Computational-aided design of melatonin analogues with outstanding multifunctional antioxidant capacity

Abstract

A set of 19 melatonin analogues, intended to be better antioxidants than the parent molecule, have been computationally designed. Eight of them were planned to have good primary antioxidant capacity (AOC), i.e., being good free radical scavengers. Seven of them were designed for their secondary AOC, being able to inhibit ˙OH production by acting as metal ion chelators. Based on their predicted behavior for the intended functions, four multifunctional melatonin analogues were proposed. They were found to be among the best peroxyl radical scavengers identified so far, in aqueous solution, at physiological pH. They were also found to be capable of turning off Cu(II) reduction by O₂˙⁻ and Asc⁻, thus fully inhibiting the associated ˙OH production. Two of them, namely the IIcD and IbG analogues, were identified as the ones with the best multifunctional AOC. They both fulfill the Lipinski's and Ghose's rules for orally active drugs. However, IIcD has been chosen as the best prospect for possible application based on potential toxicity and synthetic accessibility estimations. Hopefully, these results might provide motivation for further investigations on the subject, and the synthesis of this compound, so its potential role as protector against oxidative stress – and the associated health issues – could be experimentally confirmed or refuted.

---

Introduction

During the last decades, oxidative stress (OS) has become the focus of numerous investigations, which had demonstrated that it represents an important risk to human health. The information gathered so far provides compelling evidence on the role of OS in the onset and development of a large number of diseases. A few examples are cancer,¹ cardiovascular diseases,^2,3 and neurodegenerative disorders including Parkinson's and Alzheimer's diseases, memory loss, multiple sclerosis, and depression.^4–6 Accordingly, it is logical that searching for molecules that can offer protection against OS has become an active area of research.

Among such protectors, melatonin (N-acetyl-5-methoxytryptamine) has been proven to be particularly efficient.^7–9 There are many reasons for that. The toxicity of melatonin is very low.¹⁰ Because of its size, high solubility in lipids and partial solubility in water, melatonin can cross physiologic barriers.^11,12 Its protection against OS does not decrease after been metabolized since its metabolites also exhibit antioxidant capacity (AOC).^13–15 The combined AOC of melatonin and its metabolites, which is frequently referred to as a cascade-like protection,^16,17 implies that melatonin is able to deactivate several equivalents of oxidants. Such complex AOC can involve both primary AOC, via free radical scavenging processes, and secondary AOC through metal chelation.¹⁸ Moreover, it has been proposed that this family of compounds acts in a “task-division” way, with some members of the family being particularly efficient as free radical scavengers, and others mainly behaving as metal chelators.¹⁹

Inspired by the many appealing characteristics of melatonin, several attempts have been made to produce synthetic analogues with improved properties as antioxidants, but also for other purposes.^13,20 This strategy relies on changing the groups in the different modifiable sites of the indole ring (Scheme 1). For example several melatonin analogous have been produced by incorporating pristine and substituted benzoyl and phenyl groups at site R₁.^21–23 It was found that some of them display better anxiolytic, anticonvulsant and sedative actions than melatonin. Benzoyl-melatonin, and also acetyl-melatonin, have also been found to possess enhanced anti-inflammatory effects with respect to the parent molecule.²⁴ In addition, they have been suggested to have substantial antioxidant ability.²⁵

Scheme 1 Structures of melatonin and the indole ring with possible modifiable sites (R₁ to R₇) for producing synthetic melatonin derivatives, and some representative examples of metabolites and synthetic melatonin analogues.

Various sulfhydryl-containing melatonin derivatives have been synthesized and tested as antioxidants.²⁶ The compounds conjugating a free sulfhydryl moiety in R₇, showed improved antioxidant properties. In particular a cysteinyl-conjugated derivative was proposed as suitable for developing antioxidants species. Melatonin retinamide derivatives,²⁷ indole-3-propionamides,²⁸ 2-phenylindole derivatives,²⁹ and N-methylindole analogues of hydrazide/hydrazone melatonin derivatives³⁰ have been found to inhibit lipid peroxidation. In particular, chloroindole hydrazide/hydrazone derivatives have shown significant free radical scavenging activity.³¹ In addition, it has been reported that the lack of a methoxy group in R₃ (Scheme 1) does not significantly affect the antioxidant capacity of the melatonin derivatives.³⁰ Several melatonin analogues with modifications in the methoxy and acylaminoethyl groups were identified as better antioxidants than melatonin itself.³² Recently, a large set of indole amino acid derivatives have been synthesized, and found to be similar to melatonin in their DPPH scavenging activity, but better than the parent molecule for inhibiting lipid peroxidation.³³

Accordingly, the data gathered so far strongly indicate that melatonin analogues are very promising molecules as antioxidants. In the present work a large set of new derivatives was tested for that purpose. Their design was based on structural modifications intended to increase both the primary and secondary AOC, with respect to melatonin. In this sense they are expected to be more efficient than other melatonin derivatives that mainly present only one of these AOC, most likely free radical scavenging activity. The activity of the tested compounds was compared with those of other known antioxidants. In all the cases the group in R₇ was kept identical to that of melatonin (Scheme 1). For the primary AOC, different numbers and positions of OH groups were used as substituents in sites R₄ to R₆. This strategy was thought based on the fact that phenolic compounds are usually very efficient free radical scavengers. In fact, some hydroxylated melatonin metabolites have been proposed as very efficient for that purpose.¹⁹ For the secondary AOC, different groups containing N and O atoms were tested in sites R₁ and R₂, in such a way that their location promotes metal chelation. The compounds with the highest antioxidant activity were identified and proposed as potential candidates to be implemented as protectors against OS.

Computational details

All the electronic calculations were performed with Gaussian 09 package of programs.³⁴ The M05-2X and M05 functionals³⁵ were used for geometry optimizations and frequency calculations for the systems without and with Cu, respectively. All the calculations were carried out with the 6-311+G(d,p) basis set and the continuum solvation model based on density (SMD),³⁶ and water as solvent. The M05-2X functional has been chosen because it is recommended for kinetic calculations by its developers,³⁵ and its reliability has been independently confirmed by other authors.^37–40 It is among the best performing functionals for kinetic calculations in solution,⁴¹ and for calculating reaction energies involving free radicals.⁴² The M05 functional was chosen for the Cu involving systems because it was parameterized including both metals and non-metals, while M05-2X has double the amount of non-local exchange (2X) and was parameterized mainly for non-metals. M05 has been recommended for studies involving both metallic and non-metallic elements, and perform well not only for main-group thermochemistry but also for interactions with transition-metals.³⁵ SMD was chosen for mimicking the solvent effects because it can be consistently used for any charged or uncharged solute in any solvent or liquid medium.³⁶

Local minima and transition states were identified by the number of imaginary frequencies (0 or 1, respectively). In the case of the transition states, intrinsic coordinate calculations (IRC) were performed to verify that the imaginary frequency corresponds to the proper motion along the reaction coordinate. Unrestricted calculations were used for open shell systems. Thermodynamic corrections at 298.15 K were included in the calculation of relative energies.

pK_a calculations

The pK_a values estimated in the present work were calculated using the isodesmic method, also known as the proton exchange method, or the relative method.⁴³ It is based on the following reaction scheme:

HA + Ref⁻ ↔ A⁻ + HRef

with HRef/Ref⁻ being an acid/base pair of a reference compound. Then the pK_a can be calculated as:

For this method to be reliable there are two key factors to keep in mind. HRef should be structurally similar to the system of interest, and its experimental pK_a should be known.

Rate constants

The rate constants (k) were calculated using the conventional transition state theory (TST)^44–46 and 1 M standard state, including zero curvature tunneling corrections (ZCT).⁴⁷ For the electron transfer reactions the Gibbs free energy of activation were calculated using the Marcus theory.⁴⁸

The values of the total rate coefficients for the HT reactions (k^HT_tot) were calculated as the sum of each HT reaction path:

The overall rate coefficients (k_overall) were calculated using the molar fractions of the acid–base species involved in each chemical route, i.e., the neutral phenol for HT and the phenolate anion for SPLET, at the pH of interest:

k_overall = ^Mf_neutralk^HT_tot + ^Mf_anionk^SPLET (3)

However, the ˙OOH radical is also involved in an acid–base equilibrium, with a pK_a = 4.8. Therefore, in aqueous solution at pH = 7.4, the ˙OOH molar fraction is only 0.0025. Thus, to reproduce experimental rate constants of reactions involving this radical, under such conditions, this aspect must be considered. Accordingly, the rate coefficients that can be directly compared with the experiments are:

k′_overall = ^Mf_˙OOHk_overall (4)

The percent contributions of each reaction mechanism was then estimated as:

In addition, since several of the calculated rate constants (k) are close to the diffusion-limit, the apparent rate constant (k_app) cannot be directly obtained from TST calculations. The Collins–Kimball theory is used to that purpose,⁴⁹ in conjunction with the steady-state Smoluchowski⁵⁰ rate constant for an irreversible bimolecular diffusion-controlled reaction, and the Stokes–Einstein^51,52 approaches for the diffusion coefficient of the reactants. These computational details are in line with the quantum mechanics based test for overall free radical scavenging activity (QM-ORSA) protocol,⁵³ which has been validated by comparison with experimental results; its uncertainties have been proven to be no larger than those arising from experiments.⁵³

Metal chelation and Cu(II) reduction

Albeit different chemical routes may be involved in the Cu(II) chelation,⁵⁴ the data estimated here correspond to the coupled-deprotonation-chelation mechanism (CDCM). This is because it has been previously demonstrated to be more likely than a direct chelation mechanism when the chelation site is susceptible to deprotonation.^18,19,55 The CDCM mechanism, for a bi-dentate ligand (H_nL), can be schematically represented as:

Cu(H₂O)₄²⁺ + H_nL ⇌ Cu(H_n−1L)(H₂O)₂¹⁺ + 2H₂O + H⁺

Since it involves deprotonation, the CDCM mechanism is influenced by the pH. Under the conditions of interest in this work, i.e., physiological pH (pH = 7.4), it is possible to define a conditional Gibbs energy of reaction (ΔG′) that can be calculated, at each particular pH, according to:

ΔG′ = ΔG⁰ − 2.303RT(pH) (7)

where ΔG⁰ corresponds to the standard conditions (pH = 0, 1 M standard state). More detail on how to obtain eqn (7) can be found elsewhere.^18,56 The ΔG⁰ values were calculated using ΔG_gas(H⁺) = −4.39 kcal mol⁻¹, and ΔG_solvation(H⁺) = −265.89 kcal mol⁻¹, based on the recommendation of Camaioni and Schwerdtfeger.⁵⁷ These values lead to ΔG(H⁺) = −270.28 kcal mol⁻¹, in aqueous solution.

In addition, as shown in the above equilibrium, copper ions were modeled coordinated to water molecules in a near square-planar geometry, which has been previously demonstrated as the most likely one.^58,59 This model is more adequate to represent “free” copper under physiological conditions than the “naked” ion, since charged species are expected to be hydrated in the aqueous phase. For consistency purposes, the hydrated Cu(I) ions were modeled also with four explicit water molecules, albeit in this case the linear two-coordinate configuration is preferred.^60–62 Thus, in this model Cu(I) is coordinated to two water molecules, and the other two are solvating the system.

For calculating the reduction potential (E⁰) of Cu(II) a strategy similar to that proposed by Fu et al.⁶³ has been used. It consist on calculate:

where F is the Faraday constant (23.06 kcal mol⁻¹ V⁻¹), and ΔG⁰ is estimated as:

ΔG⁰ = G_{(oxidized form)} − G_{(reduced form)} − 4.44 (9)

The last term in eqn (9) is to express the data referred to the normal hydrogen electrode (NHE). The E⁰_Cu(II)/Cu(I) value calculated this way deviates from the experimental value (E⁰_Cu(II)/Cu(I) = 0.16 eV) by 0.52 eV. Thus this difference has been used to correct the values reported in Table 4, in such a way that the experimental E⁰_Cu(II)/Cu(I) value is reproduced.

Synthetic accessibility and toxicity

The synthetic accessibility (SA) of the designed compounds have been estimated using the SYLVIA-XT 1.4 program (Molecular Networks, Erlangen, Germany).⁶⁴ It uses several contributing criteria, including the complexity of the molecular structure and of ring systems, the number of stereo centers, and the similarity to commercially available compounds. These criteria are weighted and scaled to provide a value between 1 and 10, with the larger values corresponding to the compounds that are expected to be more difficult to synthesize. The SYLVIA program has been validated to be used for ranking virtual compound during drug discovery processes.⁶⁵

On the other hand, the Toxicity Estimation Software Tool (T.E.S.T.), version 4.1, has been used to make predictions on the potential toxicity of the investigated compounds. This program estimates several descriptors using quantitative structure activity relationships (QSAR) predictions. They are intended to be used for screening untested compounds. The toxicity descriptors used in this work are:

- LD₅₀: amount of chemical (mg kg⁻¹) body weight that causes 50% of rats to die after oral ingestion.

- M: known as Ames mutagenicity. A chemical is positive if it induces revertant colony growth in any strain of Salmonella typhimurium.

- LC^F₅₀: concentration of the chemical (mg L⁻¹), in water, that causes 50% of fathead minnow to die after 96 hours.

- LC^D₅₀: concentration of the chemical (mg L⁻¹), in water, that causes 50% of Daphnia magna to die after 48 hours.

- IGC₅₀: concentration of the chemical (mg L⁻¹), in water, that causes 50% growth inhibition to Tetrahymena pyriformis after 48 hours.

In this work the toxicity descriptors were computed using the consensus method, which makes toxicity predictions using the simple average of the toxicities predicted from several QSAR methodologies, taking into account the applicability domain of each of them.⁶⁶ The consensus method usually provides higher prediction accuracy and coverage than other methodologies.

Results and discussion

Design of melatonin analogues

Since the structural features promoting primary and secondary AOC are not necessarily the same, two separated sets of melatonin analogues were first tested for these two kinds of antioxidant protection. The first test set comprises eight melatonin analogues that were mainly designed for primary AOC (test set 1, Scheme 2). Contrary to melatonin, they present phenolic OH groups in the six member ring of the indole moiety. This particular group have been chosen based on pervious evidence that N-acetylserotonin (NAS) and 6-hydroxymelatonin (6OHM), are much better free radical scavengers than the parent molecule.¹⁹ NAS and 6OHM present an OH group in site R₃ and R₆, respectively.

Scheme 2 Test set 1 of melatonin analogues, which are designed mainly for primary antioxidant activity (R₁ = R₂ = –H). Melatonin and N-acetylserotonin (NAS) are presented as reference compounds.

The first working test set can be divided in mono-phenolic (Ia, Ib, Ic), di-phenolic (IIa, IIb, IIc), and tri-phenolic (IIIa, IIIb) indoles. In the di-phenolic and tri-phenolic species the OH group are located in such a way that they correspond to catechol and pyrogallol-like molecules. This is because such configurations are expected to be better for scavenging free radicals than the alternative ones, i.e., with multiple OH located in non-neighboring sites.

On the other hand, a molecular framework, usually referred to as M1 (Scheme 3), has been proposed as crucial for promoting metal chelation with antioxidant implications.⁶⁷ Moreover, it has been suggested that compounds with such structural feature can be considered as potentially therapeutic for the Alzheimer's disease.⁶⁷ Accordingly, seven melatonin analogues have been designed to present similar chemical frameworks, in the sense that they have a N and an O atoms located in such a way that they allow the molecule to act as a bi-dentate ligand (test set 2, Scheme 4).

Scheme 3 Structure of chelating framework M1, and some representative examples of chemical compounds presenting this structural feature.

Scheme 4 Test set 2 of melatonin analogues, which are designed mainly for secondary antioxidant activity (R₁ = R₃ = R₄ = R₅ = R₆ = –H).

After analyzing the results obtained for test sets 1 and 2, the structural features identified as most promising were combined in a third set of melatonin analogues. They are four compounds (test set 3, Scheme 5) and are intended to efficiently perform as multifunctional antioxidants, i.e. with both primary and secondary antioxidant activities. A more detailed discussion on how these species were chosen is provided in the following sections.

Scheme 5 Test set 3 of melatonin analogues which are designed for both primary and secondary antioxidant activities (R₁ = R₃ = R₆ = –H).

Both, the phenolic moieties for primary AOC and frameworks similar to M1 for secondary AOC, are investigated in this work for the first time in melatonin synthetic derivatives. These structural features have been chosen based on previous evidence regarding the AOC of unrelated compounds and have been intentionally combined in the pursuit of an efficient multifunctional antioxidant. The search of previous synthetic melatonin derivatives has been mainly focused on obtaining compounds with electron-enriched indole rings. They are expected to have increased primary AOC via electron transfer reactions from the antioxidant to the radical. On the contrary, the compounds presented here are designed to behave as multi-tasking antioxidants.

Acid–base equilibria in aqueous solution

Due to the presence of phenolic groups in many of the investigated melatonin analogues, they are expected to be involved in acid–base equilibria. In addition, such kind of equilibria have been previously proven to have an important role in the antioxidant activity of chemical compounds.^68–71 Moreover, it has been recently shown that the acid–base equilibria play an important role on the free radical scavenging activity of hydroxylated melatonin metabolites via the sequential proton loss electron transfer mechanism (SPLET).⁷²

Therefore the first phenolic pK_a of the hydroxylated melatonin analogues studied in this work have been estimated. The chemical compounds used as HRef, the calculated first pK_a values, and the corresponding molar fractions for the neutral and mono-anionic species (^Mf_neutral, ^Mf_anion) at physiological pH are reported in Table 1 for each of the tested phenolic melatonin analogs.

Table 1 First phenolic pK_a values for the melatonin analogues in test set 1, most likely deprotonation site (DP) for di-phenolic and tri-phenolic analogues, reference acid used to calculate the pK_a (HRef), and molar fractions of the neutral and mono-anionic species (^Mf_neutral, ^Mf_anion) at physiological pH (pH = 7.4)

	pKa	DP	HRef	^Mfneutral	^Mfanion
Ia	10.9		Phenol	∼1.000	<0.001
Ib	8.9		Phenol	0.972	0.028
Ic	10.1		Phenol	0.998	0.002
IIa	10.2	R4	Catechol	0.998	0.002
IIb	10.3	R6	Catechol	0.998	0.002
IIc	9.5	R5	Catechol	0.992	0.008
IIIa	9.4	R5	Pyrogallol	0.991	0.009
IIIb	11.2	R3	Pyrogallol	∼1.000	<0.001
IbD	8.0		Phenol	0.806	0.194
IbG	7.2		Phenol	0.378	0.622
IIcD	7.9	R5	Catechol	0.744	0.256
IIcG	7.7	R5	Catechol	0.648	0.352

---

It should be noted that there are several values of pK_a reported for the used HRef, i.e., phenol, catechol and pyrogallol (Table 2). The ones used in eqn (2) correspond to the average of all the previously reported ones. In addition, for di-phenolic and tri-phenolic analogues there are more than one possible deprotonation site. Therefore the relative stability of the different possible mono-anions was investigated in order to identify the most likely one in each case. The proposed deprotonation sites are reported in Table 1.

Table 2 pK_a values of phenol, catechol and pyrogallol

	pKa	Ref.
Phenol	9.89	73
	10.09	74
	9.98	75
	10.24	74
	10.05	Average
Catechol	9.25	76
	9.5	77
	9.38	Average
Pyrogallol	8.94	78
	8.905	79
	8.20	80
	8.68	Average

---

Melatonin analogues designed for primary antioxidant activity

To investigate the primary antioxidant activity of the melatonin analogues in test set 1 (Scheme 2) their reactions with the hydroperoxyl radical (˙OOH) have been used. This radical has been chosen for several reasons. The ˙OOH scavenging activity of melatonin and some of its metabolites has been already assessed. ˙OOH is the smallest member of the peroxyl family (˙OOR), and has been identify to play an essential role in the toxic side effects associated with aerobic respiration.⁸¹ ˙OOR can be successfully scavenged to retard OS,⁸² due to their moderate reactivity, which also make them appropriate for studying trends in free radical scavenging activities.^8,83

Two different reaction mechanisms have been included in this part of the study, namely the hydrogen transfer (HT) and the sequential proton loss electron transfer (SPLET). The later was proposed by Litwinienko and Ingold for the reactions of the DPPH radical with substituted phenols.⁸⁴ These mechanisms have been selected because, based on previous findings,^71,85–88 they are expected to be the most important ones for phenolic compounds. In addition, the relative importance of HT and SPLET in the antioxidant activity of phenols, in aqueous solution, is frequently ruled by the pH. When it is lower than the pK_a of the phenolic compound the dominant acid–base species is the neutral one, and the relative importance of HT is frequently higher. On the contrary, at pHs higher than the pK_a, deprotonated species prevail and the SPLET route become the most important one. However, in some cases the SPLET mechanism can be the predominant one, even if the phenolate ion is present only in minor proportions.

The HT mechanism in phenolic compounds mainly takes place from the phenolic OH:

XOH + ˙R → XO˙ + HR

On the other hand, the SPLET route, comprises two elementary reaction steps:

XOH → XO⁻ + H⁺

XO⁻ + ˙R → XO˙ + R⁻

The first step is just the deprotonation of the phenol yielding the phenolate ion, and as mentioned before it is ruled by the pH. The second step is the key one for kinetics, and involves an electron transfer from the phenolate ion to the free radical.

The Gibbs energies of reaction (ΔG), the Gibbs energies of activation (ΔG^≠), the imaginary frequencies of the transition states (TS), the tunneling corrections, and rate constants for each individual HT reaction path (k^HT_i) are provided in Table S1, ESI.† The optimized geometries of the TS are also provided as ESI.† All the modeled HT pathways were found to be exergonic, i.e., thermochemically viable, with ΔG values ranging from −14.5 to −7.1 kcal mol⁻¹, while the corresponding ΔG^≠ values are in the 7.7–12.6 kcal mol⁻¹ range. The reorganization energies and the ΔG^≠ values for the SPLET reactions are reported in Table S2 (ESI†). In this case the ΔG^≠ values range from 0.9 to 2.6 kcal mol⁻¹, indicating that the SPLET reactions are faster than the HT ones. Therefore, the pK_a values of the studied analogues would determine the relative importance of these two chemical routes, at each particular pH, regarding the peroxyl radical scavenging activity of the studied compounds.

The kinetic data and the percent contributions of the HT and SPLET routes are reported in Table 3. The data corresponding to melatonin, NAS and 6OHM was also included in the table for comparison purposes.

Table 3 Total rate coefficients of the HT reactions (k^HT_tot, M⁻¹ s⁻¹), rate constants of the SPLET reactions (k^SPLET, M⁻¹ s⁻¹), overall rate coefficients (k_overall, M⁻¹ s⁻¹), percent contributions of the HT (%^HT) and SPLET (%^SPLET) mechanisms, for the reactions of melatonin analogues in test set 1 with ˙OOH. All the data is reported at 298.15 K. Melatonin, NAS and 6OHM are included for comparison purposes

	k^HTtot	k^SPLET	koverall, pH = 7.4	k′overall, pH = 7.4	%^HT, pH = 7.4	%^SPLET, pH = 7.4
Melatonin	1.90 × 10^1	8.28 × 10^−4	1.90 × 10^1	4.75 × 10^−2	∼100	<0.1
NAS	1.17 × 10^6	7.95 × 10^9	5.50 × 10^6	1.38 × 10^4	21.3	78.8
6OH	3.62 × 10^6	8.23 × 10^9	8.39 × 10^7	2.10 × 10^5	4.3	95.8
Ia	9.81 × 10^5	7.94 × 10^9	3.42 × 10^6	8.56 × 10^3	28.6	71.4
Ib	1.43 × 10^6	7.39 × 10^9	2.07 × 10^8	5.18 × 10^5	0.7	99.3
Ic	1.58 × 10^6	7.91 × 10^9	1.68 × 10^7	4.21 × 10^4	9.4	90.6
IIa	1.47 × 10^6	7.79 × 10^9	1.33 × 10^7	3.33 × 10^4	11.0	89.0
IIb	6.69 × 10^6	8.01 × 10^9	1.80 × 10^7	4.49 × 10^4	37.2	62.8
IIc	5.12 × 10^6	8.09 × 10^9	6.46 × 10^7	1.61 × 10^5	7.9	92.1
IIIa	1.49 × 10^6	7.86 × 10^9	7.56 × 10^7	1.89 × 10^5	1.9	98.1
IIIb	2.49 × 10^7	8.09 × 10^9	2.61 × 10^7	6.51 × 10^4	95.5	4.5

---

The overall rate coefficients (k_overall) are reported at pH = 7.4. The values of k_overall, calculated using eqn (3), are also reported in Table 3 to facilitate comparisons with some data previously reported in literature, albeit k′_overall is the estimation intended to be directly compared with experiments. It was found that all the SPLET reactions occur within, or close to, the diffusion limited regime (k^SPLET ≈ 8 × 10⁹ M⁻¹ s⁻¹) while the HT route is somewhat slower with rate coefficients ranging from 9.8 × 10⁵ to 2.5 × 10⁷ M⁻¹ s⁻¹. This in line with the above discussed ΔG^≠ values. However, HT is the main mechanism involved in the ˙OOH scavenging activity of analogue IIIb, at pH = 7.4, because of its relatively high pK_a.

The most important finding, regarding primary AOC, is that all the investigated melatonin analogues are much better peroxyl radical scavengers than the parent molecule. Moreover, all the analogues except Ia, react faster with ˙OOH than NAS, which is a mono-phenolic compound with the OH group in site R₃. Comparing analogue Ia with 6OHM, the only structural difference is the group at the R₃ site. While they both have an OH group at site R₄, R₃ = –H and –OCH₃ for Ia and 6OHM, respectively. Therefore, according to the calculated data it seems that the presence of a methoxy group next to the phenolic HT site slightly promotes reactivity towards peroxyl radicals.

The only of the modeled analogues that is predicted to have better ˙OOH scavenging activity than 6OHM, is Ib. However, the overall rate coefficients of analogues IIc and IIIa are very close to that of 6OHM, and the differences are probably within the method uncertainties. Analyzing the structural features of Ib, IIc and IIIa the first thing that stands out is that they have different number of OH groups (1, 2 and 3, respectively). However, all of them have an OH group at site R₅. Thus, it seems that this particular feature is important for maximizing the primary antioxidant activity of melatonin analogues. Therefore, the Ib and IIc frameworks have been chosen for designing the test set 3, which is meant for multifunctional AOC.

Compared with other known antioxidants, the ˙OOH scavenging activity of the analogues in test set 1, except Ia, in aqueous solution at physiological pH, was found to be larger than those of Trolox,⁸⁹ caffeine,⁹⁰ capsaicin,⁹¹ α-mangostina,⁹² and gallic acid;⁷⁰ and similar to those of ascorbic acid,⁹³ resveratrol⁸⁷ and glutathione.⁹⁴ On the other hand, their primary AOC is surpassed by those of propyl gallate,⁹⁵ piceatannol,⁸⁷ edaravone⁹⁶ and caffeic acid,⁶⁹ but only moderately. Therefore, regardless of the differences among the analogues in test set 1, all of them (except Ia) can be proposed as promising primary antioxidants.

In addition, it has been previously reported that phenolic compounds can be regenerated under physiological conditions. Such regeneration takes place in such a way that these compounds can scavenge several radical equivalents in the process, two per cycle (one ˙OOH and one O₂˙⁻).^{56,87,95,97–101} This is expected to be the case for the compounds investigated here and to increase their antioxidant capacity.

Melatonin analogues designed for secondary antioxidant activity

Hydroxyl radical (˙OH) is one of the most reactive and dangerous of the oxidants present in biological systems. ˙OH is so reactive that it can damage almost any molecule in the vicinity of its site of formation.¹⁰² Therefore, the best strategy to prevent the oxidative damage caused by this radical is not scavenging it but preventing its production. One of the most important chemical routes yielding ˙OH is the Fenton reaction, or the metal catalyzed Haber–Weiss recombination (HWR). Therefore, because of the role of metal ions in such processes, the necessity of antioxidants able of chelating such metals has been pointed out.^103,104

Copper has been chosen for testing the chelating ability of the investigated melatonin analogues in test set 2 (Scheme 4). This metal has been chosen because there is evidence supporting its role in the pathogenesis of neurodegenerative disorders.¹⁰⁵ This has been related the formation of oxidative species,¹⁰⁶ in particular ˙OH.¹⁰⁵ Moreover, it has been reported that under identical experimental conditions the toxicity of Cu(II), in term of oxidative damage, is larger than that of Fe(III).^103,107 In addition, the copper chelating ability of melatonin and some of its metabolites has been previously assessed,¹⁸ thus it can be used as a reference framework.

The idea here is that while Cu(I) is required for producing ˙OH, Cu(II) is the most abundant and stable oxidative state of copper. Therefore, chelating agents capable of decreasing the viability of Cu(II) reduction should be effective for preventing, or inhibiting, the ˙OH production and the consequent oxidative stress. For that strategy to be successful it is crucial that the chelation reactions yield stable complexes, i.e. they must be exergonic. Thus, this has been the first aspect explored in this work (Table 4).

Table 4 Conditional Gibbs free energies of reaction (ΔG′, kcal mol⁻¹) and conditional equilibrium constants (K′, M), at pH = 7.4, for the Cu(II) chelation by melatonin analogues in test set 2. Reduction potential vs. NHE (E⁰, eV), Gibbs free energies (ΔG, kcal mol⁻¹) and rate constants (k, M⁻¹ s⁻¹) for the Cu(II) reduction reactions by O₂˙⁻ and ascorbate. All the data was calculated at 298.15 K

	ΔG′	K′	E^0	ΔG (O2˙^−)	ΔG (Asc^−)	k (O2˙^−)	k (Asc^−)
Cu(II)			0.16	−13.01	−2.81	6.68 × 10^8	1.83 × 10^7
A	−2.34	5.2 × 10^1	0.01	−3.87	6.33	7.13 × 10^6	7.59 × 10^4
B	1.09
C	−1.33	9.5 × 10^0	0.04	−4.73	5.47	1.07 × 10^7	1.44 × 10^5
D	−5.35	8.3 × 10^3	−0.20	1.02	11.21	3.45 × 10^5	7.89 × 10^2
E	0.61
F	−0.61	2.8 × 10^0	0.15	−7.23	2.96	6.89 × 10^6	1.63 × 10^5
G	−4.89	3.9 × 10^3	−0.16	−0.08	10.11	2.87 × 10^5	1.08 × 10^3

---

It was found that all the reactions involving Cu(II) chelation by the designed melatonin analogues in test set 2, except B and E, are exergonic at room temperature (Table 4). Therefore, analogues A, C, D, F and G are predicted to be able of acting as Cu(II) ligands. Their optimized geometries are provided as ESI.† The most stable chelates were found to be those involving analogues D and G, with conditional equilibrium constants equal to 8.3 × 10³ and 3.9 × 10³ M, respectively, at physiological pH.

Additional data, regarding Cu(II) reduction, were computed for all the chelates yield by reactions with negative ΔG values. It was found that chelation by melatonin analogues A, C, D and G significantly lower the reduction potential of Cu(II), with respect to free copper, with D and G showing the maximum effect (Table 4). Accordingly, A, C, D and G are expected to decrease the feasibility of the Cu(II) reduction, being D and G the most efficient for that purpose.

The Gibbs free energies and rate constants of the reactions of Cu(II) with two reductants, i.e., O₂˙⁻ and ascorbate (Asc⁻), have also been calculated (Table 4). The reaction with O₂˙⁻ corresponds to the first step of the HWR, and the reaction with Asc⁻ was modeled to include the possible presence of other, less strong, reductants in the biological systems. In addition, copper–ascorbate mixtures are frequently used in experiments to generate oxidative conditions. O₂˙⁻ was calculated including 4 explicit water molecules, which is a model previously validated for this species.¹⁰⁸ In the calculation of the Gibbs energies the correction to E⁰_Cu(II)/Cu(I) was included, as well as equivalent corrections for the pairs ³O₂/O₂˙⁻ and Asc/Asc⁻. These empirical corrections are expected to facilitate possible further comparisons with experiments.

The ideal behavior for a secondary antioxidant acting as a Cu(II) chelating agent should be to fully inhibit the reactions with reductants, O₂˙⁻ and Asc⁻ in this case. It was found that all the stable chelates with melatonin analogues turn the Cu(II) + Asc⁻ reaction into a non-viable process, i.e., with positive ΔG values. On the other hand, only the D analogue makes the Cu(II) reduction by O₂˙⁻ an endergonic process, albeit for analogue G the reaction becomes isoergonic (Table 4). Regarding the kinetics, these two species, D and G, are also the ones slowing the most the Cu(II) reduction. The rate constants for “free” Cu(II) were estimated to be 6.7 × 10⁸ and 1.8 × 10⁷ M⁻¹ s⁻¹, for the reactions with O₂˙⁻ and Asc⁻ respectively. According to the values in Table 4, chelation by melatonin analogs D and G decrease these values by about 2000 and 20 000 times. The reorganization energies and the ΔG^≠ values used in the calculations of the rate constants are reported in Table S3, ESI.†

In addition, the melatonin analogues that chelates Cu(II) through endergonic reactions are expected to act as ˙OH-inactivating ligand (OIL) in biological systems.¹⁰⁹ Some of them (D and G) can directly sequester metal ions from reductants. The others (A, C and F), which only downgrade the Cu(II) reduction by O₂˙⁻, should still be able of deactivating OH radicals immediately after formed via Fenton-like reactions. This is because the high reactivity of this radical towards organic frameworks would facilitate the reaction with the ligands in the Cu chelates, which would be just in the ˙OH formation site.

Analyzing altogether the data regarding the secondary AOC of the investigated melatonin analogs by chelating metal ions, it becomes evident that D and G are the most promising species for inhibiting ˙OH production, and thus the consequent oxidative stress. Therefore, their structural features have been those included in the design of melatonin analogues with multifunctional AOC.

Melatonin analogues designed for multifunctional antioxidant activity

Based on the results discussed in the two previous sections, the structural features of primary antioxidants Ib and IIc and of secondary antioxidants D and G have been included in the set of melatonin analogs designed to be multifunctional antioxidants (test set 3, Scheme 5). The IIIa structure has not been included to avoid having many H donor groups in the structure (please see the next section for more details).

The estimated first pK_a values of the analogues in test set 3 are reported in Table 5, together with the most likely deprotonation site for the two di-phenols, and the molar fractions of the neutral and mono-anionic species at physiological pH. They are all predicted to be dominantly in their neutral form at this pH, but with significant amounts of the mono-anions, except IbG, which shows the opposite trend.

Table 5 First phenolic pK_a values, most likely deprotonation site (DP) for melatonin analogues in test set 3, reference acid used to calculate the pK_a (HRef), and molar fractions of the neutral and mono-anionic species (^Mf_neutral, ^Mf_anion) at physiological pH (pH = 7.4)

	pKa	DP	HRef	^Mfneutral	^Mfanion
IbD	8.0		Phenol	0.806	0.194
IbG	7.2		Phenol	0.378	0.622
IIcD	7.9	R5	Catechol	0.761	0.239
IIcG	7.7	R5	Catechol	0.670	0.330

---

The information on the primary and secondary antioxidant activities of analogues in test set 3 is reported in Tables 6 and 7, respectively. The ΔG, ΔG^≠, imaginary frequencies of the TS, tunneling corrections and k^HT_i, are provided in Table S3, ESI.† The optimized geometries of the TS are also provided as ESI,† as well as the reorganization energies and the ΔG^≠ values for the SPLET reactions (Table S4†).

Table 6 Total rate coefficients of the HT reactions (k^HT_tot, M⁻¹ s⁻¹), rate constants of the SPLET reactions (k^SPLET, M⁻¹ s⁻¹), overall rate coefficients (k_overall, M⁻¹ s⁻¹), percent contributions of the HT (%^HT) and SPLET (%^SPLET) mechanisms, for the reactions of melatonin analogues in test set 3 with ˙OOH. All the data is reported at 298.15 K

	k^HTtot	k^SPLET	koverall, pH = 7.4	k′overall, pH = 7.4	%^HT, pH = 7.4	%^SPLET, pH = 7.4
IbD	2.47 × 10^5	5.23 × 10^9	1.02 × 10^9	2.54 × 10^6	<0.1	∼100
IbG	4.97 × 10^5	4.37 × 10^9	2.72 × 10^9	6.79 × 10^6	<0.1	∼100
IIcD	7.48 × 10^7	7.79 × 10^9	1.92 × 10^9	4.79 × 10^6	3.0	97.0
IIcG	9.61 × 10^6	7.84 × 10^9	2.59 × 10^9	6.48 × 10^6	0.2	99.8

---

Table 7 Conditional Gibbs free energies of reaction (ΔG′, kcal mol⁻¹) and conditional equilibrium constants (K′, M), at pH = 7.4, for the most likely Cu(II) chelation routes by melatonin analogues in test set 3. Reduction potential vs. NHE (E⁰, eV), Gibbs free energies (ΔG, kcal mol⁻¹) and rate constants (k, M⁻¹ s⁻¹) for the Cu(II) reduction reactions by O₂˙⁻ and ascorbate. All the data was calculated at 298.15 K

	ΔG′	K′	E^0	ΔG (O2˙^−)	ΔG (Asc^−)	k (O2˙^−)	k (Asc^−)
IbD	−5.38	8.76 × 10^3	−0.22	1.49	11.68	2.29 × 10^5	4.49 × 10^2
IbG	−7.52	3.23 × 10^5	−0.31	3.54	13.74	4.58 × 10^4	4.17 × 10^1
IIcD	−6.39	4.85 × 10^4	−0.30	3.17	13.36	4.91 × 10^4	5.46 × 10^1
IIcG	−6.47	5.55 × 10^4	−0.25	2.15	12.34	5.08 × 10^4	9.53 × 10^1

---

The overall rate coefficients for the reactions of these analogues with ˙OOH were found to be within, or near to, the diffusion limit. This indicates that they all should be excellent peroxyl radical scavengers, contrary to what has been reported for the parent molecule. Moreover, including the structural features of the secondary antioxidants D and G in the phenolic melatonin analogues does not interfere with the primary antioxidant activity. On the contrary, the analogs in test set 3 are all predicted to be better peroxyl radical scavengers than those in test set 1. For the multifunctional melatonin analogues the primary AOC is dominated by the SPLET route, with very small contributions from HT. The largest one appears to be for IIcD, and is only about 3%.

Compared to other known antioxidants, the melatonin analogues in test set 3 are predicted to be among the best peroxyl radical scavengers reported so far. They seem to be able of reacting with these radicals as fast as possible, considering the limit imposed by diffusion, at least in aqueous solution, at physiological pH. Such fast reactions are rare to find, which supports the idea that the designed melatonin analogues should be exceptionally good primary antioxidants.

Regarding their secondary AOC, the analogues in test set 3 can form different chelates with Cu(II), due to the presence of phenolic groups in the six membered ring of the indole moiety. Some of them are equivalent to the ones formed by molecules in test set 2, i.e., involving the indolic N atom, and the O atom in the R₂ group (c1). Others involve the OH in R₅ and the indolic N (c2), and for IIcD and IIcG there are also possible chelates involving the OH groups in R₄ and R₅ (c3). The Gibbs energies of all these chelation reactions, and the corresponding equilibrium constants are reported in Table S5, ESI.† It was found that, in all the cases, the c1 chelates are the lowest in energy. Thus, the following analyses are just for them, as they should be the most abundant ones.

The data on the secondary AOC of these compounds is reported in Table 7, while the reorganization energy and the ΔG^≠ values used in the calculations of the corresponding rate constants are provided in Table S4, ESI.† It was found that the reactions yielding the c1 chelates are systematically more exergonic for the molecules in the test set 3 than those involving any of the chelating agents in test set 2. This indicates that the presence of OH groups in the six membered ring of the indole moiety reinforces the chelating ability of melatonin analogues, with the exception of IbD which has a ΔG′ value at pH = 7.4 almost identical to that of the D analogue.

The same applies for all the other calculated properties. Chelates with IbD, IbG, IIcD and IIcG as ligands, all inhibit the Cu(II) reduction to a larger extent than the chelates involving the melatonin analogues in test set 2. They all turn endergonic the reactions with both O₂˙⁻ and Asc⁻ and dramatically slow them down. The most efficient as OH inhibitors, via metal chelation are expected to be IbG and IIcD. Considering that they also are excellent for scavenging free radicals, these two melatonin analogues are proposed as the most promising multifunctional antioxidants, especially IIcD due to its better secondary AOC.

Other relevant features

In addition to their reactivity towards oxidants, there are other aspects to consider regarding the potential use of the investigated melatonin analogues as potential drugs with therapeutic applications against oxidative stress and related health issues. To that purpose several molecular descriptors have been calculated. They were chosen to check if the designed compounds fulfill the Lipinski's rule of five¹¹⁰ and the Ghose's rule.¹¹¹ These two sets of rules establish some criteria that any orally active drug must have. According to Lipinski's rule they should have no more than 5 hydrogen bond donors (HB^D), no more than 10 (5 × 2) hydrogen bond acceptors (HB^A), a molecular weight (MW) under 500 (5 × 100) g mol⁻¹, and an octanol/water partition coefficient (log P) lower than 5. Molecules violating more than one of these rules may have problems with bioavailability. On the other hand the Ghose's rule establishes that orally active drugs must have a log P ranging from −0.4 to 5.6, molar refractivity (^MR) from 40 to 130, MW from 160 to 480, and a number of heavy atoms (^XAt) from 20 to 70. Otherwise the intended effects should be impeded by poor permeation or absorption. Different methods have been used to estimate log P (Table 8), and the values are all in the range 0.36–2.64, i.e., within the limits established in both Lipinski's and Ghose's rules. Discrepancies among the log P values calculated using different methods are attributed to the fact that the strategies involved in the calculations are also different. For example milog P uses molecular fragments while ALOGPS is an atomic-contribution method. More details on this topic can be found elsewhere.¹¹²

Table 8 Molecular descriptors for the whole set of melatonin analogues designed in the present work. log P, number of hydrogen bond acceptors (HB^A), number of hydrogen bond donors (HB^D), molecular weight (MW), molar refractivity (^MR), number of heavy atoms (^XAt), topological polar surface area (TPSA) and rotatable bonds (RB). Melatonin is included for comparison purposes

	log P^a	log P^b	log P^c	HB^A	HB^D	MW	^MR^d	^XAt	TPSA^e	RB
a Calculated using Crippen's method.^115b Calculated using ALOGPS 2.1.^116c Calculated using milog P 2.2 from Molinspiration online property calculation toolkit.d Calculated using the Ghose & Crippen method.^117e Reported in Å^2, calculated using the Molinspiration Property Calculation Service (http://www.molinspiration.com).
Melatonin	1.75	1.42	1.45	4	2	232	65.6	17	54.12	4
Ia	1.65	1.42	1.45	4	3	218	60.9	16	65.12	4
Ib	1.65	0.98	0.91	4	3	218	60.9	16	65.12	4
Ic	1.65	1.02	0.91	4	3	218	60.9	16	65.12	4
IIa	1.26	0.98	1.15	5	4	234	62.7	17	85.34	4
IIb	1.26	0.95	1.15	5	4	234	62.7	17	85.34	4
IIc	1.26	0.92	0.42	5	4	234	62.7	17	85.34	4
IIIa	0.87	0.87	0.65	6	5	250	64.5	18	105.57	4
IIIb	0.87	0.93	0.65	6	5	250	64.5	18	105.57	4
A	1.56	0.58	0.36	4	2	230	65.6	17	61.96	5
B	1.39	0.55	0.36	4	2	244	69.6	18	61.96	6
C	1.48	1.53	1.44	5	2	260	69.9	19	71.19	6
D	1.63	1.98	1.65	4	2	244	70.2	18	61.96	5
E	0.92	1.78	1.48	5	2	259	74.0	19	65.20	6
F	1.90	1.78	1.55	5	2	273	74.8	20	65.20	6
G	1.36	1.15	1.02	4	2	282	85.4	21	61.96	7
IbD	1.24	1.56	1.00	5	3	260	72.0	19	82.19	5
IbG	0.97	2.64	2.36	5	3	298	87.1	22	82.19	7
IIcD	0.85	1.36	1.28	6	4	276	73.8	20	102.42	5
IIcG	0.58	2.37	2.07	6	4	314	88.9	23	102.42	7

---

The values of all these molecular descriptors for the whole set of melatonin analogues designed in this work are provided in Table 8. As the values in this table show all the investigated compounds fulfill all the criteria in the Lipinski's rule. Only two exceptions were found (IIIa and IIIb), which belong to test set 1 and have 5 HB^D. That is why their structural features were not included in the test set for multifunctional antioxidant activity. Regarding the Ghose's rule, the most frequent problem is related to the ^XAt value that is slightly below 20 for most of the studied compounds. However, it is not expected to be a major inconvenience since melatonin itself has ^XAt = 17, and its beneficial effects as an oral drug has been extensively documented. The melatonin analogues that fulfill all the criteria in both Lipinski's and Ghose's rules are analogues F, G, IbG, IIcD and IIcG.

It should be noted, however, that these rules are empirical, and based on statistical analyses using the existing data on the bioavailability of oral drugs. Thus, they are meant to be guidelines, not rigorous laws. Moreover, viable drugs must also have other important features including manufacturability and safety.¹¹³ Therefore, these aspects are also addressed here. In addition, it is currently considered that physicochemical properties other than those included in Lipinski's and Ghose's rules should also be explored for newly developed drugs.¹¹⁴ Some examples are topological polar surface area (TPSA) and the number of rotatable bonds (RB). Accordingly, they have also been estimated and are included in Table 8.

The estimated synthetic accessibility, of each designed compound, is reported in Table 9 where melatonin and other – currently in use – drugs have been included for comparison purposes. The SA values can be grouped into 3 categories, using the default thresholds of the SYLVIA program: easy (SA ≤ 3), medium (3 < SA < 6), and difficult (SA ≥ 6). According to this classification all the compounds designed in this work correspond to medium SA, with values ranging from 3.5 to 4.3. Even the largest value is close enough to that of ciprofloxacin, which suggests that their manufacturability should not be problematic. In the particular case of the two compounds identified here as the most promising multifunctional antioxidants, the SA value for IIcD (3.9) is lower than that of IbG (4.2).

Table 9 Synthetic accessibility (SA), oral rat 50 percent lethal dose (LD₅₀), Ames mutagenicity (M), 96 hour fathead minnow 50 percent lethal concentration (LC^F₅₀), 48 hour Daphnia magna 50 percent lethal concentration (LC^D₅₀), and Tetrahymena pyriformis 50 percent growth inhibition concentration (IGC₅₀). Melatonin, and some other known drugs, are included for comparison purposes

	SA	LD50	M	LC^F50	LC^D50	IGC50
Melatonin	2.46	1298.11	0.05 (−)	23.51	2.25	66.62
Ia	3.50	1558.61	0.08 (−)	28.67	2.56	78.32
Ib	3.54	1526.48	−0.01 (−)	21.41	3.72	78.60
Ic	3.56	1248.74	−0.08 (−)	25.59	4.66	78.12
IIa	3.61	1061.32	0.18 (−)	20.55	4.44	58.25
IIb	3.63	1066.44	0.26 (−)	9.19	2.98	59.47
IIc	3.61	1273.36	0.12 (−)	13.71	6.64	58.17
IIIa	3.70	1646.33	0.36 (−)	5.73	6.65	52.64
IIIb	3.71	1163.93	0.41 (−)	8.09	3.17	34.20
A	3.65	1442.93	0.17 (−)	5.86	4.58	47.98
B	3.67	1732.22	0.01 (−)	3.26	N/A	34.81
C	3.86	997.81	0.03 (−)	4.04	11.85	48.55
D	3.74	1352.30	0.07 (−)	3.89	5.39	28.51
E	3.92	573.98	0.30 (−)	14.30	N/A	N/A
F	4.03	709.19	0.46 (−)	6.99	25.19	75.83
G	4.10	1135.35	0.51 (+)	0.62	0.72	2.40
IbD	3.82	1660.01	0.09 (−)	3.68	6.25	21.02
IbG	4.22	526.23	0.47 (−)	0.17	0.19	2.03
IIcD	3.90	1203.01	0.30 (−)	3.11	9.80	20.96
IIcG	4.30	265.38	0.40 (−)	0.25	0.33	2.46
Aspirin	2.20	757.21	0.43 (−)	80.28	472.51	472.51
Floxetin	3.98	1002.16	0.13 (−)	0.41	6.10 × 10^−2	2.33
Ciprofloxacin	4.26	36.25	0.70 (+)	0.11	0.16	N/A

---

Regarding the toxicity descriptors, the lower the value of M, and the larger the values of LD₅₀, LC^F₅₀, LC^D₅₀ and IGC₅₀, the lower the toxicity of the tested chemical. G is the only of the designed compounds that was identified as potentially mutagenic. Between compounds IIcD and IbG, the later have a larger value of M, but still its mutagenicity is predicted as negative and with a value similar to that of aspirin. Regarding the LD₅₀, LC^F₅₀, LC^D₅₀ and IGC₅₀ descriptors, the range taken from melatonin and other currently in use drugs are 1298.11–36.25, 80.28–0.11, 472.51–0.06, and 472.51–2.33, respectively. The values corresponding to most of the designed compounds are within these ranges. Comparing the IIcD and IbG compounds, the first one was systematically found to have the lowest predicted toxicity, regardless of the descriptor used.

Therefore, considering the values of the molecular descriptors and the primary and secondary AOC of all the investigated compounds, altogether, melatonin analogue IIcD is identified as the most promising agent against oxidative stress, and its deleterious effects. It is expected to efficiently act as a multifunctional antioxidant, contrary to what has been previously described for naturally occurring melatonin derivatives, i.e., most of them are only good as primary or secondary antioxidant, but not as both simultaneously. IIcD is also predicted to be synthetically accessible and with low toxicity. Hopefully, it might be synthesized in the near future, thus its activity could be experimentally corroborated.

Conclusions

A set of 19 melatonin derivatives, intended to be better antioxidants than the parent molecule, have been computationally designed. Eight of them were planned to be good primary AOC, i.e., good free radical scavengers. Seven of them were designed for their secondary AOC, to be able of inhibiting ˙OH production by acting as metal ion chelators. Based on their predicted behavior for the intended functions, four multifunctional melatonin analogues were proposed. They were found to be among the best peroxyl radical scavengers identified so far (in aqueous solution, at physiological pH) with rate constants within, or near to, the diffusion limited regime. They are predicted to be better antioxidants than Trolox, resveratrol, ascorbic acid, and other known antioxidants. Three of the multifunctional analogues were also found to be capable of turning off the Cu(II) reduction by O₂˙⁻ and Asc⁻, thus fully inhibiting the associated ˙OH production. Two of them, namely the IbG and IIcD were identified as the most promising multifunctional antioxidants. In addition, they fulfill both the Lipinski's and Ghose's rules for orally active drugs. However, based on potential toxicity and synthetic accessibility estimations, IIcD has been chosen as the best prospect for possible application. Hopefully, this investigation might provide motivation for further investigations in the subject, and the synthesis of this compound, so its potential role as protector against oxidative stress, and the associated health issues, could be experimentally tested.

Acknowledgements

The Laboratorio de Visualización y Cómputo Paralelo at Universidad Autónoma Metropolitana-Iztapalapa is acknowledge for computing time. This work was partially supported by project SEP-CONACyT 167491.

Notes and References

1. B. Tekiner-Gulbas, A. D. Westwell and S. Suzen, Curr. Med. Chem., 2013, 20, 4451–4459.
2. M. Matsuda and I. Shimomura, Rev. Endocr. Metab. Disord., 2014, 15, 1–10.
3. E. Eren, H. Y. Ellidag, Y. Cekin, R. U. Ayoglu, A. O. Sekercioglu and N. Yilmaz, Redox Rep., 2014, 19, 34–39.
4. B. Halliwell, Drugs Aging, 2001, 18, 685–716.
5. M. Pohanka, Curr. Med. Chem., 2014, 21, 356–364.
6. C. Pimentel, L. Batista-Nascimento, C. Rodrigues-Pousada and R. A. Menezes, Oxid. Med. Cell. Longevity, 2012, 2012, 132146.
7. M. R. Ramis, S. Esteban, A. Miralles, D. X. Tan and R. J. Reiter, Curr. Med. Chem., 2015, 22, 2690–2711.
8. A. Galano, D. X. Tan and R. J. Reiter, J. Pineal Res., 2011, 51, 1–16.
9. R. Hardeland, Endocrine, 2005, 27, 119–130.
10. G. Jahnke, M. Marr, C. Myers, R. Wilson, G. Travlos and C. Price, Toxicol. Sci., 1999, 50, 271–279.
11. L. Ceraulo, M. Ferrugia, L. Tesoriere, S. Segreto, M. A. Livrea and V. Turco Liveri, J. Pineal Res., 1999, 26, 108–112.
12. D. Bonnefont-Rousselot and F. Collin, Toxicology, 2010, 278, 55–67.
13. H. Gurer-Orhan and S. Suzen, Curr. Med. Chem., 2015, 22, 490–499.
14. A. Galano, D. X. Tan and R. J. Reiter, J. Pineal Res., 2013, 54, 245–257.
15. R. J. Reiter, D. X. Tan, M. J. Jou, A. Korkmaz, L. C. Manchester and S. D. Paredes, Neuroendocrinol. Lett., 2008, 29, 391–398.
16. D. X. Tan, L. C. Manchester, R. J. Reiter, W. B. Qi, M. Karbownik and J. R. Calvo, Biol. Signals Recept., 2000, 9, 137–159.
17. R. J. Reiter, D. X. Tan and A. Galano, Physiology, 2014, 29, 325–333.
18. A. Galano, M. E. Medina, D. X. Tan and R. J. Reiter, J. Pineal Res., 2015, 58, 107–116.
19. R. Álvarez-Diduk, A. Galano, D. X. Tan and R. J. Reiter, J. Phys. Chem. B, 2015, 119, 8535–8543.
20. S. Suzen, Curr. Drug Delivery, 2013, 10, 71–75.
21. A. Lira-Rocha, O. Espejo-González and E. B. Naranjo-Rodríguez, Eur. J. Med. Chem., 2002, 37, 945–951.
22. R. Bustamante-García, E. B. Naranjo-Rodríguez, A. S. Lira-Rocha, A. Ortiz-Osornio and O. Picazo, Behav. Brain Res., 2013, 236, 148–156.
23. R. Bustamante-García, A. S. Lira-Rocha, O. Espejo-González, A. E. Gómez-Martínez and O. Picazo, Prog. Neuro-Psychopharmacol. Biol. Psychiatry, 2014, 51, 133–139.
24. C. Phiphatwatcharaded, A. Topark-Ngarm, P. Puthongking and P. Mahakunakorn, Drug Dev. Res., 2014, 75, 235–245.
25. J. R. Johns and J. A. Platts, Org. Biomol. Chem., 2014, 12, 7820–7827.
26. P. L. Tsia and M. K. Hu, J. Pharm. Pharmacol., 2003, 55, 1655–1660.
27. Z. Ates-Alagoz, T. Coban and E. Buyukbingol, Arch. Pharm., 2006, 339, 193–200.
28. Z. Ateş-Alagöz, T. Coban and S. Suzen, Med. Chem. Res., 2005, 14, 169–179.
29. S. Suzen, P. Bozkaya, T. Coban and D. Nebioǧlu, J. Enzyme Inhib. Med. Chem., 2006, 21, 405–411.
30. H. Shirinzadeh, B. Eren, H. Gurer-Orhan, S. Suzen and S. Özden, Molecules, 2010, 15, 2187–2202.
31. A. D. Yilmaz, T. Coban and S. Suzen, J. Enzyme Inhib. Med. Chem., 2012, 27, 428–436.
32. G. Gürkök, T. Coban and S. Suzen, J. Enzyme Inhib. Med. Chem., 2009, 24, 506–515.
33. S. Suzen, S. S. Cihaner and T. Coban, Chem. Biol. Drug Des., 2012, 79, 76–83.
34. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F. Ogliaro, M. J. Bearpark, J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, Journal, 2009.
35. Y. Zhao, N. E. Schultz and D. G. Truhlar, J. Chem. Theory Comput., 2006, 2, 364–382.
36. A. V. Marenich, C. J. Cramer and D. G. Truhlar, J. Phys. Chem. B, 2009, 113, 6378–6396.
37. M. Altarawneh and B. Z. Dlugogorski, Environ. Sci. Technol., 2013, 47, 5118–5127.
38. M. Dargiewicz, M. Biczysko, R. Improta and V. Barone, Phys. Chem. Chem. Phys., 2012, 14, 8981–8989.
39. J. Murillo, D. Henao, E. Vélez, C. Castaño, J. Quijano, J. Gaviria and E. Zapata, Int. J. Chem. Kinet., 2012, 44, 407–413.
40. K. P. Prasanthkumar and J. R. Alvarez-Idaboy, RSC Adv., 2014, 4, 14157–14164.
41. A. Galano and J. R. Alvarez-Idaboy, J. Comput. Chem., 2014, 35, 2019–2026.
42. Y. Zhao and D. G. Truhlar, J. Phys. Chem. A, 2008, 112, 1095–1099.
43. J. Ho and M. L. Coote, Theor. Chem. Acc., 2010, 125, 3–21.
44. H. Eyring, J. Chem. Phys., 1935, 3, 63–71.
45. M. G. Evans and M. Polanyi, Trans. Faraday Soc., 1935, 31, 875–894.
46. D. G. Truhlar, B. C. Garrett and S. J. Klippenstein, J. Phys. Chem., 1996, 100, 12771–12800.
47. A. Kuppermann and D. G. Truhlar, J. Am. Chem. Soc., 1971, 93, 1840–1851.
48. R. A. Marcus, Pure Appl. Chem., 1997, 69, 13–29.
49. F. C. Collins and G. E. Kimball, J. Colloid Sci., 1949, 4, 425–437.
50. M. Smoluchowski, Z. Phys. Chem., 1917, 92, 129–168.
51. A. Einstein, Ann. Phys., 1905, 17, 549–560.
52. G. G. Stokes, Mathematical and Physical Papers, Cambridge University Press, Cambridge, 1903.
53. A. Galano and J. R. Alvarez-Idaboy, J. Comput. Chem., 2013, 34, 2430–2445.
54. M. Francisco-Marquez, M. Aguilar-Fernández and A. Galano, Comput. Theor. Chem., 2016, 1077, 18–24.
55. Y. Villuendas-Rey, J. R. Alvarez-Idaboy and A. Galano, J. Chem. Inf. Model., 2015, 55, 2552–2561.
56. R. Álvarez-Diduk and A. Galano, J. Phys. Chem. B, 2015, 119, 3479–3491.
57. D. M. Camaioni and C. A. Schwerdtfeger, J. Phys. Chem. A, 2005, 109, 10795–10797.
58. V. S. Bryantsev, M. S. Diallo and W. A. Goddard Iii, J. Phys. Chem. A, 2009, 113, 9559–9567.
59. J. Ortega-Castro, M. Adrover, J. Frau, J. Donoso and F. Muñoz, Chem. Phys. Lett., 2009, 475, 277–284.
60. T. Fujii, F. Moynier, M. Abe, K. Nemoto and F. Albarède, Geochim. Cosmochim. Acta, 2013, 110, 29–44.
61. J. L. Fulton, M. M. Hoffmann and J. G. Darab, Chem. Phys. Lett., 2000, 330, 300–308.
62. J. L. Fulton, M. M. Hoffmann, J. G. Darab, B. J. Palmer and E. A. Stern, J. Phys. Chem. A, 2000, 104, 11651–11653.
63. Y. Fu, L. Liu, H. Z. Yu, Y. M. Wang and Q. X. Guo, J. Am. Chem. Soc., 2005, 127, 7227–7234.
64. K. Boda, T. Seidel and J. Gasteiger, J. Comput.–Aided Mol. Des., 2007, 21, 311–325.
65. P. Bonnet, Eur. J. Med. Chem., 2012, 54, 679–689.
66. H. Zhu, A. Tropsha, D. Fourches, A. Varnek, E. Papa, P. Gramatical, T. Öberg, P. Dao, A. Cherkasov and I. V. Tetko, J. Chem. Inf. Model., 2008, 48, 766–784.
67. C. Rodriguez-Rodriguez, N. Sanchez de Groot, A. Rimola, A. Alvarez-Larena, V. Lloveras, J. Vidal-Gancedo, S. Ventura, J. Vendrell, M. Sodupe and P. Gonzalez-Duarte, J. Am. Chem. Soc., 2009, 131, 1436–1451.
68. R. M. Han, D. D. Li, C. H. Chen, R. Liang, Y. X. Tian, J. P. Zhang and L. H. Skibsted, J. Agric. Food Chem., 2011, 59, 10367–10372.
69. J. R. León-Carmona, J. R. Alvarez-Idaboy and A. Galano, Phys. Chem. Chem. Phys., 2012, 14, 12534–12543.
70. T. Marino, A. Galano and N. Russo, J. Phys. Chem. B, 2014, 118, 10380–10389.
71. A. Pérez-González, A. Galano and J. R. Alvarez-Idaboy, New J. Chem., 2014, 38, 2639–2652.
72. R. Álvarez-Diduk, A. Galano, D. X. Tan and R. J. Reiter, Theor. Chem. Acc., 2016, 135, 1–5.
73. A. Da̧browski, P. Podkościelny, Z. Hubicki and M. Barczak, Chemosphere, 2005, 58, 1049–1070.
74. J. R. Torres-Lapasió, M. C. García-Alvarez-Coque, E. Bosch and M. Rosés, J. Chromatogr., 2005, 1089, 170–186.
75. M. D. Liptak, K. C. Gross, P. G. Seybold, S. Feldgus and G. C. Shields, J. Am. Chem. Soc., 2002, 124, 6421–6427.
76. N. Schweigert, R. W. Hunziker, B. I. Escher and R. I. L. Eggen, Environ. Toxicol. Chem., 2001, 20, 239–247.
77. G. Ivanova, E. Bratovanova and D. Petkov, J. Pept. Sci., 2002, 8, 8–12.
78. H. S. Kim, T. D. Chung and H. Kim, J. Electroanal. Chem., 2001, 498, 209–215.
79. I. F. Sedeh, L. O. Ohman and S. Sjöberg, Acta Chem. Scand., 1992, 46, 933–940.
80. T. Stalin, P. Vasantha Rani, B. Shanthi, A. Sekar and N. Rajendiran, Indian J. Chem., 2006, 45, 1113–1120.
81. A. D. N. J. De Grey, DNA Cell Biol., 2002, 21, 251–257.
82. P. Terpinc and H. Abramovič, Food Chem., 2010, 121, 366–371.
83. R. C. Rose and A. M. Bode, FASEB J., 1993, 7, 1135–1142.
84. G. Litwinienko and K. U. Ingold, Acc. Chem. Res., 2007, 40, 222–230.
85. L. Cao, H. Yu, S. Shao, S. Wang and Y. Guo, Anal. Methods, 2014, 6, 7149–7153.
86. C. Iuga, J. R. Alvarez-Idaboy and N. Russo, J. Org. Chem., 2012, 77, 3868–3877.
87. M. Cordova-Gomez, A. Galano and J. R. Alvarez-Idaboy, RSC Adv., 2013, 3, 20209–20218.
88. J. M. Dimitrić Marković, D. Milenković, D. Amić, M. Mojović, I. Pašti and Z. S. Marković, RSC Adv., 2014, 4, 32228–32236.
89. M. E. Alberto, N. Russo, A. Grand and A. Galano, Phys. Chem. Chem. Phys., 2013, 15, 4642–4650.
90. J. R. León-Carmona and A. Galano, J. Phys. Chem. B, 2011, 115, 4538–4546.
91. A. Galano and A. Martínez, J. Phys. Chem. B, 2012, 116, 1200–1208.
92. A. Martínez, A. Galano and R. Vargas, J. Phys. Chem. B, 2011, 115, 12591–12598.
93. A. Galano and J. R. Alvarez-Idaboy, J. Comput. Chem., 2013, 34, 2430–2445.
94. A. Galano and J. R. Alvarez-Idaboy, RSC Adv., 2011, 1, 1763–1771.
95. M. E. Medina, C. Iuga and J. R. Alvarez-Idaboy, Phys. Chem. Chem. Phys., 2013, 15, 13137–13146.
96. A. Pérez-González and A. Galano, J. Phys. Chem. B, 2011, 115, 1306–1314.
97. A. Galano and A. Pérez-González, Theor. Chem. Acc., 2012, 131, 1–13.
98. A. Galano, M. Francisco Marquez and A. Pérez-González, Chem. Res. Toxicol., 2014, 27, 904–918.
99. M. E. Medina, A. Galano and J. R. Alvarez-Idaboy, Phys. Chem. Chem. Phys., 2014, 16, 1197–1207.
100. M. E. Medina, C. Iuga and J. R. Álvarez-Idaboy, RSC Adv., 2014, 4, 52920–52932.
101. C. Caicedo, C. Iuga, R. Castañeda-Arriaga and J. R. Alvarez-Idaboy, RSC Adv., 2014, 4, 38918–38930.
102. G. R. Buettner, Arch. Biochem. Biophys., 1993, 300, 535–543.
103. M. E. Letelier, S. Sánchez-Jofré, L. Peredo-Silva, J. Cortés-Troncoso and P. Aracena-Parks, Chem.–Biol. Interact., 2010, 188, 220–227.
104. R. R. Crichton and R. J. Ward, Met. Ions Biol. Syst., 2004, 41, 185–219.
105. G. J. Brewer, Exp. Biol. Med., 2007, 232, 323–335.
106. L. M. Gaetke and C. K. Chow, Toxicology, 2003, 189, 147–163.
107. O. I. Aruoma, B. Halliwell, E. Gajewski and M. Dizdaroglu, Biochem. J., 1991, 273, 601–604.
108. I. Janik and G. N. R. Tripathi, J. Chem. Phys., 2013, 139, 014302.
109. G. Berthon, Agents Actions, 1993, 39, 210–217.
110. C. A. Lipinski, F. Lombardo, B. W. Dominy and P. J. Feeney, Adv. Drug Delivery Rev., 2001, 46, 3–26.
111. A. K. Ghose, V. N. Viswanadhan and J. J. Wendoloski, J. Comb. Chem., 1999, 1, 55–68.
112. R. Mannhold and H. Van De Waterbeemd, J. Comput.–Aided Mol. Des., 2001, 15, 337–354.
113. P. D. Leeson and A. M. Davis, J. Med. Chem., 2004, 47, 6338–6348.
114. F. López-Vallejo, M. A. Giulianotti, R. A. Houghten and J. L. Medina-Franco, Drug Discovery Today, 2012, 17, 718–726.
115. A. K. Ghose, A. Pritchett and G. M. Crippen, J. Comput. Chem., 1988, 9, 80–90.
116. I. V. Tetko and G. I. Poda, J. Med. Chem., 2004, 47, 5601–5604.
117. A. K. Ghose and G. M. Crippen, J. Chem. Inf. Comput. Sci., 1987, 27, 21–35.

---

Footnote

† Electronic supplementary information (ESI) available: Gibbs energies of reaction, Gibbs energies of activation, imaginary frequencies, tunneling corrections, and rate constants for each individual HT reaction path. Reorganization energies and Gibbs energies of activation for the SPLET reactions. Optimized geometries of the transition states, and of the thermically viable copper chelates. See DOI: 10.1039/c6ra00549g

---