Molecular Docking and In Vitro Studies of Ochratoxin A (OTA) Biodetoxification Testing Three Endopeptidases

Ochratoxin A (OTA) is considered one of the main mycotoxins responsible for health problems and considerable economic losses in the feed industry. The aim was to study OTA’s detoxifying potential of commercial protease enzymes: (i) Ananas comosus bromelain cysteine-protease, (ii) bovine trypsin serine-protease and (iii) Bacillus subtilis neutral metalloendopeptidase. In silico studies were performed with reference ligands and T-2 toxin as control, and in vitro experiments. In silico study results showed that tested toxins interacted near the catalytic triad, similar to how the reference ligands behave in all tested proteases. Likewise, based on the proximity of the amino acids in the most stable poses, the chemical reaction mechanisms for the transformation of OTA were proposed. In vitro experiments showed that while bromelain reduced OTA’s concentration in 7.64% at pH 4.6; trypsin at 10.69% and the neutral metalloendopeptidase in 8.2%, 14.44%, 45.26% at pH 4.6, 5 and 7, respectively (p < 0.05). The less harmful α-ochratoxin was confirmed with trypsin and the metalloendopeptidase. This study is the first attempt to demonstrate that: (i) bromelain and trypsin can hydrolyse OTA in acidic pH conditions with low efficiency and (ii) the metalloendopeptidase was an effective OTA bio-detoxifier. This study confirmed α-ochratoxin as a final product of the enzymatic reactions in real-time practical information on OTA degradation rate, since in vitro experiments simulated the time that food spends in poultry intestines, as well as their natural pH and temperature conditions.


Introduction
About 100,000 fungi species have been identified, and more than 500 mycotoxins with toxigenic effects have been recognised [1] as a real public health issue that silently, still impacts, particularly in developing countries [2,3]. Mycotoxins are secondary metabolites produced by filamentous fungi, mainly of the Aspergillus, Fusarium and Penicillium genera [4] and have been found to infect fruits, grains, and seeds in their growth, harvest, drying and storage stages [5]. Their chemical structure along with the frequency of occurrence and the severity of the disease they produce, determines their importance and toxicity [6]. Throughout the more toxic mycotoxins, aflatoxins, fumonisins, zearalenone, certain ergot alkaloids, trichothecenes and ochratoxins, represented mainly by ochratoxin A (OTA), among others, have been considered [7,8]. Maximum levels have been set in European Union legislation to control these mycotoxin levels in food and feed [9]. Therefore, the Table 1. Chemical structure of tested and reference ligands.

Multiple Sequence Alignment
Results showed that the three studied enzymes were distantly distributed and in dif-

Multiple Sequence Alignment
Results showed that the three studied enzymes were distantly distributed and in different clusters, according to their genetic distance ( Figure 1). The phylogenetic tree showed four main clusters, suggesting a different degradation mechanism of OTA. In the first one, the serine endopeptidases were grouped. The second cluster contained the serine-type carboxypeptidases. The third big cluster contained the metallocarboxypeptidases and bromelain, a cysteine endopeptidase, and although they had a different catalytic mechanism, they share 28% sequence homology. According to an exploratory analysis, structures mostly shared a similar orientation of alpha and beta sheets surrounding the active site ( Figure 2). Further analysis must be performed to clarify these preliminary [56]

OTA
OTA had, on average, more negative (∆G) values when compared to T-2 and the reference ligand, showing that the binding is more spontaneous towards the bromelain binding site among all the evaluated ligands ( Table 2). OTA had approximately 1.3-fold higher affinity when compared to the reference ligand than the T-2 toxin. OTA's interactions with the bromelain binding site are mediated by a hydrogen bond (HB), aliphatic, and π-π T-shaped interactions with bromelain residues (Table 3). OTA's amide group is oriented towards bromelain's residues involved in the catalytic activity, Cys27, Gln20, His158 and Asn179 ( Figure 4).
According to docking studies, the inactivation mechanism in OTA by bromelain triggers the catalytic action of Cys26 that attacks the carbonyl of the amide group in OTA, resulting in α-ochratoxin and β-phenylalanine as the final products ( Figure 5).
Molecular dynamics showed that the bromelain-OTA complex maintained stable interactions (Table 4) within the active site until 50 (ns) when the complex started to dissociate. After 50 (ns), H-bond interactions started to diminish. The complex dissociated at 57 (ns) and OTA unbound from the enzyme ( Figure 6). Aliphatic: Ile340, Leu360, Ile366.
Although T-2 ester groups were located near the involved amino acids in the catalytic triad, given the nature of the catalytic reaction carried out by bromelain, the epoxide ring and the two ester groups found in T-2 are not going to be broken; instead, due to T-2′s high affinity (Ki = 101.78 µM) and binding probability to form a complex (ΔG = −5.45 kcal/mol) it could function as a possible competitive inhibitor ( Figure 3).  and π-π T-shaped interactions with bromelain residues (Table 3). OTA's amide group is oriented towards bromelain's residues involved in the catalytic activity, Cys27, Gln20, His158 and Asn179 ( Figure 4). According to docking studies, the inactivation mechanism in OTA by bromelain triggers the catalytic action of Cys26 that attacks the carbonyl of the amide group in OTA, resulting in α-ochratoxin and β-phenylalanine as the final products ( Figure 5).  Molecular dynamics showed that the bromelain-OTA complex maintained stable interactions (Table 4) within the active site until 50 (ns) when the complex started to dissociate. After 50 (ns), H-bond interactions started to diminish. The complex dissociated at    Moreover, in vitro differences between groups were found (ANOVA p = < 0.0001; F = 16.87). Sidak's multiple comparisons tests showed that bromelain reduced OTA concentration by 7.64 (%) at pH 4.6 (p = < 0.0001; t = 6.496; DF = 12; Figure 7). However, although the OTA ion was detected (404.08 m/z) the final product, α-ochratoxin, suggested in docking was not detected in chromatograms (212.85 m/z; Figure 8). Moreover, in vitro differences between groups were found (ANOVA p ≤ 0.0001; F = 16.87). Sidak's multiple comparisons tests showed that bromelain reduced OTA concentration by 7.64 (%) at pH 4.6 (p ≤ 0.0001; t = 6.496; DF = 12; Figure 7). However, although the OTA ion was detected (404.08 m/z) the final product, α-ochratoxin, suggested in docking was not detected in chromatograms (212.85 m/z; Figure 8).

Bovine Trypsin Serine Protease 2.5.1. T-2 Toxin
MXH reference ligand showed a higher free binding energy (∆G) compared to T-2, and its affinity (K i ) for trypsin was approximately 272-fold lower, indicating a higher affinity for the reference ligand (MXH; Table 2). T-2 toxin interactions are due to hydrogen bonds (HB), alkyl, and Van der Waals with trypsin residues (Table 3). T-2 is oriented towards trypsin's residues involved in the catalytic activity, Asp102, His57 and Ser195 (Figure 9). tested tubes (M + 1) at 404.08 (m/z). The molecular ion (M + 1) of α-ochratoxin was not detect (fragmentation at 212.85 m/z).

T-2 Toxin
MXH reference ligand showed a higher free binding energy (ΔG) compared to T and its affinity (Ki) for trypsin was approximately 272-fold lower, indicating a higher finity for the reference ligand (MXH; Table 2). T-2 toxin interactions are due to hydrog bonds (HB), alkyl, and Van der Waals with trypsin residues (Table 3). T-2 is oriented t wards trypsin's residues involved in the catalytic activity, Asp102, His57 and Ser195 (Fi ure 9). According to the docking studies, the inactivation mechanism in T-2 by trypsin tri gers the catalytic action of Ser195 that attacks the carbonyl of the ester group in OT resulting in α-ochratoxin and β-phenylalanine as the final products ( Figure 10). According to the docking studies, the inactivation mechanism in T-2 by trypsin triggers the catalytic action of Ser195 that attacks the carbonyl of the ester group in OTA, resulting in α-ochratoxin and β-phenylalanine as the final products ( Figure 10).
Molecular dynamics showed that the trypsin-T2 complex maintained stable interactions (Table 4)  Molecular dynamics showed that the trypsin-T2 complex maintained stable interactions (Table 4) within the active site until 60 (ns), when the complex started to dissociate. After this time, H-bond and Van der-Waals interactions diminish. The complex dissociated at 63 (ns) and T-2 unbound from the enzyme ( Figure 11). In vitro experiments testing trypsin's capacity to hydrolyse T-2 toxin, showed no significant changes when comparing experimental samples at different pH conditions (ANOVA p = 0.07; F = 2.748; Figure 12); however, Sidak's multiple comparisons found that the average percentage of the three repetitions at pH 4.6 had a significant decrease in T-2 concentration in 10.68% (p = 0.04; t = 2.86; DF = 12). Further, when HPLC-TOF-MS chromatograms were analysed, Both, T-2 and HT-2 corresponding ions were present, 484.25 (m/z) and 442.24 (m/z), for T-2 and HT-2, respectively ( Figure 13). Since this study is mainly related to OTA hydrolysis, the related T-2 hydrolysis by trypsin is not discussed, due to the marginal results obtained in this study, further suggesting that extensive studies are required in this regard. In vitro experiments testing trypsin's capacity to hydrolyse T-2 toxin, showed no significant changes when comparing experimental samples at different pH conditions (ANOVA p = 0.07; F = 2.748; Figure 12); however, Sidak's multiple comparisons found that the average percentage of the three repetitions at pH 4.6 had a significant decrease in T-2 concentration in 10.68% (p = 0.04; t= 2.86; DF = 12). Further, when HPLC-TOF-MS chromatograms were analysed, Both, T-2 and HT-2 corresponding ions were present, 484.25 (m/z) and 442.24 (m/z), for T-2 and HT-2, respectively ( Figure 13). Since this study is mainly related to OTA hydrolysis, the related T-2 hydrolysis by trypsin is not discussed, due to the marginal results obtained in this study, further suggesting that extensive studies are required in this regard.

OTA
MXH reference ligand showed a higher free binding energy (∆G) compared to OTA, and its affinity (K i ) for trypsin was approximately 94-fold lower, indicating a higher affinity for the reference ligand (MXH; Table 2). OTA molecular interactions are mediated by hydrogen bonds (HB), amide π, and Van der Waals interactions with trypsin residues (Table 3). OTA's amide group is oriented towards trypsin's residues involved in the catalytic activity, Ser195, His57 and Asp102 ( Figure 14).

OTA
MXH reference ligand showed a higher free binding energy (ΔG) compared to OTA, and its affinity (Ki) for trypsin was approximately 94-fold lower, indicating a higher affinity for the reference ligand (MXH; Table 2). OTA molecular interactions are mediated by hydrogen bonds (HB), amide π, and Van der Waals interactions with trypsin residues (Table 3). OTA's amide group is oriented towards trypsin's residues involved in the catalytic activity, Ser195, His57 and Asp102 ( Figure 14). According to docking studies, the inactivation mechanism in OTA by trypsin triggers the catalytic action of Ser195 that attacks the carbonyl of the amide group in OTA, resulting in α-ochratoxin and β-phenylalanine as the final products ( Figure 15). According to docking studies, the inactivation mechanism in OTA by trypsin triggers the catalytic action of Ser195 that attacks the carbonyl of the amide group in OTA, resulting in α-ochratoxin and β-phenylalanine as the final products ( Figure 15). Molecular dynamics showed that the trypsin-OTA complex maintained stable interactions throughout the 100 (ns; Table 4), despite that OTA had movement on its own axis, it always remained near trypsin's active site and the complex never dissociated ( Figure  16). Molecular dynamics showed that the trypsin-OTA complex maintained stable interactions throughout the 100 (ns; Table 4), despite that OTA had movement on its own axis, it always remained near trypsin's active site and the complex never dissociated ( Figure 16).

T-2 Toxin
T-2 toxin showed a similar free binding energy (∆G) compared to those of the reference ligand (DB0763; Table 2). However, its affinity (K i ) for neutral metalloendopeptidase was approximately 17-fold higher than the reference ligand, indicating a very low affinity for the protein. T-2 showed hydrogen bond (HB), aliphatic and π-σ-mediated interactions with amino acid residues of the ligand binding site (Table 3). Although T-2 ester groups were located near the involved amino acids in the catalytic triad, given the nature of the catalytic reaction carried out by neutral metalloendopeptidase, the epoxide ring and the two ester groups found in T-2 are not going to be broken. Instead, it could function as a possible competitive inhibitor ( Figure 19).

OTA
OTA's most negative value (∆G) value was -8.56, which is very similar to that of the reference ligand (DB07673) when both were calculated by Swiss Dock, which is considered the most precise software when metallic ions are present ( Table 2). Reference ligand showed 27.7-fold higher affinity (K i ) when compared to OTA and neutral metalloendopeptidase.
OTA's interactions with neutral metalloendopeptidase binding site residues are mediated by a hydrogen bond (HB) and aliphatic interactions (Table 3). OTA's peptide bond is oriented towards neutral metalloendopeptidase residues involved in the catalytic activity, His373, His369, His453 and Glu393 (Figure 20).
The inactivation mechanism in OTA by neutral metalloendopeptidase triggers the catalytic action of the Zn that attacks the carbonyl of the amide group in OTA, resulting in α-ochratoxin and β-phenylalanine as final products (Figure 21). Molecular dynamics showed that metalloendopeptidase-OTA complex maintained stable interactions throughout the 100 (ns; Table 4), despite OTA slightly moving from the catalytic triad, but remaining within the ligand binding site and the complex never dissociated (Figure 22). for the protein. T-2 showed hydrogen bond (HB), aliphatic and π-σ-mediated interactions with amino acid residues of the ligand binding site (Table 3). Although T-2 ester groups were located near the involved amino acids in the catalytic triad, given the nature of the catalytic reaction carried out by neutral metalloendopeptidase, the epoxide ring and the two ester groups found in T-2 are not going to be broken. Instead, it could function as a possible competitive inhibitor (Figure 19).

OTA
OTA's most negative value (ΔG) value was -8.56, which is very similar to that of the reference ligand (DB07673) when both were calculated by Swiss Dock, which is considered the most precise software when metallic ions are present ( Table 2). Reference ligand showed 27.7-fold higher affinity (Ki) when compared to OTA and neutral metalloendopeptidase.

Discussion
Biological detoxification using enzymes extracted from natural products, or produced by microorganisms, seems to be one of the most promising approaches to reduce or completely remove mycotoxin contamination from food [40]. It is especially worth trying to obtain inexpensive, easy-to-extract and resistant enzymes. In this study, the original predictions were that the proteases, bromelain cysteine from Ananas comosus, bovine trypsin serine and the neutral metalloendopeptidase from Bacillus sutbtilis would biodetoxify OTA by hydrolysing its peptide bond, according to the shared characteristics between the studied enzymes and those previously reported OTA degrading enzymes, including the type of reaction carried out by the enzyme, functional parameters, molecular properties, sequence and structural homology (See Figure 1). To our knowledge this is the first attempt to demonstrate that: (i) bromelain and trypsin are capable of hydrolysing OTA at acidic pH conditions but with low effectiveness; and (ii) the Bacillus sutbtilis neutral metalloendopeptidase was demonstrated to have a high efficacy as an OTA biodetoxifier. Moreover, in all the studied enzymes, both results obtained by docking and in vitro, were consistent, although in vitro experiments showed differences in enzymatic efficiency to hydrolyse OTA. Besides, α-ochratoxin was confirmed as a final product of the enzymatic reaction with trypsin and the metalloendopeptidase due to OTA's degradation and not due to adsorption. These results are of practical use because pH conditions and contact time are comparable to the digestion process of poultry, constituting an important contribution of this study. This section will discuss the most important findings and their implications are discussed.
This study is the first attempt at evaluating the capacity of bromelain for OTA detoxification, despite being an easy-to-obtain, low-cost and highly resistant enzyme. Although findings by docking showed it was likely that OTA might be detoxified by bromelain, and despite experiments being performed at three different pH levels, results were only significant at the most acidic pH (4.6); however, hydrolysis per cent was low. Since bromelain is a cysteine/thiol-type enzyme containing a Cys-His catalytic dyad, this group of proteases is generally cut from the left side [59]. Thus, catalysis is unlikely to occur if OTA couples backwards to the active site. Moreover, as a protease, bromelain acts specifically, and the peptide bond of Arg-X is reported as its preferable cleavage site [60]. It also needs to recognise the Arg side chain. Since OTA structures do not have these similarities with Arg, it is, therefore, possible that the enzyme may not recognise the cleavage site efficiently. Results showed that although OTA reduced its concentration in the presence of bromelain, there was a lack of detection of the α-ochratoxin ion. This could be because the bromelain degradation rate might be very low and does not accumulate enough α-ochratoxin to be detected. However, further studies are needed to better clarify these assertations including a wide range of pH's, temperatures, incubation times, substrate concentrations, and other cofactors. Moreover, molecular dynamics studies showed that the OTA-bromelain complex had stability until 50 (ns), suggesting that interactions in the binding site are strong enough to keep the ligand near to the catalytic triad during the time required to hydrolyse OTA. Further studies are needed to delve into this matter.
The results demonstrated that trypsin was effective in hydrolysing OTA at pH 4.6. Since chick's intestinal pH is acidic, our results suggest that trypsin may be able to hydrolyse OTA in vivo, although OTA's degradation percentage was below 20%. The only previous study in which OTA hydrolysis using trypsin was evaluated at pH 7.5 and 25 • C, found no OTA hydrolysis [26]. Future studies using trypsin in OTA biotransformation should develop in vivo assays. Further, molecular dynamics demonstrated that OTA was always next to the trypsin active site. This result suggest that the protein-ligand complex is therefore stable and its low degradation rate as demonstrated in in vitro experiments may be due either to non-optimum experimental conditions, or due to OTA being a long time near the active site, as shown in the dynamics, it could be hydrolysed and the product could be behaving as an inhibitor if it has a high affinity for the trypsin active site. However, this statement requires further investigation.
Results showed that the neutral metalloendopeptidase from Bacillus sutbtillis was able to degrade OTA in acidic and neutral pH conditions. Similar values have been previously reported for other metallopeptidases such as carboxypeptidases, from Bacillus amyloliquefaciens [31] and from Aspergillus niger [36] where OTA was found to be degraded by 72 and 99%, respectively, and others [26,27]. Alternatively, the similarity of results found in this study between the pH conditions 4.6 and 5, can be explained in terms of OTA's ionisation value (4.4 pKa) [61], in which, regardless of OTA's carboxyl ionisation state, non-significant differences in the mycotoxin transformation were observed. In fact, in the proposed degradation mechanism of OTA, the carboxyl does not take part in the hydrolysis/ionisation. Moreover, although the neutral metalloendopeptidase from Bacillus sutbtilis in this study, degraded OTA at pH 5, it was greater at pH 7, which agrees with those found for a carboxypeptidase by Abrunhosa et al. (2006) [36]. This study reported as optimal pH conditions, values between 6.5 and 7.5. However, they did not obtain OTA degradation testing this enzyme at 50 ( • C). Results showed that the metalloendopeptidase effectively degrades OTA at 41 ( • C). Further, this is the first attempt to report a metalloendopeptidase with the ability to hydrolyse OTA. According to molecular dynamics results, the metalloendopeptidase-OTA complex stays stable over time, what matches with the high amount of OTA degradation, obtained in vitro.
Results showed that the final OTA metabolite was α-ochratoxin as a product of the reaction in the three tested enzymes. OTA's enzymatic detoxification could be achieved by the hydrolysis of either: (i) the amide bond to generate α-ochratoxin, and L-phenylalanine by using an amido-hydrolase, or (ii) the lactone ring by using an ochratoxin-lactonase [17,19,24]. However, when compared with the lactone ring opening, the hydrolysis of the amide bond between the phenylalanine and α-ochratoxin, as obtained in this study, has been widely recognised as essentially non-toxic products. α-ochratoxin is the isocoumarin section of OTA. Transforming OTA into α-ochratoxin is an efficient way to reduce not only its concentration but its toxicity [40], since α-ochratoxin has been described as the less toxic member of ochratoxins, where OTA, the most toxic is followed by OTC, OTB and finally by α-ochratoxin [62][63][64][65]. In fact, α-ochratoxin has been found to be 1000-times less toxic in brain cell cultures and its elimination half-life in vivo to be ten-times faster than for OTA [36], and to date, has been further chemically identified, as an OTA degradation product by microorganisms and enzymes [41].
Further, the (2S)-2-methyl-3-phenylpropanoic acid (DB07673), the chosen reference ligand for the neutral metalloendopeptidase protease docking, had the highest affinity (K i ) when compared with the tested ligands (T-2 and OTA). This molecule is an inhibitor by nature, due to its relatively small size, its phenyl and carboxylic acid groups, and the fact that at pH 7 it gets deprotonated. Such traits allow it to easily enter the catalytic site, where its deprotonated carboxylic acid group can perfectly coordinate with the zinc metal ion [66]. Other metallopeptidase inhibitors have also observed the same behaviour [67][68][69][70].
Results showed that trypsin hydrolyses T-2 toxin, even though it was with a low efficiency. However, molecular docking studies showed that T-2 is near the catalytic triad, and molecular dynamics results showed that the trypsin-T2 complex was stable, on the whole, further suggesting that optimal experimental conditions have not yet been reached for a higher hydrolysis yield. Moreover, considering that two out of the three enzymes studied, do not break ester bonds, the T-2 toxin was used as a negative control. However, given its arrangement within both, bromelain and neutral metalloendopeptidase binding Sites, as found in docking, it is therefore possible that the toxin could act as a natural inhibitor. If this statement were true, the amount of enzyme could be increased so it can be inhibited by T-2 leaving the rest free so it might be available to act with OTA. Further, if T-2 is interacting with the enzyme, then although the enzyme cannot degrade it, it would be limited and could not be adsorbed in the animal's intestine. Since mycotoxins often co-occur in food and feed, it has been shown that they can act harmfully due to interactions between them [71,72]; a stronger synergistic effect has been found for mixtures containing T-2 toxin and OTA [73]. Therefore, results obtained in this study further highlight the importance of continuing to seek effective, environmentally and specific enzymatic options that can detoxify not only one, but several mycotoxin types.
Molecular docking was performed by using blind and directed approaches. Blind docking was used to explore binding sites in the whole crystal to identify or discard putative allosteric binding sites using a grid box of 120 × 120 × 120 Å. Directed docking was used to evaluate ligand interactions directly in the ligand binding site of each protein, and a grid-box of 60 × 60 × 60 Å was centred on the site where their reference ligand was removed from the crystallised protein. While a Lamarckian genetic algorithm with an initial population of 100 random individuals and 1 × 10 7 iterations was used for Autodock Vina and AutoDock 4.2. [83], an "accurate" docking type was selected in SwissDock. Results were analysed for affinity estimation values and interactions with AutoDockTools 1.5.6 [84]. Ligand affinity for the protein was evaluated by calculating binding free energy values (∆G) and were calculated in Autodock Vina, SwissDock and Autodock4, where the more negative the value, the more spontaneous is the binding between ligand-protein. The theoretical value of inhibition (K i ) was calculated in Autodock4, where smaller values indicate a higher ligand affinity for the protein. Images were performed in Discovery Studio ® [85] and PyMOL ® [86]. To validate the docking procedure, the re-docking of the removed ligand on the enzyme was performed and evaluated.

Molecular Dynamics
To perform the equilibrium molecular dynamics (EMD), files from Autodock 4 tool ligand-protein complexes (.pdb) were prepared using Visual Molecular Dynamics software (VMD v.1.9.4). A water box 10 Å spaced from the protein edge was constructed. The potential CHARMM36m was used to represent the molecules' force field [87], and the system was neutralized using NaCl ions (0.15 mM). A Langevin dynamic was established to maintain 310 K (36.85 • C). While in the minimization process 2500 steps were used, for the equilibration process (NVT) 2,500,000 steps were (5 ns), and for the final simulation time (NPT) 50,000,000 steps 100 (ns). The time step was set to 2 fs/step. EMD were performed using the NAMD 2.14 software [88]. Trajectories were analysed by calculating the mean root square deviation (RMSD) with the RMSD plugin using the Visualizer Tool of VMD 1.9.4. Finally, trajectories were plotted using the Xmgrace software.

Residual OTA Analysis
Initial OTA quantification was assessed using a reverse phase with an HPLC (Agilent Technology 1100 series) coupled to a fluorescence detector (Perkin Elmer, Beaconsfield, Buckinghamshire, UK; LS50B), at an excitation wavelength of 345 (nm) and an emission wavelength of 455 (nm). A column DISCOVERY C-18 (250 × 4.6 mm; 5-micron, Supelco, USA) was used at a temperature of 25 ( • C). The mobile phase consisted of acetonitrile: 0.25 N phosphoric acid (50:50). The run was isocratic with a constant flow of 1 (mL/min). Data collection was done by FL Winlab (version 2.0; The Perkin-Elmer Corporation). Samples were injected in triplicate and chromatograms were compared with those of their own control.
To identify final metabolites, when the tested enzyme significantly reduced OTA in samples compared with their own control, a reverse phase HPLC-ESI-TOF-MS (High Pressure Liquid Chromatography-Electrospray-Time of Flight-Mass spectrometry) was used, in positive ion mode, using an Agilent 1260 Infinity HPLC system equipped with a C-18 column (2.1 × 100 mm; 1.8 micron, ZORBAX Eclipse Plus, CA, USA); 1.8 Micron (Agilent Technologies, CA, USA) at a temperature of 25 ( • C). The mobile phase consisted of water: formic acid (1%) and acetonitrile: formic acid (1%) with a flow rate of 0.15 (mL/min) and an injection volume of 40 (µL). The HPLC was coupled to a TOF/MS (Agilent 66230B) with an electrospray interface [89]. Gas temperature was 350 ( • C), gas flow 6 (L/min) and nebuliser pressure 50 (psig), shredder 100 (V), skimmer 65 (V) and OCT RF vpp 750 (V), capillary voltage 3500 (V). Final precursor ions were identified at 404.05 and 212.84 (m/z), OTA and α-ochratoxin, respectively. The total range examined was from 100 to 1000 (m/z). Samples were injected in triplicate and data were registered with the acquisition software Mass Hunter Workstation LC/MS Data Acquisition version B.05.01.

Residual T-2 Analysis
T-2 quantification was carried out by reverse phase HPLC-ESI-TOF-MS, in a positive ion mode using an Agilent 1260 Infinity HPLC (Agilent Technologies, Yishun, Singapore) system equipped with an RRHD Eclipse Plus C-18 column (1.8 µm, 2.1 × 100 mm; Agilent Technologies, USA); at 25 ( • C). The mobile phase consisted of (A) ammonium acetate (10 mM) and (B) HPLC-grade methanol. The gradient started with 80 (%) A and 20 (%) B. Flow rate was set to 0.15 (mL/min) and 40 (µL) for the sample injection volume. The HPLC was coupled to an Agilent 66230B TOF/MS with an electrospray interface. The gas temperature was at 350 ( • C), gas flow 6 (L/min) and nebuliser pressure 50 (psig), shredder 105 (V), skimmer 60 (V) and Oct RF 750 (V), capillary voltage 4000 (V). The final precursor ion for T-2 was identified at 484.25 (m/z) and 442.24 (m/z) for HT-2. The total range examined was from 100 to 1000 (m/z). Samples were injected in triplicate. Data were analysed in the Mass Hunter Data Acquisition software for 6200 series (version 5.01.5125; Agilent Technologies, USA) and Qualitative Analysis (version 6.0.633.10 (Aligent Technologies, 2017).

Data Analysis
To determine whether differences in OTA and T-2 concentrations (µg/L) exist between control and tested tubes under different pH and temperature conditions, one-way multiple analysis of variance (ANOVA) was performed, followed by Sidak's post hoc tests [90,91]. A p < 0.05 was considered significant. Analyses and figures were performed using Prism 8 ® Version 8.4.0 for Mac OS, (GraphPad Software, Inc., La Jolla, CA 92037, USA).

Conclusions
Overall, this study provides consistent in silico and in vitro evidence, that bromelain and trypsin are capable of hydrolysing OTA in acidic pH conditions at a low efficiency; The Bacillus sutbtilis neutral metalloendopeptidase has a great effectiveness as an OTA bio detoxifier, confirming α-ochratoxin as a final product of the enzymatic reactions and providing real-time practical information on OTA degradation rate.