A Theoretical Study of 8-Chloro-9-Hydroxy-Aflatoxin B1, the Conversion Product of Aflatoxin B1 by Neutral Electrolyzed Water

Theoretical studies of 8-chloro-9-hydroxy-aflatoxin B1 (2) were carried out by Density Functional Theory (DFT). This molecule is the reaction product of the treatment of aflatoxin B1 (1) with hypochlorous acid, from neutral electrolyzed water. Determination of the structural, electronic and spectroscopic properties of the reaction product allowed its theoretical characterization. In order to elucidate the formation process of 2, two reaction pathways were evaluated—the first one considering only ionic species (Cl+ and OH−) and the second one taking into account the entire hypochlorous acid molecule (HOCl). Both pathways were studied theoretically in gas and solution phases. In the first suggested pathway, the reaction involves the addition of chlorenium ion to 1 forming a non-classic carbocation assisted by anchimeric effect of the nearest aromatic system, and then a nucleophilic attack to the intermediate by the hydroxide ion. In the second studied pathway, as a first step, the attack of the double bond from the furanic moiety of 1 to the hypochlorous acid is considered, accomplishing the same non-classical carbocation, and again in the second step, a nucleophilic attack by the hydroxide ion. In order to validate both reaction pathways, the atomic charges, the highest occupied molecular orbital and the lowest unoccupied molecular orbital were obtained for both substrate and product. The corresponding data imply that the C9 atom is the more suitable site of the substrate to interact with the hydroxide ion. It was demonstrated by theoretical calculations that a vicinal and anti chlorohydrin is produced in the terminal furan ring. Data of the studied compound indicate an important reduction in the cytotoxic and genotoxic potential of the target molecule, as demonstrated previously by our research group using different in vitro assays.


Introduction
The contamination of food and feed with aflatoxins (a family of foodborne carcinogenic mycotoxins) is a serious problem for human and livestock health, and consequently has serious effects for agricultural economics. Once a commodity has been identified as being contaminated beyond a level fit for human or animal consumption, the problem arises as to what should be done with it. The producer of a highly aflatoxin-contaminated commodity may be faced with the problem of its disposal, unless some treatment may be found. Many experiments have therefore been performed to reduce the level of aflatoxins in contaminated materials using physical, biological or chemical approaches [1].
The value of the aflatoxins decontamination depends on the method used and the remaining cytotoxic and genotoxic potential of the treated material. The "ideal" decontamination process should have the following characteristics: easy to use, inexpensive, and free of the potential to produce compounds that remain or may reverse to form the parent mycotoxin, as in the case of nixtamalized products [2]. Recently, an environmentally friendly aflatoxin detoxification process was proposed in which maize grains were soaked in neutral electrolyzed water (NEW), filtered and dried at room temperature [3]. NEW treatment offers multiple advantages not found in other detoxification methods, including less chemical residues, more security, energy-savings and cost-effectiveness. NEW is produced by electrolysis of water-with no added chemicals, except for sodium chloride-in an electrolytic cell. In the anode, chloride ions and water molecules are transformed into chlorine oxidants such as hypochlorous acid (HOCl), hypochlorite ions (ClO´) and chlorine (Cl 2 ) [4]. The antimicrobial mechanism of NEW depends mainly on three physicochemical properties: pH, oxidation-reduction potential (ORP) and available chlorine concentration (ACC) [5].
In a previous work, our research group proposed that the detoxification process eliminates the aflatoxin-associated cytotoxicity and genotoxicity effects due to the fact that NEW reacted with the double bond in the terminal furan ring of the aflatoxin B 1 (1) molecule to yield 8-chloro-9-hydroxy-aflatoxin B 1 (2) [3]. It was hypothesized that the most important factor in 1 transformation was the high level of ACC, taking into account that NEW contained primarily hypochlorous acid («95%), hypochlorite ions («5%) and trace amounts of Cl 2 . Although the analyses of 2 with Fourier transform ion cyclotron resonance mass spectrometry and 1 H nuclear magnetic resonance have been reported [6], there is no specific information available concerning the structural, electronic and spectroscopic properties of the target molecule. Thus, the goal of this article is to inform novel and interesting results of 2 related to a theoretical study by applying quantum chemistry methods: (i) conformational analysis; (ii) the fully optimized molecule structure using Density Functional Theory (DFT) calculations at the level of theory defined by the Becke's three-parameter hybrid functional (B3LYP), employing the 6-311++G(d,p) basis set; (iii) determination of the structural-geometrical parameters including bond lengths and angles; and (iv) calculation of electronic properties such as natural atomic charges, thermo chemistry, highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO), performed both in gas-phase and water as the solvent. Moreover, the molecular prediction about toxicological risks and some physicochemical properties of 2 were obtained using the OSIRIS-Property-Explorer (Actelion Pharmaceuticals Ltd, Allschwil, Switzerland). After a careful analysis of the obtained data, a plausible mechanism is now proposed in order to explain the formation of 2 in the process of aflatoxin detoxification with NEW.

Results and Discussion
As it is well known, haloydrins are common chemical addition products of alkenes [7][8][9]. In the case of 1, the reaction to generate the corresponding chlorohydrin was previously reported by Xiong et al. [6] and Jardon-Xicotencatl et al. [3], as a strategy for aflatoxin detoxification, specifying the achievement of 8-chloro-9-hydroxy-aflatoxin B 1 stereoisomer (Figure 1).

Aflatoxin B1 Chlorohydrin Optimal Structure
As stated above, sixteen possible isomers (stereoisomers and regioisomers) of 2 were fully optimized in the first phase of this study using the DFT level. The observed facts for these isomers were the chlorine and hydroxyl connectivity in position C8 or C9, and the conformational arrangement syn or anti regarding the hydrogen atoms H9a and H6a, respectively ( Figure 2

Aflatoxin B 1 Chlorohydrin Optimal Structure
As stated above, sixteen possible isomers (stereoisomers and regioisomers) of 2 were fully optimized in the first phase of this study using the DFT level. The observed facts for these isomers were the chlorine and hydroxyl connectivity in position C 8 or C 9 , and the conformational arrangement syn or anti regarding the hydrogen atoms H 9a and H 6a , respectively ( Figure 2).  Figure 1. Formation of the 8-chloro-9-hydroxy aflatoxin B1 compound (2) from aflatoxin B1 (1) and hypochlorous acid from neutral electrolyzed water (NEW).

Aflatoxin B1 Chlorohydrin Optimal Structure
As stated above, sixteen possible isomers (stereoisomers and regioisomers) of 2 were fully optimized in the first phase of this study using the DFT level. The observed facts for these isomers were the chlorine and hydroxyl connectivity in position C8 or C9, and the conformational arrangement syn or anti regarding the hydrogen atoms H9a and H6a, respectively ( Figure 2   The energy values of each isomer at the same level are summarized in Table 1. The less stable isomer corresponded to structure 2N (-1,030,730.4 kcal/mol), as a consequence of the steric hindrance by the nearest atoms. In contrast, the most stable isomer was 2H (-1,030,762.1 kcal/mol), with the chlorine and hydroxyl groups as expected according to this class of addition -anti position, with an energy difference of 31.77 kcal/mol. Thus, the structure 2H was used in subsequent calculations and this isomer is named as 2.

Reaction Mechanism
In order to elucidate the formation process of 2, two reaction pathways were considered according to the previously reported typical addition mechanism [10], one with only ionic species of the HOCl (Path A), and the second, considering the entire HOCl molecule (Path B). These two pathways are shown in Figure 3. Additionally, it is worth mentioning that both pathways were studied in gas and solution phases. The energy values of each isomer at the same level are summarized in Table 1. The less stable isomer corresponded to structure 2N (-1030730.4 Kcal/mol), as a consequence of the steric hindrance by the nearest atoms. In contrast, the most stable isomer was 2H (-1030762.1 Kcal/mol), with the chlorine and hydroxyl groups as expected according to this class of addition -anti position, with an energy difference of 31.77 kcal/mol. Thus, the structure 2H was used in subsequent calculations and this isomer is named as 2.

Reaction Mechanism
In order to elucidate the formation process of 2, two reaction pathways were considered according to the previously reported typical addition mechanism [10], one with only ionic species of the HOCl (Path A), and the second, considering the entire HOCl molecule (Path B). These two pathways are shown in Figure 3. Additionally, it is worth mentioning that both pathways were studied in gas and solution phases.   The reaction begins-in path A and B-with an electrophilic attack by the double bond of 1 to the chlorenium ion (Path A: ionic), or hypochlorous acid (path B: molecular) succeeding the first ionic activated state (TSi1 for path A) or the molecular activated state (TSm1 for path B), followed by the production of the chloronium ion reactive intermediate (1a). In a second step, the hydroxide ion attained a nucleophilic attack to 1a, which produced the second activated state (TS2), and concluded with the formation of 2 [11,12].

Geometrical Parameters Analysis of the Structures Involved in the Mechanism
Regarding the active species generation process, the optimized structures are shown in Figure 4. Selected geometrical parameters of the structures are summarized in Tables 2 and 3. The reaction starts as suggested, with the interaction between the double bond at C 8 -C 9 atoms of 1 and the chlorine atom (chlorenium ion or hypochlorous acid) to form the expected chloronium ion. The bond lengths ( Table 2) of TSi1 show an increase of the distance between C 8 -C 9 atoms of 1. Additionally, the distance of C 8 -C 9 with the chlorine atom (1.95 and 2.35 Å, respectively) demonstrates a bond formation. However, a minor distance between C 8 -Cl atoms was observed, indicating a stronger interaction. Thus, a slight decrease in the distance and angle ( Table 2) between C 9b and C 9 atoms was also observed. The reaction begins-in path A and B-with an electrophilic attack by the double bond of 1 to the chlorenium ion (Path A: ionic), or hypochlorous acid (path B: molecular) succeeding the first ionic activated state (TSi1 for path A) or the molecular activated state (TSm1 for path B), followed by the production of the chloronium ion reactive intermediate (1a). In a second step, the hydroxide ion attained a nucleophilic attack to 1a, which produced the second activated state (TS2), and concluded with the formation of 2 [11,12].

Geometrical Parameters Analysis of the Structures Involved in the Mechanism
Regarding the active species generation process, the optimized structures are shown in Figure 4. Selected geometrical parameters of the structures are summarized in Tables 2 and 3. The reaction starts as suggested, with the interaction between the double bond at C8-C9 atoms of 1 and the chlorine atom (chlorenium ion or hypochlorous acid) to form the expected chloronium ion. The bond lengths ( Table 2) of TSi1 show an increase of the distance between C8-C9 atoms of 1. Additionally, the distance of C8-C9 with the chlorine atom (1.95 and 2.35 Å, respectively) demonstrates a bond formation. However, a minor distance between C8-Cl atoms was observed, indicating a stronger interaction. Thus, a slight decrease in the distance and angle ( Table 2) between C9b and C9 atoms was also observed.        In the same way, a similar behavior in the activated state in path B was found. The bond at C 8 -C 9 atoms increased in similar value to that of TSi1. On the other hand, the distance of C 8 -Cl atoms was shorter, while C 9 -Cl was longer compared to the same bonds in TSi1, indicating the preference of the chlorine addition at the C 8 atom. The rest of the bonds in TSi1 and TSm1 agree with the experimental values and theoretical calculation for 1. After the addition of the chlorine atom on the double bond, a reactive intermediate is produced. However, the proposed structure of the chloronium ion ( Figure 3) was not obtained; instead, a non-classic carbocation was formed ( Figure 5) [13].  In the same way, a similar behavior in the activated state in path B was found. The bond at C8-C9 atoms increased in similar value to that of TSi1. On the other hand, the distance of C8-Cl atoms was shorter, while C9-Cl was longer compared to the same bonds in TSi1, indicating the preference of the chlorine addition at the C8 atom. The rest of the bonds in TSi1 and TSm1 agree with the experimental values and theoretical calculation for 1. After the addition of the chlorine atom on the double bond, a reactive intermediate is produced. However, the proposed structure of the chloronium ion ( Figure 3) was not obtained; instead, a non-classic carbocation was formed ( Figure 5) [13]. The change in the structure of the reactive intermediate 1a is a consequence of the interaction of the nearest aromatic system, which opens the chloronium ion and delocalizes the positive charge in the carbons of the benzene ring [13,14]. The electrostatic potential molecular surface (Figure 5b) showed a blue surface denoting the cationic character of 1a; furthermore, the zones of the benzene and lactone rings had the most intense color confirming the electronic delocalization in these moieties. This process produced a cyclopropane moiety between C9, C9a and C9b with a value of 1.47 Å for C9-C9a atoms, shorter than in 1, and a value of 1.53 Å for C9a-C9b atoms, which is longer in relation to the same bond in 1. Finally, the bond length value for C9-C9b atoms corresponds to 1.58 Å. The change in the structure of the reactive intermediate 1a is a consequence of the interaction of the nearest aromatic system, which opens the chloronium ion and delocalizes the positive charge in the carbons of the benzene ring [13,14]. The electrostatic potential molecular surface (Figure 5b) showed a blue surface denoting the cationic character of 1a; furthermore, the zones of the benzene and lactone rings had the most intense color confirming the electronic delocalization in these moieties. This process produced a cyclopropane moiety between C 9 , C 9a and C 9b with a value of 1.47 Å for C 9 -C 9a atoms, shorter than in 1, and a value of 1.53 Å for C 9a -C 9b atoms, which is longer in relation to the same bond in 1. Finally, the bond length value for C 9 -C 9b atoms corresponds to 1.58 Å. Moreover, the angle for C 9 -C 9a -C 9b decreases the value to 63.6˝, which is consistent with the expected values for cyclopropane structures. The next reaction step is the nucleophilic attack of the hydroxide ion at C 9 atom to form 2 via the second activated state (TS2). In this sense, the closeness of the hydroxide ion to C 9 (2.51 Å) causes the enlargement of the C 9 -C 9b bond (from 1.582 in 1a to 1.680 Å); additionally, the angle value at C 9 -C 9a -C 9b also increases, signaling the breakage of the bond. The other bond angles and lengths values do not show significant changes. The attack of the hydroxide ion in the cyclopropane ring (instead of other sites with positive charge) is caused by the Bayer's stress of the cycle and the high stability of the aromatic ring. Additional information presented in Tables 2 and 3 are the geometrical properties of 2. This molecule does not have experimental values of X-ray diffraction reported yet, though the theoretical values for 2 was compared with 1, showing good congruity considering the structural differences.

Natural Charge in the Structures of the Reaction Mechanism
The natural populations of charges provide evidence of the electronic delocalization and the reactive site in the molecule. The charge analysis of the species involved in the reaction mechanism was also studied. The first step consisted of the interaction of the chlorine atom with the double bond to form either: TSi1 or TSm1, causing a reduction of the positive charge on C 8 atom. Meanwhile, C 9 atom changed to positive charge as a consequence of partial bonds formed with the chlorine atom.
The interaction between C 9b and C 9 atoms was observed by a change in the distance and angle values in TSi1 and TSm1, respectively. In addition, the natural charge analysis from TSm1 (Table 4) showed that an increment in the negative charge on C 9b atom (´0.202) and the positive charge on C 9 atom (0.108) was greater than that on C 8 atom (0.097), assuming a possible interaction among them. It is observed that when the C 9b atom forms a bond with the C 9 atom, the electronic density in C 9c decreases. Thereby, O 10 shares electronic density from unshared electrons, causing a drop in the negative charge of O 10 . The resonance effect delocalizes electrons, provoking an increment in the positive value at C 4 (from 0. 398 in 1 to 0.478 in 1a). The electrodonating effect of the methoxy group is observed in decrement of the negative charge in O 13 (from´0.539 to´0.479). Finally, C 5a also changes the value from 0.386 to 0.471. Considering the charge values, the atoms in the aromatic ring also change their values more closely to the 1, recovering the original aromatic state.

Thermochemistry of the Reaction
The relative Gibbs free energy profiles for these processes are plotted in Figure 6. The energy calculations for all species were made in aqueous solution; therefore, two profiles were constructed. state 1 for path A, TSm1: molecular activated state 1 for path B, TS2: activated state 2.
It is observed that when the C9b atom forms a bond with the C9 atom, the electronic density in C9c decreases. Thereby, O10 shares electronic density from unshared electrons, causing a drop in the negative charge of O10. The resonance effect delocalizes electrons, provoking an increment in the positive value at C4 (from 0. 398 in 1 to 0.478 in 1a). The electrodonating effect of the methoxy group is observed in decrement of the negative charge in O13 (from −0.539 to −0.479). Finally, C5a also changes the value from 0.386 to 0.471. Considering the charge values, the atoms in the aromatic ring also change their values more closely to the 1, recovering the original aromatic state.

Thermochemistry of the Reaction
The relative Gibbs free energy profiles for these processes are plotted in Figure 6. The energy calculations for all species were made in aqueous solution; therefore, two profiles were constructed. The path A ( Figure 6, profile a) starts with the interaction of the double bond and the chlorine atom via activated state TSi1 (Path A), which has an energy barrier in gas phase of 173.0 kcal/mol. However, considering the solvent effect, the energy barrier decreases to 50.4 kcal/mol, thus indicating a faster and feasible process. This change is a consequence of the stabilization of the ionic activated state (TSi1) by a polar solvent such as water [15][16][17], resulting in a faster reaction. The gas phase does not have this effect; consequently, an energy increment was observed. Since 1a is an ionic reactive intermediate, the stability in water solution was greater (30.9 kcal/mol) than in the gas phase (150.0 kcal/mol) by the solvating effect. In the second step, the activated state (TS2) decreased the charge of the molecule (closely to the neutral charge) provoking a similar energy barrier in gas (62.2 kcal/mol) and solution (46.2 kcal/mol). Finally, the product is considered a polar molecule, which has a hydroxyl group. Thereby, the interaction of this group with water reduces the energy of the product up to −45.8 kcal/mol.
On the other hand, the thermochemical calculations from path B were performed in analogue form, considering the interaction between the double bond of 1 and the hypochlorous acid. In the calculations of gas and solution phases, both values were similar; the relative activation energy of the first activated state (TSm1) was 68.3 kcal/mol, while the energy barrier was 54.1 kcal/mol in The path A ( Figure 6, profile a) starts with the interaction of the double bond and the chlorine atom via activated state TSi1 (Path A), which has an energy barrier in gas phase of 173.0 kcal/mol. However, considering the solvent effect, the energy barrier decreases to 50.4 kcal/mol, thus indicating a faster and feasible process. This change is a consequence of the stabilization of the ionic activated state (TSi1) by a polar solvent such as water [15][16][17], resulting in a faster reaction. The gas phase does not have this effect; consequently, an energy increment was observed. Since 1a is an ionic reactive intermediate, the stability in water solution was greater (30.9 kcal/mol) than in the gas phase (150.0 kcal/mol) by the solvating effect. In the second step, the activated state (TS2) decreased the charge of the molecule (closely to the neutral charge) provoking a similar energy barrier in gas (62.2 kcal/mol) and solution (46.2 kcal/mol). Finally, the product is considered a polar molecule, which has a hydroxyl group. Thereby, the interaction of this group with water reduces the energy of the product up to´45.8 kcal/mol.
On the other hand, the thermochemical calculations from path B were performed in analogue form, considering the interaction between the double bond of 1 and the hypochlorous acid. In the calculations of gas and solution phases, both values were similar; the relative activation energy of the first activated state (TSm1) was 68.3 kcal/mol, while the energy barrier was 54.1 kcal/mol in solution. The less polarity of TSm1 in comparison with TSi1 allows more stability in gas phase and minor energy; however, partial charges formed were more stable in solution. Similar to path A, the reactive intermediate 1a is produced and reacts with hydroxide ion to form TS2, with the same energy barriers mentioned previously. The results in gas phase showed a great difference in the activation energies (profiles not showed), in contrast with the solution process; for this reason, the process in gas phase is currently under study using other functional and basis sets.

Frontier Orbitals Analysis
In order to describe the reactivity of the species involved in the reaction, a frontier molecular orbital analysis was made (Figure 7). HOMO-LUMO energy gaps of gas phase and solution were calculated and compared; the most feasible process has the smaller value. Thus, the interaction among the HOMO of 1 with the LUMO of HOCl or Cl + in gas phase yielded energy gaps of 3.74 eV and´12.66 eV, respectively. In contrast, values of 3.73 eV and´6.12 eV were found in solution, which showed higher reactivity in aqueous medium. Although the interaction of 1 with the chlorenium ion has a minor value, this specie is unstable (gap value of 1.08 eV) and less probable to exist in the medium; thereby, the study was centered on path B.
solution. The less polarity of TSm1 in comparison with TSi1 allows more stability in gas phase and minor energy; however, partial charges formed were more stable in solution. Similar to path A, the reactive intermediate 1a is produced and reacts with hydroxide ion to form TS2, with the same energy barriers mentioned previously. The results in gas phase showed a great difference in the activation energies (profiles not showed), in contrast with the solution process; for this reason, the process in gas phase is currently under study using other functional and basis sets.

Frontier Orbitals Analysis
In order to describe the reactivity of the species involved in the reaction, a frontier molecular orbital analysis was made (Figure 7). HOMO-LUMO energy gaps of gas phase and solution were calculated and compared; the most feasible process has the smaller value. Thus, the interaction among the HOMO of 1 with the LUMO of HOCl or Cl + in gas phase yielded energy gaps of 3.74 eV and −12.66 eV, respectively. In contrast, values of 3.73 eV and −6.12 eV were found in solution, which showed higher reactivity in aqueous medium. Although the interaction of 1 with the chlorenium ion has a minor value, this specie is unstable (gap value of 1.08 eV) and less probable to exist in the medium; thereby, the study was centered on path B.   Figure 8 shows the HOMO and LUMO of reactive species. In 1, the HOMO (Figure 8a) is mainly distributed around the double bond of C 8 -C 9 atoms and the aromatic system confirms the electrodonating effect of this moiety towards the LUMO of HOCl (Figure 8b). The LUMO of 1a (Figure 8c) is concentrated in the aromatic ring, as a consequence of delocalization of the positive charge, and the cyclopropane motif reacts with the HOMO of the hydroxide ion (Figure 8d). Regarding product 2, the distribution of its HOMO and LUMO is quite similar to the frontier orbitals of 1.
charge, and the cyclopropane motif reacts with the HOMO of the hydroxide ion (Figure 8d). Regarding product 2, the distribution of its HOMO and LUMO is quite similar to the frontier orbitals of 1.

Bond Order and Reaction Mechanism
It is noteworthy to mention that, based on theoretical results, it is possible to propose the reaction mechanism with the activated states and intermediates (Figure 9). In summary, the double bond of 1 makes a nucleophilic attack on the chlorine atom of the hypochlorous acid breaking the Cl-OH bond; however, by assistance of the nearest aromatic system, the chloronium ion is not formed and a non-classic carbocation is obtained [13].

Bond Order and Reaction Mechanism
It is noteworthy to mention that, based on theoretical results, it is possible to propose the reaction mechanism with the activated states and intermediates (Figure 9). In summary, the double bond of 1 makes a nucleophilic attack on the chlorine atom of the hypochlorous acid breaking the Cl-OH bond; however, by assistance of the nearest aromatic system, the chloronium ion is not formed and a non-classic carbocation is obtained [13]. charge, and the cyclopropane motif reacts with the HOMO of the hydroxide ion (Figure 8d). Regarding product 2, the distribution of its HOMO and LUMO is quite similar to the frontier orbitals of 1.

Bond Order and Reaction Mechanism
It is noteworthy to mention that, based on theoretical results, it is possible to propose the reaction mechanism with the activated states and intermediates (Figure 9). In summary, the double bond of 1 makes a nucleophilic attack on the chlorine atom of the hypochlorous acid breaking the Cl-OH bond; however, by assistance of the nearest aromatic system, the chloronium ion is not formed and a non-classic carbocation is obtained [13]. Figure 9. Reaction mechanism proposed with activated states and intermediates. Figure 9. Reaction mechanism proposed with activated states and intermediates.
The bond order values of activated state TSm1 were compared to that of the reactive 1 (Table 5). In TSm1, the C 8 -C 9 bond had a bond order of 1.04 in contrast with 1, which had a value of 1.869.
In this same sense, the bond order among C 9b and C 9 increased its value in TSm1, indicating a possible bonding interaction, coinciding with the decrease in the distance and bond angle between these atoms. On the other hand, the partial bond formed within C 8 -Cl and C 9 -Cl has values of 0.819 and 0.315, respectively, showing a stronger interaction among the C 8 and Cl atoms. The bond orders in the aromatic ring carbon atoms were slightly smaller, and this fact is explained by the increment in bond length (Table 3). The same analysis was made for 1a, where it was possible to observe the bonding formation value between C 9b -C 9 (0.822), a decrement in the bond of C 11 -O 10 and C 9b -C 9c , and an increment in the bond order among C 9c -O 10 , when compared to 1. All variations are in close agreement with the geometrical parameters formerly described. Finally, it is observed that bond order values of 2 are quite similar to those obtained for 1, only with a significant decrement by the double bond break at C 8 -C 9 atoms.

Spectroscopy Properties of 1 and 2
A complementary and interesting result was obtained from the calculation of Nuclear Magnetic Resonance (NMR) theoretical chemical shift for 1 H, the coupling constant for three bonds (J 3 ), and its correlation with those previously reported [6,18].

Chemical Shift Prediction
The calculated values by the Gauge-Invariant Atomic Orbital method (GIAO) 1 H NMR chemical shifts were plotted with and without the solvent effect versus the experimental data reported for 1 ( Figure 10). describe the fit is: where T is a chemical shift predicted based on the experimental values; ex, the slope, and the intercept is ppm, having a standard deviation of 0.0469 and 0.2432, respectively. The same analysis was made considering the solvent effect with a regression coefficient of 0.988 and standard deviation of 0.192 ppm, the equation to describe the fit is: with a standard deviation of 0.0466 and 0.2421, respectively. The results of the calculated values for 1 and 2 were in close agreement with the experimental values. Linear regression analysis of the data set of 1 H NMR shifts in gas phase provided the following results: a regression coefficient of 0.988 and a standard deviation of 0.193 ppm. The equation to describe the fit is:

Theoretical Coupling Constant
BT " 1.041Bex´0.2395 ppm (1) where BT is a chemical shift predicted based on the experimental values; Bex, the slope, and the intercept is ppm, having a standard deviation of 0.0469 and 0.2432, respectively. The same analysis was made considering the solvent effect with a regression coefficient of 0.988 and standard deviation of 0.192 ppm, the equation to describe the fit is: BT " 1.034Bex´0.2077 ppm (2) with a standard deviation of 0.0466 and 0.2421, respectively.
The calculated values in gas and solution phases of 1 H chemical shift for 2 were also correlated with the experimental values. The regression analysis showed r 2 coefficients of 0.882 in gas phase and 0.894 in solution; the standard deviations were 0.542 and 0.572 ppm, respectively. The equations to define these trends are: BT " 0.9875Bex`0.0088 ppm in gas phase BT " 0.9933Bex´0.0291 ppm in solution (4) The results of the calculated values for 1 and 2 were in close agreement with the experimental values.

Theoretical Coupling Constant
The theoretical determination of the coupling constant (J 3 ) was made for 1 and 2. Comparing with the experimental values (Table 6), the predicted values are very close to the experimental values. In conclusion, both analyses reflected an adequate description of the experimental chemical shifts of 1 H and J 3 coupling constant by the selected method, theory level and basis set.

Predicted Toxicological Properties for 1 and 2
The toxic, teratogenic and mutagenic effects of 1 have been amply studied [19][20][21]. However, these effects are a consequence of the epoxidation of the double bond at C 8 -C 9 atoms and the covalent bonding of the epoxide to guanidine nucleotide in the DNA. The properties of 1, 2 and the epoxide were predicted using the OSIRIS-Property-Explorer. The drug likeness may be defined as a complex balance of various molecular properties and structural features that determine whether a particular molecule is similar to the known drugs. The OSIRIS calculations for 1 and 2 and the epoxide are summarized in Table 7. Results of the toxicity risk predictor showed that the compound with less risk of undesirable effects is 1, which do not present risks of mutagenicity and tumorigenicity; however, this compound presented high irritating and reproductive effects. On the contrary, the epoxide presented medium risk for mutagenicity and tumorigenicity, though existing high risk to present irritating and reproductive effects. Finally, 2 does not present mutagenicity, which is in close agreement with previous reports [3,6], but the risk in other effects was high. The physicochemical properties of the compounds were also estimated. clog P, the logarithm of its partition coefficient between n-octanol and water log(c octanol /c water ), is a property that describes the molecular hydrophobicity and varied from 1.56 to 4.95 (<5) [22]. In this research, the less bioavailable compound was 2. As a consequence, the compound has poor permeability, which agrees with previous work [3,6], while the epoxide and 1, had similar values. Drug solubility (expressed by log S) is an important factor to describe the absorption process. Poor solubility leads to poor absorption and biodisponibility [22]. The most soluble compound was 2, indicating that this compound possesses the best absorption, movement in the blood stream, and better disposal by the urinary tract. The drug score (DS) is the combination of drug likeness, clog P, log S, molecular weight and toxicity risks in one handy value that may be used to judge the compound overall potential to qualify as a drug [22].

Optimization of the Structure Involved in the Mechanism
This study considered the AFB 1 molecule's maximum stability stereoisomer previously reported by our research group [23]. The connectivity of Cl´and OH´ions suggests sixteen possible stereoisomers for 2 due to the addition of chlorine and hydroxyl groups at C 8 and C 9 atoms. At a first stage, 2 was built with standard bond lengths and bond angles, using the PC Spartan 06 program [24]. Therefore, the first task was to establish the conformation of maximum stability. Thus, the geometry for each stereoisomer was fully optimized using DFT calculations, which were carried out using the Gaussian 09 program [25]. These calculations were carried out defined by the Becke's three parameter hybrid functional (B3LYP) [26,27], which include a mixture of Hartree-Fock exchange with DFT exchange-correlation. The used basis set includes the split-valance and diffuse functions, 6-311++G(d,p) [28][29][30][31][32]. The default convergence criteria were also employed. The minima were verified performing a vibrational analysis. The structural parameters (bond lengths and bond angles) were analyzed at the same level, for intermediates and activated state (TS) of 1, which were confirmed by frequency calculations. In all cases, intrinsic reaction coordinate (IRC) calculations were performed to test that the determined TSs connect with the proper reactants and products.

Atomic Charge Analysis
Natural Bond Orbital (NBO) was used for electron natural population analysis in the Gaussian program. Natural Population Analysis (NPA) was used for comparing differences rather than determining absolute atomic charges. This analysis was performed to investigate the electronic properties in the reaction mechanism [33]. The atomic charge was calculated at the same level of theory for all species involved in the reaction mechanism.

Thermochemical Parameters
Thermochemical values were estimated from frequency calculations, which included a thermochemical analysis of the system considering 298 K, 1 atm of pressure and the principal isotope for each element [34], all at the same theory level. The zero point correction to the electronic energy (ZPE) of the molecule was used to calculate the values of enthalpy and Gibbs free energy. The solvent effect was also calculated by using the self-consistent reaction field (SCRF) method and considering the Tomasi's polarizable continuum model (PCM) at the same theory level, using water as a medium. For all energies, ZPE corrections were taken into consideration [35,36].

Frontier Orbital Analysis
The highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) gap is a typical quantity used to describe the dynamic stability of molecules [37]. The values of the orbital energy and the surface of the frontier orbitals were calculated using the same level of theory.

Bond Order
The natural bond orbital analysis provides an efficient method for studying intra and intermolecular bonding as well as its interaction. The calculation of bond order was developed to investigate the bond length before and after chemical transformation. The electron density of double and single bonds of the furan ring, clearly demonstrates a strong delocalization inside the molecule [33]. The calculation of bond order was developed using a natural bond orbital (NBO) analysis for all signaled species to investigate the electronic properties in the reaction mechanism [38][39][40].

Spectroscopy Properties
In addition, the 1 H NMR chemical shifts were calculated using the Gauge Invariant Atomic Orbital method (GIAO method) taking into account the solvent effect and using tetramethylsilane as a reference. In the case of the coupling constant, the spin-spin option was employed at the same theory level [41,42].

Biological Properties
The toxicological risk and some physicochemical properties were obtained using the OSIRIS-Property-Explorer (Actelion Pharmaceuticals Ltd, Allschwil, Switzerland). The toxicological risk prediction process relies on a precompiled set of structural fragments that give rise to toxicity alerts in case they are encountered in the structure currently drawn. clog P and log S calculation method of OSIRIS is implemented as an increment system adding contributions of every atom based on its atom type. The drug likeness approach is based on a list of 5300 distinct substructure fragments with associated drug likeness scores. The drug likeness was calculated employing the score values of those fragments that are present in the molecule under investigation [43][44][45].