**Modulation and Protection E**ff**ects of Antioxidant Compounds against Oxidant Induced Developmental Toxicity in Zebrafish**

#### **Nuria Boix 1,2,\*, Elisabet Teixido 1,2, Ester Pique 1,2, Juan Maria Llobet 1,2 and Jesus Gomez-Catalan 1,2**


Received: 22 July 2020; Accepted: 4 August 2020; Published: 8 August 2020

**Abstract:** The antioxidant effect of compounds is regularly evaluated by in vitro assays that do not have the capability to predict in vivo protective activity or to determine their underlying mechanisms of action. The aim of this study was to develop an experimental system to evaluate the in vivo protective effects of different antioxidant compounds, based on the zebrafish embryo test. Zebrafish embryos were exposed to tert-butyl hydroperoxide (tBOOH), tetrachlorohydroquinone (TCHQ) and lipopolysaccharides from *Escherichia coli* (LPS), chemicals that are known inducers of oxidative stress in zebrafish. The developmental toxic effects (lethality or dysmorphogenesis) induced by these chemicals were modulated with n-acetyl l-cysteine and *N*ω-nitro l-arginine methyl ester hydrochloride, dimethyl maleate and dl-buthionine sulfoximine in order to validate the oxidant mechanism of oxidative stress inducers. The oxidant effects of tBOOH, TCHQ, and LPS were confirmed by the determination of significant differences in the comparison between the concentration–response curves of the oxidative stress inducers and of the modulators of antioxidant status. This concept was also applied to the study of the effects of well-known antioxidants, such as vitamin E, quercetin, and lipoic acid. Our results confirm the zebrafish model as an in vivo useful tool to test the protective effects of antioxidant compounds.

**Keywords:** oxidative stress; zebrafish embryo; in vivo model; antioxidant effect

#### **1. Introduction**

Reactive oxygen species (ROS) and reactive nitrogen species (RNS) are products of cellular metabolism, which play a dual role in beneficial and deleterious effects over different organs [1]. Aerobic organisms have antioxidant defenses to protect cells from oxidative damage. These defenses can be enzymatic (antioxidant enzymes) or non-enzymatic (antioxidant compounds) [2]. The imbalance between reactive metabolite production and antioxidant defenses in the organism is denominated oxidative stress (OS) and can produce potential detrimental effects in the organisms [3]. The consequences of OS can be very variable depending on the reactive species implicated, the subcellular structure where they are generated, the organs or tissues implicated in the effect, the genetic characteristics of the organism or developmental stage, among other factors. It is a phenomenon which has been related to different processes (aging, cancer, diabetes, cardiovascular and neurodegenerative diseases, etc.) as it can damage and inhibit the normal function of lipids, proteins, and DNA [4].

Antioxidants are chemicals that can inhibit or prevent oxidation processes. Such compounds can be produced within the human body or absorbed from dietary intake [5]. The antioxidant capacity of compounds is usually evaluated by in vitro techniques as the oxygen radical absorbance capacity (ORAC) or the total radical-trapping antioxidant parameter (TRAP), which are useful for the high-throughput screening of the antioxidative or radical-scavenging capacities of the compounds [6]. There are also cell-culture approaches, such as the cell-based antioxidant assay (CAA), which uses Caco-2 cells that allow for the study of intracellular influence of antioxidant chemicals [7]. These in vitro assays and biological techniques, which are regularly used to evaluate antioxidant capacity, do not have predictive capability for the protective activity that natural compounds have in vivo, or to determine their underlying mechanisms of action [8]. In vivo assays of antioxidant capacity of natural compounds have been performed in mice [9], in rats [10], and using other animal models, such as *Caenorhabditis elegans* [11] and adult zebrafish [12]. However, until now none of these in vivo models have been established and validated to systematically evaluate the protective effects of natural compounds.

Zebrafish (*Danio rerio*, ZF) is a tropical fish of the *Cyprinidae* family. The ZF embryo is considered a potential tool for investigating environmental exposures with direct relation to human health [13]. The ZF embryo presents multiple advantages, from which it can be highlighted that it is an in vivo model which studies the whole organism, with the main characteristics of an in vitro model: easy maintenance, large number of offspring, rapid embryonic development, possibility to combine with other biochemical, cellular and molecular techniques, screening of compounds, application to high-throughput methods, etc. [14–16]. The ZF embryo has been used as a model to study alterations and diseases related to OS mechanisms: inflammation [17], senescence [18], teratogenicity [19], neurodegenerative [20], and cardiovascular diseases [21]. Furthermore, ZF presents antioxidant genes and enzymes to protect them against OS effects. These defenses are analogous to mammalian antioxidant systems [22,23]. The protective effects of some antioxidants against exposure to OS inducers in ZF embryos have been studied with the objective to investigate the antioxidant mechanisms of action and demonstrate the usefulness of these antioxidants against oxidative damage [19,24,25].

The aim of the present work was to design an experimental system based on the ZF embryo test, which could be the basis for the study of in vivo protective effects of chemicals with antioxidant activity against oxidant-induced developmental toxicity in ZF embryos.

#### **2. Materials and Methods**

#### *2.1. Chemicals and Solution Preparation*

Tetrachlorohydroquinone (TCHQ, CAS number 87-87-6), lipopolysaccharides from *Escherichia coli* 0111:B4 (LPS), *N*ω-nitro l-arginine methyl ester hydrochloride (L-NAME, CAS number 51298-62-5), DL-buthionine sulfoximine (BSO, CAS number 5072-26-4), (±)-α-tocopherol (vitamin E, CAS number 10191-41-0), (±)-α-lipoic acid (lipoic acid, CAS number 1077-28-7), and quercetin hydrate (quercetin, CAS number 337951) were obtained from Sigma-Aldrich, Madrid, Spain. Tert-butyl hydroperoxide (tBOOH, CAS number: 75-91-2) was acquired from TCI Europe and n-acetyl-l-cysteine (NAC, CAS number 616-91-1) and diethyl maleate (DEM, CAS number 141-05-9) were obtained from Cymit Química, Barcelona, Spain.

tBOOH, LPS, NAC, and DEM were directly dissolved in 0.3X Danieau's buffer (17.4 mM NaCl; 0.23 mM KCl; 0.12 mM MgSO4·7 H2O; 0.18 mM Ca(NO3)2; 1.5 mM HEPES (N-(2-hydroxyethyl) piperazine-N- -(2-ethanesulfonic acid); pH 7.4). TCHQ, vit. E, quercetin, and lipoic acid were dissolved in 100% dimethyl sulfoxide (DMSO, Sigma-Aldrich, Madrid, Spain) and subsequently diluted in 0.3× Danieau's buffer to a final DMSO concentration of 0.05 % (*v*/*v*).

Our previous experience with 0.05 % DMSO in 0.3× Danieau's buffer clearly indicates that it does not produce any effects in lethality or dysmorphogenesis in ZF embryos, and it was not expected to modify the toxicity of the compounds. Moreover, DMSO is only expected to modify the permeability of chemicals if used at higher concentrations > 0.1% [26].

Concentrations of all chemicals are expressed in molarity, except for LPS that is given as μg/mL, due to the variable molecular mass of LPS, as it is part of the outer membrane of bacteria—in this case, *Escherichia coli*.

#### *2.2. Animals and Embryo Production*

Adult wild type ZF (BCN Piscicultura Iberica; Terrassa, Spain) were kept in aquariums with a closed flow-through system at 26 ± 1 ◦C and 10–14 h constant dark–light cycle. Females and males were housed separately and fed with commercial flakes and brine shrimp (Ocean Nutrition, San Diego, USA). The day before the experiments, females and males were transferred to a breeding tank (10 females; 8 males). ZF embryos were collected within 1 h after the onset of lights in the morning. They were extensively cleaned, and fertilized eggs were staged according to [27] and selected for subsequent exposure under a dissection stereomicroscope (Motic SMZ168, Motic China Group, LTD., Luwan, Shanghai, China). The study was approved by the Ethic Committee for Animal Experimentation of the University of Barcelona and by the Department of Environment and Housing of the Generalitat de Catalunya with license number DAAM 7971.

#### *2.3. Exposure of Zebrafish Embryos to Oxidative Stress Related Compounds*

To characterize the effects on embryonic development produced by compounds related to OS, ZF embryos were exposed to OS inducers, modulators of antioxidant status and antioxidants. For compounds which were dissolved in DMSO and diluted with Danieau's buffer, a vehicle negative control group with 0.05% DMSO in 0.3× Danieau's buffer was assayed.

Exposures to antioxidants and to modulators were performed from 2 to 26 h post-fertilization (hpf) in order to select, for each of the compounds of study, the highest concentration at which any effect in lethality or in embryonic development was observed (maximum tolerable concentration, MTC). From 26 to 50 hpf, embryos were incubated with 0.3× Danieau's buffer with or without DMSO, depending on the dissolution of the compound of study.

Exposure to OS inducers was conducted from 26 to 50 hpf to select the working concentrations of these compounds for the experimental design. In this case, from 2 to 26 hpf, embryos were incubated with Danieau's buffer with or without DMSO, depending on the dissolution of the compound of study.

Exposure of ZF embryos was semi-static and was carried out in 6-well plates (Greiner Bio-one, Frickenhausen, Germany). Ten embryos per group were selected and randomly distributed into the wells and filled with 5mL of the corresponding solution of the compound. Embryos were incubated at 26 ± 1 ◦C with a dark–light cycle of 10–14 h. Renewal of the medium and of the solutions was made every 24 h. Evaluation of the embryos was performed at different time points. Lethality was determined at 8, 26, and 50 hpf based on egg coagulation, the absence of tail detachment, or somite formation and the absence of heartbeat [28]. Dysmorphogenic effects were evaluated at 50 hpf by the total morphological score system described by [29]. For each compound of study, at least three independent experiments were performed using embryos from different spawning events (*n* = 3).

The percentage of lethality and of dysmorphogenesis was calculated per compound at every tested concentration, and the concentration–response curves for these effects were plotted. From these curves the concentration, which produced mortality to 50% of the embryos (lethal concentration 50, LC50), and the concentration at which 50% of the embryos presented at least one dysmorphogenic feature (effective concentration 50 for dysmorphogenesis, EC50), were calculated.

#### *2.4. Pre-Exposure of the Embryos to Modulators of Antioxidant Status* + *Exposure to OS Inducers*

To elucidate the OS role in the developmental effects produced by OS inducers, another assay was performed by modulating the ZF embryos' OS responses through pre-exposure to the MTC of compounds which can affect OS conditions, and the posterior exposure to the working concentrations of OS inducers. NAC and L-NAME were used to potentiate antioxidant status, as NAC increases

glutathione levels [30] and L-NAME inhibits nitric oxide production [31]. On the other hand, DEM and BSO were used to inhibit glutathione synthesis [32] by increasing the sensitivity of the embryos to OS.

At 2 hpf, embryos were pre-exposed to the MTC of modulators of antioxidant status for 24 h, and then washed with 0.3× Danieau's buffer. At 26 hpf, embryos were exposed to OS inducers at the selected working concentrations. Lethality and dysmorphogenesis were evaluated as previously described, and concentration–response curves were plotted. A comparison of the concentration–response curves of pre-exposure to modulators of antioxidant status + exposure to OS inducers with concentration–response curves of OS inducers exposure was performed.

#### *2.5. Pre-Exposure of the Embryos to Antioxidant Compounds* + *Exposure to OS Inducer*

To detect the protective effects of chemicals against oxidant induced developmental toxicity in ZF embryos, different compounds with well determined antioxidant activity were assayed.

A pre-exposure to the MTC of vitamin (vit.) E, lipoic acid and quercetin was performed from 2 to 26 hpf, followed by a washing step with 0.3× Danieau's solution and the exposure to the working concentrations of the selected OS inducer for 24 h. Evaluation of the embryos was performed as described before, and concentration–response curves were graphically represented. A comparison between the concentration–response curves of pre-exposure to antioxidants + exposure to the selected OS inducer with the concentration–response curve of the exposure to the selected OS inducer was performed.

#### *2.6. Data Evaluation*

Comparison of categorical variables was performed with the Fisher's exact test. Concentration– response curves for lethality and dysmorphogenesis were fitted to all the data using the Hill model in GraphPad Prism 6 software and compared with the extra sum-of-squares F test, which compares the parameters fit to datasets (GraphPad Software, La Jolla, CA, USA). Confidence intervals were set at 95% and a probability of *p* < 0.05 was considered as statistically significant.

#### **3. Results**

#### *3.1. Characterization of the E*ff*ects of Oxidative Stress Related Compounds in Zebrafish Embryos*

The results of the characterization of the lethal and dysmorphogenic effects produced by ZF embryos exposure to OS inducers, modulators and antioxidants are shown in Table 1.


**Table 1.** Characterization of lethality and dysmorphogenesis in zebrafish embryos, produced by oxidative stress-related compounds.

Range of tested concentrations, maximum tolerable concentration (MTC), lethal concentration 50 (LC50), effective concentration 50 for dysmorphogenesis (EC50) and exposure window for each of the studied compounds. n.d.: data were not determined. a: MTC was not determined because the compound produced lethal or dysmorphogenic effects at all the studied concentrations. b: LC50 was not calculated because no lethal effects were observed until the highest concentration, where lethality was of 100%. c: LC50 or EC50 was not calculated because no significant effects in the mortality of the embryos were observed. d: LC50 and EC50 were not calculated because the compounds did not produce lethal or dysmorphogenic effects at any of the tested concentrations. e: Quercetin solution precipitated from 30 μM. It was not possible to evaluate the effects at higher concentrations.

OS inducers produced developmental effects in zebrafish embryos (lethality and dysmorphogenic effects), which were concentration-dependent. Modulators of antioxidant status and antioxidants did not produce lethality at the studied concentrations, and the dysmorphogenic effects observed in the embryos exposed to the tested compounds were mainly developmental delay, cardiac oedema, and brain necrosis, which were not specific alterations. The only compound-specific effect was observed in TCHQ exposure, which produced an effect in the pigmentation of the embryos.

#### *3.2. Pre-Exposure to Modulators of Antioxidant Status* + *Exposure to OS Inducers*

We attempted to modulate the embryotoxic and lethal effects produced by OS inducers in zebrafish embryos by pre-exposing them to a set of known modulators of antioxidant status in zebrafish (Table 2), in order to evaluate if the effects produced by OS inducers were caused by an OS mechanism.


**Table 2.** Lethality and dysmorphogenesis effective concentration values in zebrafish embryos on the modulation of developmental effects produced by OS inducers.

Lethal concentration 50 (LC50), effective concentration 50 for dysmorphogenesis (EC50) and 95% confidence interval. Statistically significant differences with respect to the group, which was not exposed to any modulator: \*: *p* < 0.05; \*\*: *p* < 0.01; \*\*\*: *p* < 0.001; n.d.: no lethality was observed; <sup>1</sup> A unique tBOOH and LPS concentration–response curve was generated with the dissolution of the compounds in Danieau's buffer without DMSO and compared to all the concentration–response curves of the groups pre-exposed to chemicals (initially dissolved or not in DMSO) due to the lack of effect of DMSO in the embryonic development of ZF.

In embryos which were exposed to tBOOH, a pre-exposure to NAC and L-NAME significantly drifted the tBOOH concentration–response curves to higher concentrations (Figure 1), the fact that, at the NAC and L-NAME pre-exposure group, no significant effects in mortality of the embryos were observed being of special importance. On the contrary, when ZF embryos where pre-exposed to DEM and BSO, a significant shift in the concentration–response curves to lower concentrations of tBOOH was observed (Figure 1). As described before, the tBOOH concentration–response curve was generated after a pre-incubation of the embryos for 24 h in 0.3× Danieau's buffer without DMSO, due to the lack of effects of DMSO in ZF development.

**Figure 1.** Concetration–response curves for lethality and dysmorphogenesis of tert-butyl hydroperoxide (tBOOH) alone or in combination with modulators of antioxidant status.

The modulation of antioxidant status in embryos exposed to TCHQ presented similar results to tBOOH. When ZF embryos were pre-exposed to NAC and L-NAME, the concentration–response curves for lethality and dysmorphogenesis were significantly shifted to higher concentrations of TCHQ (Figure 2). On the other hand, assays conducted with pre-exposure to DEM and BSO produced a statistically significant drift in the concentration–effect curves for lethality and dysmorphogenesis to lower concentrations of TCHQ (Figure 2).

**Figure 2.** Concentration–response curves for lethality and dysmorphogenesis of tetrachlorohydroquinone (TCHQ) alone or in combination with modulators of antioxidant status.

Pre-exposure of ZF embryos to NAC, DEM, L-NAME, and BSO followed by LPS exposure at the selected working concentrations shifted the lethality concentration–effect curves significantly. In the analysis of dysmorphogenic effects in ZF embryos, no significant effects were observed in embryos pre-exposed to NAC, DEM, and BSO, and subsequently exposed to LPS. Only a significant reduction in dysmorphogenic effects was observed in L-NAME pre-exposed embryos (Figure 3). As described in the previous section, embryos were pre-incubated with 0.3× Danieau's buffer without DMSO, followed by LPS exposure and calculation of the concentration–response curve, due to the lack of effects of DMSO in ZF development.

**Figure 3.** Concentration–response curves for lethality and dysmorphogenesis of lipopolysaccharides of *Escherichia coli* 0111:B4 (LPS) alone or in combination with modulators of antioxidant status.

The modulation of antioxidant status in embryos exposed to tBOOH produced more consistent results than other OS inducers. The observed effects in the embryonic development were general alterations not compound-specific. tBOOH was selected as the general OS inducer for the study of protective effects of antioxidant compounds.

#### *3.3. Detection of Protective E*ff*ects of Antioxidant Compounds in Zebrafish Embryos*

The second part of the study consisted in the use of tBOOH as a general OS inducer for the detection of compounds with very well-known antioxidant capacity. ZF embryos were exposed from 2 to 26 hpf to antioxidant compounds (vit. E, lipoic acid, and quercetin), before exposing them to tBOOH from 26 to 50 hpf.

In all cases, pre-exposure to the studied compounds produced a significant drift in the concentration–response curves of lethality and dysmorphogenesis to higher concentrations of tBOOH (Figure 4), which may indicate an antioxidant effect.

**Figure 4.** Concentration–response curves for lethality and dysmorphogenesis of tBOOH alone or in combination with different antioxidant compounds.

The LC50, after tBOOH exposure, was 2.38mM, and values obtained after vit. E, lipoic acid, and quercetin exposure were 2.83 mM, 3.72 mM, and 3.26 mM, respectively. For the EC50 values, the situation was similar, from tBOOH exposure, the EC50 was 1.64 mM, while pre-exposure to the studied compounds returned an EC50 of 2.42 mM for vit. E, 3.70 mM for lipoic acid and 3.05 mM for quercetin (Table 3).


**Table 3.** Effects of antioxidant compounds in lethality and dysmorphogenesis of zebrafish embryos exposed to tBOOH.

Lethal concentration 50 (LC50), effective concentration 50 for dysmorphogenesis (EC50) and 95% confidence interval. Statistically significant differences with respect to the group which was not exposed to any antioxidant compound: \*\*\*: *p* < 0.001; <sup>1</sup> A unique tBOOH concentration–response curve was generated and compared to all the concentration–response curves of the antioxidants pre-exposure groups (initially dissolved or not in DMSO) due to the lack of effect of DMSO in the embryonic development of ZF.

#### **4. Discussion**

Oxygen is an essential element for cell life and, from its metabolism, some toxic derivatives are produced, such as ROS, which are highly reactive to biological molecules and can produce OS [33]. An important factor that could prevent OS effects is the alimentary antioxidants intake. For this reason, the study of antioxidant capacity of compounds has been gaining interest in the past few years. It has been postulated that, in order to evaluate the antioxidant potential, a method which includes in vivo techniques would have more impact on the results because OS implies mechanisms which depend on many system conditions, especially the kinetic part of the reactions [34]. We have proposed the ZF embryo test, which could be a valuable in vivo method to test the antioxidant capacity of compounds, with the main advantages of an in vitro technique.

In the first part of this study, we characterized the embryotoxic and dysmorphogenic effects of several compounds, which have an OS-related mechanism of action on the ZF embryos: tBOOH, TCHQ, and LPS. The induction of OS by tBOOH is due to its capacity to generate butoxyl radicals which deplete antioxidant systems and lead to cell death [35], and it has been previously used in ZF embryos to induce OS [36]. TCHQ can induce OS by producing superoxide radicals, favoring the depletion of the reduced glutathione concentrations [37], and it has also been observed that TCHQ can produce DNA strand breakage in cells [38]. LPS is a microbial product of bacteria and its contribution to ROS production has been studied as a secondary effect to inflammation [39]. It has been used as an OS inducer in different in vitro and in vivo models [40]. All the studied OS inducers produced a significant increase in lethality and in the production of dysmorphogenesis in the exposed ZF embryos.

In order to check if the observed effects in ZF embryos could be produced by an OS mechanism, we performed assays of modulation of the embryos' antioxidants statuses with compounds related to OS. The modulation was carried out through raising or decreasing the antioxidant defenses of the embryos with NAC and L-NAME, and DEM and BSO, respectively. NAC is an antioxidant compound, which is a rate-limiting substrate in glutathione synthesis, and it can also act as a scavenger of free radicals [41]. L-NAME is an inhibitor of nitric oxide synthase, the enzyme responsible for nitric oxide synthesis. As a result of this inhibition, it reduces the production of endogenous nitric oxide, which is a compound that can produce reactive nitrogen species and consequently, OS [30]. DEM is an alkylating agent that can produce a conjugation and depletion of glutathione [42], and it can also activate the nuclear factor (erythroid-derived 2)-like 2 (Nrf2) pathway [22], and BSO is an antioxidant molecule suppressor, which specifically inhibits γ-glutamyl cysteine synthetase, the enzyme for glutathione biosynthesis, and causes the depletion of glutathione levels [43].

In general terms, we have demonstrated that the lethal and dysmorphogenic effects of tBOOH and TCHQ were significantly reduced when the embryos were pre-treated with antioxidant compounds (NAC and L-NAME). From the opposite position, the observed effects in mortality and in dysmorphogenesis were significantly increased when ZF embryos were pre-exposed to compounds which decrease the antioxidant status (DEM and BSO). We could conclude that tBOOH and TCHQ produced their embryolethal and dysmorphogenic effects in ZF embryos by an OS mechanism of action. No significant effects in dysmorphogenesis related to OS were observed in the LPS treatment group. The observed effects in the lethality of ZF embryos exposed to LPS could be more related to its mechanism as an inflammation inducer [44] than as an OS mechanism. Nevertheless, the effects of these compounds, associated with an OS mechanism, should be verified by analyzing parameters directly related to OS, like evaluation of the expression of OS-related genes in the exposed ZF embryos.

Because of its consistent results, tBOOH was selected as the OS inducer to be used to evaluate the antioxidant potential of compounds. To validate the use of tBOOH to detect the protective effects of antioxidant compounds, ZF embryos were pre-exposed to diverse compounds with a well-established antioxidant capacity (vit. E, lipoic acid, and quercetin) and posteriorly exposed to tBOOH. In addition, a statistical analysis was performed by comparing the concentration–effect curves for lethality and for dysmorphogenesis obtained in both experiments: tBOOH alone and antioxidants + tBOOH.

Vit. E is a compound with free-radical scavenging activity, which leads to an antioxidant action that has been demonstrated in vitro [45]. Lipoic acid is a thiol regenerating compound, which increases the level of glutathione. It inhibits the formation of hydroxyl radicals, and it also scavenges ROS [46]. Quercetin is a flavonol found in apples, tea, and onions, and exerts its antioxidant effect through different bioactive effects. Its main antioxidant mechanism of action is through quenching different radicals, such as hydroxyl, peroxyl, and superoxide, as well as nitric oxide and lipid oxidation [5]. Quercetin can induce antioxidant gene expression through the activation of Nrf2 [47]. Among these, quercetin can also modulate mitochondrial biogenesis by reducing ROS production in various cell types [48]. The pre-exposure of the embryos to vit. E, quercetin, and lipoic acid, followed by the exposure to the OS inducer, has confirmed the protective effects of well-known antioxidant compounds against oxidant-induced developmental toxicity in ZF. In all the cases, the pre-exposure of ZF embryos to the compounds followed by the exposure to the selected concentrations of tBOOH produced a significant shift of the concentration–effect curves of lethality and dysmorphogenesis. These results indicated the preventive effect of vit. E, lipoic acid, and quercetin against the toxic effects of tBOOH, which were related to an OS mechanism of action. The antioxidant effect of these compounds versus oxidant effects produced by the OS inducer should be confirmed by the application of antioxidant capacity evaluation methods.

The ZF embryo test has been widely used to study different types of compounds, including OS-related chemicals. Recently, a new stable transgenic line has been developed for the rapid detection of oxidative stress, although it has not been systematically tested to evaluate the antioxidant capacity of chemicals [49]. The results of our study are similar to those observed by the authors in [25], in which they observed the protective effect of vit. E in ZF embryos exposed to PCB126, which causes OS. There are other studies in which they evaluated the effects of compounds, which may have part of its mechanism of action related to oxidative injury, such as ethanol, in ZF embryos [19]. In this case, they analyzed and confirmed the partial prevention of ethanol-induced cardiovascular disfunction by lipoic acid in ZF embryos. Natural antioxidant compounds, such as quercetin, have demonstrated their antioxidant capacity and their protective activity against different diseases using the ZF embryo test [12], reinforcing the results obtained in our study.

#### **5. Conclusions**

The ZF embryo has been established as the basis for the study of the modulative and protective effects of antioxidant compounds in oxidant induced developmental toxicity in ZF. An experimental design using tBOOH as an OS inducer has been developed in the present study. The evaluation of the OS-related effects produced by tBOOH was estimated by a modulation of the antioxidant status assay with NAC, L-NAME, DEM, and BSO. The study of the protective effects of antioxidant compounds was performed with pre-exposure of ZF embryos to vit. E, lipoic acid, and quercetin, which are compounds with a well-established antioxidant capacity, and the protective effect of these compounds on developmental effects in the embryos was confirmed.

Our experimental system could be used as a valuable in vivo tool for testing compounds with presumable antioxidant activity, with advantages in respect to other techniques used in the evaluation of the antioxidant capacity (analytical or cell-based assays).

Further studies should be done to extensively characterize the effects of tBOOH as an OS inducer, as well as to evaluate the antioxidant capacity of compounds, in order to establish an OS model based on ZF embryos to study new antioxidant compounds and the mechanism of action by which they exert their antioxidant activity.

**Author Contributions:** Conceptualization, N.B., E.T., E.P., J.M.L., and J.G.-C.; Formal analysis, N.B., E.T., E.P., and J.G.-C.; Funding acquisition, J.M.L.; Investigation, N.B. and J.G.-C.; Methodology, N.B., E.T., and E.P.; Project administration, J.M.L. and J.G.-C.; Resources, J.M.L. and J.G.-C.; Supervision, E.P., J.M.L., and J.G.-C.; Validation, N.B., E.T., E.P., J.M.L., and J.G.-C.; Visualization, N.B., E.T., E.P., J.M.L., and J.G.-C.; Writing—original draft, N.B.; Writing—review and editing, N.B., E.T., E.P., J.M.L., and J.G.-C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Antioxidative Action of Ellagic Acid—A Kinetic DFT Study**

**Jelena Tošovi´c 1,2 and Urban Bren 1,3,\***


Received: 8 June 2020; Accepted: 3 July 2020; Published: 6 July 2020

**Abstract:** Although one can find numerous studies devoted to the investigation of antioxidative activity of ellagic acid (EA) in the scientific literature, the mechanisms of its action have not yet been fully clarified. Therefore, further kinetic studies are needed to understand its antioxidative capacity completely. This work aims to reveal the underlying molecular mechanisms responsible for the antioxidative action of EA. For this purpose, its reactions with HO• and CCl3OO• radicals were simulated at physiological conditions using the quantum mechanics-based test for overall free-radical scavenging activity. The density functional theory in combination with the conductor-like polarizable continuum solvation model was utilized. With HO• radical EA conforms to the hydrogen atom transfer and radical adduct formation mechanisms, whereas sequential proton loss electron transfer mechanism is responsible for scavenging of CCl3OO• radical. In addition, compared to trolox, EA was found more reactive toward HO•, but less reactive toward CCl3OO•. The calculated rate constants for the reactions of EA with both free radicals are in a very good agreement with the corresponding experimental values.

**Keywords:** QM-ORSA; antioxidative mechanisms; reaction rate constants; physiological conditions; polyphenols

#### **1. Introduction**

In recent years, therapeutic applications of non-drug substances such as functional foods, are progressively increasing. Therefore, studies on functional foods represent a cutting-edge topic among nutritional scientists. The significance of bioactive compounds as functional supplements of foods has been well established due to their effectiveness in health promotion by disease prophylaxis or treatment. Special attention has been devoted to investigations of polyphenolic compounds, known for their various nutritional, biologic, and pharmacological effects. In order to provide the health benefits to the consumers, functional foods and nutraceuticals have been supplemented with these compounds in recent years [1].

Among polyphenolic compounds, ellagic acid (EA) attracts an ever-increasing interest due to its great potential in food technology, as well as in pharmaceutical, medical and cosmetic industries [2]. EA, a dimeric derivative of gallic acid, arises from acidic hydrolysis of ellagitannins. It represents a planar molecule which contains four hydroxyl and two lactone groups (see Figure 1). This dietary polyphenol can be found in a wide variety of fruits. Raspberries, cranberries, strawberries, grapes, as well as pomegranate seeds, are known for example for their high content of EA [3]. Other sources include pecans, walnuts and distilled beverages [4].

**Figure 1.** Optimized structures of neutral ellagic acid and its monoanion. Carbon atoms are depicted in gray, oxygen atoms in red, chlorine atoms in green and hydrogen atoms in white color. The atom labeling scheme and color coding are applied throughout the study.

Like other dietary polyphenols, EA possesses a wide range of biological activities suggesting that it can exert strong beneficial effects on human health. In many epidemiological and experimental studies, anticarcinogenic, anti-inflammatory, antiviral, antibacterial, anti-atherosclerosis, antihypertensive, antihyperglycemic, cardioprotective and anti-fibrosis actions of EA have been demonstrated [5–11]. It can inhibit carcinogenesis by occupying sites (i.e., microsomal *P*-450 enzymes, glutathione-S -transferase or DNA), that would normally interact with ultimate carcinogens, through several mechanisms [12,13]. The anticarcinogenic effect of ellagic acid has been studied in various cancer cells. There it exhibits antiproliferative activity, combined with the ability to cause cell cycle arrest and to induce apoptosis [14]. The anticarcinogenic effects of ellagic acid have been observed in several cancer types: prostate, skin, esophageal and colon cancers [15]. Moreover, EA causes cell-specific responses, meaning that tumor cells are more susceptible to EA than normal cells [10]. In addition, EA prevents metabolic activation of aflatoxin B1, polycyclic aromatic hydrocarbons (PAHs) and nitroso compounds into ultimate carcinogens that cause DNA damage [12]. Due to its beneficial effects against a wide range of diseases, EA represents a great candidate for a therapeutic and chemopreventive agent, especially in the form of functional food supplements [16].

It has been shown that the high free radical scavenging activity of EA may be at least partially responsible for the observed in vivo biologic effects [14]. The presence of four hydroxyl groups enables EA to scavenge numerous reactive oxygen and nitrogen species and makes this compound a powerful antioxidant [17,18]. EA represents also a very efficient inhibitor of lipid peroxidation even at micromolar concentrations [19]. Furthermore, studies of Hassoun et al. showed that EA exhibits a better antioxidative efficacy against oxidative stress and lipid peroxidation than vitamin E [19]. Finally, besides numerous beneficial effects on human health, EA as a strong antioxidant can prolong shelf life and preserve the quality of foods [1].

In scientific literature, one can find a few theoretical studies devoted to the examination of the antioxidative activity of EA through thermodynamic and kinetic approaches. Markovi´c et al. have shown that the thermochemical viability of different antioxidative mechanisms depends on the deprotonated portion of EA, the polarity of reaction media, as well as on the properties of the free radical [20]. Based on the calculated thermodynamic parameters, they have suggested that the hydrogen atom transfer (HAT) is the most favorable mechanism in nonpolar media, whereas sequential proton loss electron transfer (SPLET) is preferred in polar media, which is in agreement with results reported by Mazzone et al. [21]. Utilizing the transition state theory, Tiwari and Mishra have determined the rate constants for the reactions of EA (as well as its monomethyl and dimethyl derivatives) with hydroxyl (HO•), methoxy (CH3O•) and nitrogen dioxide (NO2 •) radicals [22]. However, the calculated rate constants of HO• and CH3O• radicals have been overestimated by several orders of magnitude in comparison with the experimentally obtained values [17,18]. Galano et al. have also investigated several aspects related to the antioxidant activity of EA [17]. They have demonstrated that the

free radical scavenging activity of EA does not decrease upon metabolism and provides continuous protection against oxidative stress. To the best of our knowledge, the results of Tiwari and Mishra and Galano et al. represent the only theoretical studies dedicated to kinetic investigations of antioxidative activity of EA [17,22].

However, one can find numerous experimental studies devoted to the examination of EA as an important component of various foods and beverages, its antioxidative mechanisms have not been fully clarified. Elucidation of the mechanisms by which dietary polyphenols prevent and suppress various diseases represents an important step in understanding their effects in vivo and may help the design of novel strategies for disease prophylaxis and treatment. Therefore, further kinetic investigations are needed to reveal and to fully understand the underlying molecular mechanisms responsible for the antioxidative action of EA. Consequently, the hydrogen atom transfer (HAT), radical adduct formation (RAF), sequential proton loss electron transfer (SPLET) and single electron transfer (SET) mechanisms [23–25] were studied by simulating the reactions of EA with two free radicals, HO• and CCl3OO•, at physiological conditions (pH = 7.4 in aqueous solution) using quantum-chemical methods. An additional goal was to determine the relative antioxidative activity of EA, using trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid, Tx) as a reference compound.

#### **2. Materials and Methods**

#### *2.1. Computational Methods*

All results were obtained from calculations using the density functional theory (DFT) approach. Full geometry optimizations and subsequent frequency calculations were performed using the hybrid meta M06-2X functional in conjunction with flexible 6-311++G(d,p) basis set and conductor-like polarizable continuum model (CPCM) [26], as implemented in the Gaussan 09, Revision D.01, software package [27]. Implicit water solution (dielectric constant, ε = 78.3553) was employed to mimic the physiological aqueous environment. The M06-2X functional was developed for studying main-group thermochemistry and kinetics [28,29]. In addition, this theoretical model has recently demonstrated robustness and very good overall performance in investigations of several related polyphenolic systems [30–34]. Restricted and unrestricted calculations were applied for the closed-shell and open-shell structures, respectively. The nature of the reactive species was confirmed by analyzing the results of the subsequent frequency calculations in the harmonic approximation: only real frequencies for equilibrium geometries and exactly one imaginary frequency for transition states (TSs) were obtained. The intrinsic reaction coordinate (IRC) calculations were additionally performed to verify each transition state. IRC represents the minimum energy reaction pathway (MERP) in mass-weighted cartesian coordinates between the transition state and the corresponding reactants and products. Moreover, the natural bond orbital (NBO) analysis was applied for all structures to obtain the corresponding partial atomic charges [35]. The IRC and NBO analyses were performed using default settings.

#### *2.2. Quantum Mechanics-Based Test for Overall, Free-Radical Scavenging Activity*

Four antioxidative mechanisms—HAT, RAF, SPLET and SET—were examined following the quantum mechanics-based test for overall free-radical scavenging activity (QM-ORSA) protocol [36], which was designed for studying free-radical reactions in solutions of different polarities. QM-ORSA represents a universal and quantitative method of evaluating the free radical scavenging activity of chemical compounds, that is, their primary antioxidant activity. This methodology involves revealing of all thermodynamically feasible reaction pathways included in the antioxidative process, which are subjected to subsequent kinetic investigations. By calculating the reaction pathways for all present acid–base forms of the investigated compound, QM-ORSA takes into account also the influence of pH. Namely, at a particular pH, the antioxidant can be present in different acid–base forms (cationic, neutral, monoanionic, dianionic, etc.) depending on its *pKa* values. The reliability of the QM-ORSA

protocol was confirmed on the set of test reactions, where the correlation between the logarithms of the calculated and experimental rate constants was excellent (the R value is very close to one (0.99), the slope is very close to one (0.99), and the intercept is very close to zero (0.06) [36]. Moreover, the absolute error of the Gibbs activation free energies of 1.213 kJ mol−<sup>1</sup> was significantly lower than the accepted computational accuracy of 4.184 kJ mol−1. Finally, this protocol has been successfully applied for several investigations of antioxidative activity in the scientific literature [31,32,34,37–42].

#### 2.2.1. Thermodynamic Considerations

The thermochemical viability of all possible reaction pathways and reaction sites included in the antioxidative process was investigated in terms of the reaction Gibbs free energies (Δ*Gr*). The free energies of the examined reactions were determined at T = 298.15 K and P = 101,325 Pa. The exergonic (Δ*Gr* < 0) and isoergonic (Δ*Gr* ≈ 0) reaction paths were subjected to further kinetic calculations.

#### 2.2.2. Kinetic Considerations

Depending on the type of the mechanism, the reaction rate constants were obtained in two different ways. In the case of HAT and RAF mechanisms, where the transformation of reactants to products occurs over energy barriers, the Eckart method [43] also known as zero-tunneling method (ZCT-0) was applied. This method uses the Eckart function for generating the ground-state potential energy function based on information on the stationary points (reactants, transition state and products) along MERP. To perform the Eckart method calculations TheRate program [44] was utilized. For the electron transfer reactions involved in SPLET and SET mechanisms, the Marcus theory [45] was applied.

The overall reaction rate constants (*k*overall), which correspond to experimentally observed reaction rates of specific free-radical reactions, were calculated. The *k*overall values were obtained as a sum over all acid–base species (*i*) present at the physiological pH (7.4) of the total reaction rate constant values (*k*TOT) multiplied by the corresponding molar fractions (*f*):

$$k\_{\text{overall}} = \sum\_{\substack{i = \{\text{acid} - \text{base}\} \\ \text{species}}} f(i) \times k\_{\text{TOT}}(i) \tag{1}$$

The *k*TOT values for all acid–base species were obtained as sums of the reaction rate constants corresponding to each antioxidative mechanism (*j*):

$$k\_{\text{TOT}} = \sum\_{j = \{\text{antioxide}\,\text{time}\}} k\_{\text{mech}}(j) \tag{2}$$
 
$$\text{mechanism}$$

The *k*mech is defined as a sum of reaction rate constants (*k*) belonging to the same antioxidative mechanism calculated at different reactive sites (*l*):

$$k\_{\text{mech}} = \sum\_{l=\text{(antixidative})} k(l) \tag{3}$$
 
$$\text{pathway}$$

The antioxidative pathway belongs to a specific antioxidative mechanism at a specific reactive site. To determine the relative contribution of an antioxidative pathway (*l*), the branching ratio, Γ(*l)*, was calculated using the following relation:

$$
\Gamma(l) = \frac{k(l)}{k\_{\text{overall}}} \tag{4}
$$

#### 2.2.3. Relative Antioxidative Activity

The relative antioxidative activity of **EA** (*r*T) was calculated by dividing *k*overall of EA with *k*overall of **Tx**:

$$r^T = \frac{k\_{\text{overall}}^{\text{EA}}}{k\_{\text{overall}}^{\text{Tx}}} \tag{5}$$

#### **3. Results and Discussion**

#### *3.1. Thermodynamic Considerations*

In the previous study of Galano at al. it was found that the dominant species of EA present at physiological conditions (pH = 7.4) are neutral (~10.7%) and monoanionic (EA−, ~89.3%) forms, which is also in accordance with the reported *pKa* values (Figure 1) [17].

To select favorable mechanistic pathways for further kinetic investigations of the antioxidative action of EA the Gibbs free energies of the following reactions of neutral species:

$$\text{HAT} \colon \text{EA} + \text{R}^\bullet \to \text{EA}^\bullet + \text{RH} \tag{6}$$

$$\text{RAF} \colon \text{EA} + \text{R}^\bullet \to \text{ [EA}-\text{R}]^\bullet \tag{7}$$

$$\text{SPLET (I step)}: \text{EA} + \text{HO}^- \rightarrow \text{EA}^- + \text{H}\_2\text{O} \tag{8}$$

$$\text{SPLET (II step)}: \text{ EA}^- + \text{R}^\bullet \rightarrow \text{ EA}^\bullet + \text{R}^- \tag{9}$$

$$\text{NET}: \text{EA} + \text{R}^{\bullet} \to \text{EA}^{\bullet+} + \text{R}^{-} \tag{10}$$

as well as of monoanionic species:

$$\text{HAT} \colon \text{EA}^- + \text{R}^\bullet \to \text{EA}^{\bullet-} + \text{RH} \tag{11}$$

$$\text{RAF} \colon \text{EA}^- + \text{R}^\bullet \to \text{ [EA}-\text{R}]^{\bullet-} \tag{12}$$

$$\text{SPLET (I step)}: \text{ EA}^- + \text{HO}^- \rightarrow \text{ EA}^{2-} + \text{H}\_2\text{O} \tag{13}$$

$$\text{SPLET} \left(\text{II}\,\text{step}\right) : \text{EA}^{2-} + \text{R}^{\bullet} \to \text{EA}^{\bullet-} + \text{R}^{-} \tag{14}$$

$$\text{SET} : \text{EA}^- + \text{R}^\bullet \to \text{EA}^\bullet + \text{R}^- \tag{15}$$

had to be examined first. In reactions (6)–(15) R• stands for HO• or CCl3OO•. The HO• represents the most electrophilic among the oxygen-centered radicals capable of reacting immediately after its formation with almost any molecule in the vicinity. It is responsible for 60% to 70% of the tissue damage caused by ionizing radiations and most oxidative damage to DNA [23]. CCl3OO• is generated in the organism during the metabolism of CCl4, a well-known liver toxin. As most of the oxygen radicals, CCl3OO• reacts with various biomolecules such as proteins, DNA and lipids [46,47]. In addition, CCl3OO• was specifically selected because it is often used in experimental studies to imitate larger peroxyl radicals [48]. A wide variety of experimental studies have been indeed conducted in order to elucidate an effective scavenger of this radical, especially among the naturally occurring antioxidants. However, to the best of our knowledge, computational investigations regarding the reactivity of CCl3OO• remain surprisingly scarce.

Structures of EA and its monoanion employed in the present study are consistent with the structures published in previous papers [17,20]. The calculated reaction free energies are summarized in Table 1. In the case of the neutral species, only half of the positions in the molecule must be considered explicitly due to the symmetry, whereas in the case of the monoanion the symmetry is broken and all the sites must be considered explicitly.


**Table 1.** Gibbs energies Δ*Gr* (kJ mol<sup>−</sup>1) of the reactions of ellagic acid (EA) and its monoanion (EA<sup>−</sup>) with HO• and Cl3COO•; HAT, RAF, SPLET and SET denote hydrogen atom transfer, radical adduct formation, sequential proton loss electron transfer and single electron transfer mechanisms, respectively.

According to the highly exergonic Δ*Gr* values, EA and EA<sup>−</sup> can scavenge both free radicals through HAT reaction pathways. In the case of EA, all four positions (1a = 1a' and 2a = 2a') are equally feasible, whereas in the case of EA− the reaction pathway at position 2a' becomes the most favorable. In the case of the RAF mechanism, the reaction pathways at positions 7 and 7 are excluded from the examination. Namely, the significant partial positive charge of carbonyl carbons, makes these positions unsuitable for the attack of the studied electrophilic free radicals (Figure S1). All remaining reactive positions with HO• radical are exergonic or isoergonic (Δ*Gr* ≈ 0), suggesting that the RAF mechanism is thermodynamically favorable. As for CCl3OO• radical, we were unable to locate the corresponding radical adduct for the reaction with EA− at position 1 and all remaining positions are endergonic. These findings imply that RAF mechanism cannot be responsible for the antioxidative action of EA in the case of CCl3OO• radical.

The basic environment provides conditions for proton loss from EA and EA−, to form EA− and EA2−, respectively, which is reflected in the negative Δ*Gr* values for the first step of the SPLET mechanism. Considering, that EA2<sup>−</sup> represents the dominant form only at higher pH values (pH > 10), it is reasonable to assume that SPLET mechanism cannot be responsible for the antioxidative action of EA− toward the studied selected free radicals [32,49]. On the other hand, the second step of the SPLET mechanism of EA deserves a careful inspection. Namely, electron transfer reaction is endergonic in the case of the highly reactive HO•, whereas it is isoergonic in the case of CCl3OO• and should, therefore, be further examined. The higher reactivity of CCl3OO• in comparison to HO• during the electron transfer reaction can be explained by the strong negative inductive effect of the three chlorine atoms which increases the electron affinity of the radical. High Δ*Gr* values for the SET reactions between EA and both studied free radicals indicate that this mechanism does not occur, whereas the SET reaction pathway of EA− is identical to the second step of the SPLET mechanism of EA.

#### *3.2. Kinetic Considerations*

All exergonic and isoergonic reaction pathways were subjected to kinetic examination aimed at revealing the TSs and at calculating the corresponding activation free energies and reaction rate constants (Table 2).


**Table 2.** Activation energies Δ*G*‡ *<sup>a</sup>* (kJ mol<sup>−</sup>1) and rate constants *k* (M−<sup>1</sup> s<sup>−</sup>1) for exergonic reaction pathways of the reactions of EA and EA<sup>−</sup> with HO• and CCl3OO•.

All our attempts to locate TSs for the HAT reactions of EA and EA− with HO• were unsuccessful, so it was reasonable to assume that all these processes are barrierless. To confirm this assumption, each reactive position was further investigated in the following manner. HO• radical was positioned in the vicinity of the corresponding H atom and then allowed to approach the reactive center up to the formation of the products. Dependence of the total energy on the corresponding scan coordinate (HO•–H distance) was analyzed. Based on the monotonous decrease of total energy with decreasing of HO•–H distance it was concluded that these reactions are indeed barrierless and therefore diffusion-controlled, with the corresponding reaction rate constant of 1.91 <sup>×</sup> 109 M−<sup>1</sup> s−1. Two representative total energy profiles (one for the neutral form EA and one for the monoanion EA−) for HAT reactions are depicted in Figure 2.

A majority of TSs involved in the RAF mechanism of EA and EA− with HO• were successfully allocated (Cartesian coordinates of all TSs are provided in the Supplementary Materials). Two exceptions represent the reaction pathways at positions 2 and 6 of EA−, for which we were not able to locate TSs, despite numerous attempts. For this reason, the relaxed scan procedure applied for the HAT reaction pathways has also been employed in these two cases. The total energy of the system indeed monotonously decreases with decreasing HO•–C distance, so it can be concluded that these two processes are also diffusion-controlled (Figure S2). The increased affinity of C2 and C6 atoms toward HO• in comparison to other positions is not surprising due to the strong mesomeric activating effect of O− at ortho and para positions. The rate constants for the reactions at other positions are correspondingly reduced mostly by two orders of magnitude. The mutual characteristics of TSs obtained for the studied RAF reactions include relatively strong interactions of hydroxyl radical with π electrons of the aromatic ring, as well as the preserved planarity of the molecule (Figure 3a). Slower rates are observed for the reactions at positions 4 = 4 of EA and 4of EA−, whereas the

slowest reactions are those in positions 5 = 5 of EA, as well as 5 and 5 of EA−. In the first case, the π-interactions between the hydrogen of the hydroxyl radical and the aromatic ring is lost in TS (Figure 3b), whereas in the second, the reacting system becomes nonplanar and therefore less stable (Figure 3c). Additionally, the results of the IRC calculations for the representative TSs for the RAF mechanism with HO• are shown in Figure S3.

**Figure 2.** Dependence of total energy on the characteristic HO•–H distance during the hydrogen atom transfer between ellagic acid (top) or its monoanion (bottom) and HO•.

− For the HAT reactions between EA and CCl3OO• both TSs were successfully located (Figure 4). The results of the IRC calculations are shown in Figure S4. In both TSs, the planarity of the system is preserved. As expected, the HAT reaction pathways with CCl3OO• are slower than the corresponding reactions with HO• (Table 2). On the other hand, we have encountered significant difficulties to locate TSs for the reactions of EA<sup>−</sup> with CCl3OO•. Only one approximation of TS was revealed, using a similar procedure described in detail in a previous study of Tošovi´c and Markovi´c [32]. Namely, the energy profile of the reaction in 2a' position is characterized by an extremely steep decrease to the energy minimum (Figure S5). It is worth pointing out that the corresponding energy

maximum is characterized by a single desired strong imaginary vibrational frequency (1339.94i cm<sup>−</sup>1). Considering that the calculated Δ*G*‡ *<sup>a</sup>* value is extremely high (~200 kJ mol<sup>−</sup>1) and the corresponding rate constant is tremendously small, the contribution of this reaction pathway to the overall antioxidative capacity of EA toward CCl3OO• remains negligible. It is reasonable to assume that similar results would be observed in the case of the two remaining HAT reaction paths (at positions 2a and 1a'), so they were not considered further.

− − − **Figure 3.** Representative examples of transition states obtained for the RAF reaction of EA and EA− with HO• at positions: (**a**) 1- (EA−), (**b**) 4=4- (EA) and (**c**) 5 (EA−). All distances are reported in Å. − −

−

**Figure 4.** Optimized geometries of transition states for the hydrogen atom transfer (HAT) reaction pathways of ellagic acid with CCl3OO• at positions 1a=1a' and 2a=2a'. All distances are reported in Å.

Figure S6 demonstrates a barrierless formation of EA− in a proton loss reaction of the SPLET mechanism of EA, whereas the rate constant value of 1.56 <sup>×</sup> 10<sup>9</sup> M−<sup>1</sup> s−<sup>1</sup> for the second step of the SPLET mechanism, i.e., the electron transfer reaction, indicates that this reaction is also practically diffusion controlled.

The obtained *<sup>k</sup>*overall values amount to 9.70 <sup>×</sup> 109 and 3.71 <sup>×</sup> 108 M−<sup>1</sup> s−<sup>1</sup> for the reactions with HO• and CCl3OO• (Table 2), respectively and it is very interesting to compare them with the existing experimental results. In the study of Priyadarsini et al. [18], the rate constants for these reactions were determined in aqueous solution at pH=7 using pulse radiolysis technique and amount to 8.9 <sup>×</sup> <sup>10</sup><sup>9</sup> and 1.4 <sup>×</sup> 10<sup>8</sup> M−<sup>1</sup> s−<sup>1</sup> for the HO• and CCl3OO• radicals, respectively. Considering that the agreement between experimental and calculated reaction rate constants is very good, it can be concluded that the utilized computational approach successfully quantified reactivity of EA toward both studied free radicals.

To estimate the importance of each individual path to the overall antioxidative capacity of EA, the branching ratios were calculated (Table S1). The greatest Γ values in the case of HO• belong to the diffusion-controlled reactions of monoanion, i.e., all HAT reaction pathways and two specific RAF reactions. On the other hand, the highest Γ values were obtained for the SPLET reaction paths between CCl3OO• and the neutral form of EA.

Galano et al. reported the overall rate constant values for the reactions of EA with HO• and CCl3OO• radicals (among others) calculated solely based on the mechanisms in which the electron transfer reactions are involved, i.e., SPLET and SET, using a different theoretical model [17]. Our work suggests that in the case of CCl3OO• radical the electron transfer reaction is indeed the predominant antioxidative pathway and the comparison of our results with the study of Galano et al. for overall rate gives a good agreement. On the other hand, our results indicate that SPLET and SET mechanisms are not favorable for scavenging of HO• and no meaningful comparison with the work of Galano et al. can be made.

#### *3.3. Relative Antioxidative Activity*

According to the QM-ORSA protocol, a thermodynamic and kinetic study needs to also be performed for the reference compound, Tx, to determine relative antioxidative value, *r*T. The *k*overall value for the reaction of Tx with HO•, calculated using an identical methodology and theoretical model under equal conditions, has been recently reported and amounts to 1.94 <sup>×</sup> 109 M−<sup>1</sup> s−<sup>1</sup> [32]. On the other hand, to the best of our knowledge, the *k*overall value for the reaction of Tx with Cl3COO• is yet unknown. For this reason, all necessary calculations regarding this reaction had to be performed. The corresponding results and short discussion are provided in the Supplementary Materials. The obtained *<sup>k</sup>*overall value for Tx reacting with Cl3COO• is equal to 1.91 <sup>×</sup> 109 <sup>M</sup>−<sup>1</sup> <sup>s</sup><sup>−</sup>1.

Based on the calculated *k*overall values for the reactions of EA and Tx with HO• and CCl3OO• in aqueous solution the *r*<sup>T</sup> values were determined. The obtained *r*<sup>T</sup> values of 5.00 and 0.19 for the reactions with HO• and CCl3OO•, respectively, imply that EA is more reactive toward HO•, but less reactive toward CCl3OO• in comparison to Tx.

#### **4. Conclusions**

Antioxidants represent an important group of functional compounds that possess the ability to extend shelf life and maintain the quality of foods. More important, in biologic systems, antioxidants protect against oxidative stress and consequently help to prevent numerous diseases.

In this work, we investigated the antioxidative mechanisms of a dietary polyphenol EA by utilizing the QM-ORSA methodology. For this purpose, the reactions of EA with HO• and CCl3OO• radicals were simulated.

Highly exergonic Δ*Gr* values indicate that EA and EA<sup>−</sup> can scavenge both investigated free radicals through HAT reaction pathways. The RAF reaction pathways are thermodynamically possible in the case of the reactions with HO•, whereas the SPLET reaction mechanism is thermodynamically feasible in the case of CCl3OO• radical. High Δ*Gr* values for the SET reactions between EA and both studied free radicals indicate that this mechanism does not play a vital role.

Based on the obtained kinetic results, EA can scavenge HO• primarily through HAT and RAF mechanisms, whereas SPLET mechanism is responsible for scavenging of the CCl3OO• radical. Moreover, based on the calculated *r*<sup>T</sup> values, EA is more reactive toward HO•, but less reactive toward CCl3OO• than Tx.

Last but not least, the calculated overall reaction rate constants, *k*overall, for the reactions of EA with HO• and CCl3OO•, respectively, are in a very good agreement with the experimental values, indicating that the applied computational methodology successfully quantified the reactivity of EA toward both investigated free radicals. Considering that antioxidative mechanisms in aqueous environments are extremely complex, the consensus between the calculated and available experimental data strongly supports the reaction mechanisms proposed in this work.

**Supplementary Materials:** The supplementary materials are available online at http://www.mdpi.com/2076-3921/ 9/7/587/s1.

**Author Contributions:** Methodology, investigation, data analysis, visualization, writing—original draft preparation, J.T.; conceptualization, supervision, writing—review and editing, U.B. Both authors have read and agreed to the published version of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Ministry of Education, Science and Sport of the Republic of Slovenia through Project Grant AB FREE as well as through Slovenian Research Agency grants J1-6736 and P2-0046.

**Acknowledgments:** The authors are grateful to Svetlana Markovi´c for her useful suggestions regarding this work.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
