Elimination of Deoxynivalenol, Aflatoxin B1, and Zearalenone by Gram-Positive Microbes (Firmicutes)

Mycotoxin contaminations in the feed and food chain are common. Either directly or indirectly, mycotoxins enter the human body through the consumption of food of plant and animal origin. Bacteria with a high mycotoxin elimination capability can reduce mycotoxin contamination in feed and food. Four Gram-positive endospore-forming bacteria (Bacillus thuringiensis AMK10/1, Lysinibacillus boronitolerans AMK9/1, Lysinibacillus fusiformis AMK10/2, and Rummeliibacillus suwonensis AMK9/2) were isolated from fermented forages and tested for their deoxynivalenol (DON), aflatoxin B1 (AFB1), and zearalenone (ZEA) elimination potentials. Notably, the contribution of bacterial cell wall fractions to the observed outstanding ZEA elimination rates was demonstrated; however, the ZEA elimination differed considerably within the tested group of Gram-positive bacteria. It is worth noting that the purified cell wall of L. boronitolerans AMK9/1, L. fusiformis AMK10/2 and B. thuringiensis AMK10/1 were highly efficient in eliminating ZEA and the teichoic acid fractions of B. thuringiensis AMK10/1, and L. fusiformis AMK10/2 could also be successfully used in ZEA binding. The ZEA elimination capacity of viable R. suwonensis AMK9/2 cells was outstanding (40%). Meanwhile, R. suwonensis AMK9/2 and L. boronitolerans AMK9/1 cells produced significant esterase activities, and ZEA elimination of the cell wall fractions of that species did not correlate with esterase activity. DON and AFB1 binding capabilities of the tested bacterial cells and their cell wall fractions were low, except for B. thuringiensis AMK10/1, where the observed high 64% AFB1 elimination could be linked to the surface layer (S-layer) fraction of the cell wall.


Introduction
Mycotoxin production by molds is a global problem that cannot be solved easily in the field, even with the introduction of highly efficient proper agricultural practices. Environmental conditions affected by climate change and improper storage conditions may facilitate mold growth and secondary metabolite production [1,2]. These fungi can infect many plants and a wide array of food and agricultural products, and they can also produce a broad spectrum of harmful secondary metabolites, such as carcinogenic aflatoxins (AFs) [3,4]. Another dangerous mycotoxin, zearalenone (ZEA), is produced mainly by Fusarium species [5], and due to its estrogenic character, ZEA and its derivatives may cause severe reproductive and sexual dysfunctions [6][7][8]. The co-occurrence of ZEA with another Fusarium mycotoxin deoxynivalenol (DON) is common [9,10], and DON is a potent inhibitor of the protein synthesis in Eukarya [11], causing nausea, vomiting, diarrhoea, or anorexia in animals [12].
Not surprisingly, removing these harmful mycotoxins from feed and food is a crucial issue worldwide. In addition to the extensive range of available physical and chemical mycotoxin decontamination methods, biological technologies often based on food-grade microorganisms are gaining ground. The tested biological detoxification methodologies typically incorporate biodegradation and biosorption processes [13][14][15][16]. Mycotoxin elimination by bacteria and yeasts can happen through binding to biopolymers (e.g., peptidoglycan, PG), degradation by enzymes (e.g., esterase), or enzymatic conjugation to various molecules, giving rise to so-called masked or bound mycotoxins that make the original mycotoxin forms hidden to classic chemical analytical methods [17].
Bacteria are distinguished tools in the biological control of mycotoxigenic molds and the mycotoxins themselves. For example, bacterial cells can adsorb mycotoxins, e.g., aflatoxins are bound effectively by various cell wall components [18], or can detoxify these contaminants via enzymic degradation [15,19], or can inhibit the growth of molds by different antifungal metabolites, e.g., bacteriocins and organic acids [20].
Gram-positive lactic acid bacteria (LAB) typically have an excellent mycotoxin elimination capability, and both viable and nonviable LAB can eliminate aflatoxin B1 (AFB1). However, the elimination depends on the genus, pH, and bacterial density [21]. The binding of some major mycotoxins, including AFB1 [22], ZEA [23], and certain trichothecenes [24] by some probiotic LAB, was also demonstrated in vitro. Furthermore, some non-lactic acid bacteria, including Bacillus spp., were also investigated for their possible biotechnological application in mycotoxin elimination [18].
The bacterial cell wall consists of several biopolymers that can absorb different toxic substances, such as mycotoxins. Therefore, differences between Gram-negative and Grampositive bacterial cell walls ( Figure 1) would allow for predicting different adsorption capabilities in mycotoxin elimination due to the apparent differences in the thickness of the PG layer and the presence of the outer membrane. The surface layer, which is the outermost cell wall layer in many Bacteria and Archaea cells, attaches to the PG by electrostatic interactions and possesses inherent, entropy-driven affinities to self-assemble with each other [25] or, in the Gram-negative bacterial cell wall, to attach to outer membrane lipopolysaccharides (LPS) [26].
In this work, bacteria and bacterial cell wall preparations were screened for mycotoxin elimination using a unique collection of endospore-forming Gram-positive bacteria isolated from forages in Hungary. Novel mycotoxin-eliminating bacteria were aimed to find and shed light on cell wall fractions and other cell constituents, such as enzymes, which have the most critical role in mycotoxin elimination.  In this work, bacteria and bacterial cell wall preparations were screened for mycotoxin elimination using a unique collection of endospore-forming Gram-positive bacteria isolated from forages in Hungary. Novel mycotoxin-eliminating bacteria were aimed to find and shed light on cell wall fractions and other cell constituents, such as enzymes, which have the most critical role in mycotoxin elimination.

Identification of Isolated Bacteria
Bacterial isolates from fermented forages were identified with higher than 97% homology using 16S rRNA gene sequences (Table 1). All bacteria were Gram-positive endospore-forming organisms and taxonomically belonged to the phylum Firmicutes. Except for Rummeliibacillus suwonensis (Planococcaceae), all isolates were placed in the family Bacillaceae.

Identification of Isolated Bacteria
Bacterial isolates from fermented forages were identified with higher than 97% homology using 16S rRNA gene sequences ( Table 1). All bacteria were Gram-positive endosporeforming organisms and taxonomically belonged to the phylum Firmicutes. Except for Rummeliibacillus suwonensis (Planococcaceae), all isolates were placed in the family Bacillaceae.
All samples showed negligible DON elimination under the tested experimental conditions. Furthermore, AFB1 elimination rates were low and typically under 20% by all tested bacteria and their cell wall fractions except the S-layer fraction of B. thuringiensis AMK10/1, which eliminated 64% of AFB1 ( Figure 2). AMK10/2, and B. thuringiensis AMK10/1, as well as their cell wall fractions, were supplemented with zearalenone (ZEA), aflatoxin B1 (AFB1), and deoxynivalenol (DON), respectively, to study the mycotoxin elimination capability.
All samples showed negligible DON elimination under the tested experimental conditions. Furthermore, AFB1 elimination rates were low and typically under 20% by all tested bacteria and their cell wall fractions except the S-layer fraction of B. thuringiensis AMK10/1, which eliminated 64% of AFB1 ( Figure 2). The elimination of ZEA by the bacteria and their cell wall preparations showed remarkable species-specific differences ( Figure 3). In many cases, bacterial cells and their cell wall preparations showed remarkable ZEA elimination rates ( Figure 3A), but the eliminated ZEA could only be partially recovered from the bacteria and cell wall pellets (Figure 3B). In addition, for some bacteria and cell wall preparations, the quantity of the recovered ZEA remained even below the detection limit (LOD, 2.6 µg/L).
Interestingly, ZEA elimination by viable R. suwonensis AMK9/2 cultures and its cell wall preparations showed approximately the same values: about 40% ( Figure 3A). Significantly, purified cell wall, PG, and S-layer protein fractions of L. boronitolerans AMK 9/1 nearly eliminated ZEA, and the remaining ZEA concentrations in the supernatants were below LOD. ZEA concentrations recovered from the various cell wall fraction pellets were 1-12% of the starting values (cell wall fraction: 1%, PG: 12%, S-layer proteins: 6%) ( Figure  3B).
L. fusiformis AMK10/2 cell debris, purified cell wall, PG fraction, and S-layer protein fraction were valuable for eliminating ZEA ( Figure 3A). However, only small portions of the eliminated ZEA could be recovered from the fractions, except for the S-layer fraction of AMK10/2, where 25% of the original ZEA could be extracted ( Figure 3B).
Considering B. thuringiensis AMK10/1, viable cells, cell debris, purified cell wall, and teichoic acid fractions were suitable for efficiently eliminating ZEA from the liquid phase ( Figure 3A). Interestingly, the elimination of ZEA by PG fraction of the AMK10/1 strain The elimination of ZEA by the bacteria and their cell wall preparations showed remarkable species-specific differences ( Figure 3). In many cases, bacterial cells and their cell wall preparations showed remarkable ZEA elimination rates ( Figure 3A), but the eliminated ZEA could only be partially recovered from the bacteria and cell wall pellets ( Figure 3B). In addition, for some bacteria and cell wall preparations, the quantity of the recovered ZEA remained even below the detection limit (LOD, 2.6 µg/L).
Interestingly, ZEA elimination by viable R. suwonensis AMK9/2 cultures and its cell wall preparations showed approximately the same values: about 40% ( Figure 3A). Significantly, purified cell wall, PG, and S-layer protein fractions of L. boronitolerans AMK 9/1 nearly eliminated ZEA, and the remaining ZEA concentrations in the supernatants were below LOD. ZEA concentrations recovered from the various cell wall fraction pellets were 1-12% of the starting values (cell wall fraction: 1%, PG: 12%, S-layer proteins: 6%) ( Figure 3B).
L. fusiformis AMK10/2 cell debris, purified cell wall, PG fraction, and S-layer protein fraction were valuable for eliminating ZEA ( Figure 3A). However, only small portions of the eliminated ZEA could be recovered from the fractions, except for the S-layer fraction of AMK10/2, where 25% of the original ZEA could be extracted ( Figure 3B).
Considering B. thuringiensis AMK10/1, viable cells, cell debris, purified cell wall, and teichoic acid fractions were suitable for efficiently eliminating ZEA from the liquid phase ( Figure 3A). Interestingly, the elimination of ZEA by PG fraction of the AMK10/1 strain was almost negligible, while, for the S-layer fraction, it was about 38%. Meanwhile, the teichoic acid fraction was the most important cell wall fraction for ZEA elimination ( Figure 3A). Nevertheless, AMK10/1 cells and cell wall fractions released ZEA under extraction ( Figure 3B) and the most significant release was achieved from the total cells.

Esterase Activity
Since esterase activity is connected to ZEA degradation in living bacterial cells [19], this enzyme activity was measured and varied in a wide range. R. suwonensis AMK9/2 and L. boronitolerans AMK9/1 produced high esterase activities compared to L. fusiformis AMK10/2 and B. thuringiensis AMK10/1, the highest enzyme activity was measured in R. suwonensis AMK9/2 cell debris (1.98 ± 0.3 mM p-nitrophenol released/min) (Figure 4). Correlation analysis revealed a correlation coefficient of 0.618 between enzyme activity and extracted ZEA from cell debris and −0.748 between enzyme activity and the ZEA remaining in the cells' supernatant. AMK 9/2 cells and AMK 9/2 cell wall fractions eliminated ZEA with the same ratio. However, esterase activity could not be measured from the cell wall fractions because those samples were produced using different chemicals wall fractions because those samples were produced using different chemicals and heat treatment. Therefore, esterase activity could not explain ZEA elimination by AMK 9/2 cell wall fractions.

Discussion
Mycotoxin elimination technologies can use viable and dead bacterial biomasses and various cell fractions. The mycotoxin adsorption capability of bacteria depends primarily on cell wall components [23,24], where mainly polysaccharides, such as peptidoglycan (PG) and teichoic acids, play a critical role in mycotoxin binding [31]. Besides adsorption, enzymatic degradation (e.g., [14]) is possible, making a cell-free biotechnological process for mycotoxin elimination achievable.
Elimination of zearalenone (ZEA), deoxynivalenol (DON), and aflatoxin B1 (AFB1) using forage-based isolates of the Gram-positive R. suwonensis, L. boronitolerans, L. fusiformis, and B. thuringiensis was targeted. For R. suwonensis AMK9/2 and L. boronitolerans AMK9/1, no mycotoxin elimination-related studies are published thus far, to the best of our knowledge. For L. fusiformis, there was only one study on AFB1 elimination [32] but it was only performed with viable bacteria and not with cell fractions; meanwhile, for B. thuringiensis, there are several papers available on its antagonistic effects on patulin-producing fungi and its lactonase production, which degrades patulin mycotoxin (e.g., [33]). In the present study, viable cells, cell wall debris, purified cell walls, teichoic acid, PG, and S-layer fractions were produced with heat and different chemical treatments and tested in mycotoxin elimination experiments.
Both heat and organic acids affect the adsorption of mycotoxins by the cell walls because both factors reduce the PG's thickness, increasing the pore size of the structure [22,23]. In addition, heat treatment disrupts the glycosidic bonds of cell wall polysaccharides, such as PG, and organic acids break the amide ligaments in the structure of PG [22,23]. These treatments also denature proteins and split them into smaller peptides by breaking peptide bonds and exposing more binding sites [34]. Protein denaturation caused by high temperature is also a part of pore generation, resulting in enhanced permeability in living cells [35].
Under the experimental conditions, the trichothecene mycotoxin DON was not eliminated by bacterial cells and cell fractions, unlike in other Gram-positive cells [21,36,37]. Success in controlling DON-producing fungi by microbial antagonists and detoxifying DON with microbes and enzymes was reported in the past several years [16]. El-Nezami

Discussion
Mycotoxin elimination technologies can use viable and dead bacterial biomasses and various cell fractions. The mycotoxin adsorption capability of bacteria depends primarily on cell wall components [23,24], where mainly polysaccharides, such as peptidoglycan (PG) and teichoic acids, play a critical role in mycotoxin binding [31]. Besides adsorption, enzymatic degradation (e.g., [14]) is possible, making a cell-free biotechnological process for mycotoxin elimination achievable.
Elimination of zearalenone (ZEA), deoxynivalenol (DON), and aflatoxin B1 (AFB1) using forage-based isolates of the Gram-positive R. suwonensis, L. boronitolerans, L. fusiformis, and B. thuringiensis was targeted. For R. suwonensis AMK9/2 and L. boronitolerans AMK9/1, no mycotoxin elimination-related studies are published thus far, to the best of our knowledge. For L. fusiformis, there was only one study on AFB1 elimination [32] but it was only performed with viable bacteria and not with cell fractions; meanwhile, for B. thuringiensis, there are several papers available on its antagonistic effects on patulin-producing fungi and its lactonase production, which degrades patulin mycotoxin (e.g., [33]). In the present study, viable cells, cell wall debris, purified cell walls, teichoic acid, PG, and S-layer fractions were produced with heat and different chemical treatments and tested in mycotoxin elimination experiments.
Both heat and organic acids affect the adsorption of mycotoxins by the cell walls because both factors reduce the PG's thickness, increasing the pore size of the structure [22,23]. In addition, heat treatment disrupts the glycosidic bonds of cell wall polysaccharides, such as PG, and organic acids break the amide ligaments in the structure of PG [22,23]. These treatments also denature proteins and split them into smaller peptides by breaking peptide bonds and exposing more binding sites [34]. Protein denaturation caused by high temperature is also a part of pore generation, resulting in enhanced permeability in living cells [35].
Under the experimental conditions, the trichothecene mycotoxin DON was not eliminated by bacterial cells and cell fractions, unlike in other Gram-positive cells [21,36,37]. Success in controlling DON-producing fungi by microbial antagonists and detoxifying DON with microbes and enzymes was reported in the past several years [16]. El-Nezami et al. [24] reported on significant differences in the ability of bacteria to bind trichothecenes in vitro. Despite heat and acid treatments, which significantly enhanced the ability of Grampositive lactic acid bacteria to remove DON from the MRS medium [38], these parameters did not increase DON elimination yields. Results on presented Gram-positive bacteria or cell wall fractions suggest the lack of degrading enzymes or adsorption.
In Lacticaseibacillus rhamnosus, aflatoxin binding occurred through the carbohydrate and protein contents of the cell wall [22]. For B. thuringiensis AMK10/1, the S-layer fraction had the highest potential in AFB1 elimination. Lahtinen et al. [39] emphasized the significant role of the carbohydrate moieties of PG and other structures closely associated with PG in the binding process. Moreover, the teichoic acid fraction of B. thuringiensis showed a more significant AFB1 eliminating potential than the intact bacterial cells [39]. Similar observations were reported by Hernandez-Mendoza et al. [40] for Lactobacillus reuteri NRRL14171 and Lactobacillus casei Shirota [40], where the teichoic acid-deficient microbes bound significantly lower amounts of AFB1. Other lactic acid bacteria showed that the higher the D-alanine or teichoic acid contents, the higher level of AFB1 bound [41].
Teichoic acids may constitute up to 50% of the bacterial cell wall [42], but its amount is species-specific. Some reports indicated that the extraction of teichoic acids is likely influenced by the extraction time and the applied trichloroacetic acid (TCA) concentration [43]. Therefore, teichoic acid-based elimination of aflatoxins should be carefully interpreted when comparing various bacterium species. Adebo et al. [32] also suggested enzymatic degradation of AFB1 by an extracellular enzyme produced by L. fusiformis. However, the enzymatic reactions were carried out at an AFB1 concentration of 2200 ppb (3-24 h incubations), almost two orders of magnitude higher than the AFB1 concentration employed in this study.
Considering the elimination of ZEA, L. fusiformis AMK10/2 viable cells eliminated less mycotoxin than dead bacteria, indicating massive, heat-induced changes in the conformation of cell wall biopolymers, increasing the number and availability of ZEA binding sites [23]. Similarly, L. rhamnosus cell wall polysaccharide components showed good ZEA adsorption potential, and both heat and acid treatments significantly enhanced the ZEA adsorption capability of the fractions [24]. In addition, ZEA was bound predominantly by carbohydrate components of the cell wall of lactic acid bacteria [44]. Interestingly, the ZEA elimination capability of R. suwonensis AMK9/2 was identical by both viable and heat-treated bacterial cells and their fractions. Therefore, we can assume that the applied treatments did not affect cell wall structures to modify ZEA elimination yields.
Enzymatic degradation can also play a role in ZEA elimination. Tinyiro et al. [45] demonstrated that the quantities of ZEA bound by autoclaved and acid-treated cells of a Bacillus strain were identical, while suggesting that a metalloenzyme was responsible for ZEA degradation by a Bacillus natto strain. Several recent studies supported the view that bacterial esterase activities could contribute to ZEA degradation. For example, Wang et al. [19] selected ZEA-eliminating bacteria based on their esterase activities. In this study, R. suwonensis AMK9/2 showed the highest cell wall-bound esterase activity, but the esterase activities did not correlate to the eliminated ZEA in cell wall fractions. It was also demonstrated that even much higher esterase activities did not result in significantly improved ZEA eliminations (14-52%, depending on the strains) under ZEA expositions of Lactiplantibacillus plantarum [46]. Esterase activities of ruminal bacteria were also tested for deacetylating Fusarium T-2 toxin and showed varying success rates [47]. In Lactobacillus, ZEA elimination was suggested to occur through adsorption and not enzymatic degradation [23]. In the case of B. thuringiensis AMK10/1, the amounts of ZEA eliminated by the cell fractions were below those eliminated by the viable cells. The most significant ZEA release was also achieved from the viable cells, suggesting adsorption without enzymatic degradation.

Conclusions
Application of cell wall fractions for mycotoxin elimination can be more successful in some cases compared to viable cells. Cell-free technologies applying only cell fractions have a future in postharvest technologies. Industries can use bioadsorption techniques without producing metabolites with unknown physiological effects. Besides preharvest biocontrol techniques, bioadsorption can be a suitable method for decreasing mycotoxin presence. Species-specific elimination abilities could be originated from the diverse available adsorption surfaces of the microbial cell wall. However, the effect of the chemical treatment (which resulted in new surfaces and binding sites) should also be considered for its positive effect on making more potent microbial bioadsorption matrixes.

Isolation of Bacteria
Fermented forages were collected at the final stage of the fermentations (after 4-6 weeks) from different Hungarian dairy cattle farms in 2019-2020. The producers collected ten parallel samples per site from freshly opened silos and bales, combined them in sterile Velcro bags (min. 5 kg), and transferred them to the analytical laboratory for further analysis [48].
Fermented forage samples (100 g) placed in sterile homogenizing Stomacher bags were suspended in a 1:9 ratio buffered peptone water (BPW) solution (Scharlab, Barcelona, Spain) and homogenized with a Stomacher masticator homogenizer (IUL Instruments, Barcelona, Spain). Following that, decimal dilutions were made of the suspensions, and total microbial counts were determined on plate count agar (Scharlab, Barcelona, Spain) medium applying the pour plate method. Inoculated solid agar media were incubated at 30 • C for three days under either anaerobic or aerobic conditions as the standard method [49]. Bead-based lysis was carried out for 20 s at 6500 rpm with Precellys 24 homogenizer (Peqlab Biotechnologie Ltd., Erlangen, Germany). Cell debris of lysed bacteria was separated by centrifugation (10 min at 14,000 rpm). To each 600 µL aliquot of the supernatant was added 240 µL of chloroform, stirred for 30 s, and centrifuged (20 min at 14,000 rpm). Aliquots (400 µL each) of the upper aqueous phase were mixed with equal volumes of isopropanol and were centrifuged again (10 min, 14,000 rpm). After discarding the supernatants, the pellets were washed with 500 µL aliquots of 70% ethanol, re-centrifuged, and left to dry at room temperature. Genomic DNA was re-suspended in 500 µL of water and was purified further using an Amicon Ultra filtration kit (Merck Millipore, Darmstadt, Germany) following the manufacturer's instructions.
DNA amplicons were cleaned in a 0.8% agarose gel (BIOLINE), Nucleo SpinGel, and PCR clean-up column (Macherey-Nagel, Düren, Germany) kit to the protocols of the Toxins 2022, 14, 591 9 of 12 manufacturers. The PCR products were sequenced by BIOMI Ltd. (Gödöllő, Hungary). The sequences were submitted to the National Library of Medicine at the National Center for Biotechnological Information (NCBI) under accession numbers OP183257-OP183263.

Bacterial Cell Fractions
Bacterium strains were grown in nutrient broth (Scharlab; 16 h at 30 • C), and exponential phase cultures were centrifuged (8000 rpm, 10 min, 4 • C), the supernatants were removed, and the cell pellets were washed three times with 200 µL aliquots of sterile phosphate-buffered saline solution (PBS). After washing, the aliquots of the pellets were exposed to various chemicals to get different cell wall fractions based on Niederkorn et al. [31] and Goh et al.

Mycotoxin Elimination
Deoxynivalenol (DON), zearalenone (ZEA), and aflatoxin B1 (AFB1) (Biopure, Romer Labs, Tulln, Austria) calibration solutions were purchased and used in appropriate dilution for mycotoxin elimination tests. Mycotoxins were diluted to the final concentration in phosphate-buffered saline (PBS) and were added at the following concentrations to living bacterial cells and different bacterial cell wall preparations: AFB1-24 µg/L, DON-700 µg/L, ZEA-100 µg/L. The toxin concentrations applied were based on literature search, the average toxin content of the forage samples, and preliminary mycotoxin tolerance studies [48]. All mycotoxin-supplemented samples were incubated in PBS for 1 h at 25 • C with shaking (250 rpm), centrifuged (8000 rpm, 10 min, 4 • C), and the supernatants were removed, extracted, and analyzed by HPLC. All assays were performed in triplicate, and positive controls (without cells or cell wall fractions) and negative controls (without mycotoxin) were included. HPLC detection of DON was carried out on Hitachi Elit LaChrom HPLC (San Jose, CA, USA) equipment. For deoxynivalenol (DON) measurement, the filtrated supernatant samples were loaded onto a Phenomenex (Torrance, CA, USA) RP-C18 column (125 × 4 mm, 5 µm) and detected with a diode array detector in UV 218 nm with acetonitrile: water (10:90) eluent.
For AFB1, DON, and ZEA, the limits of detection (LOD) were 0.02 µg/L, 0.05 mg/L, and 2.6 µg/L, respectively. The retention times were 10.3 min, 3.2 min, and 8.9 min, respectively. The coefficient of variation within test repetitions was calculated and found to be under 15% in all cases.

Statistical Analysis
Data analyses were done in Windows Excel Analysis ToolPac (Microsoft), where a t-probe (at p ≤ 0.05) was performed for the significance analysis. A correlation (Pearson) test of esterase activity and mycotoxin elimination was also done in Windows Excel data Analysis ToolPac (Microsoft).