Comparative In Vitro Assessment of a Range of Commercial Feed Additives with Multiple Mycotoxin Binding Claims

Contamination of animal feed with multiple mycotoxins is an ongoing and growing issue, as over 60% of cereal crops worldwide have been shown to be contaminated with mycotoxins. The present study was carried out to assess the efficacy of commercial feed additives sold with multi-mycotoxin binding claims. Ten feed additives were obtained and categorised into three groups based on their main composition. Their capacity to simultaneously adsorb deoxynivalenol (DON), zearalenone (ZEN), fumonisin B1 (FB1), ochratoxin A (OTA), aflatoxin B1 (AFB1) and T-2 toxin was assessed and compared using an in vitro model designed to simulate the gastrointestinal tract of a monogastric animal. Results showed that only one product (a modified yeast cell wall) effectively adsorbed more than 50% of DON, ZEN, FB1, OTA, T-2 and AFB1, in the following order: AFB1 > ZEN > T-2 > DON > OTA > FB1. The remaining products were able to moderately bind AFB1 (44–58%) but had less, or in some cases, no effect on ZEN, FB1, OTA and T-2 binding (<35%). It is important for companies producing mycotoxin binders that their products undergo rigorous trials under the conditions which best mimic the environment that they must be active in. Claims on the binding efficiency should only be made when such data has been generated.


Introduction
Mycotoxins are toxic, low-molecular weight compounds produced as secondary metabolites by several fungi species belonging mainly to Aspergillus, Fusarium, Penicillum, Alternaria and Clavicep genera [1]. Under favourable environmental conditions such as moisture and temperature, these fungi can invade crops and proliferate (during growth, transportation and storage) to produce mycotoxins [2]. Other factors including climate change, poor harvesting practices, improper drying, handling and packaging may also predispose crops to fungal invasion and subsequent mycotoxin production [3]. Mycotoxins appear in the food and feed chain because forages and cereals, which are most susceptible crops to these fungi, are utilised as the main components of animal feed. Among the more than 400 mycotoxins currently identified, aflatoxin B1 (AFB1), deoxynivalenol (DON), zearalenone (ZEN), ochratoxin A (OTA), fumonisins B1 (FB1) and trichothecenes T-2/HT-2 toxin are considered the most (HSCAS), with mycotoxin binding claims [24,25]. Many researchers have investigated mycotoxin sequestering potentials of some of these products, which are commercially available worldwide. However, most of the studies are focused on only one or two mycotoxins, particularly AFB1 [26][27][28]. As mycotoxins are often co-occurring in animal feed [29,30], the current study aims to evaluate and compare the efficacy of ten commercial feed additives with multi-mycotoxin binding claims on DON, T-2, ZEN, OTA, FB1 and AFB1 using an in vitro model simulated to mimic the gastro-intestinal tract (GIT) of a monogastric animal. Results showed that only one of the products (a modified yeast cell wall) effectively adsorbed more than 50% of DON, ZEN, FB1, OTA, T-2 and AFB1, in the following order: AFB1 > ZEN > T-2 > DON > OTA > FB1. The remaining products were able to moderately bind AFB1 (44-58%) but had less, or in some cases, no effect on ZEN, FB1, OTA and T-2 binding (<35%).

Results and Discussion
Contamination of different agricultural commodities with multi-mycotoxins, as well as adverse health effects and reduction in animal performance, due to mycotoxicosis are still prevalent, despite the prevention strategies currently employed [31][32][33]. Additives are added to the diets of livestock animals to bind mycotoxins and reduce their bioavailability in GIT and distribution to blood and target organs. The most prevalent adsorbing agents are polymers, yeast cell wall, cholestyramine and clay minerals [21,22]. In vitro analysis of mycotoxin adsorption is a very useful tool for rapid screening and identification of agents that may have mycotoxin sequestering potentials [22]. Several researchers have investigated different mycotoxin binders, however, most of the studies have focused on a single mycotoxin and carried out using buffer solutions mostly at pH 3 and 7, to simulate physiological pH in stomach and intestine, respectively. This does not truly reflect the conditions in a farm animal GIT as other factors including temperature, digestive enzymes, feed, bile salts and nutrients may interfere with the adsorption (ion-exchange) process [21].
In the current study, ten commercial feed additives with adsorption, inactivation or detoxification claims on DON, ZEN, FB1, OTA, T-2 and AFB1 were obtained and categorised into three groups based on their composition. Their capacity to simultaneously bind or adsorb DON, ZEN, FB1, OTA, T-2 and AFB1, which often co-occur in complete feed or feed ingredients such as maize, wheat and barley was assessed and compared. The ratio of additive:binder used in this study is based on the maximum permitted/guidance levels for mycotoxins in European pig feed [34] and the conventional binder inclusion level of 2 g/kg feed [35]. In order to assess the mycotoxin binding capacity of the adsorbents, an in vitro system with buffer solutions at pH 3 and 7-to simulate stomach and intestine respectively, was used to study mycotoxin adsorption/desorption. Furthermore, a robust in vitro model relative to the GIT of a monogastric animal in terms of compartment, enzymes, feed, gastric fluids, temperature, pH and transit time was designed, to investigate the adsorption efficacy of the feed additives. The percentage adsorption of DON, ZEN, FB1, OTA, T-2 and AFB1 by various feed additives in buffer solutions as well as in vitro GIT model are presented in Tables 1 and 2, respectively.

Inorganic Additives
Aluminosilicate constitute the most abundant group of rock-forming minerals [36]. The basic structural unit of silicate clay minerals consists of the combination of aluminium octahedral and silica tetrahedral sheets, both with hydroxyl and oxygen groups [37]. Most studies (both in vivo and in vitro) on mycotoxin binders using clay minerals have focused on aluminosilicates such as bentonite, montmorillonite, zeolite and hydrated sodium calcium aluminosilicates (HSCAS). They possess high cation exchange capacity, pore volume and large surface area, which enable them to adsorb low-molecular weight compounds such as mycotoxins to their surfaces, edges and interlayer spaces [27]. Four commercial clay-based products (1, 2, 3 and 4) were investigated for their multi-mycotoxin binding potentials. Results obtained for in vitro buffer solutions showed that all the 4 products bound DON, ZEN, FB1, OTA, T-2 and AFB1 at adsorption rates of 29-58%, 27-42%, 29-47%, 5-40%, 9-38%, 51-68% respectively (Table 1). Product 1 and 3 had a significant adsorption on AFB1 (68% and 61%), DON (53% and 49%) and ZEN (42% and 46%) respectively, compared to product 2 and 4 (p < 0.05). There was no significant difference (p > 0.05) in the adsorption of DON, ZEN, FB1, OTA, T-2 and AFB1 by product 2 and 4, as they sequestered <34% of DON, ZEN and FB1; <13% of OTA and T-2, and approximately 50% of AFB1. Within this category, AFB1 was the most adsorbed mycotoxin followed by DON, ZEN, FB1, OTA and T-2 at pH 3 and 7. Several studies on adsorption of mycotoxins using buffer solutions at different pH (mostly 3, 5 and 7) have shown that AFB1 is highly adsorbed by clay minerals at acidic and alkaline pH with little to no adsorption of other mycotoxins [26,38,39]. A recent study on the efficacy of commercial clay minerals to sequester 0.1 µg/mL of AFB1, DON and ZEN showed that 1% of a commercial smectite and an aluminosilicate clays significantly adsorbed AFB1 (95-100%) and ZEN (56-82%) in acidic pH, with no significant effect on DON (<10%) [40]. Similarly, 50 mg of a commercial bentonite adsorbed 99% of 10 ng/mL AFB1 and 1% of 250 ng/mL DON in buffer solution (pH 5) [41]. The difference in the ability of clay minerals to sequester mycotoxin has been attributed to their origin and physiochemical properties such as cation exchange capacity, pore volume and expandability [21,22].
Although, the adsorption capacity of product 1 and 3 were reduced in the GIT model, they still significantly adsorbed DON, ZEN, FB1, OTA, T-2 and AFB1 (p < 0.05) when compared with product 2 and 4. These products (1 and 3) are chemically modified clay minerals. Modified clay minerals have been shown to possess high mycotoxin-sequestering ability compare to natural clay minerals (product 2 and 4). Modified adsorbents are prepared by alteration of surface properties such as cation exchange capacity using acids, alkalis, organic compounds and heat, that consequently increase their contaminant removal capacity and efficacy [44]. Nevertheless, their safety and interaction with nutrients and veterinary substances remain a concern [45].
Organic additives such as yeast cell wall and glucommanan have been shown to have a high binding activity across a wide spectrum of mycotoxins compare to inorganic minerals [48]. The cell wall of the yeast Saccharomyces cerevisiae is composed of lipids, protein and polysaccharide fraction, with glucans and mannans being the two main constituents of the latter fraction [47]. Glucomannan is a water-soluble polysaccharide composed of hemicellulose, it is present in the cell wall of some plant species. Several authors have suggested that the cell wall components of these substances could be responsible for the adsorption of mycotoxins through non-covalent, hydrogen bonds, ionic or hydrophobic interactions [49][50][51]. Rignot et al. [50] showed that β-D-glucans are the yeast component largely responsible for the complexation of mycotoxins, and that the reticular organization of β-D-glucans and the distribution between β-(1,6)-D-glucans and β-(1,3)-D-glucans plays a vital role in mycotoxin adsorption [50]. Furthermore, Van der Waals forces and weak hydrogen bonding maybe involved in the adsorption of mycotoxins by β-D-glucans [49]. The efficacy and type of mycotoxins a yeast cell wall product can adsorb is dependent on the origin of yeast, strain, pH, binding sites or accessible surface area, growth condition and percentage of cell wall components (mannoproteins, chitins, lipids and β-glucan) [51]. Glucomannan is commonly used as a dietary fibre, however, there are very limited studies published regarding the types of mycotoxins adsorbed and mechanisms of adsorption.
Generally, all the commercial binder or feed additive products assessed for their capacity to bind multi-mycotoxin in the current study adsorbed DON, ZEN, FB1, OTA, T-2 and AFB1 simultaneously at different rates under both acidic and alkaline pH. However, percentage adsorption at pH 3 was more significant when compare with adsorption at pH 7 (p < 0.05), this indicates that products investigated can form a stable mycotoxin-binder complex at pH 3 and to some extent at pH 7. Under in vitro GIT model, adsorption efficacies of all the products were reduced (except product 6, for ZEN) possibly due to interaction of binder products with other components of GIT such as pepsin, HCl and feed [43,52,53]. For instance, Barrientos et al. [43] showed that the adsorption of a globular protein (pepsin) by a smectite clay significantly reduced the adsorption rate of AFB1 in simulated acidic gastrointestinal fluid [43]. Also, a corn protein interfered with AFB1 adsorption to a smectite clay in corn fermentation solution [53].
Mycotoxin binders adsorb mycotoxin at the surface, to form a mycotoxin-binder complex, the bound mycotoxins are then excreted along with the binder in animal faeces. The adsorption capacity and stability of the complex through the GIT is influenced by physiochemical properties of the binder including polarity, size of the pores and accessible surface area as well as physicochemical properties of mycotoxins including polarity, solubility, size and charge [38]. AFB1 is relatively hydrophilic with aromatic planar molecules, therefore it is easily bound by most binders, particularly clay minerals, under both acidic and alkaline conditions, by formation of a coordination bonds with the beta-carbonyl system [26]. However, other mycotoxins-ZEN, OTA, FB1, T-2 and DON range from being moderately hydrophilic to high hydrophobic compounds, therefore being very difficult to adsorb [26]. However, emerging nanocomposites [54], modified organic and inorganic adsorbents [55] are being used to sequester these mycotoxins.

Conclusions
In light of the high co-occurrence of fungi and mycotoxins in agricultural commodities, exacerbated by climate change, products with wide spectrum mycotoxin adsorption or detoxification are in great demand from farmers and animal feed producers, to minimise the economic losses caused by mycotoxicosis. In the current study, an in vitro GIT model was designed to assess and compare the efficacy of ten commercially available binder products with multiple mycotoxin claims on DON, ZEN, FB1, OTA, T-2 and AFB1. Results showed that most of the products were able to significantly bind DON, ZEN, FB1, OTA, T-2 and AFB1 in both alkaline and acidic buffer solutions. However, under the in vitro model simulating the conditions in the GIT of monogastric animals such as chicken and pig, the efficacy of all the products were significantly reduced and only one of the products tested (6-a modified yeast cell wall) was still able to simultaneously adsorb more than 50% of DON, ZEN, FB1, OTA, T-2 and AFB1, in the following order AFB1 > ZEN > T-2 > DON > OTA > FB1. The remaining products were able to moderately bind AFB1 (44-58%) but had less than 35% or in some cases no binding effect on ZEN, FB1, OTA and T-2 binding. A robust method that mimics the GIT condition of a farm animal must be used to study the efficiency of a potential mycotoxin binder, not the conventional use of buffers at different pH. Furthermore, producers of feed additives with mycotoxin binding claims should ensure appropriate and detailed labelling of their products such as the composition, physicochemical properties, mode of action, dosage and importantly the specific mycotoxin(s) their product can bind, adsorb or detoxify, to ensure farmers and animal nutrition companies are not misled.

Feed Additives
Ten commercially available products claiming multiple mycotoxin adsorption or binding on DON, T-2, ZEN, AFB1 and FB1 were obtained and categorised into three groups (inorganic, organic and mixture of additives) based on their main functional composition. The products were coded with numbers to preserve the confidentiality of the source. Products 1, 3 and 4 were purchased online, while products 2, 5, 6, 7, 8, 9 and 10 were obtained directly from the companies. Product details including mode of action and main composition (as stated on the product labels and manufacturers' websites) are listed in Table 3. Table 3. Composition and mode of actions (as stated on the product labels and manufacturers' websites) of commercial feed additives claiming multiple-mycotoxin binding. Modified yeast cell wall Adsorption 7

Category Product Main Composition Mode of Action
Esterified glucomannan *** Mixed adsorbent 8 Mixed silicates and yeast cell wall *** 9 Aluminosilicate and enzyme Adsorption and biotransformation 10 Natural minerals and algae Adsorption and degradation *** no information provided.

Buffer Solution
Mycotoxin stock solutions (1 mg/mL) of AFB1, DON, ZEN, T-2 and OTA were prepared by dissolving pure solid standards in methanol and FB1 in acetonitrile/water (50:50, v/v). A mixed-mycotoxin working solution was prepared in 10 mL acetonitrile and stored at −20 • C until use. To evaluate adsorption efficacy of the binding products and stability (adsorption/desorption) of mycotoxin-binders complex in both acidic and alkaline conditions, adsorption capacity of each product was studied at pH 3 and 7 to simulate physiologic pH in the stomach and intestine of monogastric animal respectively. The buffer solutions (pH 3 and 7) were prepared by using 0.1 M citrate and 0.2 M phosphate buffers. Each product (20 mg) was weighed into a 30 mL flask containing 10 mL of buffer solution; 20 µL of multi-mycotoxin working standard solution was added to reach a final concentration of 20 ng/mL, 50 ng/mL, 900 ng/mL, 5000 ng/mL, 100 ng/mL and 250 ng/mL for AFB1, OTA, DON, FB1, ZEN and T-2 respectively, this was performed for each pH in triplicate. A blank control was prepared using only multi-mycotoxin working solution in buffers without any mycotoxin binder. The flasks were shaken and incubated for 3 h at 37 • C in an incubator shaker. Thereafter, samples were centrifuged (30 min, 1000× g, 25 • C), and 1 mL of supernatant was mixed with an equal volume of acetonitrile and evaporated to dryness under gentle nitrogen stream (40 • C). The residue was reconstituted in 1 mL of methanol, filtered through 0.2 µm PTFE filter and transferred to a glass vial for LC-MS/MS analyses.

In Vitro Gastrointestinal Model
An artificially contaminated feed was made by spiking 1 g of finely ground feed material with 200 µL of multi-mycotoxin stock solution to reach approximately the following mycotoxin concentrations, based on EU permitted/regulated limit for mycotoxins in pig feed: AFB1 (21.2 µg/kg), OTA (48.9 µg/kg), DON (997.2 µg/kg), FB1 (5582.3 µg/kg), T-2 (243.1 µg/kg) and ZEN (152.8 µg/kg). The spiked material was incubated overnight in the dark at 40 • C to evaporate to dryness. To check the homogeneity of the batch, three samples taken randomly were extracted and analysed for multi-mycotoxins using a previously validated QuEChERS-based LC-MS/MS method [56]. To assess the efficacy of commercial feed additives to adsorb multiple mycotoxins, an in vitro model was designed to simulate the GIT conditions of monogastric animal using an automated dissolution USP Apparatus 2 (Vankel VK 7010, Erweka, Germany) with an auto-controlled multi-channel peristaltic pump (Vankel VK 810, England). Temperature of 40 • C and rotation speed of 100 rpm were used throughout the experiment. The first GIT compartment simulated was the crop/oesophagus, 1.0 g of multi-mycotoxin contaminated feed and 20 mg of each feed additive were mixed with 40 mM of acetic acid, 0.2 M Na 2 HPO 4 and 5 M NaCl buffer. Each tube was mixed to reach a pH value of 4.5-5.3 and incubated for 60 min. Subsequent stomach/proventriculus simulation was performed by addition of 0.23 M HCl, 0.034 M NaCl and 5000 U of purified pepsin derived from pig stomach mucosa, to reach a pH between 1.9 and 3.7, tubes were further incubated for 90 min. The final GIT compartment simulated was the intestine; here, 0.05 M NaHCO 3 , pancreatin (0.5 mg/mL) and 0.4% bile salt were added to the tubes, the pH was increased and ranged between 5.3 and 7.5. All samples were incubated further for 120 min. The total incubation time for the in vitro digestion was 4 h and 30 min. Blank controls were prepared without the addition of any feed additive, and all experiments were performed in quintuplicate. After incubation, 1 mL of sample was withdrawn and mixed with 1 mL of 0.01% formic acid in acetonitrile, followed by a rigorous vortex and centrifugation at 10,000× g for 30 min. Subsequently, 1 mL of supernatant was dried under gentle nitrogen stream (35 • C) and residue was re-dissolved in 0.5 mL of methanol, filtered through 0.2 µm PTFE filter and transferred to a glass vial for LC-MS/MS analysis.

Method Performance
For quantification, a seven-point calibration curve was prepared for each mycotoxin in the following concentration range: 0.5−50 ng/mL for AFB1, 10−5000 ng/mL for FB1, 5-500 ng/mL for ZEN, T-2 and OTA, and 10−2000 ng/mL for DON. To evaluate the effects of the matrix on MS quantification, matrix-induced suppression/enhancement (SSE) was determined by comparing the response of matrix spiked with seven different concentrations of each mycotoxin to a neat solvent standard at the same concentrations. The experiment was performed for buffer solutions (pH 3, pH 7) and gastrointestinal fluid (GF) in triplicate at three different times. SSE was calculated as the ratio of calibration curve slope for matrix-matched standards and neat solvent standards multiplied by 100. Limits of detection and limits of quantification (LOQ) of each mycotoxin were calculated at a signal-to-noise ratio of 3:1 and 10:1, respectively, based on a matrix-matched calibration in buffer solutions and GF. The coefficients of determination (R 2 ) for selected mycotoxins in the three matrices (pH 3, pH 7 and GF) ranged from 0.9901 to 0.9995. The retention times of the analyte in the sample extract were checked to correspond to that of the calibration standards and was within a tolerance of ± 0.1 min. Also, the ion ratios were within 25% of that obtained from the calibration standard for all analytes. SSE and LOQs values obtained for mycotoxins in the three matrices are reported in Table 4. Figure 1 shows chromatograms obtained for the six mycotoxins in a spiked feed sample. Table 4. Signal suppression-enhancement/relative standard deviation (SSE/RSD) and limit of quantification (LOQ) obtained for deoxynivalenol (DON), zearalenone (ZEN), fumonisins B1 (FB1), ochratoxin A (OTA), T-2 and aflatoxin B1 (AFB1) and validated matrices-pH 3, pH 7 and gastrointestinal fluid (GF).

Quantification of Mycotoxins and Statistical Analysis
The percentage adsorption of mycotoxins by each product was calculated as follows: Adsorption (%) = (C b − C t )/C t × 100.
where C b is the mycotoxin concentration in blank spiked buffer solutions (ng/mL); and C t , the amount of mycotoxin in the supernatant of sample (ng/mL). The data obtained were analysed using TargetLynx processing software (Waters, Wilmslow, UK) and Prism ® version 8 (San Diego, CA, USA). The means of treatments showing significant differences in two and three-way ANOVA were compared using Tukey's honestly significance difference multiple-comparisons post-test. All statements of significance are based on the 0.05 level of probability.