#### **8. E**ff**ects of Processing on Fumonisin**

Fumonisins are known to be comparatively heat-stable and affected only when heated above 150–200 ◦C during food processing techniques like baking, frying, roasting, or extrusion cooking. The degree of reduction in their chemical structure and toxicity depends on the cooking conditions and the composition of the food matrix [149]. However, this reduction could be due to the structural modifications of fumonisins while interacting with other components of food that leads to the conjugate's formation [150]. FB1 interacts with reducing sugars to form strong covalent bonds during heat treatments. For instance, FB1 reacts with D-glucose of corn grits during the extrusion cooking at 160–180 ◦C and forms a reaction product, N-(carboxymethyl) fumonisin B1 (NCM) [151]. However, the condensation reaction of FB1 and D-glucose forms N-(deoxy-Dfructos-1-yl) FB1 (NDF) [152].

Besides this, the wet milling causes the reduction of fumonisins to some extent in steep water. Further industrial milling processes reduce the fumonisin content significantly such that the fractions obtained (gluten, fiber, germ, and starch) are suitable for animal and human consumption [153]. However, during the dry milling process, there is a negligible reduction in fumonisin content as the fumonisins are embedded in the germ and pericarp in higher concentrations than in the endosperm and its derivatives [72,154,155]. Fumonisins are variably distributed in cereals and the fractions depending upon the type of cultivars, agricultural practices, and the method of milling processes [153,156]. The toxins might be degraded or modified during the processing of Tortillas at high temperatures and pH [157]. However, the industrial processing methods like roasting, frying, and extrusion cooking are effective in reducing the fumonisins to significantly low levels [158].

#### **9. E**ff**ects of Environmental Temperature on Fumonisin Production**

The two main factors impacting on the growth of fungus and the production of fumonisin are temperatures and water potential [159]. Therefore, the toxins are predominant in temperate and Mediterranean climatic regions [160–163]. The Mediterranean climate regions experience extreme temperature, rainfall patterns, as well as longer durations of drought. These conditions might lead to variation in the population of mycotoxigenic fungi and the fumonisin production by them which ultimately impacts the control strategies [164]. The infection of maize by *F. verticillioides* and accumulation of fumonisins is determined by the climatic conditions, insect damage, as well as the plant characteristics. The ear rot infection by *F. verticillioides* occurs during the flowering stage and is favored by warm and dry conditions; however, both warm and wet conditions following silking have been found to be favorable for disease development [165]. The weather conditions are critical for toxin accumulation during flowering as well as prior to harvesting [166,167]. It has been found that the less rainfall with maximum temperatures of 30–35◦C during flowering induces disease development [168].

Cendoya et al. [169] evaluated the effect of different levels of temperature and water activity (aw) on the fungal growth and fumonisin biosynthesis in wheat using three strains of *F. proliferatum.* Temperatures of 15, 25, and 30 ◦C and aw of 0.99, 0.98, 0.96, 0.94, 0.92, and 0.88 were evaluated. They found maximum growth of fumonisins at the highest aw of 0.99 at 15 ◦C for two strains while for the third strain, the maximum growth was observed at 25 ◦C at the same aw level. Furthermore, environmental factors like light along with nutrients available impacted the growth of *F. proliferatum* and the production of fumonisin [48,170]. In addition, Li et al. [171] evaluated the impact of pH levels on the growth of *F. proliferatum* culture. It was found that the toxin production was significantly inhibited in culture maintained at pH 5 compared to the culture at pH 10. However, the acidic pH 3–4, was found to enhance FB1 production by the fungus *F. proliferatum* [172].

#### **10. Detection Techniques**

The FB1 presence was detected by the Association of Official Analytical Chemists (AOAC) official method in food and feed samples. The derivatization was done using precolumn with ortho-phthaldialdehyde (OPA) and the detection by chromatographic techniques like HPLC (high-performance liquid chromatography) coupled with a fluorescence detector (HPLC-FLD). However, the drawback of this method is the use of high sample size (around 50 g), more extraction solvent (methanol:water), and solid-phase extraction (SPE) cartridges [173]. Therefore, methods like QuEChERS (Quick, Easy, Cheap, Effective, Rugged, and Safe) proved to be ideal for the detection of

FB1 [174–176]. Some of the commonly used techniques for fumonisin extractions include: (i) solid-liquid extraction (SLE) [177–180], (ii) liquid-liquid extraction (LLE) [181,182], (iii) matrix solid-phase dispersion (MSPD) [87,183,184], and (iv) dispersive liquid-liquid microextraction (DLLME) [185]. Recently, it was observed that the extraction yields were higher in finer flours indicating the importance of sample particle size on the recovery of fumonisins [11].

The traditional analytical methods to detect and quantify fumonisin include HPLC or UPLC (ultra-performance liquid chromatography) coupled with detectors such as UV–Vis spectrophotometric [186], fluorescence [187,188], and mass spectrometry (MS) [176,189–191]; liquid chromatography-mass spectrometry (LC-MS), and thin-layer chromatography (TLC) [192–194]. As these methods are expensive, tedious, and time-consuming [195], other advanced methods like the detection of mycotoxins producing fungi, enzyme-linked immunosorbent assay (ELISA), surface plasmon resonance (SPR), lateral flow immunoassay (LFI), immunosensors, electronic nose, and hyperspectral imaging are found to be more efficient [194,196]. Fumonisins producing genes have been amplified by PCR to detect *Fusarium* species in freshly harvested maize kernels [197]. PCR-based methods are used for the detection of mycotoxins producing fungal genera *Fusarium, Aspergillus*, and, *Penicillium* [198,199].

Recently, Nagaraj et al. [196] used a multiplex PCR technique to detect fumonisin producing *F. verticillioides* strains. ELISA coupled with PCR, i.e., PCR-ELISA by Omor et al. [200] for the detection of *F. verticillioides* based on the FUM21 gene in corn. In addition to this, a highly sensitive indirect competitive enzyme-linked immunosorbent assay (icELISA) and gold nanoparticle-based gray imaging quantification immunoassay (GNPs-GI) has been developed to detect FB1 in agricultural products [201]. Another important and non-destructive way of identifying toxigenic fungi in maize is by the application of hyperspectral imaging processes [202,203]. Besides this, the color-encoded lateral flow immunoassay (LFIA) has emerged as a leading technique for simultaneous detection of aflatoxin B1 and type-B fumonisins in a single test line [204]. Nowadays, electrochemical immunosensors are employed for rapid and sensitive detection of FB1 [205]. Furthermore, a rapid and ultrasensitive molecularly imprinted photoelectrochemical (MIP-PEC) sensing technique has been recently developed to measure FB1 [206].

#### **11. Masked Mycotoxins as a major concern in detection**

The masked mycotoxins issue was initially seen during the mid-1980s due to several mysterious cases of mycotoxicosis occurrence; however, the symptoms of mycotoxins in affected animals did not connect with the low mycotoxins content detected in their feed. At the same time, the metabolic biotransformation of deoxynivalenol (DON) to the less toxic derivatives *in planta* was first reported to appear in corn inoculated with *F. graminerium* [207] and also in naturally infected winter wheat [208]. In vivo studies for masked mycotoxins were carried out in pig and reported that zearalenone-14-glucoside was decomposed during the digestion process and zearalenone (ZEN) and zearalenol (ZEL) were detected in urinary and fecal metabolites [209].

During infection in plants, the mycotoxins produced by fungi are modified by plant enzymes and often conjugated to more polar substances, like sugars. These form of toxins are often less toxic metabolites stored in the vacuole in the soluble form or bound to macromolecules and are not detectable during routine analysis processes; therefore, referred to as masked mycotoxins [210]. These mycotoxins may not be a homogeneous group of contaminants but somewhat a complex mixture of different plant metabolites of various classes of mycotoxins and they are overall termed as the 'maskedome' [211]. Detection of masked mycotoxins is difficult as they change the physiological properties of their molecules leading to modified chromatographic behavior [212]. Due to less detectability, these toxins are a serious concern for food safety and these toxins may be converted back to the parent toxin forms during the food digestion process [213]. De Boevre et al. [214] analyzed cereal-based food products and raw feed materials for the presence of mycotoxins including deoxynivalenol, 3-acetyldeoxynivalenol, 15-acetyldeoxynivalenol, zearalenone, α-zearalenol (α-ZEL), β-zearalenol, and their respective

masked forms like α-zearalenol-1-3glucoside, zearalenone-4-glucoside, α-zearalenone-4-glucoside, β-zearalenone-4-glucoside, and zearalenone-4-sulfate in fiber-enriched bread, bran-enriched bread, cornflakes, popcorn, and oatmeal. Binder et al. [215] evaluated the absorption, distribution, metabolism, and excretion (ADME) of plant (ZEN-14-Glc, ZEN-16-Glc) and fungal (ZEN-14-S) ZEN metabolites in pigs and found that the total amounts of ZEN-14-GlcA, ZEN, and α-ZEL were excreted into urine after 0–48 hours of administration.

#### **12. Degradation Kinetics**

The degradation of FB1 was first revealed by Duvick et al. [216] to occur by microbes like *Exophiala spinifera, Rhinocladiella atrovirens,* and *Sphingomonas* or *Xanthomonas* having the capacity to metabolize FB1. These microbes were isolated from various tissues of maize. Further, the fumonisin metabolism by *E. spinifera* and the bacterium (deposited as ATCC55552 with the American Type Culture Collection) was studied by radiochemical and chromatographic (e.g., thin layer chromatography, TLC) methods. The initial two steps of biodegradation of FB1 were revealed to be due to de-esterification by a carboxylesterase releasing two tricarballylic acid (TCA) moieties leading to the formation of hydrolyzed FB1 (HFB1). The bacterial strain ATCC55552 further metabolized 14C-FB1 with the release of 14 molecules of CO2. However, *E. spinifera* could not further metabolize the TCA moieties. Blackwell et al. [217] later studied the oxidative deamination process of HFB1 by *E. spinifera* through TLC and mass spectrometry. They found that the HFB1 gets converted to Nacetyl HFB1 and 2-oxo-12,16-dimethyl-3,5,10,14,15-icosanepentol hemiketal. A cluster of genes in the bacterium ATCC55552 responsible for the degradation of fumonisin is mentioned in a patent, WO 00/04158 by Duvick et al. [218].

FB1 can be degraded to the less toxic form of hydrolyzed FB1 (HFB1) by an enzymatic process which could be used to reduce intestinal inflammation in pigs [219]. Further, the gene that catalyzes the oxidative deamination process of HFB1 in *E. spinifera* was revealed; however, the responsible enzyme for the deamination reaction is still unknown [218]. Later, Benedetti et al. [220] screened and isolated a bacterium related to the Delftia/Comamonas group (known as NCB 1492) from the soil. It was able to hydrolyze and deaminate FB1, but still, the sequences of the responsible genes are unknown. A year before, *Sphingomonas* sp. MTA144 was shown to have fumonisin degrading activity [221]. Further, Heinl et al. [222] identified two genes (carboxylesterase and aminotransferase) having prominent fumonisin-degrading activity. In addition to this, essential oils from plants were found to inhibit as well as degrade FB1 for example anise, camphor, cinnamon, citral, clove, eucalyptus, *Litsea cubeba*, and spearmint [223,224].

#### **13. Management and Control Strategies**

#### *13.1. Management and Control using Agricultural Practices*

As the crop plants like maize are infected by fumonisins during their growth in fields [225], the implementation of good agricultural practices (GAP), good storage practices (GSP), and good manufacturing practices (GMP) can mitigate the fumonisin contamination [226]. Harvesting the crop at earlier stages could be one of the strategies to control fumonisin contamination [227]; however, this cannot be applied to crops that need to be harvested at full maturity. Instead, the early harvest can be done for forage maize to increase the digestibility of silage. These practices require careful study as the farmers prefer a delayed harvest because of advancement in technologies. For instance, the use of kernel processors during forage harvesting leads to the production of digestible silage from maize when harvested at later stages [228].

Recently, the Codex Alimentarius Commission has set maximum levels of fumonisin at 4000 μg/ kg and 2000 μg/kg, respectively for raw maize and for maize flour and meal which have been implemented in South Africa. However, the lowering of fumonisin exposure in subsistence farmers need an integrated approach, and this cannot be solely achieved by regulatory measures [229]. Besides

these approaches, the use of nanotechnology and genetic engineering should be encouraged in the field of agriculture to develop resistant varieties of crops to get rid of *Fusarium* infection and FB contamination. The creation of drought and insect-resistant crops can also play a significant role in the fumonisin control as these factors are responsible, in one or the other way, for the fungal infection [230]. In addition to this, educating the farmers about the importance of drying and sorting out of the contaminated kernels from the crops can manage and control the risk of infection to some extent [231]. The in vitro study of combinations of fungicides (fludioxonil + metalaxyl-M) showed that it was not sufficient in the growth inhibition of *F. verticillioides* and even the increase in the production of FB1 by their strains [232]. A similar study also showed that these fungicides inhibit the growth and extracellular material formation but enhance the sporulation and fumonisin production in liquid culture of *F. verticillioides* [233]. Masiello et al. [234] reported that prothioconazole and thiophanate-methyl were effective in reducing the *F. graminearum* (52% and 48%) and *F. proliferatum* contamination (44% and 27%) under the field trial.

Fumonisin production and *Fusarium* growth are the result of interactions with various biotic and abiotic factors. In the case of abiotic factor temperature, water stress was the most significant environmental factor which influenced the fumonisin production and *Fusarium* growth. Several other stress conditions such as osmotic stress, pH, and fungicides were reported for the production mycotoxins [235,236]. *F. verticillioides* isolates were found to exhibit better performance at higher temperatures and under water stress conditions in comparison to *F. proliferatum*, another fumonisinproducing species. Marin et al. [237] suggested that environmental conditions leading to water stress (drought) might result in an increased risk of fumonisin contamination of maize caused by *F. verticillioides*. Drought stress and excess irrigation favor *Fusarium* infection. Drought stress should be avoided during the period of wheat seed development and maturation [238]. Excess moisture during the flowering seasons and early grain-fill periods also supports the *Fusarium* infection and moisture also increases the DON contamination [239]. Fungicide treatments were found to be effective against wheat *Fusarium* infection and DON contaminations [240,241]. Azole fungicides were found to be effective in the reduction of DON and other emerging and modified mycotoxins [242]. Therefore, an integrated approach, involving good agricultural management practices, hazard analysis, and critical control point production, storage management along with selected biologically based treatments, and mild chemical and physical treatments could reduce the fumonisin contamination effectively [243].

#### *13.2. Management and Control using Mycotoxin Binder*

Mycotoxin binders or adsorbents are substances that bind to mycotoxins and prevent them from being absorbed through the gut and prevent their entrance into the blood circulation. The mycotoxin binders can be helpful and utilized when other preventive measures fail against molds and mycotoxins [244]. The main aim of mycotoxin binders is to prevent the absorption of the mycotoxins from the intestinal tract of animals by absorbing the toxin to their surface. These binders may be organic or inorganic in nature, such as clay and yeast derived products, respectively [245]. However, mycotoxin modifiers are used to alter the chemical structure of mycotoxins and reduce their toxicity. These are microbiological in origin containing whole bacterial and yeast culture and specifically extracted compound such as enzymes [246]. In the field during harvesting of the crop, the production of mycotoxins can be reduced by choosing varieties that are adapted to the growing area and have resistance to fungal diseases. Mycotoxin production can also be reduced in the field by proper irrigation and balanced fertilizer applications [247]. These binders bind to the mycotoxins strong enough to prevent toxic interactions with the consuming animals and their absorption across the digestive tract. Potential absorbent materials include activated carbon, aluminosilicates (bentonite, clay, montmorillonite, zeolite, pollyosilicates etc.), complex indigestible carbohydrates (Cellulose, polysaccharides in the cell wall of yeast and bacteria such as glucomannans, petidoglycans) and other synthetic polymers such as cholestryamine and polyvinylpyrrolidone and derivatives [247]. De Mil et al. [248] characterized 27 feed additives marketed as mycotoxin binders and screened them

for their in vitro zearalenone (ZEN) adsorption. Recent studies showed that the addition of the commercial toxin binders to the aflatoxin B1 (AFB1) containing diets reduced the adverse effects of AFB1 and could be helpful as a solution to the aflatoxicosis problem in young broiler chicks [249].

#### **14. Conclusion**

The contamination of food and feed by fumonisin is a serious threat for disease outbreaks worldwide. The various techniques ranging from physical to biochemical as well as genetic engineering can be utilized in an efficient manner to mitigate fumonisin contamination of foods. However, a major issue of concern lies with the development of fungal and insect resistant crops to combat the fungal infection and fumonisin contamination. The naturally occurring soil microorganisms have been reported to have an immense capability of degrading and reducing the biosynthesis of fumonisins and its contamination in various agricultural crops. Moreover, the application of nanotechnology and genetic engineering should be given more emphasis to develop resistant varieties of crops and ensure the safety and quality of food for future generations.

**Author Contributions:** P.K. conceived and designed the manuscript; D.K.M., M.K., and P.K. wrote the manuscript; K.E.L. and S.D. helped in the editing of the manuscript; P.K. and S.G.K. critically reviewed the manuscript and did the required editing.

**Funding:** This work was supported by the Sunforce Inc. Republic of Korea.

**Acknowledgments:** All authors are highly grateful to the authority of the respective department and Institution for their support in doing this research. Author (P.K.) would like to thank the DST-SERB (file no ECR/2017/001143) and DBT-Twinning (No. BT/PR24741/NER/95/659/2017) for their financial support. This work was supported by the Sunforce Inc. Republic of Korea.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*

## **Co-Occurrence and Combinatory E**ff**ects of** *Alternaria* **Mycotoxins and Other Xenobiotics of Food Origin: Current Scenario and Future Perspectives**

**Francesco Crudo 1, Elisabeth Varga 2, Georg Aichinger 2, Gianni Galaverna 1, Doris Marko 1,2, Chiara Dall'Asta <sup>1</sup> and Luca Dellafiora 1,\***


Received: 6 September 2019; Accepted: 31 October 2019; Published: 3 November 2019

**Abstract:** Mycotoxins are low-molecular weight compounds produced by diverse genera of molds that may contaminate food and feed threatening the health of humans and animals. Recent findings underline the importance of studying the combined occurrence of multiple mycotoxins and the relevance of assessing the toxicity their simultaneous exposure may cause in living organisms. In this context, for the first time, this work has critically reviewed the most relevant data concerning the occurrence and toxicity of mycotoxins produced by *Alternaria* spp., which are among the most important emerging risks to be assessed in food safety, alone or in combination with other mycotoxins and bioactive food constituents. According to the literature covered, multiple *Alternaria* mycotoxins may often occur simultaneously in contaminated food, along with several other mycotoxins and food bioactives inherently present in the studied matrices. Although the toxicity of combinations naturally found in food has been rarely assessed experimentally, the data collected so far, clearly point out that chemical mixtures may differ in their toxicity compared to the effect of toxins tested individually. The data presented here may provide a solid foothold to better support the risk assessment of *Alternaria* mycotoxins highlighting the actual role of chemical mixtures on influencing their toxicity.

**Keywords:** *Alternaria* mycotoxins; combinatory effects; food safety; combined toxicity; co-occurrence; bioactive compounds

**Key Contribution:** This work provides for the first time an extensive and critical analysis of the most relevant literature concerning the occurrence and toxicity of *Alternaria* mycotoxins, studied either individually or in combination with other mycotoxins or bioactive compounds of food origin. Overall, this review pinpoints the need to investigate the simultaneous occurrence of diverse mycotoxins in food and to assess their combined toxicity to better support the risk assessment of *Alternaria* mycotoxins.

#### **1. Introduction**

Mycotoxins are low-molecular-weight toxic compounds synthetized by different types of molds belonging mainly to the genera *Aspergillus*, *Penicillium*, *Fusarium* and *Alternaria* [1]. They may enter the food chain worldwide as a consequence of the ability of mycotoxin-producing molds to infect a wide number of crops and food commodities [2]. It has been reported that up to 25 % of world crops may be contaminated with mycotoxins and over 4.5–5.0 billion people are thought to be chronically exposed

to these food contaminants [3]. However, a much higher prevalence of detected mycotoxins can be found depending either on the considered mycotoxin or crop (up to 80% in certain circumstances), as recently reported [4]. Although the highest levels of food contamination are more frequently found in low-income countries, mycotoxins actually represent a growing threat also on account of climate changes [5]. The contamination of food and feed by mycotoxins results in significant economic losses worldwide, not only in terms of food and feed spoilage, but also in terms of a burden on human health, animal productivity and international trade [6]. In particular, mycotoxins may pose a toxicological concern for humans and animals since they may exert a wide number of effects including acute toxic, mutagenic, carcinogenic, teratogenic, estrogenic and immunotoxic actions [7]. Among the various categories of mycotoxins, those produced by the genus *Alternaria* are gaining increasing interest due to their frequent occurrence in food, the recent insights on their genotoxic potential and mechanisms of action, and their consequent possible effects on human health [2,7]. The *Alternaria* toxins belong to the group of the so called "emerging" mycotoxins. They are compounds of possible concern due to their abundance, occurrence or toxicity, but the limited available data do not allow a comprehensive risk assessment with an acceptable degree of certainty.

*Alternaria* species are ubiquitous plant pathogens and saprophytes that may contaminate a wide variety of crops and raw materials due to their environmental adaptability, particularly to their tolerance to low temperature and water stress conditions. They produce a cocktail of secondary metabolites and more than 70 *Alternaria* toxins have been characterized so far [2]. Based on their chemical structures, *Alternaria* toxins may be divided into five groups (Figure 1): (i) dibenzo-α-pyrones, including alternariol (AOH), alternariol monomethyl ether (AME), and altenuene (ALT); (ii) perylene quinones, including the altertoxins I, II, III (ATX-I, ATX-II and ATX-III, respectively), stemphyltoxin I and III (STTX-I and STTX-III, respectively), and alterperylenol/alteichin (ALP); (iii) tetramic acid derivatives, including tenuazonic acid (TeA) and iso-tenuazonic acid (iso-TeA); (iv) *A. alternata* f. sp. *lycopersici* toxins, which includes several phytotoxins such as AAL-TA and ALL-TB sub-groups (v) miscellaneous structures, as tentoxin (TEN), which has a cyclic tetrapeptidic structure [2,8]. However, many other mycotoxins might be produced by *Alternaria* spp. such as dihydrotentoxin, isotentoxin, altenuisol (ALTSOH), altenusin, infectopyrone, altersetin, macrosporin A, altersolanol A, monocerin, altenuic acids I, II, and III [9].

Due to the broad spectrum of adverse effects observed in vitro (e.g., genotoxic, mutagenic, clastogenic, androgenic, and estrogenic effects) and in vivo (e.g., fetotoxic and teratogenic effects), some of the *Alternaria* mycotoxins most frequently found in food may pose a severe threat to human health, especially for the most exposed categories such as infants, toddler and vegetarians [10]. Nevertheless, for most *Alternaria* mycotoxins, neither the toxicity nor the occurrence in food is adequately described. The current limitation of data hinders the proper assessment of risks to human health and, consequently, it prevents the establishment of specific regulations [11]. Therefore, the need of additional representative data to support the proper risk assessment of *Alternaria* toxins, especially for AOH, AME, TeA, TEN and ALT, was claimed by the expert Committee "Agricultural Contaminants" of the EU commission in 2012 [12]. In 2016, a call to collect data for the human exposure assessment to *Alternaria* toxins (AOH, AME, TeA and TEN) was published by the European Food Safety Authority (EFSA) [13].

In this respect, the chemical risk assessment of food-related compounds is currently based on the integration of knowledge about the single exposure to a given substance and its potential to individually cause harmful effects [14]. However, food is typically contaminated simultaneously by more than one mycotoxin. It is noteworthy that the simultaneous occurrence of compounds (either toxicants or bioactive food constituents) may lead to combinatory interactions (namely, additive, synergistic or antagonistic effects) that may significantly change the final toxicological outcome depending on the overall composition of chemical mixtures (see Section 3.2). In addition, mycotoxins may be present in food along with a high number of bioactive compounds, showing a huge variety of chemical structures and mechanisms of action, which may further modify their toxic impact. On this basis, risk assessment

studies should take into account this complexity rather than relying on individual evidences, to better evaluate the overall risk associated with the consumption of mycotoxins-contaminated food.

Therefore, in the framework of supporting a better risk assessment of *Alternaria* mycotoxins, this work aims at consolidating the current knowledge on occurrence and combined actions of *Alternaria* mycotoxins. The relevance of investigating the effects and occurrence of chemical mixtures to support the thorough assessment of the actual risk this class of mycotoxins may pose to humans is pointed out. In more detail, this work presents the current state-of-the art in terms of co-occurrence and combinatory effects of: (i) different *Alternaria* toxins; (ii) *Alternaria* toxins in combination with other mycotoxins; (iii) *Alternaria* toxins in combination with bioactive compounds of food origin.

**Figure 1.** Chemical structures of the main *Alternaria* mycotoxins. AOH – alternariol; AME – alternariol monomethyl ether; ALT – altenuene; ATX-I, ATX-II, ATX-III – altertoxin I, II and III; STTX-III – stemphyltoxin III; TeA – tenuazonic acid; Iso-TeA – iso-tenuazonic acid; TEN – tentoxin; AAL-TA1-2 *Alternaria alternata* f. sp. *lycopersici* toxins sub-group A 1 and 2; AAL-TB1-2 *Alternaria alternata* f. sp. *lycopersici* toxins sub-group B 1 and 2; TCA - tricarballylic acid.

#### **2. Natural Occurrence and Co-Occurrence of** *Alternaria* **Mycotoxins in Food**

The occurrence of *Alternaria* mycotoxins in food and feed has been reviewed over the years [8,15–18]. However, in most cases, the occurrence and the relative concentrations of single or a small group of toxins has been reported, whilst the simultaneous co-occurrence of a high number of mycotoxins likely co-occurring together was not systematically assessed.

This section presents a collection of the co-occurrence of multiple *Alternaria* toxins in food commodities. In addition, data on the co-occurrence of *Alternaria* mycotoxins along with other mycotoxins and food constituents are reviewed. The key references covered in this review addressing the natural co-occurrence of different *Alternaria* mycotoxins are summarized in Table 1, while a schematic overview of the literature concerning the study of the co-occurrence of *Alternaria* and other mycotoxins is provided in Table 2. Detailed information concerning the number of samples analyzed, mycotoxin concentrations, as well as the methods and instruments used are reported in the Supplementary Materials (Table S1).

#### *2.1. Co-Occurrence of Di*ff*erent Alternaria Toxins in Food*

With regard to the co-contamination of food by different *Alternaria* toxins, AOH, AME, ALT, TeA, TEN, and ATX-I are the most frequently investigated compounds, while broader sets of compounds, including for instance ATX-II, IsoALT, AAL-TA1, AAL-TA2, ALP, macrosporin, ALTSOH, and Val-TeA, are rarely reported.

As shown in Table 1, the presence of *Alternaria* mycotoxins has been well-documented both in fresh and processed food, including fruits and vegetables, nuts, seeds, cereals, and fermented beverages. Among the food commodities investigated so far, apples, tomato, and their derivative products have been more frequently explored than other types of fruits and vegetables. Notably, most of them were found simultaneously contaminated by both AOH and AME, and, in some cases, also by up to five different mycotoxins. One of the first investigations were performed by Stinson and co-workers [19] who reported the contamination of apples and tomatoes with several *Alternaria* toxins already back in 1981. The observed contamination determined by HPLC-UV was in the low mg/kg range for AOH and AME, and in the μg/kg range for ALT and TeA in the case of apple samples. In tomatoes, TeA showed the highest contamination levels with up to 139 mg/kg. Furthermore, the presence or absence of ATX-I was assessed by thin layer chromatography. In the last ten years, multi-analyte measurements using liquid chromatography coupled to mass spectrometry became more and more important. The contamination with seven different *Alternaria* mycotoxins (AOH, AME, ALT, TeA, TEN, ATX-I, and ALP) and two phase-II metabolites (AOH-3-sulfate and AME-3-sulfate) was reported in tomato sauce, sunflower seed oil and wheat flour samples by Puntscher et al. [20]. In this study, the simultaneous contamination in the μg/kg range was reported in sample(s) from Austria, Croatia and Italy.

Infant foods were also found to be contaminated by multiple *Alternaria* mycotoxins. As an example, Gotthard and co-workers reported that tomato sauce and apple-pear-cherry puree were simultaneously contaminated by AOH, AME, TeA, and TEN [21]. In addition, those mycotoxins were also found in cereal-based infant formulas and they were reported along with ATX-I in wheat- and spelt-based food. These results are particularly relevant considering that the young population (infants and toddlers) show a higher exposure to *Alternaria* toxins in comparison to the other population categories due to their high food consumption in relation to body weight [10]. The most important dietary contributors to these mycotoxins were fruits and fruit products, vegetable oil, cereal-based foods and fruiting vegetables (tomatoes) wherein multiple mycotoxins were often found simultaneously, as shown in Table 1.

This scenario is further complicated by the possible presence of so called "masked mycotoxins". This term refers to modified forms of mycotoxins as a result of their metabolic transformations in plants. Masked mycotoxins have been reported to abundantly co-occur in contaminated food and raw materials along with their respective parent counterparts [22]. The most common masked mycotoxins covalently link sulfate or glucoside groups as a result of plant phase-II metabolism [23]. After ingestion, these phase II plant metabolites can be hydrolyzed during the digestion releasing the respective toxic parental compounds [1]. The transformation of masked mycotoxins to metabolites with higher toxicity than the parent compounds was also described in vitro [24,25], further highlighting the toxicological potential of the masked forms of mycotoxins (referred to as "maskedome"). Nevertheless, masked mycotoxins are not routinely screened, and this may result in an underestimation of the actual amounts of mycotoxins in foods. In this respect, Puntscher et al. [26] reported the presence of some modified forms of AOH and AME (i.e., AOH-3-glucoside, AOH-9-glucoside, AOH-3-sulfate and AME-3-sulfate) in tomato

sauce samples from Italy. In particular, one sample was found contaminated not only with AOH, AME, TeA and TEN, but also with AOH-3-glucoside, AOH-3-sulfate and AME-3-sulfate. Similarly, Walravens and co-workers found tomato products (juices, sauces and concentrates) contaminated with AOH-3-sulfate and AME-3-sulfate, with a prevalence ranging from 11% to 26% and from 32% to 78%, respectively [27]. The authors reported the highest prevalence of AOH and AME in tomato sauces (86% and 78%, respectively), while ALT was most frequently detected in tomato concentrates (56%). In addition, a prevalence of TEN-contaminated products, ranging from 21% to 64% (in sauces and juices, respectively), was also reported and, interestingly, all the tested samples showed a high contamination with TeA. More recently, another study highlighted the contamination of both fresh and dried tomato samples by different *Alternaria* toxins, among which TeA was found the most frequent and abundant compound [28].

The frequent co-occurrence of multiple *Alternaria* mycotoxins was also described in many other foods, including peppers. As an example, Gambacorta and co-workers [29] analyzed samples of fresh, dried, grounded, and fried sweet pepper, wherein AOH, AME, TeA, and TEN were found together (limit of quantifications in the low μg/kg range). In particular, TeA was detected in all samples, while AOH was detected in 86%, 43%, 100% and 14% of fresh, dried, grounded and fried products, respectively. Fresh pepper samples were mostly contaminated by AME (57% of fresh pepper samples), while fried peppers were the least AME-contaminated samples (14% of fried peppers samples). ALT was detected only in 43% and 13% of fresh and grounded samples, respectively.

Beside fruits and vegetables, cereals and derived products play an important role in the exposure to *Alternaria* toxins, representing the main source of exposure for infants and toddlers [10]. According to EFSA [2], the highest mean concentrations of AOH, AME, TeA and TEN in grains were observed as follows: AOH (spelt, oats, rice); AME (oats, rice); TeA (wheat, barley, rye, spelt, oats and rice); TEN (rye). Nevertheless, in addition to the above-mentioned mycotoxins, some authors reported also the presence of other compounds in grains, although the actual co-occurrence was not clearly specified. Specifically, ragi, sorghum and spelt were found contaminated by ALT [30,31], while ATX-I was detected in spelt and wheat [20,21]. Among the least investigated mycotoxins, macrosporin, which is produced primarily by the *Stemphylium* genus but it can be produce by *Alternaria* spp. too [32], was found in corn and wheat silage [33], while ALP was detected in wheat flour samples [20]. The presence of macrosporin was also detected in dried fruits and nuts, such as almonds, dried grape berries, hazelnuts, peanuts, and pistachios [34], often in combination with other *Alternaria* mycotoxins. In a study performed by Mikušová et al. [35], dried grape berries from three Slovak winemaking regions were simultaneously contaminated by up to eight *Alternaria* mycotoxins, i.e., AOH, AME, ALT, TeA, TEN, ATX-I, ATX-II, and macrosporin, whose highest concentrations were 1308 μg/kg, 776 μg/kg, 4120 μg/kg, 159.6 μg/kg, 43.1 μg/kg, 31175 μg/kg, 624 μg/kg, and 762 μg/kg, respectively. Notably, TEN was detected in all the analyzed samples.

*Alternaria* toxins can be found also in beverages such as fruit juices, beers and wines [36–41], as well as in food supplements used for various purposes [42]. Milk thistle-based supplements for liver diseases were simultaneously contaminated by AOH, AME, TEN, and TeA with maximum concentrations of 4560 μg/kg, 3200 μg/kg, 1280 μg/kg, and 2140 μg/kg, respectively. The same mycotoxins were detected, even though at a lower concentration, in supplements used to treat menopause symptoms (containing red clover, flax seeds and soy) or for general health support (containing among others green barley, nettle, goji berries and yucca). The maximum concentration of TeA was found in supplements for general health support (6780 μg/kg), while milk thistle-based supplements showed the highest average concentrations of all mycotoxins. Notably, the beneficial effects of health-promoting compounds of food supplements might be impaired to various extents by the presence of mycotoxins. In addition, taking into account that food supplements are thought to supply specific deficiencies, the presence of mycotoxins might have a higher impact on specific categories of consumers. These aspects require urgent investigations to timely support the enforcement of specific regulations.


**Table 1.** Co-occurrence of *Alternaria* toxins in food.

Tomato (dried) Tomato (puree and ketchup)

Tomato (sun-dried)

Tomato juice Tomato sauce Tomato sauce (puree infant formula) Tomato soup (puree infant formula)

Vegetable juice

 X a

 X

 X

 –

 X

 X

 –

 X

 X

 X

 X

 X

 X

 –

 –

 X

 –

 ALP (–)

 ALP (–)

 X

 X a

 X

 X X

 X a

X

 ALP (X), AOH-3-S (X a), AME-3-S (X)

 X

 X

 X

 X

 –

 X

 X a

–

 Iso-ALT (–), AAL TA1 (–), AAL TA2(–)

X a

X a

X a

[52]

 [46] [45]

 [38]

 [20]

 [21]

 [21] [36]




co-occurrence;

co-occurrence;

 –:

 not


#### *2.2. Co-Occurrence of Alternaria Toxins with Other Mycotoxins*

As discussed above, many food categories may be contaminated by more than one *Alternaria* mycotoxin. However, food commodities can be simultaneously contaminated by a high number of different mycotoxins produced by molds other than *Alternaria*. In particular, mycotoxins produced by *Aspergillus*, *Fusarium,* and *Penicillium* genera frequently co-occur with *Alternaria* mycotoxins (Table 2). Among them, the most investigated and frequently detected were those produced by the genera *Fusarium* and *Aspergillus* [e.g., aflatoxins, enniatins (ENNs) and beauvericin], while the least frequently examined or detected were ochratoxins (ochratoxin A, OTA; ochratoxin B, OTB).

A study conducted by Gambacorta et al. [29] investigated the co-occurrence of 17 different mycotoxins in fresh, fried, dried or grounded sweet pepper products. Notably, all of them were contaminated by more than one mycotoxin simultaneously. In more detail, 6 out of 39 samples contained 2, 3 or 4 different mycotoxins, while the remaining samples were positive for a number of mycotoxins ranging from 5 to 16. The fried peppers showed the lowest average level of contamination (with an average mycotoxin contamination of 231 μg/kg), while the fresh pepper samples were the most contaminated (27,280 μg/kg). TeA was the most frequently detected mycotoxin (100% of samples) with an average concentration of 4817.9 μg/kg. With regard to the other *Alternaria* toxins, 93%, 56%, 33%, and 9% of pepper samples were found to be contaminated by TEN, AOH, AME and ALT, respectively. These compounds (except for ALT) were found to co-occur along with 7 other *Fusarium* mycotoxins (nivalenol, HT-2 toxin, T-2 toxin, fumonisin B1, fumonisin B2, deoxynivalenol (DON) and zearalenone (ZEN)), 4 other *Aspergillus* mycotoxins (the aflatoxins B1, B2, G1, and G2), and OTA in the most contaminated sample. It is worth mentioning the average low level of contamination of fried samples. In this respect, the frying process might have a role in lowering the content of *Alternaria* mycotoxins, though it was not directly assessed by the authors. It would be in agreement with other studies pointing to a significant reduction of mycotoxin content upon fry cooking [62]. In addition, high-temperature treatments already proved to be effective in mitigating the content of certain *Alternaria* mycotoxins [63], supporting the possible role of fry cooking in reducing the content of *Alternaria* mycotoxins. The effects of three extrusion processing parameters (moisture content, feeding rate and screw speed) on the degradation of TeA, AOH and AME in whole wheat flour have been investigated. With the optimal parameters, a reduction of 65.6, 87.9 and 94.5% was achieved for TeA, AOH and AME, respectively [63]. As a general remark, the thermal stability of *Alternaria* mycotoxins needs to be further investigated, along with the possible formation of toxic by-products, to identify effective food processing for reducing their content in food.

The co-occurrence of AOH with the *Fusarium* mycotoxins ZEN and DON and, with the ergot alkaloid ergometrine was described in beer [64]. In particular, ergometrine, a toxin produced by *Claviceps* spp. used in pharmaceutical applications [65], was detected at low concentrations in 93% of the beer samples (0.07–0.47 μg/L, median 0.15 μg/ L). AOH (0.23–1.6 μg/L, median 0.45 μg/L) and ZEN (0.35–2.0 μg/L, median 0.88 μg/L) were detected in all the beer samples, while DON was found in 75% of samples (2.2–20 μg/L, median 3.7 μg/L). In the light of the low concentrations reported above, the authors concluded that beer should not be considered among the most important source of dietary intake of AOH, ZEN and DON.

In another study, 253 samples of dried fruits and nuts were analyzed for the presence of 16 mycotoxins (aflatoxins, ochratoxins, *Alternaria* toxins and trichothecenes) [57]. The authors reported that 124 samples were contaminated with at least one mycotoxin, while more than half (66 out of 124 samples) were contaminated by at least two mycotoxins. AME was the most frequently detected mycotoxin (44/124), followed by AOH (found in 31 out of 124 samples) and enniatin B1 (found in 30 out of 124 samples). The most contaminated sample contained eight different mycotoxins (i.e., aflatoxins B1 and B2, enniatins B and B1, beauvericin (BEA), TEN, AOH, and AME). Among the number of combinations found, the most common were binary (such as BEA + AME) and tertiary (such as BEA + AME + AOH) combinations. Ochratoxin B was found occurring along with the *Alternaria* toxins AOH, AME and TEN only in two samples.

The co-occurrence of ochratoxin A with AOH and aflatoxin B2 was described with a low frequency in berry juice (only 1 out of 32 samples was found positive) [37]. Additionally, although 47% of berry juices were negative for all the investigated mycotoxins, at least one mycotoxin was present in 53% of the samples, with percentage distributions of 9%, 9%, 22%, and 13% for 1, 2, 3 and 4 co-occurring mycotoxins, respectively. Moreover, TEN and aflatoxin B1 were not detected in any of the analyzed samples, while aflatoxin B2 + aflatoxin G2 + AME + AOH and aflatoxin G2 + AME + AOH were the most frequently found combinations. Importantly, in 87% of the contaminated samples at least one *Alternaria* mycotoxin was detected: AOH was most frequently found (73%; concentrations from 2.5 to 85 ng/mL) followed by AME (67%; concentrations from 267 to 308 ng/mL). Similarly, the co-occurrence of *Alternaria* toxins with other mycotoxins was also reported in dried fruit samples from China (apricots, raisins, dates, and wolfberries) [58]. In particular, 64.6% of the samples were contaminated by at least one mycotoxin, while 31.4% of the samples were contaminated with two to four compounds. TeA was the most abundant (from 6.9 to 5665.3 μg/kg) and frequently detected compound, followed by TEN (20.5% of samples) and mycophenolic acid (MPA; 19.5% of samples). MPA is produced by various *Penicillium* species and it is used as an immunosuppressant drug to prevent organ rejection after transplantation. In terms of safety, its occurrence in food may raise concern on account of its potential to predispose susceptible individuals to infectious diseases [58]. The combinations TeA + TEN and TeA + MPA were found with a prevalence of 13.2% and 11.4%, respectively [58]. In addition, TeA was simultaneously detected along with OTA in 7% of samples, with an apparently inverse relationship: the higher the concentration of TeA, the lower the concentration of OTA. This might be due to competition phenomena between mycotoxin-producing fungi or due to degrading processes, as reported by Müller et al. [66]. They described an inverse correlation between the increase of AOH, AME and TeA production and the decrease of *Fusarium* toxins (DON and ZEN) possibly due to the degradation of the latter by *Alternaria* strains. In this context, in vitro studies on the synthesis of mycotoxins during the co-incubation of *Alternaria* strains with other fungi may be useful to investigate the existence of a possible mutual influence, which seems likely to exist on the basis of low level of co-occurring mycotoxins reported so far in the literature.

As already reported in Section 2.1, food supplements might be highly contaminated by *Alternaria* toxins. However, *Alternaria* toxins can be found in food supplements also along with other mycotoxins. As an example, Veprikova and co-workers found 66 out of 69 samples contaminated by more than one mycotoxin. Specifically, 58% of milk thistle-based supplements contained more than 12 different mycotoxins simultaneously, while one of the most contaminated samples contained 14 different mycotoxins, i.e., AOH, AME, TEN, 3-acetyl-DON, beauvericin, fusarenon-X, ZEN, HT-2 toxin, T-2 toxin and enniatins B, B1, A and A1 [42]. The most common combinations described were ENNs + HT-2/T-2 + AOH + AME + TEN and ENNs + AOH + AME + TEN + MPA. As a general remark, the state-of the-art of food supplements contamination warns about a potentially dangerous scenario. Indeed, although to date no maximum limits of *Alternaria* mycotoxins have been defined for food, the relatively high concentrations of mycotoxins occasionally detected in food supplements might suggest the need to perform dedicated risk assessment studies. Therefore, further occurrence and exposure studies have to be done urgently paving the ground to timely enact specific regulations for food supplements.


195

n.a.: information not available.

Co-occurrenceof*Alternaria*toxinswithother

*Toxins* **2019** , *11*, 640

#### **3. Individual Toxicity of Main** *Alternaria* **Toxins and Combined Toxicity with Other Mycotoxins and Bioactive Compounds of Food Origin**

*Alternaria* species may produce a huge variety of different mycotoxins showing a great variability in terms of chemical structures [68]. AOH, AME, TeA, ALT, and altertoxins (I, II, III) are considered the most relevant for food toxicology, taking into account their occurrence and/or toxicity. Nevertheless, in vivo toxicological data currently available are not adequate for a proper risk assessment and, therefore, they are not sufficient to define toxicological standard values for the establishment of maximum limits in food and feed. At present, the only LD50 values are available, even if they refer to a limited number of compounds (Table 3).


**Table 3.** LD50 values of *Alternaria* mycotoxins currently available.

<sup>1</sup> LD50 values of AOH and AME were not reached at the maximum dose tested, corresponding to 400 mg/kg, <sup>2</sup> Unit of measurement: μg/egg, <sup>3</sup> Information about sex not available.

As a general remark, except for these few mycotoxins, very few data are available for the other members of the *Alternaria* mycotoxin family, which still remain largely uncharacterized in terms of toxicity and mechanisms of action.

As already discussed, the simultaneous occurrence of more than one *Alternaria* mycotoxin, also in combination with other mycotoxins produced by different fungi, is common in food. In this respect, it is important to remark that the risk assessment of mycotoxins currently relies on single substance effects [2,74], neglecting any possible mutual combined actions due to simultaneous exposure. These mycotoxin-mycotoxin interactions might modify the individual toxicity of compounds, likely resulting in a final toxic outcome different from the single compound tested alone. In addition, it must be considered that many extra-nutritional constituents (such as bioactive food constituents) are widely present in food, and their biological activity may also interfere with mycotoxin activity. The combined actions can be referred to as: (i) additive effects, when the final toxicity is the sum of the individual toxic effects of compounds; (ii) synergistic effects, when the resulting total toxicity is greater than the sum of individual effects or iii) antagonistic effects, when the opposite is the case and the combinatory effect is less than additive [75]. Several mathematical models and methods are commonly used to evaluate the nature of the combined effects of toxic compounds. Among them, the most common are the independent joint action model and the combination index-isobologram method. The first one allows to calculate an expected additive value from the effects of the single compounds [76] that can in turn be compared to a measured combinatory effect. The combination index-isobologram method

allows to take into account the shape of dose-response curves when determining the type of interaction (synergism, additive effect and antagonism) [75,77]. This is considered the state-of-the-art model; however it can be challenging to meet the requirements to apply it.

As shown below, the evidence collected so far clearly states that synergistic effects of mycotoxins in mixtures with other compounds (either mycotoxins or other food components) may have important consequences on the single-compound activity. This might have an impact on the assessment of risk related to the presence of *Alternaria* mycotoxins in food, which should consider the mixtures, rather than focusing on single-compound evidences. The individual toxicity of the main *Alternaria* mycotoxins and the effects of their combination with other mycotoxins or food constituents are reported in the following sections.

#### *3.1. Individual Toxicity of Alternaria Mycotoxins*

#### 3.1.1. Genotoxic Effects

Among the best characterized *Alternaria* toxins, those with genotoxic properties are considered of most concern for human health by regulatory authorities. This particularly applies to AOH and AME, for which the EFSA concluded that "the estimated mean chronic dietary exposures at the upper bound and 95th percentile dietary exposures exceeded the TTC value" in their latest exposure assessment [10], and thus called for more data regarding exposure and toxicity of those metabolites [13].

In human cells, both AOH and AME have been reported to induce DNA strand breaks in the comet assay at concentrations ≥1 μM [78], to act clastogenic at ≥2.5 μM [79] and to possess mutagenic potential at ≥10 μM, as measured by HPRT and TK gene mutation assays [80]. An in vivo study on mice did not find AOH to cause systemic DNA damages in liver tissue and bone marrow [81]. However, the authors argue that any toxicity of the substance would probably be limited to the gastrointestinal tract due to poor bioavailability, but did not include corresponding organs in their survey.

Concerning the mechanisms of action, both AOH and AME were found to act as a topoisomerase (TOP) poison at micromolar concentrations, affecting the activity of both TOP I and TOP II, with a certain preference for the α isoform of TOP II [78]. Those enzymes are needed to untangle the DNA for replication or transcription, a process which involves the induction of a transient DNA strand break that is re-ligated at the end of the catalytic cycle. Poisoning of these enzymes by small molecules results in a toxin-dependent stabilization of the covalent DNA–topoisomerase complex (i.e., the so-called "cleavable complex"). Stabilization of the cleavable complex by TOP "poisons" hinders release of TOP in the catalytic cycle and re-ligation of the DNA, thus resulting in a persistence of the initially induced strand break. Thus, TOP poisons are commonly described to act genotoxic [82].

An additional mechanism contributing to the toxicity of *Alternaria* toxins is the induction of intracellular reactive oxygen species (ROS), which indicate oxidative stress. ROS production induced by AOH and AME might play an important role in the inhibitory effects on cell proliferation observed in different cellular models [83,84].

Of note, ALT and iso-ALT were not found to affect topoisomerase activity [78], probably due to their less planar structure not allowing for DNA intercalation in comparison to AOH/AME [85].

However, it was observed that extracts from cultured *Alternaria* strains by far exceeded the genotoxicity of their dibenzo-α-pyrone contents [86]. This has led to the discovery of the epoxide-carrying perylene quinone ATX-II as a major contributing factor to the genotoxicity of naturally occurring mixtures of *Alternaria* toxins [87,88]. Later on, not only ATX II, but also the structurally related STTX-III was found to be more mutagenic then AOH. Regarding their mode of action, these mycotoxins were also found to act as inhibitors of TOPs at high concentrations. However, their main genotoxic mode of action is thought to be the formation of DNA adducts, a hypothesis which still awaits experimental confirmation [87–89].

Of note, there is speculation that yet not characterized secondary metabolites might also possess genotoxic properties, as an *Alternaria* extract very low on dibenzo-α-pyrones, which was additionally stripped off ATX-II and STTX-III, still maintained substantial DNA-damaging properties [90].

#### 3.1.2. Endocrine-Modulating and Other Toxic Effects

AOH and AME, as well as other related metabolites, were reported to elicit estrogenic effects in cellular systems. In particular, AOH was described to be able to activate both ER-α and β but with a greater affinity (approximately ten-fold higher) for ER-β [79,91], although the binding strength is 10,000-fold weaker than the endogenous hormone estradiol. AME was found to be slightly more potent than AOH at 10 μM, and the methylation at the 9-OH group was thought to improve the molecular fitting within the estrogen receptor pocket [92]. AOH was additionally found to induce androgenic effects in the yeast androgen bioassay [93]. Recently, computational studies reported that mutations of the androgen receptors might affect the capability of AOH to bind and possibly stimulate the activation of receptors [94]. Moreover, increases in progesterone and estradiol levels, as well as in progesterone receptor expression, were reported in human adrenocarcinoma H295R cells treated with AOH, supporting its actual role as endocrine disruptor [95]. However, in naturally occurring mixtures of *Alternaria* toxins, endocrine-disrupting effects of AOH and related metabolites might be "quenched" by cytotoxic and anti-estrogenic properties of co-occurring compounds, as recently demonstrated in Ishikawa cells [90].

In addition to the above listed toxic effects, AOH and AME were found to modulate innate immunity in both human bronchial epithelial BEAS-2B cells and mouse macrophages RAW264.7, through the suppression of the lipopolysaccharide-induced innate immune responses [96]. This activity was also confirmed in THP-1 derived macrophages by Kollarova et al. [97]: AOH, in fact, suppressed lipopolysaccharide (LPS)-induced NF-κB pathway activation, induced transcription of the anti-inflammatory cytokine IL-10, and reduced the transcription of the pro-inflammatory cytokines IL-8, IL-6 and TNF-α.

TeA deserves a particular mention as, unlike the other *Alternaria* mycotoxins, it exerts toxic effects mainly by inhibiting the release of proteins from the ribosome. Although a low toxicity of this mycotoxin has been reported in vitro [86,98], in vivo studies carried out on several animal models highlighted more severe effects such as emesis, tachycardia and haemorrhages [18].

#### *3.2. Combinatory E*ff*ects of Alternaria Mycotoxins*

There are only a few studies investigating the combinatory effects of *Alternaria* mycotoxins, though food may be quite often simultaneously contaminated by more than one single compound as shown, for instance, for AOH and AME (Section 2.1). Notably, these two mycotoxins are not of particular concern in terms of cytotoxic effects, also on account of the high concentrations required to cause harmful effects when tested individually. However, the simultaneous exposure to AOH and AME may have significant effects on the overall toxicity in respect to their individual testing. In more detail, their combined effects (1:1 concentration ratio) were invested by Bensassi et al. on the human intestinal cell line HCT-116 [99]. No significant difference in cell viability was detected at 25 μM up to 24 h of exposure when mycotoxins were tested either individually or in combination. Conversely, both mycotoxins reduced cell viability about 30% after 24 h of exposure when tested individually, while they reduced viability about 50% when tested in combination. In this study, the nature of interactive effects was described to be additive, while Fernández-Blanco and co-workers reported synergistic effects in Caco-2 cells after 24 h of exposure to AOH and AME in a 1:1 binary combination and in a concentration range from 3.125 to 30 μM [83]. Moreover, the AOH-AME binary combination reduced cell proliferation to a greater extent than AOH alone at all tested concentrations, while it had greater effects than AME alone at 15 and 30 μM. The binary mixture also caused a greater dose-dependent reduction of cell proliferation after 48 h of incubation (in the concentration range 7.5–30 μM) than AOH

or AME tested alone. In this case, the nature of the interactive effects was described as synergistic or additive at small or higher fraction affected, respectively.

The effects exerted by the simultaneous exposure to AOH and the genotoxic *Alternaria* mycotoxin ATX-II were investigated by Vejdovszky et al. [100] on two intestinal (HT-29, HCEC-1CT) and one hepatic (HepG2) cell line. Seven different concentrations, ranging from 500 nM to 10 μM for ATX-II and from 5 μM to 100 μM for AOH were tested for binary combinations (constant ratio of 1:10, ATX-II:AOH). As a result, the HT-29 cell line was found to be the least sensitive to cytotoxic effects mediated by the two tested mycotoxins and significant differences in cell viability were found starting from the combination 5 μM:50 μM (ATX-II:AOH). Among the different concentrations tested, the highest decrease in cell viability observed was of nearly 40%. Notably, the cell treatment at low mycotoxin concentrations led to an increase in mitochondrial activity in the co-treated samples. HepG2 cells were found the most sensitive to the cytotoxic effects exerted by AOH, while HCEC-1CT cells proved to be the most sensitive to the effects of ATX-II. Combining these two mycotoxins, an increased sensitivity to cytotoxic effects was also found in the HepG2 cell line, leading to a reduction in cell viability starting from the combination 1 μM:10 μM (ATX-II:AOH). Although most of the tested 1:10 combinations showed additive effects, antagonistic effects were reported in HCEC-1CT and HepG2 cell lines, while only one of the combinations analyzed showed synergistic effects on HepG2 cell line (750 nM ATX-II:750 nM AOH, 1:1 ratio). Modifications of microRNAs expression profile after incubation of HepG2 cells with the mixture 10 μM AOH:1 μM ATX-II may partially explains such effects. The combined exposure caused a significant increase of miR-224 expression after 12 h of exposure, which was no longer over-expressed after 24 h, while miR-192 and miR-29a were respectively down-regulated and up-regulated after 24 h. In addition, miR-29a was up-regulated also in samples treated with AOH alone, suggesting a possible role in the up-regulation of this miRNA by the binary mixture. Interestingly, these three microRNAs are involved in the regulation of apoptotic processes and the observed modifications led the authors to conclude that such miRNAs may be in part involved in the antagonistic effects observed for some of the combinations tested.

As previously described, *Alternaria* mycotoxins are often found in food commodities along with *Fusarium* mycotoxins. In a recent study [101], the cytotoxic effects and the type of interactions of AOH combined with *Fusarium* mycotoxins enniatin B and DON were evaluated after 24, 48, and 72 h of exposure in Caco-2 cells. For binary and tertiary combinations, five different concentrations, ranging from 0.3125 to 5 μM for enniatin B and DON, and from 1.875 to 30 μM for AOH, were tested. The binary combinations enniatin B + AOH (1:6 ratio) led to higher cytotoxic effects compared to AOH tested alone at all the timepoints and concentrations tested. However, no difference between enniatin B tested alone and in mixture was observed, suggesting that the cytotoxic effects were mainly mediated by enniatin B. With regard to the binary combinations DON + AOH (1:6 ratio), the resulting cytotoxicity after 24 h of exposure was lower than that exerted by DON tested alone. On the contrary, an opposite trend was observed after 48 and 72 h of exposure. As expected, the tertiary mixtures enniatin B + DON + AOH (ratio 1:1:6) led to a greater decrease, albeit of slight intensity, of cell viability compared to the binary combinations. Although the pattern was not uniform along the fraction affected, the application of the isobologram analysis described the interactions in the binary mixtures as additive and synergistic, depending on the concentrations and timepoints tested. Interestingly, the ternary combinations showed antagonistic effects, which were described as due to competition mechanisms at the same receptor site. In this respect, it is worth mentioning the marked diversity of these mycotoxins in terms of chemical structures. Taking into account that the competition to the same protein site usually requires strict conservation of key structural motifs [102], the inherent structural heterogeneity among enniatin B, DON and AOH is not fully compatible with their capability to physically compete with the same site. Therefore, both the molecular mechanisms and the network of biological targets involved in such antagonistic behavior need to be precisely described to better understand the effects of the enniatin B/DON/AOH ternary combination.

The effects of binary and ternary combinations of AOH with the DON's acetylated derivatives 3-ADON and 15-ADON were also investigated on HepG2 cells up to 72 h of incubation [103]. Constant ratios of 16:1 (AOH: 3-acetyl-ADON and AOH:15-acetyl-DON) and 16:1:1 (AOH:3-acetyl-DON:15-acetyl-DON) were chosen to test these mixtures, with concentrations ranging from 3.2μM to 24μM for AOH, and from 0.2 μM to 1.5 μM for DON's derivatives. Cytotoxicity ranking was found to be the same for all tested time points (AOH+3-acetyl-DON + 15-acetyl-DON > AOH + 3-acetyl-DON > AOH + 15-acetyl-DON) and a concentration-dependent decrease in HepG2 cell viability was found in all tested mixtures. The effects caused by binary and ternary mixtures were described to be mainly synergistic, but some exceptions were found for AOH + 3-acetyl-DON at 72 h (where additive effects were observed at higher fraction affected), and for AOH + 15-acetyl-DON (where additive or antagonistic effects were observed depending on the concentration and timepoint tested).

Binary effects of TeA with the *Fusarium* mycotoxins enniatin B, ZEN, DON, nivalenol and aurofusarin (AURO) were also evaluated on Caco-2 cells with two different concentration sets, named "low concentrations" (none or slight cytotoxic effect) and "high concentrations" (pronounced cytotoxic effect) [104]. TeA combinations at "low concentrations" of mycotoxins did not show significant differences between the measured and expected effects (calculated on the basis of the Independent Joint Action model). This indicates that the combinations of TeA at "low concentrations of mycotoxins" only determined additive effects. On the contrary, binary combinations at "high concentrations" led to lower cytotoxic effects then the calculated additive effects. Additional investigations allowed getting more details about the type of interactions between TeA and *Fusarium* mycotoxins. No difference in cytotoxicity was found in samples co-treated with enniatin B and ZEN keeping the concentration of *Fusarium* mycotoxins constant (from 5 to 50 μM depending on the mycotoxin) and varying that of TeA (from 1 μM to 250 μM). Indeed, the cytotoxicity of binary mixtures with TeA was found to be equivalent to the toxicity of toxins tested individually. Notably, the toxic effect induced by 10μM DON was reduced in a concentration-independent manner by the combination with TeA at concentrations between 10 μM and 200 μM. A similar trend was found for the combination with 10 μM nivalenol, although differences were not statistically significant. Keeping in mind that TeA and the *Fusarium* mycotoxins DON and nivalenol are known to inhibit protein synthesis in vitro [104], the lower cytotoxic effects of binary mixtures might be due to a molecular interplay at the level of protein synthesis inhibition. Nevertheless, considering that nivalenol and DON inhibit protein synthesis by different mechanisms (i.e., by inhibiting the initiation or elongation-termination steps, respectively) [105], the observed effects cannot be straightforwardly explained in terms of mechanisms of action pointing out the need of investigating further the molecular basis of such interaction. In this respect, the inhibition of protein synthesis by TeA may modify the expression of specific factors, including metabolizing enzymes, and consequences on the pattern of metabolites produced by cells are thought likely. This is of particular relevance as some trichothecenes metabolites might be involved in mediating ribotoxic effects of parent mycotoxins, as supported recently by in silico studies [106]. On this basis, TeA might have indirect effects on trichothecenes toxicity acting on their metabolism and changing the relative abundance of ribotoxic metabolites produced.

Recently, an interesting study was performed by Solhaug et al. that investigated the ability of AOH, DON and ZEN in binary and tertiary mixtures to affect immune response checking the differentiation of monocytes to macrophages [107]. The differentiation process leads to several changes, including modifications of the expression of some cell surface markers such as CD14, CD11b and CD71. AOH, DON and ZEN were able to modify the expression of these markers in THP-1 monocytes, but with some differences: while AOH affected the expression of the all set of markers, DON did not modify the expression of CD71 and ZEN altered only the expression of CD-14. Since CD-14 was the only marker modified by all the three mycotoxins, its expression was used to evaluate the type of interactions in binary and ternary mycotoxins mixtures by applying the "Concentration Addition" (CA) and the "Independent Joint Action" (IA) models. Since authors did not find significant differences

between the experimental data and the predicted models, the type of interaction was described to be additive. Remarkably, at the lowest concentrations of the AOH + ZEN combination, the confidence interval of the predicted CA model did not overlap with the experimental values, suggesting a possible synergistic effect. The same results were obtained for the binary combinations through the application of the isobologram analysis. To verify if the observed inhibitory effects of AOH, DON, and ZEN on the up-regulation of CD14 led to a real reduction in macrophage activation, the pro-inflammatory cytokine TNFα and its gene expression were quantified after incubation with single mycotoxins. Contrary to what was observed for AOH and ZEN, DON induced an increased secretion of TNFα following the increase of TNFα gene expression, in spite of its inhibitory action on the up-regulation of CD14. The expression of NF-kB, a protein complex involved in TNFα expression, might provide a plausible explanation to these differences. Indeed, ZEN was reported to reduce the expression of NF-kB [108], and, recently, also AOH showed the ability to suppress the lipopolysaccharide-induced NF-kB pathway activation, resulting in the reduction of TNFα [97]. In contrast, DON was found to induce both NF-kB activation and TNF-α expression, but the signaling pathway was different from those activated by ZEN and AOH [109].

#### *3.3. Combined E*ff*ects with Bioactive Food Constituents*

Beside the combined action of the different members of *Alternaria* mycotoxins group, also in combination with mycotoxins produced by fungi other than *Alternaria*, it is important to take into consideration even the complex interactions that these mycotoxins may have with the other bioactive compounds of food origin.

In this contest, Vejdovszky et al. recently investigated the combinatory estrogenic effects of the isoflavone genistein (GEN) in combination with ZEN and AOH [110]. To elucidate the combinatory effects, the human endometrial adenocarcinoma Ishikawa cell line was chosen as a model system and the phosphatase alkaline (ALP) activity assay was used to measure estrogen receptor activation. The xenoestrogens under investigation were tested at different concentrations (ranging from pM to μM) after 48 h of incubation. All of them increased the ALP activation when tested individually, with the following order of potency in terms of EC50: E2 (17β-estradiol; used as positive control) > ZEN > GEN > AOH. Moreover, these xenoestrogens did not only differ in terms of potency, but also in terms of efficacy as none of them (at any concentration) was able to determine the same effects induced by 1 nM E2. A possible explanation for this finding is that AOH, ZEN and GEN might act as partial agonists. The lower capability to satisfy the pharmacophoric requirements of estrogen receptors pockets in comparison to E2 [111,112] might provide a structural rational to explain such evidence. With regards to binary mixtures of GEN with ZEN or AOH, some of them resulted in significantly higher effects than the respective compounds tested individually, clearly pointing out the existence of synergistic effects. However, combinations of GEN-AOH activated ALP to a lower extent than ZEN-AOH mixtures. It must be highlighted that in many studies ZEN was found to be more estrogenic than AOH, and this could partly justify the lowering of estrogenic effect observed in combinations [110]. In addition, while the authors noted the preference of AOH and GEN to ERβ, ZEN was previously described with a higher affinity for ERα [113]. The simultaneous activation of both α and β estrogen receptor isoforms in the ZEN-GEN and ZEN-AOH binary mixtures may explain the stronger synergistic effects observed. Although some GEN-AOH combinations showed synergistic effects, other combinations at very low doses led to antagonistic effects. Indeed, anti-estrogenic effects were found testing the combination 0.001 μM GEN-0.1 μM AOH and observing a reduction of ALP activation (10.9%) compared to the control (vehicle). A subsequent more-in-depth analysis of the combinatory effects, performed through the combination index and the isobologram method, allowed to determine the type of interactions occurring in the different combinations. Both methods showed that the combinatory effects of GEN and ZEN in the constant ratio of 1000:1 were mainly synergistic and, only at very low or very high effect levels, additive or antagonistic effects were observed. In the constant ratios of 100:1 and 10:1, the substances led to a strong antagonism at low effect levels, and to a strong synergism at higher effects. Comparable outcomes were reported for the 1:10 GEN:AOH ratio (which showed antagonistic or synergistic effects at low or high effect levels, respectively), while the 1:5 combination ratio determined mainly antagonistic effects. Additionally, the 1:1 GEN:AOH ratio resulted in the onset of synergistic effects up to about 65% of the maximum ALP activation observed (E2 1nM). Above, additive or antagonistic effects were observed depending on the concentrations tested. Thus, the nature of the interactions seemed to depend on both the ratio of substances and the specific concentrations tested.

It was also established that AOH is able to cause oxidative stress and to exert genotoxic effects in different cellular models, mainly by acting as a topoisomerase poison [78]. Aichinger et al. investigated the effects of AOH in combination with the two polyphenols GEN and delphinidin (DEL) [114]. These two compounds are known for their antioxidant effects at specific concentrations, although pro-oxidant effects at certain concentrations were also demonstrated [115,116]. Both GEN and DEL were found to interact, albeit with different mechanisms, with topoisomerases: while GEN usually acts as a topoisomerase poison, turning the enzyme into a DNA-damaging agent, DEL acts as a catalytic inhibitor of topoisomerase hindering the formation of the TOP-DNA intermediate. Therefore, considering both the antioxidant effects and the interaction with the topoisomerases, a modification of the effects induced by AOH may be expected when the mycotoxin is combined with these two polyphenols. Preliminary investigations on the combinatory cytotoxic effects were conducted in HT-29 colon carcinoma cells with concentrations ranging from 1 to 100 μM (1:1 ratio): cytotoxicity was observed starting from 25 μM for AOH and GEN, and from 50 μM for DEL. Both AOH/GEN and AOH/DEL combinations led to cytotoxic effects starting from 25 μM (1:1 ratio) and the type of interactions was described as synergistic, with a tendency to lose synergism when increasing cytotoxic effects. DNA strand breaks and oxidative DNA damages of the combinations of AOH (50 μM) with DEL (10–100 μM) or GEN (25–250 μM) were evaluated by performing an alkaline comet assay with or without treatment with formamidopyrimidin-DNA-glycosylase (FPG). When combined, DEL and AOH showed marked antagonistic effects at 50 μM in the FPG-untreated samples, while lower oxidative DNA damages were observed at 25 and 100 μM. Similar results were found for the combination AOH/GEN at 25 and 100 μM, which showed a lower oxidative damage than AOH tested individually. The authors also evaluated the influence of the co-incubations on the stabilization of the topoisomerases/DNA intermediate (the so-called "cleavable complexes"), which is typically due to the action of topoisomerase poisons (such as AOH). The AOH/GEN combination did not increase the formation of cleavable complexes, rather an antagonistic effect was found at the highest GEN concentration tested (100 μM). Antagonistic effects were also found in AOH/DEL combinations starting from 25 μM. These results were partially attributed to the dual anti-oxidant or pro-oxidant properties of the polyphenols. In this respect, simultaneous short-time incubations with AOH and DEL led to a reduction of AOH-induced ROS generation at concentrations of DEL starting from 1μM. On the contrary, GEN induced oxidative stress per se and did not suppress the pro-oxidative effects induced by AOH. Moreover, 24-h pre-incubations with polyphenols followed by incubation with AOH, did not result in any change in pro-oxidant effects of AOH. This evidence led to exclude any possible modulations of anti-oxidant defense systems as a mechanism underlying the observed antagonistic effects. Therefore, direct anti- or pro-oxidant activities are reasonably as the base of the effects observed during the co-incubations with DEL and GEN. On this basis, DEL could help in preventing the genotoxic effects of AOH, but, considering the low systemic bioavailability of DEL, these protective effects may be limited to the gastrointestinal tract only [117,118].

The same authors also investigated the effects of DEL in combination with ATX-II, one of the most genotoxic *Alternaria* toxins [119]. As reported for the combination with AOH, DEL reduced both DNA strand breaks and oxidative damage in HT-29 cells after short-time co-incubation with ATX-II. The type of interaction was found to be antagonistic according to the applied "independent joint action" model. The production of ROS induced by 10 μM ATX-II was also reduced by DEL in concentrations from 1 μM to 100 μM, but these reductions cannot fully explain the huge reduction of genotoxic effects observed following the co-incubation with DEL. Indeed, no increase of ROS production was observed

at the concentration of ATX-II used in the comet assay (1 μM). In cell-free conditions, a reduction of the concentration of ATX-II was found upon co-incubation with DEL. The authors suggested that DEL, after being degraded to phloroglucinol aldehyde (PGA) and gallic acid (GA), might react with ATX-II neutralizing its epoxy group, which is the reactive chemical moiety presumably responsible for genotoxicity. Considering the hypothesis that PGA can react with ATX-II, it is important to underline that the reduction of adverse effects of this mycotoxin may actually occur in subjects that follow diets with a high content of anthocyanins as they are prone to release PGA during digestion.

In the context of the evaluation of combinatory effects between *Alternaria* toxins and bioactive compounds of food origin, polyphenols represent a class of compounds of great interest since they are widely distributed in those food categories which are prone to contamination with *Alternaria* mycotoxins. Quercetin (QUE) is one of the most abundant flavonoids in human diets. QUE has been previously associated to several potential health benefits mainly related to its antioxidant properties, although pro-oxidant effects at certain concentrations have also been described [120]. The potential ability of QUE to reduce the cytotoxicity of AOH and AME was investigated by Fernàndez-Blanco et al. [83]. Although cytoprotective effects were attributed to QUE [121], simultaneous exposure of Caco-2 cells to AOH and QUE (at concentrations ranging from 3.125 to 100 μM) did not result in any cytoprotective effect. In particular, no significant differences were found between the QUE-AOH combination and AOH tested alone after 48 h of exposure. However, the combination significantly affected cell viability at 24 h of treatment in comparison to AOH tested alone. Similarly, no difference between the binary combinations QUE + AME and AME tested alone were detected and, additionally, no cytoprotective effect was found in the tertiary combination AOH + AME + QUE at any of the tested concentrations. Therefore, QUE was not effective in reducing the effects of AOH and AME.

Possible cytoprotective properties of food components against the effects of AOH were also evaluated by Vila-Donat et al. in Caco-2 cells [122]. Keeping in mind that AOH may contaminate legumes (including soybeans and lentils), the authors investigated the effects of AOH in combination with soy saponin I (Ss-I), which was previously found to possess antioxidant activity, or with a lentils extract. In particular, the authors used two different approaches to evaluate the effects of Ss-I and lentil extract: (i) the first one consisted in pre-treating cells with Ss-I (6.25 μM) or with the extract, and then refreshing the growth medium and testing different dilutions of AOH (ranging from 3.125 to 50 μM); (ii) the second approach aimed at evaluating the combinatory effects and the type of interaction co-incubating AOH with Ss-I or lentil extract. By using the first approach, no differences were found between samples pre-treated with Ss-I and samples treated only with AOH. As an exception, the highest AOH concentration tested (50 μM) caused an increase in cell viability in the pre-treated samples. In contrast, co-treatments with AOH + Ss-I (1:1 ratio) above 6.25 μM resulted in an increase in cell viability compared to AOH tested alone. These results suggested that Ss-I likely acted via a direct interaction rather than modulating intracellular defense systems. With regard to cytoprotective effects of the lentils extract, only one single combination was tested and about 30% increase in cell viability was found in comparison to AOH tested alone.

#### **4. The Key Role of Bioactive Compounds**

Food is a complex matrix composed by macro- and micro-nutrients, containing also a huge number of non-nutrient compounds that may exert several biological activities. These compounds can interfere at different levels with mycotoxin activities. For instance, they can: (i) activate or inhibit enzymes involved in the metabolism of xenobiotics; (ii) act as anti-oxidant or pro-oxidant compounds; (iii) act as receptor agonists or antagonists targeting, in some cases, the same biological targets of mycotoxins; (iv) modify the expression of genes encoding proteins involved in the regulation of important physiological functions. On this basis, bioactive compounds of food origin may determine the onset of additive, synergistic or antagonistic effects when combined with *Alternaria* mycotoxins. Keeping in mind that the application of mitigating strategies along the food chain are supposed to progressively reduce the dietary exposure to toxicants, the assessment of combinatory effects

of mycotoxins with other food constituents will be the most accurate and realistic, but also highly challenging tasks to achieve in the next decades. The challenge will be even harsher taking into account that many food constituents potentially interplaying with mycotoxins are generally recognized as health promoting (i.e., polyphenols) and the consumption of foods rich in such compounds is typically recommended in healthy diet habits. In this framework, this section focuses on the modulation of *Alternaria* mycotoxins toxicity by bioactive compounds.

One of the best characterized toxicological endpoints of *Alternaria* mycotoxins likely affected by food constituents is the estrogenic activity. As a matter of fact, estrogenic and anti-estrogenic effects of bioactive compounds might markedly modify the overall estrogenicity of the *Alternaria* mycotoxins AOH and AME. In terms of risk characterization, this might change the toxicological relevance of such mycotoxins case by case, though they show a weak estrogenicity per se, depending on the composition of chemical mixtures in given foods. In this respect, foods prone to *Alternaria* contamination with a high content of potentially interfering constituents (e.g., polyphenolic phytoestrogens) are legumes (especially soy) and some alcoholic beverages (especially wine and beer). In particular, soybeans and derived products are among the richest dietary sources of phytoestrogens, and many of the isoflavones of soy (including genistein, daidzein, glycitein, and coumestrol) induce estrogen-receptor dependent estrogenic stimuli [123]. As a matter of fact, combinations of GEN-AOH at specific concentrations have been demonstrated to determine synergistic or antagonistic effects in Ishikawa cell line [110]. Similarly to soybeans, hops used to produce beer is characterized by the presence of some prenylflavonoids (e.g., naringenin, 8-prenylnaringenin, 6-prenylnaringenin, 6,8-diprenylnaringenin, and 8-geranylnaringenin) that are potent phytoestrogens with a dual effect being able to bind both estrogen receptor isoforms and to inhibit specific enzymes involved in the estrogenic cellular responses [124,125]. In this context, Aichinger et al. [126] demonstrated the ability of the phytoestrogens from hops xanthohumol and 8-prenylnaringenin to antagonize the estrogenic effects of the *Fusarium* mycotoxins ZEN and α-ZEL. Therefore, possible interactions can be expected also in combination with the estrogenic *Alternaria* toxins AOH and AME. Other important food constituents able to modulate estrogen receptor activity are resveratrol and β-sitosterol, whose primary dietary sources are peanuts, grapes, and wine. Resveratrol, in particular, may exhibit a super-agonist activity inducing a stimulation higher than the endogenous ligand 17β-estradiol in estrogenic gene report assay, even if anti-estrogenic effects were found in the MCF-7 cell line [127]. Although evidences have been not yet collected, these compounds are likely to affect the estrogenicity of *Alternaria* mycotoxins.

Another focal point of the cross-talk between mycotoxins and food components that requires further investigations is the modulation of the aryl hydrocarbon receptor (AhR) [128]. The cascade of events following the activation of AhR is of particular interest in toxicological investigations as it modulates the expression of genes involved in detoxification and transport of various xenobiotics, including the expression of cytochrome P450 family members. Interestingly, AOH and AME were able to bind and activate AhR, causing the increase of CYP1A1 expression and promoting their own metabolism [129]. This process was not affecting the mycotoxin-dependent production of ROS in murine hepatoma cells (Hepa1c1c7). In addition, the authors showed that mycotoxins reduced the number of cells via an AhR-independent process, although the apoptotic phenotype was found only in cells with functional AhR and ARNT [129]. With regard to the ability of AOH to suppress the lipopolysaccharide-induced inflammation previously mentioned, Grover & Lawrence did not find any correlation between AOH-mediated AhR activation and the suppression of the inflammation found in BEAS-2B cells [96]. Thus, despite the increased metabolism of AOH and AME, AhR activation does not seem to raise much concern for ROS production, cytotoxic and immunosuppressive effects, further studies are needed to determine the toxicity of hydroxylated metabolites (e.g., estrogenic properties). In particular, Dellafiora and co-workers showed that hydroxylated forms of AOH and AME cannot interact with estrogen receptors in vitro, pointing to the relevance of phase-I metabolism to modify the toxicodynamic of these mycotoxins. However, methylation of respective catecholic metabolites might reactivate the estrogenic potential [92].

Besides AOH and AME, many food constituents have been described to activate or inhibit AhR. Thus, they are likely to interfere with the ability of AOH and AME to bind AhR and/or with the metabolic processes following the activation of AhR. Foods consumed worldwide such as potatoes, cruciferous, bread, hamburgers, and citrus juices were investigated for the presence of natural AhR-agonists (NAhRAs) [128]. Among these, indole-3-carbinol, and many polyphenols and furocoumarins were found to be responsible for the activities shown by cruciferous vegetables (Brussels sprouts, broccoli, cabbage) and citrus juices, respectively. On the contrary, the activation of AhR induced by the baked or fried foods tested is thought due to secondary chemicals originating from the high-temperature processing, such as polycyclic aromatic hydrocarbons (PAHs), heterocyclic amines or Maillard products [128]. In addition, many dietary flavonoids showed a significant context–dependent AhR agonist or antagonist activities, depending on the concentration and cell types tested [130]. As an example, galangin, GEN, daidzein, and diosmin were found to be AhR agonist only in Hepa-1 cells, while cantharidin acted as an agonist only in human HepG2 and MCF-7 cells. On the contrary, AhR antagonist activities were shown both in MCF-7 and HepG2 cells by luteolin, while the antagonistic activity of kaempferol, quercetin and myricetin was strictly dependent on the cell context [130]. Many other flavones, flavonols, flavanones, isoflavones, and catechins also showed a high affinity to the AhR at dietary exposure levels [131]: apigenin, luteolin, quercetin, kaempferol, and myricetin were found to inhibit the activation of AhR induced by the most potent AhR activator identified so far (2,3,7,8-tetrachlorodibenzo-*p*-dioxin at 5 nM in MCF-7 cells) [131]. Taken together, these findings suggest that the AhR-dependent effects of food constituents strongly depend on both the chemical environment (which may significantly change among the different type of food) and on the cell type tested. Therefore, both the metabolism of *Alternaria* mycotoxins in vivo and their ability to modulate AhR could change depending on the food-specific chemical mixture.

An additional noteworthy activity of AOH and AME common with a number of food components is the capability to poison topoisomerases. In particular, many food bioactives naturally occurring in fruits, vegetables and legumes have been shown to affect the activity of both topoisomerase I and II. Taking into consideration that some *Alternaria* mycotoxins exert important genotoxic effects via either the inhibition or poisoning of these enzyme (see Section 3 for further details), the co-occurrence of other compounds targeting topoisomerases may reasonably change the overall topoisomerase-dependent genotoxic effects of *Alternaria* mycotoxins. Several studies demonstrated the ability of some polyphenols to poison topoisomerase I and/or topoisomerase II, albeit their specific mechanism of action has been poorly investigated. Kaempferol and quercetin were reported to be, at specific concentrations, non-covalently binders of topoisomerase IIα, while myricetin showed the ability to covalently bind to topoisomerase IIα and cleaving DNA in a redox-dependent way [132]. Additionally, the flavonoids quercetin, myricetin, fisetin, and apigenin were highlighted by other authors as poisons of topoisomerase I [133], whilst genistein, daidzein, biochanin A, chrysin, have shown poisoning effects also against topisomerases II [134]. Interestingly, genistein and especially delphinidin (that acts as catalytic topoisomerase inhibitor) were found to protect cells from AOH-induced genotoxicity [114]. Grapes and red wines are characterized by a large amount of resveratrol, belonging to polyphenols' stilbenoids group, which has always been regarded to have beneficial effects thanks to its manifold activity. Nevertheless, the capability to establish non-covalent cross-linking interactions with both topoisomerase II and DNA leading to cell death was described too [135]. An influence of these compounds on poisoning and/or inhibition of topoisomerases by *Alternaria* mycotoxins, also diversifying the outcomes in vivo in a mixture-dependent way, appears therefore to be possible.

On the basis of the data reported above, *Alternaria* mycotoxins and a wealth of food constituents may interfere to each other, mutually influencing their final effects. Moving further steps toward a more precise molecular-oriented understanding of the food-specific and mixture-dependent outcomes in vivo will allow mapping those categories of food might pose a higher risk for specific toxicological endpoints. In the near future, adopting such an approach will effectively pave the ground to set personalized risk/benefit assessment studies of food prone to be contaminated by *Alternaria* mycotoxins.

#### **5. Conclusions and Future Perspectives**

*Alternaria* mycotoxins are frequently occurring in various fresh and processed foods such as cereals, fruits, vegetables, nuts, fruit and vegetable juices, seeds and oils. In many cases, contaminated foods have been found to simultaneously contain more than one *Alternaria* mycotoxin. In addition, the co-occurrence of *Alternaria* mycotoxins along with *Fusarium*, *Penicillium* and *Aspergillus* mycotoxins is also well documented, though not routinely checked. In addition, mycotoxins co-occur with the huge number of food constituents inherently present in contaminated foods. Notably, a growing number of data pointing to significant effects of chemical mixtures of mycotoxins in combination with each other or with food components is available. On this basis, a more precise description of mycotoxin contamination in food, detailing both the co-occurrence of mycotoxins and the types of co-contaminated food categories, is urgently required to better support risk assessment studies.

In this respect, the current risk assessment of mycotoxins is mostly based on human exposure data and animal toxicity evidences of individual compounds, while the evaluation of possible effects due to chemical mixtures is only occasionally assessed. Studies on the combinatory effects of different *Alternaria* mycotoxins, also in combination with other mycotoxins, have already shown that the co-exposure may result in either additive, synergistic, or antagonistic effects, depending on the doses, time of exposure or type of combinations assessed. In addition, recent findings have shown that mycotoxins may interplay with other food constituents, with different outcomes depending on the nature of combinations tested. Taken together, these results show that the toxicity of mycotoxins may significantly change depending on the composition of chemical mixtures, whereby not only co-contaminants but also food bioactives might act as contributors. This evidence pointed out the need to carefully check the multiple co-occurrence of mycotoxins, also in combination with the other food constituents. On the other side, it is crucial to characterize the effects the various combinations with the other food constituents may cause on the toxicity of mycotoxin mixtures. However, the evaluation of combinatory effects is not easy to perform since the toxic action exerted by individual mycotoxins is often strictly dependent on the cellular model and the concentrations tested. In addition, the use of different cellular models and different tested concentrations makes the inter-laboratory comparison of results difficult. Moreover, from a practical point of view, the number of food constituents possibly co-occurring with mycotoxins and potentially able to modulate their toxicity is so huge to make the systematic assessment of any possible combination unaffordable. Therefore, the definition of a consensus to define the combinations that really deserve investigations is strongly suggested. From a toxicological point of view, the use of the Adverse-Outcome-Pathway (AOP) approach or the adoption of grouping criteria, such as read-across methodologies or other computational-based categorizing methods, might provide a convincing rational to support the early definition of combinations to be tested. Moreover, in order to improve the interpretability of the data, homogeneity in the expression of the results, as well as in the tested concentrations, used cellular models, and applied methods, should become a common objective for researchers dealing with these issues in the future.

In summary, *Alternaria* toxins in food are not yet regulated mainly as a consequence of the shortage of toxicological occurrence and exposure data. A more in-depth elucidation of their toxicity, taking into account the effects of chemical mixtures, will ensure a more precise evaluation of their effects on human health eventually resulting in a more reliable assessment of risks with an overall lower degree of uncertainty. In this framework, this review collected the main data available so far in terms of occurrence and combined actions of *Alternaria* mycotoxins and it highlighted that chemical mixture may significantly change the individual toxicity of mycotoxins. Notably, most of the combinations found naturally in food still need to be tested in terms of toxicity. Therefore, it is hard to infer with precision the actual toxicological effects due to the consumption of food contaminated by *Alternaria* mycotoxins. Nonetheless, the data presented here may serve as a ground to design further studies to deepen the knowledge about the toxicity of this class of mycotoxins and to support the assessment of risk taking into account the actual role of chemical mixtures. The proposed paradigm can be logically extended to the risk assessment of other mycotoxins, as the relevance of mixtures has been described also for other classes of mycotoxins.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6651/11/11/640/s1, Table S1: Co-occurrence of *Alternaria* toxins in food.

**Author Contributions:** Conceptualization: F.C., C.D., L.D.; formal analysis: F.C.; data curation: F.C. and L.D.; writing—original draft preparation: F.C. and L.D.; writing—review and editing: G.G., G.A., E.V., D.M., C.A.; supervision: D.M., C.D.

**Funding:** This research received no external funding.

**Acknowledgments:** The study was partially supported by Fondazione Cariparma, under the TeachInParma Project. Moreover, the authors acknowledge the financial support of the Center for Studies in International and European Affairs of the University of Parma.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Toxins* Editorial Office E-mail: toxins@mdpi.com www.mdpi.com/journal/toxins

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18