Molecular design of curcumin analogues with potent antioxidant properties and thermodynamic evaluation of their mechanism of free radical scavenge

Background: Free radical attack of cellular structures in the human system is the major cause of various forms of degenerative diseases. Consequently, recent research has been focused on the development of new antioxidants with more efficient free radical scavenging potentials. Ligand-based virtual screening was employed in the rational design of potent antioxidant derivatives of curcumin by the density functional theory method. Various antioxidant descriptors that characterize the three major mechanisms of free radical scavenge, namely, hydrogen atom transfer (HAT), single electron transfer followed by proton transfer (SET-PT), and sequential proton loss electron transfer (SPLET), were calculated. Also, the spin density distribution on the generated radicals and the frontier orbital distribution and energy of the studied compounds were evaluated in order to gain further insight on the reaction mechanism. The reaction Gibbs free energy for scavenging the two important peroxyl radicals (HOO· and CH3OO·) was calculated in order to evaluate the preferred mechanism of free radical scavenge by these compounds. Results: The investigated compounds were able to scavenge HOO· and CH3OO· radicals by HAT and SPLET mechanisms in the gas phase and aqueous solution, based on the computed results of reaction enthalpies and Gibbs free energy. The SET-PT mechanism for these compounds was observed to be thermodynamically unfeasible in the gas phase. However, the thermodynamic feasibility of free radical scavenge by SET-PT mechanism was observed in aqueous solution. Among the investigated compounds, MCC 009 (1E,4E)-1-(3-(aminomethyl)-4hydroxyphenyl)-5-(4-hydroxy-3-((hydroxy (methyl)amino)methyl)phenyl)penta-1,4-dien-3-one at the 19-OH position possessed the highest capacity to scavenge both HOO· and CH3OO· radicals by HAT, SET-PT, and SPLET mechanisms. The reaction Gibbs free energy of scavenging HOO· radical by this molecule in the gas phase and aqueous solution is ΔrGBDEgas = − 58.18, ΔrGBDEH2O 1⁄4 − 73:77, ΔrGAIPgas = 611.48, ΔrGAIPH2O 1⁄4 − 74305:49, ΔrGPDEgas = − 669.66, ΔrGPDEH2O 1⁄4 74231:72, ΔrGPAgas = − 271.40, ΔrGPAH2O 1⁄4 − 73:09,ΔrGETEgas = 213.22, and ΔrGETEH2O 1⁄4 − 0:69. Conclusion: New set of curcumin derivatives with potent free radical scavenging properties was successfully designed, and their mechanism of free radical scavenging evaluated by thermodynamic studies. This research is a gateway to the exploitation of the considered curcumin derivatives in food chemistry and pharmacy.

Antioxidants are substances that have the ability to scavenge free radicals, thereby preventing the onset of oxidative stress (Alisi et al. 2019a). Antioxidants employ various mechanisms to exert their influence. Three major mechanisms of free radical scavenge by antioxidants have been recognized. These include hydrogen atom transfer (HAT), single electron transfer followed by proton transfer (SET-PT), and sequential proton loss electron transfer (SPLET) (Galano et al. 2016;Vo et al. 2018).
Free radicals are molecules or their fragments that contain unpaired electrons in their atomic or molecular orbitals. Free radicals are highly reactive species and have the ability to initiate chain reactions that propagates their molecular damage on proteins and lipids (Alisi et al. 2018a). When the level of free radicals in the human system exceed a threshold level, oxidative stress results. Oxidative stress plays a vital role in the pathogenesis and pathophysiology of various chronic health disorders such as osteoporosis, cancer, diabetes, and neurodegenerative conditions (Pisoschi and Pop 2015;Tan et al. 2018;Yeung et al. 2019). These conditions with time develop resistance to most drugs used for their treatment. This trend has aroused a great need to develop highly efficient free radical scavengers. A method widely employed in the design of new chemical entities with potent biological activities is in silico design by virtual screening through the method of quantitative structure activity relationship (QSAR).
In silico design by virtual screening has been employed in the rational design of various compounds with potent antioxidant activities such as hydrazones (Alisi et al. 2019b), 1,3,4-oxadiazoles , Schiff base-1,2,4-triazoles bearing butylated hydroxytoluene moiety (Yehye et al. 2016), and phenolic derivatives bearing NO donor groups (Mitra et al. 2011a). Recently, it has also been employed in the design of electrolytic solvents (Marcou et al. 2019), membranolytic anticancer peptides (Gabernet et al. 2019), graphene oxide membranes with variable water content and flake oxygen content (Williams et al. 2019), and linear and cyclic pentapeptide ligands potentially active against the influenza A virus (Perrier et al. 2019).
QSAR is based on the principle that the biological activities of compounds are a function of their molecular structure (Fourches and Ash 2019). QSAR is a valuable approach that has been employed extensively to investigate the structure property relationships of various compounds and their associated biological activities (Alisi et al. 2018b;Gul et al. 2019;Yousefinejad et al. 2019).
Density-functional theory (DFT) is a computational quantum mechanical modeling method employed to investigate the structural, magnetic, and electronic properties of matter. The results of DFT computations have been employed to investigate the antioxidant properties and energetics of free radical scavenge of various compounds (Hernandez et al. 2020;Yusuff et al. 2019;Boulebd et al. 2020;Zhou et al. 2019).
In the present research, the earlier developed QSAR model for curcumin antioxidants was employed in the design of a new set of 1,5-diphenylpenta-1,4-dien-3-one derivatives. Their antioxidant activities were subsequently determined by the leverage approach using the same model. Furthermore, the mechanism of free radical scavenge by these compounds was investigated via thermodynamic studies by the (DFT) method.

Design of new compounds
The design of a new set of curcumin antioxidants was accomplished by ligand-based virtual screening using the earlier developed QSAR model for curcumin antioxidants (Alisi et al. 2018c). This was performed by insertion, deletion, and substitution of various substituents on the template molecule using the curcumin antioxidant model as a basis (Asadollahi et al. 2011;Mitra et al. 2011b). This model has an applicability domain with a leverage threshold value of 0.649. In the present research, compound M29 ( Fig. 1) listed in Table 1 of (Alisi et al. 2018c) was chosen as a template based on its impressive antioxidant activity (pIC 50 = 6.260).
All molecular structures were drawn using Chem Draw Program (Li et al. 2004). While geometry optimization of the relevant chemical structures was accomplished using Spartan 14 program (Shao et al. 2006), this was executed at the DFT level using the Becke three-parameter Lee-Yang-Parr hybrid functional (B3LYP) in combination with 6-311G* basis set without symmetry constraints. The optimized structures were converted to Sdf files and submitted to the PADEL program package version 2.20 (Yap 2011). The above procedures resulted in the generation of quantum chemical and molecular descriptors for each molecular structure. The antioxidant activities of the modified structures were further calculated using the curcumin antioxidant model. Subsequently, their applicability domain was accessed by the leverage approach.

Computation of antioxidant descriptors
Various antioxidant descriptors that characterize the various mechanisms of free radical scavenge were calculated in the gas phase and aqueous solution as presented below.

The homolytic bond dissociation enthalpy (BDE)
This descriptor was calculated under standard conditions of 1 atm and 298.15 K using Eq. (1). It represents the standard enthalpy change at a given temperature when a particular chemical bond is broken under standard conditions (Ruscic 2015). Its value determines the stability of the corresponding hydroxyl group. The lower BDE value for a given antioxidant indicates the low stability of the O-H bond. This results in a higher tendency to break the corresponding O-H bond (Sun and Jin 2013). This results in better antioxidant activity of the considered molecule.
Adiabatic ionization potential (AIP) values represent the ability of the antioxidant to transfer electrons to the free radical. It describes the process of electron donation by the antioxidant. The lower the AIP value for a given molecule, the easier is the ability to transfer electrons. Molecules with lower AIP values have higher tendency to undergo ionization. Such molecules have been observed topossess stronger antioxidant activities (Özbakır Işın 2016). The AIP values were estimated using Eq. (2): The proton dissociation enthalpy (PDE) describes the ability of the cationic radical of the antioxidant to donate a proton to the free radical. The computation of PDE values was accomplished using Eq. (3). Antioxidants with lower PDE values are more susceptible to proton abstraction (Mikulski et al. 2014).
Proton affinity (PA) for a given molecule is the negative of its molar enthalpy change at 298.15 K. lower PA values result in higher antioxidant activity of molecules. The PA value was estimated according to Eq. (4): The electron transfer enthalpy (ETE) represents the ability of the antioxidant anion to transfer electrons to the free radical. The lower the ETE value for a given molecule, the more active is the resulting phenoxide anion. Equation (5) was employed in the computation of the ETE values.
where H radical is the total enthalpy of phenoxyl radical, H H is the total enthalpy of the hydrogen atom, H neutral is the total enthalpy of neutral compound, H H þ is the total enthalpy of the proton, H cation radical is the total enthalpy of the cation radical, H electron is the total enthalpy of the electron, and H anion is the total enthalpy of the anion.
In this study, the total enthalpy of the species was obtained as presented in Eq. (6). Also, energy was converted to enthalpy by adding RT (PV-work) term (Najafi et al. 2011).
where E 0 is the total energy at 0 K, ZPE is the zeropoint vibrational energy, and H trans , H rot and H vib are the translational, rotational, and vibrational contributions to the enthalpy respectively.

Investigation of the thermodynamically favored mechanism
The reaction Gibbs free energy (Δ r G) was employed to determine the thermodynamically favored mechanism (Amic et al. 2017;Zheng et al. 2017). In this research, the change in Gibbs free energy between the reactants and products for the studied mechanisms of the reactions between the antioxidants and the peroxyl radicals (HOO· and CH 3 − OO·) was estimated in vacuum and water.
A thermodynamically favorable reaction is one whose Gibbs free energy change is exergonic (Eq. 7). In the present context, the exergonicity of the reaction between HOO· and CH 3 − OO· radicals and curcumin antioxidants was investigated.
The Gibbs free energy change for the HAT mechanism is given by Δ r G BDE (Eq. 8).
The Gibbs free energy change for the SET-PT mechanism is given by Δ r G AIP (Eq. 9) and Δ r G PDE (Eq. 10), while that of the SPLET mechanism is given by Δ r G PA (Eq. 11) and Δ r G ETE (Eq. 12).
where G(H n Antiox) is the Gibbs free energy of neutral antioxidant, G(H n − 1 Antiox ⦁ ) is the Gibbs free energy of phenoxyl radical,G(R ⦁ ) is the Gibbs free energy of free radical, G(HR) is the Gibbs free energy of product formed by hydrogen abstraction to free radical,G(H n − 1 Antiox ⦁+ ) is the Gibbs free energy of cation radical, G(H + ) is the Gibbs free energy of proton, G(R − ) is the Gibbs free energy of free radical anion,and G(H n − 1 Antiox − ) is the Gibbs free energy of anion.
The Gibbs free energy of the electron (e − ) and proton (H + ) employed in gas phase computations is −3.72 kJ/ mol and −26.28 kJ/mol, respectively. However, in aqueous solution, −156.8 kJ/mol and −1104.5 kJ/mol were used as the Gibbs free energy for the electron and proton, respectively (Tissandier et al. 1998;Hwang and Chung 2005).

Molecular design
The pIC 50 standard residuals and leverage values of the newly designed curcumin antioxidants are presented in Table 1. The applicability domain of curcumin antioxidants model has a leverage threshold, h * value of 0.6486. From Table 1, we observe that all the designed molecules have leverage values less than the threshold leverage (h < h * ). This is an indication that no structural outliers exist between the designed molecules. Also, the curcumin derivatives with best antioxidant activities include MCCM 07, MCCM 09, and MCCM 15 with pIC 50 values of 6.334, 6.275, and 6.282, respectively. Based on the above results, MCC 07, MCC 09, and MCC 15 were subjected to quantum chemical calculations in order to investigate their mechanism of free radical scavenge. The molecular structure and carbon atom numbering of these molecules are presented in Fig. 2.

Analysis of the HAT mechanism
The results of BDE (kJ/mol) at various OH positions for MCC 007, MCC 009, and MCC 015 in the gas phase and aqueous solution are presented in Hydrogen atom transfer from these positions is the most difficult in comparison to other -OH positions. Subsequently, the primary target of free radical attack in MCC 007 is the 16-OH position which has the lowest BDE value. Similarly, for MCC 009, the primary site for free radical attack is the 19-OH position, while that of MCC 015 is the 3-OH position (Table 2). When the BDE results for these compounds at their respective preferred sites of free radical scavenge are compared with that of phenol (327.550 kJ/mol) which is usually chosen as a reference compound at the same level of theory, MCC 007 16-OH, MCC 009 19-OH, and MCC 015 3-OH have lower BDE values. This is an indication that these compounds possess higher tendency to undergo free radical scavenge through the HAT mechanism than phenol at Thus, these molecules undergo HAT in the gas phase more readily than in aqueous solution.

Analysis of the spin density distribution
The spin density distribution is of great use in rationalizing the stability of the antioxidant radical (Zheng et al. 2017;Xue et al. 2014;Wang et al. 2015). The BDE in conjunction with the stability of the radicals is an important factor that influences the antioxidant activity of phenolics. Lower spin density distribution values result in higher delocalization of the spin density in a given radical. Such radicals possess higher stability, which is associated with lower BDE for a given antioxidant. The radical spin density results (IsoVal 0.002) for MCC 007, MCC 009, and MCC 015 are presented in Table 2. For MCC 007, MCC 009, and MCC 015, lowest radical spin density values of 0.000223, 0.000230, and 0.000214 were observed at the 16-OH, 19-OH, and 3-OH, respectively. For each of these molecules, radicals obtained at these sites are the most stable. From Table 2, it is evident that these sites also possess the lowest BDE values for each of the molecules. Consequently, these are the preferred sites of free radical scavenge based on the HAT mechanism.

Analysis of SET-PT mechanism
The computed results of adiabatic ionization potential (AIP) and proton dissociation enthalpy (PDE) for MCC 007, MCC 009, and MCC 015 molecules at their various free radical scavenging sites in vacuum and water are presented in Table 2. The first step of the SET-PT mechanism for antioxidants is characterized by AIP. In the gas phase, the AIP results for the considered molecules in vacuum are appreciably lower than that of phenol (572.776 kJ/mol generated at the same level of theory) which is usually considered as a reference (Table  2). This indicates that these molecules are better electron donors compared to phenol in the gas phase. For MCC 007, MCC 009, and MCC 015 molecules, the sites with the best electron-donating ability are the 4-OH, 4-OH, and 16-OH positions with AIP values of 314.61 kJ/mol, 381.14 kJ/mol, and 315.61 kJ/mol, respectively, in vacuum. However, in aqueous solution, the preferred site of electron donation is the 16-OH position in each of the considered molecules. This position in each of these molecules has the lowest AIP value. The AIP value in aqueous solution is lower than the value in vacuum of each molecular site. Consequently, electron donation by these antioxidants is more favored in aqueous solution than in vacuum. Also, Table 2 indicates that the trend of AIP values of these curcumin derivatives is different from that of the computed BDE values. A possible explanation for this discrepancy is that AIP values are influenced by the structure of the entire molecule, such as the extended delocalization and conjugation of the π-electrons, whereas BDE of a given molecule is influenced by its local environment due to the substituents (Xue et al. 2014;Wang et al. 2015). The PDE characterizes the second step of the SET-PT mechanism for antioxidants. MCC 007 has the lowest PDE values of 1275.23 kJ/mol and 1612.42 kJ/mol at the 16-OH position in vacuum and aqueous solution, respectively. For MCC 009 and MCC 015 molecules, the values are 1104.08 kJ/mol and 1475.71 kJ/mol and 1100.03 kJ/mol and 1474.80 kJ/mol at the 19-OH and 3-OH positions, respectively. These are the most favored sites for proton dissociation by SET-PT mechanism. The observed trend of PDE is analogous to those of BDE. This is not surprising since proton dissociation results in the generation of the antioxidant radical like bond dissociation. From the perspective of thermodynamics, the first step (in this case AIP) is the most important for reaction that involves multiple mechanisms such as the SET-PT.

Analysis of SPLET mechanism
The results of proton affinity (PA) and electron transfer enthalpy (ETE) for MCC 007, MCC 009, and MCC 015 in vacuum and water are presented in The ETE value describes the ability of the antioxidant anion to transfer electrons to the free radical. The most favored sites for the second step of the SPLET mechanism in MCC 007, MCC 009, and MCC 015 are the 16-OH, 19-OH, and 3-OH positions in both vacuum and aqueous solution. These positions recorded the lowest ETE values in each molecule (Table 2). However, deprotonation and electron transfer reactions of the SPLET mechanism occur more readily in aqueous solution than in the gas phase. A comparison of the generated results for ETE and AIP indicates that the ETE values are lower than those of AIP at a given scavenging site for each molecule in vacuum. This could be attributed to the fact that single electron transfer from the neutral form is less favored to that from the anionic form. This result is in agreement with previous research reported in the literature (Naik et al. 2011;Rimarcik et al. 2010Xue et al. 2014).

Evaluation of thermodynamically preferred mechanism
The observed results of the Gibbs free energy change for scavenging HOO· and CH 3 OO· radicals via HAT, SET-PT, and SPLET mechanisms by the considered curcumin antioxidants are presented in Tables 3 and 4, respectively. The thermodynamically preferred reaction is represented by the most negative reaction Gibbs free energy change (Δ r G) result. This implies that in the assignment of the free radical scavenging potency of molecules, the Δ r G result could very helpful.
In the gas phase, MCC 007 16-OH, MCC 009 19-OH, MCC 015 3-OH, and MCC 015 16-OH molecules possess good potency to scavenge HOO· via the HAT and SPLET mechanism, because these processes are exergonic (Table 3). A similar trend was observed for the free radical scavenge of CH 3 OO· (Table 4). For the SPLE T mechanism in vacuum, the exergonicity of Δ r G PA in the first step overwhelms the endergonicity of Δ r G ETE in the second step at these positions. This makes the entire SPLET mechanism thermodynamically feasible at these positions, because a thermodynamically unfeasible reaction could be driven by a thermodynamically feasible reaction that is coupled to it. The SET-PT mechanism in vacuum was thermodynamically unfeasible at all Table 3 Gibbs free energy changes (Δ r G in kJ/mol) of scavenging HOO· radical by curcumin derivatives at the B3LYP/6-311G*/6-31G* level in vacuum and water positions, because the first step of this pathway (Δ r G AIP ) is endergonic and overwhelms the exergonicity of Δ r G PDE . In addition, for a reaction that involves multiple mechanisms, the first step, in this case, Δ r G AIP is more important.
In aqueous solution, a similar trend was observed for the free radical scavenge of HOO· and CH 3 OO· radicals by HAT and SPLET mechanisms. In addition, the SET-PT mechanism was thermodynamically feasible at the MCC 007 16-OH, MCC 009 19-OH, MCC 015 3-OH, and MCC 015 16-OH positions in aqueous solution.
With respect to the chemical structures of the considered compounds, the Δ r G is observed to be the most exergonic in situations where the OH group is attached to a nitrogen atom that also contains a methyl group. For instance, the molecule of MCC 009 is found to be the most exergonic in its reactions at the 19-OH position. A similar trend is observed for MCC 015 at the 3-OH position. Consequently, this molecular structure confers improved antioxidant activity on these compounds. Comparison of the results in Tables 3 and 4 shows that the investigated curcumin antioxidants possess higher potential to scavenge HOO· than CH 3 OO· Table 4 Gibbs free energy changes (Δ r G in kJ/mol) of scavenging CH 3 OO· radical by curcumin derivatives at the B3LYP/6-311G*/6-31G* level in vacuum and water based on the exergonicity of their HAT, SPLET, and SET-PT mechanisms. The molecule MCC 009 at the 19-OH position exhibited the highest potential to scavenge HOO· and CH 3 OO· by HAT, SPLET, and SET-PT mechanisms. Consequently, the mechanisms of free radical scavenge by this molecule by HAT, SPLET, and SET-PT mechanisms are presented in Scheme 1a-c, respectively.
Analysis of frontier molecular orbitals Figure 3 gives the frontier orbital distribution and energy for MCC 007, MCC 009, and MCC 015 computed at the B3LYP/6-311G* level in the gas phase. The lowest unoccupied molecular orbital (LUMO) in the three molecules presents similar distributions and localized mainly on the isolated carbons that contains the ketone group. The highest occupied molecular orbital (HOMO) on the other hand is delocalized on the carbon atoms of the two phenyl rings, A and B, and on the oxygen atoms that contain the hydroxyl group at the 4-OH and 16-OH sites of each molecule. These are the preferred sites of electron loss by these molecules. The 4-OH site which has a high density of HOMO distribution also has the lowest AIP value among the considered sites for each molecule (Table 2) The results of HOMO-LUMO energy gap (E gap) for the considered curcumin antioxidants are also presented in Fig. 3. The reactivity of a molecule is reflected in its E Gap value. The lower E gap value for a given molecule is an indication of higher reactivity. Based on the E Gap results, MCC 009 has the lowest E gap value of 3.55 eV. This is in line with the sequence of predicted antioxidant activity of these compounds presented in Table 1. In this Table, MCC 009 has the best antioxidant activity with a pIC 50 value of 6.260. MCC 09 was also predicted to have the highest potential to scavenge free radicals (Tables 3  and 4).

Conclusion
This research employed the technique of virtual screening to design a new set of curcumin antioxidants whose free radical scavenging potency was tested using the developed QSAR model. Three of these compounds, namely, MCCM 07, MCCM 09, and MCCM 15 showed better antioxidant activities compared to the template molecule. They were subsequently subjected to thermodynamic studies through the computation of their reaction enthalpies of free radical scavenge such as BDE, AIP, PDE, PA and ETE, and Gibbs free energy in the gas phase and aqueous solution. The three major mechanisms of free radical scavenge, namely, HAT, SET-PT, and SPLET were considered. The BDE revealed that MCCM 09 at the 19-OH position has the greatest ability to donate hydrogen atoms to the free radical in comparison to the other molecules, including phenol that was used as the reference molecule. The sequence of BDE value in these molecules was similar to that of PA, but different from that of AIP. Also, it was observed that single electron transfer from the neutral form is less favored to that from the anionic form as reflected in the computed ETE and AIP values.
In the molecules of MCCM 07 and MCCM 09, the thermodynamic feasibility for scavenging HOO· and CH 3 OO· radicals was observed at the 16-OH and 19-OH sites, respectively. For MCCM 15 molecule, 3-OH and 16-OH sites were observed to be thermodynamically feasible. Also, these molecules showed greater scavenging potency for HOO· than CH 3 OO· by HAT, SET-PT, and SPLET mechanisms. In vacuum, these molecules could scavenge HOO· and CH 3 OO· radicals by HAT and SPLET mechanisms. The SET-PT mechanism was thermodynamically unfeasible in vacuum. In aqueous solution, MCCM 07, MCCM 09, and MCCM 15 could scavenge these peroxyl radicals by the three considered mechanisms. The exploration of the free radical scavenging potency of these curcumin derivatives in this research will arouse subsequent exploitation of these compounds in food chemistry and pharmacy.