**Human Keratinocyte UVB-Protective E**ff**ects of a Low Molecular Weight Fucoidan from** *Sargassum horneri* **Purified by Step Gradient Ethanol Precipitation**

**Ilekuttige Priyan Shanura Fernando 1, Mawalle Kankanamge Hasitha Madhawa Dias 2, Disanayaka Mudiyanselage Dinesh Madusanka 2, Eui Jeong Han 2, Min Ju Kim 2, You-Jin Jeon 3, Kyounghoon Lee 4, Sun Hee Cheong 1,2, Young Seok Han 5, Sang Rul Park <sup>6</sup> and Ginnae Ahn 1,2,\***


Received: 30 March 2020; Accepted: 19 April 2020; Published: 21 April 2020

**Abstract:** Ultraviolet B (UVB) radiation-induced oxidative skin cell damage is a major cause of photoaging. In the present study, a low molecular weight fucoidan fraction (SHC4) was obtained from *Sargassum horneri* by Celluclast-assisted extraction, followed by step gradient ethanol precipitation. The protective effect of SHC4 was investigated in human keratinocytes against UVB-induced oxidative stress. The purified fucoidan was characterized by Fourier-transform infrared spectroscopy (FTIR), 1H nuclear magnetic resonance (NMR), agarose gel-based molecular weight analysis and monosaccharide composition analysis. SHC4 had a mean molecular weight of 60 kDa, with 37.43% fucose and 28.01 ± 0.50% sulfate content. The structure was mainly composed of α-L-Fucp-(1→4) linked fucose units. SHC4 treatment dose-dependently reduced intracellular reactive oxygen species (ROS) levels and increased the cell viability of UVB exposed HaCaT keratinocytes. Moreover, SHC4 dose-dependently inhibited UVB-induced apoptotic body formation, sub-G1 accumulation of cells and DNA damage. Inhibition of apoptosis was mediated via the mitochondria-mediated pathway, re-establishing the loss of mitochondrial membrane potential. The UVB protective effect of SHC4 was facilitated by enhancing intracellular antioxidant defense via nuclear factor erythroid 2–related factor 2 (Nrf2)/heme oxygenase-1 (HO-1) signaling. Further studies may promote the use of SHC4 as an active ingredient in cosmetics and nutricosmetics.

**Keywords:** oxidative stress; ultraviolet B; gradient ethanol precipitation; fucoidan; HaCaT keratinocyte; heme oxygenase-1; nutricosmetic

#### **1. Introduction**

Ultraviolet B (UVB)-irradiation (280–320 nm) primarily induces cell damage by augmenting intracellular reactive oxygen species (ROS) levels. The radiation easily penetrates the stratum corneum, where it increases the risk of photoaging, with symptoms characterized by thickening of

the epidermis, discoloration, skin wrinkling, loss of elasticity and skin cell growth retardation [1]. Recent developments in marine bioresource technology have substantiated the therapeutic potential of marine natural products to be used as cosmeceuticals, functional foods, new drugs and drug leads. Fucoidan is a fucose-rich sulfated polysaccharide found in brown algae that is renowned for its versatile biologic activities. High molecular weight fucoidans closely resemble the structure of heparin and are effective anticoagulants [2]. Nevertheless, low molecular weight fucoidans are reported to possess antioxidant, anti-inflammatory, anticancer and anti-microbial activities [3,4]. Fucoidan mainly contains α-1,3 and α-1,4-linked fucose units with sulfate ester substituents at the 2, 3 and/or 4 positions. However, the chains could also contain monosaccharides such as glucose, mannose, xylose, arabinose and hexuronic acid. The monomer composition, connectivity, branching, sulfation pattern, and molecular weight are species-related [2]. The conventional method of refining fucoidan involves hot water extraction of dry algal powder under mildly acidic pH, followed by ethanol precipitation. The precipitate undergoes additional fractionation via ultrafiltration, size exclusion or anion exchange chromatography. This method lacks efficiency based on the low extraction yield, convenience, time consumption and selectivity over molecular weights.

As fucoidan is mainly localized to the cell wall of brown algae, degradation of the cell wall using enzymes is a well-planned extraction strategy conserving polymer integrity, and thereby its biofunctional properties. The use of enzymes, including Celluclast, Termamyl, Ultraflo, Amyloglucosidase, and Viscozyme, is advisable to obtain fucoidan whose biofunctional properties are preserved [5]. Gradient alcohol precipitation is a fractionation method used for fractionating polysaccharides with gradually narrowed down molecular weight distributions [6]. It is a relatively inexpensive and simple method compared to ultrafiltration and chromatography. The precipitation is achieved by incorporating organic solvents such as methanol, ethanol, isopropanol, 1-butanol and acetone or concentrated inorganic salts such as ammonium sulfate ((NH4)2SO4) and polymerizing agents such as polyethylene glycol (PEG). The use of this method has been previously reported for fractionating fucoidan [7]. The present study was undertaken as a part of a project to investigate possible industrial uses of *S. horneri* as a sustainable approach for managing its large biomass. The extraction of fucoidan enriched crude polysaccharides followed an optimized green approach using enzymes, while the fractionation followed a step gradient ethanol precipitation.

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

Fucoidan standard, KBr (FTIR grade), deuterium oxide, 2- ,7- -dichlorodihydrofluorescein diacetate (DCFH2-DA) and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), o-Toluidine blue, trifluoroacetic acid and 2-mercaptoethanolwere purchased from Sigma-Aldrich (St. Louis, MO, USA). Celluclast was obtained from Novozyme Co. (Bagsvaerd, Denmark). Chloroform, methanol and ethanol were of analytical grade. Dulbecco's Modified Eagle Medium (DMEM), fetal bovine serum (FBS) and penicillin/streptomycin mixture were purchased from GIBCO INC., (Grand Island, NY, USA). Primary and ary antibodies were purchased from Cell Signalling Technology, Inc. (Beverly, MA, USA) and Santa Cruz Biotechnology (Santa Cruz, CA, USA).

#### *2.1. Preparation of Fucoidan Fraction*

#### 2.1.1. Extraction of Fucoidan Enriched Polysaccharide

Washed and dried *S. horneri* samples collected off Jeju coast were provided to us by Seojin Biotech Company limited. The samples were pulverized using an MF 10 basic, IKA microfine grinder (Werke, Germany). Depigmentation was carried out using a solvent system of chloroform and methanol 1:1. Next, the dried powder was soaked in a solution of ethanol containing 10% formaldehyde for 3 h at 40 ◦C. The dried powder was washed twice with 80% ethanol. After evaporating off any remaining solvent, the sample powder was suspended in 5 L of deionized water at a 1:10 (kg/L) ratio. The pH was adjusted to 4.5 by adding diluted HCl while equilibrating at 50 ◦C in a shaking incubator for 1 h. Celluclast was added at a 0.5% sample ratio and kept for 8 h under continuous agitation at 50 ◦C. The mixture was filtered through a muslin cloth. Celluclast was heat-denatured at 100 ◦C for 10 min. The extract was neutralized at room temperature by adding diluted NaOH and centrifuged at 5000× *g* for 20 min to remove unfiltered particles. The supernatant (4.5 L) was frozen and lyophilized to reduce the volume to 1 L.

#### 2.1.2. Step Gradient Ethanol Precipitation

The ratio of ethanol was determined following optimization studies. As the first step in gradient ethanol precipitation, 250 mL of ethanol was gently added to 1 L of the extract while stirring. The mixture was incubated at 4 ◦C for 12 h, allowing it to equilibrate while precipitating the polysaccharides (Figure 1). After, the mixture was centrifuged at 5000× *g* for 20 min at 4 ◦C to obtain the first precipitate designated as SHC1. Sequentially the second, third and fourth precipitates were collected by, respectively adding 500 mL, 1 L and 2 L of ethanol to the supernatant after each precipitation step. All precipitates were dually washed with 95% ethanol (homogenization) and centrifuged to recover the polymer. Finally, the precipitates were dissolved in deionized water and dialyzed using 3.5-kD molecular weight cutoff dialysis membranes (Spectra/Por, Los Angeles, CA, USA). Polysaccharide fractions were lyophilized and stored at −20 ◦C for proceeding experiments.

**Figure 1.** The procedure of sample pretreatment, enzyme-assisted extraction, and fractionation by gradient ethanol precipitation.

#### *2.2. Analysis of Molecular Weight (MW) Distribution*

Approximate molecular weight distribution, homogeneity, and separation efficiency of the polysaccharide fractions were analyzed by an agarose gel electrophoresis method [3]. Briefly, markers and samples (1 mg mL<sup>−</sup>1) were electrophoresed in 1% agarose gels in Tris-Borate-EDTA running buffer (pH 8.3) at 100 V for 20 min. The gel was stained with 0.02% toluidine blue and 0.5% Triton X-100 in 3% acetic acid and de-stained with 3% acetic acid.

#### *2.3. Fourier-Transform Infrared Spectroscopy (FTIR) and Monosaccharide Composition Analysis*

Polysaccharide powders were cast into KBr pellets and analyzed by a VERTEX 70v FTIR spectrometer (Bruker, Germany) [3]. For the monosaccharide composition analysis, polysaccharides were hydrolyzed with 4 M of trifluoroacetic acid and separated on a CarboPac PA1 column integrated to a Dionex ED50 Detector (HPAEC-PAD) (Dionex, Sunnyvale, CA, USA). A standardized monosaccharide mixture was used as the reference standard [3].

#### *2.4. H<sup>1</sup> Nuclear Magnetic Resonance (NMR) Analysis*

The selected polysaccharide fraction, SHC4, was deuterium exchanged by co-lyophilizing with deuterium oxide, dissolved in deuterium oxide, and analyzed by a JNM-ECX400, 400 MHz spectrometer (JEOL, Tokyo, Japan).

#### *2.5. In Vivo Cell Culture*

#### 2.5.1. Cell Maintenance

HaCaT keratinocytes (Korean Cell Line Bank, Seoul, Korea) were maintained in DMEM containing 10% FBS and 1% penicillin-streptomycin. Cells were sub-cultured once every two days. Exponentially growing cells were used for seeding (2 <sup>×</sup> <sup>10</sup><sup>5</sup> cells mL<sup>−</sup>1). After 24 h, wells were treated with different sample concentrations and further incubated for 24 h. Cytotoxicity was measured by MTT assay. Briefly, cells were incubated with MTT for 4 h, dissolved in dimethyl sulfoxide (DMSO), and the absorbance readings were taken at 540 nm using a SpectraMax® M2 system (Molecular Devices, Sunnyvale, CA, USA).

#### 2.5.2. UVB Exposure and Oxidative Stress Analysis

HaCaT keratinocytes were incubated for 2 h with different concentrations of samples. The cell culture media was withdrawn from the wells, once washed, and then filled with PBS. Wells except the control were exposed to UVB (50 mJ cm−2) by a UVP CL-1000L ultraviolet cross-linker (Upland, CA, USA). Immediately after the irradiation, PBS was withdrawn, once washed, and replaced with serum-free culture media. Intracellular ROS levels were measured after incubating for an hour (DCFH2-DA assay) and cell viability was measured by MTT assay after incubating for 24 h [1]. SHC4 that indicated superior antioxidant and cytoprotective effects were further verified by DCFH2-DA staining of cells and analysis by fluorescence microscopy (EVOS FL Auto 2 Imaging, ThermoFisher Scientific, CA, USA) and flow cytometry (CytoFLEX, Beckman Coulter, PA, USA) based on methods described in our previous publications [8].

#### 2.5.3. Analysis of Nuclear Morphology

Cells were stained with either Hoechst 33342 or mixture of acridine orange and ethidium bromide according to the method described in our previous publication [9]. Nuclear morphologies were visualized on an EVOS FL Auto 2 Imaging microscope (Thermo-Fisher, Waltham, MA, USA).

#### 2.5.4. Western Blot Analysis

HaCaT cells after initial sample treatment and UVB-exposure were used for the experiments following a 24 h incubation period. Western blot analysis was done following our previously reported method [10]. A nuclear and cytoplasmic extraction kit, NE-PER® (Thermo Scientific, Rockford, IL, USA), was used to isolate cytosolic and nucleic proteins. Proteins were standardized using a BCATM protein assay kit (Pierce, Rockford, IL, USA). Electrophoresis was carried out using 10% SDS-polyacrylamide gels. SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo, Burlington, ON, Canada) was used for the detection of protein bands on a Core Bio DavinchChemi imaging system (Seoul, Korea).

#### 2.5.5. Cell Cycle Analysis

Harvested cells were permeabilized in 70% ethanol for 30 min, incubated in PBS containing EDTA, RNase and propidium iodide at 37 ◦C for 30 min, and analyzed by a Beckman Coulter CytoFLEX flow cytometer (Brea, CA, USA). A dual-region gating approach, initial FS vs. SS to remove derbies and FL3-area vs. FL3-hight to eliminate doublets, was applied. Sub-G1 apoptotic cell population percentages were used for data compression.

#### 2.5.6. JC-1 Assay

Mitochondria membrane potential was ratiometrically measured by flow cytometry using MitoProbe JC-1 Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer's instructions. Cultured cells were initially subjected to sample treatment and UVB-exposure. After 4 h, cells were harvested for the assay.

#### 2.5.7. Comet Assay

DNA damage in individual cells was evaluated by alkaline comet assay following the procedure outlined in our previous publication [9]. Briefly, cells were harvested and suspended in low-melting agarose at 37 ◦C and gently applied on the surface of agarose coated slides. The slides were overnight immersed in alkaline lysis buffer at 4 ◦C. Electrophoresis was performed at 4 ◦C in an alkaline running buffer. Finally, the slides were submerged in a chilled neutralization buffer and stained with ethidium bromide. Comet tails were pictured on an EVOS FL Auto 2 Imaging microscope and tail DNA contents were quantified using OpenComet plugin in NIH Image J software (US National Institutes of Health, Bethesda, MD, USA).

#### *2.6. Statistical Analysis*

The data from experiments are presented as mean ± standard error (SE). Statistical comparisons were carried out by one-way analysis of variance by Duncan's test using PASW Statistics 19.0 software (SPSS, Chicago, IL, USA). The level of significance was set at \* *p* < 0.05 and \*\* *p* < 0.01.

#### **3. Results**

#### *3.1. Extraction E*ffi*ciency and Proximate Compositions*

The enzyme (Celluclast)-assisted extraction yield of *S. horneri* powder was 20.30 ± 0.32%. Yields corresponding to precipitated polysaccharides during step gradient ethanol precipitation are provided in Table 1. The highest yield was observed from fraction SHC2, obtained by incorporating 500 mL of ethanol. The proximate chemical compositions of the subsequent fractions indicated a gradual increase in the degree of sulfation in polysaccharides. The polyphenol, protein and ash content were comparatively low in all fractions, demonstrating the efficiency of the procedure in obtaining sulfated polysaccharides.


**Table 1.** Yield and proximate composition of the major components in the fractions.

SHC1-SHC4 denote different polysaccharide fractions obtained via step gradient ethanol precipitation. Data represent the mean ± standard deviation of triplicate determinants (*n* = 3).

#### *3.2. Molecular Weights Distribution of Polysaccharide Fractions and Their Vibrational Spectra*

Agarose gel electrophoresis has previously been used to estimate the weight distributions of anionic polysaccharides [11,12]. The heterogeneous nature of polysaccharides results in a weight distribution rather than a single band. As Figure 2A indicates, the molecular weight distributions of the polysaccharide fractions are gradually reduced in subsequent fractions. The approximate weight distributions were calculated using the Image J software based on the mean values of weight distributions of the molecular weight markers. Accordingly, the mean molecular weights of the fractions, SHC1, SHC2, SHC3 and SHC4, were estimated to be 230, 205, 90 and 60 kDa, respectively. The fingerprint region of the polysaccharide vibrational spectra (wavenumbers 2000–500 cm−1) is depicted in Figure 2B. The peak at 840 cm−<sup>1</sup> represented C-O-S bending vibrations of axial sulfate substituents on the carbon at the 4 position (C-4) of fucose units. The prominent, broad peak between 1200–970 cm−<sup>1</sup> occurred from the overlapping of peaks corresponding to C-C and C-O stretching vibrations of pyranoid rings and C-O-C stretching of glycosidic bonds, which are common to all polysaccharides. The major peak between 1220–1270 cm−<sup>1</sup> and its minor encounter at 585 cm−<sup>1</sup> were attributed to the O=S=O stretching vibrations of sulfates, which is a characteristic feature of sulfated polysaccharides such as fucoidans. The intense peak at the 1620 cm−<sup>1</sup> region corresponds to the bending vibration of O-H moieties [13–15].

**Figure 2.** Characterization of polysaccharide fractions (SHC1–SHC4) obtained by step gradient ethanol precipitation. (**A**) Molecular weights (MW) distribution analysis of polysaccharide fractions compared to 50–500-kDa, 60-kDa, 50-kDa and 8-kDa MW standards and (**B**) vibrational spectra of polysaccharide fractions compared to commercial fucoidan standard.

#### *3.3. SHC4 Increased the Protective E*ff*ects against UVB-Induced Oxidative Stress*

The cytotoxicity of polysaccharide fractions within 25–200 μg mL−<sup>1</sup> concentrations was examined prior to the evaluation of their bioactivities. None of the fractions indicated a significant decrease in cell proliferation within the tested concentration range (Figure 3A). Hence, the above concentrations were considered safe for the subsequent analysis of UVB protective effects. Based on DCFH2-DA and MTT assays, UVB irradiation significantly increased the intracellular ROS generation and inhibited cell proliferation (Figure 3B). All fucoidan fractions dose-dependently attenuated the effects of UVB irradiation. The best UVB protective effects were observed from the SHC4 fraction (25–100 μg mL<sup>−</sup>1): it significantly reduced UVB-induced intracellular ROS levels and increased cell proliferation. The 200 μg mL−<sup>1</sup> concentration of SHC4 deviated from the dose-dependent response observed for the 25–100 μg mL−<sup>1</sup> concentration range. Due to the increase seen in intracellular ROS levels and the affiliated inhibition of cell viability, the 200 μg mL−<sup>1</sup> concentration was not used

in subsequent analysis. Fluorescence-activated cell sorting (FACS) analysis using the DCFH2-DA fluoroprobe was considered reliable compared to the DCFH2-DA assay as it would omit the error caused by the decrease in fluorescence intensity due to cell death. As Figure 3C indicates, UV irradiation causes the peak to shift to higher intensity (Y-axis—FITC A) compared to the control. The effects were dose-dependently recovered with increasing concentrations of SHC4. Fluorescence microscopy analysis of DCFH2-DA fluoroprobe-treated cells indicated increased green fluorescence for ROS in UV-exposed cells compared to the control (Figure 3D), whereas the fluorescence was dose-dependently decreased with SHC4 treatment. Results of each ROS assay showed a moderate correlation, indicating the antioxidant effects of SHC4.

**Figure 3.** Protective effects of low molecular weight fucoidan fraction (SHC4) against ultraviolet B (UVB)-induced oxidative stress in HaCaT keratinocytes. (**A**) Cytotoxicity dose responses of polysaccharide fractions. (**B**) Analysis of intracellular reactive oxygen species (ROS) levels and cell viability after UVB exposure. Analysis of intracellular ROS levels by (**C**) flow cytometry and (**D**) fluorescence microscopy. HaCaT cells were treated with different doses of polysaccharide fractions for 2 h and exposed to UVB radiation. Measurement of intracellular ROS were performed 1 h after UVB exposure. Data represent the mean ± standard deviation of triplicate determinants (*n* = 3). \* *p* < 0.05 and \*\* *p* < 0.01 are significantly different compared with group indicated by "#".

#### *3.4. Characterization of SHC4 by NMR and Monosaccharide Composition Analysis*

Integrated with FTIR spectra, monosaccharide composition analysis and NMR spectra provide reliable evidence to characterize fucoidans. The NMR spectrum of SHC4 (Figure 4A) was obtained with a relatively low resolution, mainly because of the heterogeneous nature of the polymer. However, we were able to identify some characteristic peaks in the spectrum, signifying fucoidans. The intense signals obtained between 1.1–1.2 ppm represented protons of methyl groups in fucopyranose. Based on previous records, the occurrence of peaks in the 1.1–1.2 ppm area is attributable to an α-L-Fucp-(1→4) connectivity pattern [16]. The proton signals between 2.0–2.2 ppm represent alcoholic protons [16], while signals between 3.5–4.0 ppm were attributed to H2-H6 sugar residues. The prominent signal at 4.6 ppm, 1[H] can be attributed to protons at the anomeric center of 3-linked d-galactopyranosyl residues [16,17]. The signal at 6.25 ppm indicates α-anomeric protons of l-fucopyranosyl units [15]. The obtained results were similar to findings recorded from other brown seaweeds [3,15–17]. HPAEC-PAD analysis of monosaccharide composition (Figure 4B) proposed higher fucose (81.45%) and mannose (8.63%) content, which suggested that SHC4 was a mannofucan.

**Figure 4.** Characterization of fraction SHC4 by nuclear magnetic resonance (NMR) and monosaccharide composition analysis. SHC4 (**A**) 1H NMR spectrum and (**B**) monosaccharide composition analysis.

#### *3.5. SHC4 Reduced UVB-Induced Apoptotic Body Formation and DNA Damage*

Cell death due to UVB-induced oxidative stress is reported to proceed via apoptosis [18]. Thus, we assessed apoptotic body formation and DNA damage. Hoechst 33,342 and nuclear double staining with ethidium bromide and acridine orange are the methods of choice for studying nuclear morphology. Hoechst 33,342 binds specifically to DNA, allowing the identification of apoptotic nuclei with DNA fragmentation and chromatin condensation [19]. A prompt increase of chromatin condensation and DNA fragmentation was seen in UVB-stimulated cells compared to the control (Figure 5A). The occurrence of orange-colored nuclear fragments in UVB-exposed cells upon nuclear double staining was indicative of late apoptotic events. SHC4 treatment dose-dependently suppressed the formation of apoptotic bodies, demonstrating its protective effects. According to Figure 5C, the increase in the Sub-G1 population (23.88%) further indicates UVB-induced apoptosis compared to non-irradiated controls (0.83%). SHC4 treatment dose-dependently reduced the Sub-G1 population. Additionally, DNA damage was examined by comet assay (Figure 5D). The increased tail DNA

content was an indication of UVB-induced DNA damage. SHC4 treatment dose-dependently reduced UVB-induced cell damage.

**Figure 5.** Effects of SHC4 on ultraviolet B (UVB)-induced apoptotic body formation and DNA damage. Nuclear morphology analysis of apoptotic body formation by (**A**) Hoechst 33342 and (**B**) nuclear double staining with ethidium bromide and acridine orange. (**C**) Cell cycle analysis of Sub-G1 apoptotic cell accumulation. (**D**) Analysis of DNA damage by comet assay. Comet-tail DNA contents were quantified using OpenComet plugin in ImageJ software. HaCaT cells were treated with different doses of polysaccharide fractions for 2 h and exposed to UVB radiation. Cells were harvested for experiments 1 h after UVB exposure. Repeatability of results was validated with three independent determinations (*n* = 3).

#### *3.6. SHC4 Exerted Its Protective E*ff*ects against UVB-Induced Oxidative Stress by Inhibiting Mitochondria-Mediated Apoptosis Signaling*

The loss of the mitochondrial membrane potential due to pro- and anti-apoptotic Bcl-2 family proteins cause a subsequent discharge of caspase activators inducing apoptosis. The collapse of the mitochondrial inner transmembrane potential (ΔΨm) causes the opening of permeability transition pores resulting in an inward flux of water, which causes swelling and disruption of the outer membrane [20]. Based on a flow cytometry JC-1 assay (Figure 6A), a higher proportion of JC-1 aggregates with 530 nm emission (unhealthy mitochondria) were observed in UVB-exposed cells (78.77%), which was analogous to carbonyl cyanide m-chlorophenyl hydrazone (CCCP) treated cells. CCCP is a potent disruptor of the mitochondrial membrane potential. The membrane potential was dose-dependently attenuated following SCOC4 treatment, as evidenced by the increased emission at 590 nm and decreased emission at 530 nm. Key molecular mediators of the mitochondria-mediated apoptosis pathway were studied by western blot analysis (Figure 6B). An immediate increase was observed for BAX, caspases-3 and 9, p53, cleaved P poly (ADP-ribose) polymerase (PARP) and cytochrome C upon UVB-stimulation, whereas the anti-apoptotic proteins, Bcl-xL and Bcl2 levels

were decreased. SHC4 dose-dependently attenuated levels of molecular mediators, exhibiting its protective effects. Intracellular antioxidant enzymes, such as heme oxygenase-1 (HO-1), glutathione peroxidase (GPx), catalase (CAT) and superoxide dismutase (SOD) play a crucial role in maintaining the redox balance in cells. Compounds that can upregulate the transcription of the above antioxidant enzymes offer protective effects against oxidative stress, which can lead to apoptosis [21]. According to Figure 6C, UVB exposure slightly increases Nrf2 levels in the nucleus as well as HO-1 levels in the cytosol. SHC4 treatment further induced Nrf2 and HO-1 levels in UVB exposed keratinocytes in a dose-dependent manner.

**Figure 6.** Western blot analysis of the protective effects of SHC4 against UVB-induced apoptosis mediators. Analysis of the UVB protective effects of SHC4 on (**A**) changes in mitochondria inner transmembrane potential by JC-1 assay, (**B**) mediation of mitochondria-mediated apoptotic pathway proteins against UVB-induced apoptosis and (**C**) effects on intracellular antioxidant enzymes. HaCaT cells were treated with different doses of SHC4 for 2 h and exposed to UVB radiation. Cells were, respectively harvested after 4 h and 24 h for the JC-1 assay and western blot analysis. Repeatability of results was validated with three independent determinations (*n* = 3).

#### **4. Discussion**

The edible brown seaweed *Sargassum horneri* has long been acknowledged for its biofunctional effects. It is abundant along the Jeju island coasts in South Korea. Currently, the native *S. horneri* population is threatened by the massive invasion of *S. horneri* from the east coast of China [22]. This not only threatens the coastal ecosystem but also reduces the attraction for tourists. Climate changes have contributed to warmer temperatures and increased rainfall, which increase fertilizer runoff from paddy fields and the aquaculture industry, adding nutrients to coastal areas. The drastic change observed in *S. horneri* growth pattern is an example of how human activities alter the environmental balance. Other than clearing up the coasts, the utilization of *S. horneri* in the food industry and for obtaining functional ingredients such as polyphenols, fucoidan and alginic acid could be an effective and beneficial strategy.

Celluclast is a food-grade cellulase that hydrolyzes cell wall cellulose, resulting in a higher extraction yield of plant materials. The water-based extract under pH 4.5 may contain ions and polar metabolites such as phenols, soluble polysaccharides and proteins. Alternatively, extraction at acidic pH greatly reduces alginate contamination. Based on preliminary investigations, the addition of CaCl2 was not required as the formation of calcium alginate was negligible. Pretreatment with a mixture of chloroform and methanol (1:1) is reported to reduce lipophilic contaminants [23]. The chloroform and methanol (1:1) extract is currently studied for its potential bioactive natural products. Pretreatment with 10% formaldehyde in ethanol is reported to reduce the polyphenolic contamination of polysaccharides [23]. Otherwise, the contamination could lead to unreliable observations when assessing the bioactivities of polysaccharides. Further, washing with 80% ethanol not only removes any remaining formaldehyde but also removes polar compounds, such as phenolic acids. The reduced levels of polyphenol, protein and ash content (a measure of mineral content) in polysaccharide fractions obtained by step gradient ethanol precipitation suggested the efficacy of the reported method. The current pretreatment and extraction strategy was developed based on observations by several previous studies and preliminary analysis [3,11,14,24].

Crude fucoidan is generally obtained by adding a large volume of ethanol to the water-based extract. Subsequent fractionation is done by filtering precipitated polysaccharides through a series of molecular weight cutoff membranes (ultrafiltration) or by chromatography using size exclusion or anion exchange columns [23]. Ultrafiltration and chromatography are not suitable for industrial-scale preparations considering the investment, cost of equipment and operation delays. The step gradient ethanol precipitation method offers convenience over conventional fucoidan purification techniques. However, the technique has not widely been used for fractionating fucoidans except for that in one reported study [7]. The gradient alcohol precipitation method is commonly used in dextran fractionation and less frequently for purifying other polysaccharides such as starch, hemicellulose, glucan, fructan, pectin, arabinan and pullulan [6]. Based on agarose gel electrophoresis, four different fractions with approximate molecular weight distributions of 230, 205, 90 and 60 kDa were successfully obtained using step gradient ethanol precipitation. The high yield of the SHC2 fraction suggested that most fucoidans in *S. horneri* had a molecular weight range of approximately 205 kDa. Furthermore, low molecular weight polysaccharide fractions indicated a higher degree of sulfation. This further relates to the solubility of polysaccharides, where the ionic characteristics of polysaccharides increase with an increased degree of sulfation. Hence, the step gradient ethanol precipitation method could be said to separate fucoidans based on both the molecular weight and solubility.

The structural features of the fucoidan heteropolysaccharide vary considerably depending on seaweed species and environmental conditions. These features include monosaccharide composition, connectivity, chain length, degree of sulfation and connectivity of sulfate groups. Due to its heterogeneous nature, a full-scale analysis of connectivity remains challenging. Based on the present observations, the fractions were identified as fucoidans by FTIR and chemical composition analysis. Though it is not a generally accepted explanation, we propose that fucoidans could be ranked based on their quality, with the major parameters including the amount of fucose and degree of sulfation. According to many studies, the fucose amount is directly proportional to the degree of sulfation, which suggests that most sulfate groups remain attached to fucose units in fucoidan [3]. The fucoidan fraction that had the most prominent antioxidant effects, SHC4, was found to contain 45.96 ± 0.36% polysaccharides, of which, 81.45% was fucose, which corresponded to 37.43% of the fraction weight with a sulfate content of 28.01 <sup>±</sup> 0.50%. 1H NMR data revealed the presence of <sup>α</sup>-l-Fucp-(1→4) residues, which are a major feature of fucoidans. Hence, we designated SHC4 as a high-quality fucoidan. The lack of data on monosaccharide and sulfate group connectivity patterns is a limitation of the present study. The connectivity "methylation analysis" requires the fucoidan to be further fractionated, obtaining a less heterogeneous polymer mix.

Fucoidan is renowned for its potential antioxidant properties among a wide range of bioactivities [23]. The dose-dependent reduction of intracellular ROS levels in all fractions indicated the protective effect of fucoidans. Considering the slope observed for each fraction dose, the rate of the intracellular ROS reduction increased in succession, with the SHC4 fraction showing the most reduction. This suggested that low molecular weight and high-quality fucoidans had superior antioxidant properties. The increments observed for cell viability suggested that the protective effects of SHC4 were within the 25–100 μg mL−<sup>1</sup> concentration range. The minor reduction in cell viability at 200 μg mL−<sup>1</sup> concentration could be associated with toxic effects of sulfated polysaccharides under high concentrations. Further bioassays were conducted to assess the protective effects of SHC4 against UVB-induced oxidative stress and apoptosis. There was a reduction in UVB-induced apoptotic body formation, accumulation of Sub-G1 apoptotic cells and DNA damage, clarifying the potential antioxidant activity of SHC4. Further studies were then performed to understand the effects of SHC4 on the regulation of apoptosis.

The mitochondria-mediated apoptosis pathway is considered a major signaling route mediating apoptosis; the underlying mechanism can vary based on the type of cell, stimulus and other factors. Oxidative stress resulting from multiple factors, including the induction of UVB-radiation is reported to activate the mitochondria-mediated apoptosis pathway [18]. The pathway is initiated with the permeabilization of the mitochondrial outer membrane. The permeabilization is controlled by Bcl-2 family proteins, which includes anti-apoptotic proteins (Bcl-2 and Bcl-xL) and pro-apoptotic proteins (Bax, Bak, Bok, Bad, Bid, Bik, Bim, Bmf, Puma and Noxa). Their primary function is disrupting the functions of the Bcl-2 family proteins, thereby promoting the release of apoptogenic proteins present in the intermembrane space. Released apoptogenic proteins include cytochrome c, apoptosis-inducing factor (AIF) and endonuclease G. Cytochrome c with pro-caspase 9 and apoptosis protease activating factor (APAF-1) form an 'apoptosome'. The apoptosome promotes caspase 9 activation, in turn activating effector caspases, which collectively execute apoptosis. Both AIF and endonuclease G contribute to DNA fragmentation and chromosomal condensation, features that are a hallmark of apoptosis [25]. Apart from the aforementioned mediators, permeabilization of the mitochondrial outer membrane causes the release of numerous proteins that antagonize the activation of caspases. Pro-apoptotic stimuli inducing p53 further aggravate mitochondrial pathway-mediated apoptosis. p53 plays a crucial role in ultraviolet-induced apoptosis in HaCaT keratinocytes [26]. Under physiological conditions, p53 is maintained at a low concentration, inhibiting its transcriptional activity. Activation and post-translational stabilization of p53 drives the expression of pro-apoptotic factors, provoking death pathways. Caspases, consisting of initiator and effector caspases, are proteases that mediate cell death. Initiator caspases, such as caspase-9, activate death receptors, whereas effector caspases directly mediate the cleavage of numerous cytoplasmic and nuclear substrates. Cytochrome c activates caspase-9 and subsequently activates downstream effector caspases such as caspase-3, −6 and −7 [9]. The cleavage of PARP, a nuclear enzyme that maintains DNA stability, repair and transcription, is another crucial feature of apoptosis. Effector caspases cause the cleavage of PARP, thereby inhibiting its catalytic activity [9]. The present observations clarified the effect of UVB on the initiation of mitochondria-mediated apoptosis proteins and the dose-dependent attenuation effects of SHC4. There could be numerous alternative pathways, other than the prominent mitochondria-mediated

apoptosis that regulate oxidative stress-induced cell death. Hence, further studies could broaden the understanding of the effects of fucoidan on UVB-induced apoptosis.

Based on previous studies, fucoidans are reported to upregulate gene expression of HO-1 and SOD-1 in HaCaT keratinocytes via upregulating the transcription factor Nrf2 [21]. Similar results were observed during the present analysis, where SHC4 treatment caused a dose-dependent increase of HO-1 production in the cytosol and Nrf2 levels in the nucleus. This suggested that the protective effects of fucoidan against UVB-induced oxidative stress were mediated by Nrf2/HO-1 signaling.

#### **5. Conclusions**

Based on the present analysis, step gradient ethanol precipitation was identified as a desirable approach to obtain fucoidan fractions with different molecular weight distributions. The fucoidan fraction with a mean molecular weight of 60 kDa from *S. horneri* was mainly composed of α-L-Fucp-(1→4) linked fucose units with 37.43% fucose content and a 28.01 ± 0.50% sulfate content. The above fucoidan fraction, SHC4, demonstrated significant protective effects against UVB-induced photodamage in dermic keratinocytes by enhancing intracellular antioxidant defense. In the future, SHC4 could be studied for developing cosmetics with UV protective effects.

**Author Contributions:** Conceptualization, I.P.S.F. and G.A.; Methodology, I.P.S.F.; Software, I.P.S.F.; Validation, I.P.S.F., E.J.H. and M.J.K.; Formal analysis, I.P.S.F.; Investigation, I.P.S.F., D.M.D.M. and M.K.H.M.D.; Resources, G.A., K.L., Y.S.H. and Y.-J.J.; Data curation, E.J.H. and M.J.K.; Writing—Original draft preparation, I.P.S.F.; Writing—Review and editing, G.A.; Visualization, D.M.D.M. and M.K.H.M.D.; Supervision, G.A.; Project administration, S.H.C., S.R.P. and G.A.; Funding acquisition, S.H.C., S.R.P. and G.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work (Grants No. M01201920180359) was supported by the Korea Institute of Marine Science & Technology Promotion (KIMST).

**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* **Computational Study of** *Ortho***-Substituent E**ff**ects on Antioxidant Activities of Phenolic Dendritic Antioxidants**

**Choon Young Lee 1,\*, Ajit Sharma 1, Julius Semenya 1, Charles Anamoah 1, Kelli N. Chapman <sup>1</sup> and Veronica Barone 2,\***


Received: 7 February 2020; Accepted: 19 February 2020; Published: 25 February 2020

**Abstract:** Antioxidants are an important component of our ability to combat free radicals, an excess of which leads to oxidative stress that is related to aging and numerous human diseases. Oxidative damage also shortens the shelf-life of foods and other commodities. Understanding the structure–activity relationship of antioxidants and their mechanisms of action is important for designing more potent antioxidants for potential use as therapeutic agents as well as preservatives. We report the first computational study on the electronic effects of *ortho*-substituents in dendritic tri-phenolic antioxidants, comprising a common phenol moiety and two other phenol units with electron-donating or electron-withdrawing substituents. Among the three proposed antioxidant mechanisms, sequential proton loss electron transfer (SPLET) was found to be the preferred mechanism in methanol for the dendritic antioxidants based on calculations using Gaussian 16. We then computed the total enthalpy values by cumulatively running SPLET for all three rings to estimate electronic effects of substituents on overall antioxidant activity of each dendritic antioxidant and establish their structure–activity relationships. Our results show that the electron-donating *o*-OCH3 group has a beneficial effect while the electron-withdrawing *o*-NO2 group has a negative effect on the antioxidant activity of the dendritic antioxidant. The *o*-Br and *o*-Cl groups did not show any appreciable effects. These results indicate that electron-donating groups such as *o*-methoxy are useful for designing potent dendritic antioxidants while the nitro and halogens do not add value to the radical scavenging antioxidant activity. We also found that the half-maximal inhibitory concentration (IC50) values of 2,2-diphenyl-1-picrylhydrazyl (DPPH) better correlate with the second step (electron transfer enthalpy, ETE) than the first step (proton affinity, PA) of the SPLET mechanism, implying that ETE is the better measure for estimating overall radical scavenging antioxidant activities.

**Keywords:** antioxidant; dendrimer; electronic effect; hydrogen atom transfer (HAT); single electron transfer-proton transfer (SET-PT); sequential proton loss electron transfer (SPLET); DPPH

#### **1. Introduction**

The use of antioxidants to combat oxidative damage caused by excess free radicals is important for food, medical, cosmetics and other industries. Hence, synthesis of new potent antioxidants to tackle specific oxidative problems in these areas is very much needed. Phenolic antioxidants have attracted much attention in recent years, since many natural antioxidants contain the phenolic moiety.

Recently, we reported a new class of phenolic antioxidants that we called dendritic antioxidants. Syringaldehyde and vanillin, very weak natural antioxidants, were used as building blocks to assemble

potent phenolic antioxidant dendrimers with half-maximal inhibitory concentration (IC50) values significantly smaller than the building blocks. For example, dendrimers with 4, 6 and 8 syringol units had 27-, 100-, and 170-fold higher scavenging activities, respectively, than the syringaldehyde building block [1–3]. Besides enhancing radical scavenging, the dendritic architecture allowed metal chelation, thereby preventing potentially deleterious pro-oxidant effects of the antioxidants. We have also previously shown the ability of these dendritic antioxidants to protect biomolecules such as DNA and low-density lipoproteins from radical damage. Dendritic antioxidants offer several advantages similar to dendrimers. For example, their size, solubility, chelation ability, and antioxidant activity may be precisely manipulated by using appropriate cores and building blocks. Therefore, understanding the structure–activity relationship of dendritic antioxidants and their mechanism of action is paramount.

Substituents, such as electron-donating groups (EDGs) and electron-withdrawing groups (EWGs) on the phenol ring significantly affect the antioxidant activity of a phenolic antioxidant [3–5]. Several computational and experimental studies suggest that EDGs (e.g., OH, NH2, OCH3, CH3) decrease O–H bond dissociation enthalpy (BDE) of the phenol, which implies increased antioxidant activity [6–10]. On the other hand, EWGs showed varying results. Some computational studies reveal that EWGs, such as NO2, CN, CF3, F, Cl, and Br increase BDE of the phenol (lower antioxidant activity) [9,10]. However, experimental studies on halogens report different activities depending on the position relative to the phenol OH or the number of halogen substituents. For example, 6-chromanol with di-*o*-Cl substitution showed lower galvinoxyl radical scavenging activity but better than mono-*o*-Cl and tri-Cl substitution (two *o*-Cl groups and one *m*-Cl) compared to the unsubstituted 6-chromanol [5]. Hydroxycinnamic acids with mono-*o*-Br showed no effect in 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging activity [4].

Antioxidant action for free radical scavenging can occur in several different ways, but the ultimate outcome is the donation of a hydrogen atom (H•) to the radical. The hydrogen atom transfer can occur in multiple different ways. It can be (1) direct H• atom transfer to radicals, (2) electron transfer, followed by H<sup>+</sup> loss, or (3) proton loss, followed by electron transfer. The proposed mechanism for (1) is the hydrogen atom transfer (HAT) mechanism in which the hydrogen atom (H•) is transferred from the phenolic OH to the radical. In the HAT mechanism, the BDE is the most important parameter in estimating the antioxidant activity. The mechanism for (2) is the single electron transfer-proton transfer (SET-PT), in which an electron is first transferred from PhOH to the radical. The PhOH then becomes a phenoxy radical cation (PhOH•+), which in turn deprotonates during the second step to form a phenoxy radical (PhO•). In the SET-PT mechanism, the radical transfer in the 1st step is measured as the ionization potential (IP) and the deprotonation in the 2nd step corresponds to the proton dissociation enthalpy (PDE). The proposed mechanism for (3) is the sequential proton loss electron transfer (SPLET), which also proceeds via two steps. The 1st step involves the deprotonation of PhOH, forming a phenoxide ion (PhO−). Subsequently, the PhO− ion transfers an electron to the radical and becomes a phenoxy radical (PhO•). The enthalpy of the 1st step is denoted as the proton affinity (PA) and the 2nd step corresponds to the electron transfer enthalpy (ETE). HAT has been reported to be the most favorable antioxidant mechanism in non-polar solvents or in gas phase, while SET-PT and SPLET are the more prevalent mechanisms in polar solvents [9–15]

Computational studies of the electronic effects on dendritic antioxidants with multiple free radical scavenging sites have not been reported before. In order to study the effects of EDGs and EWGs on dendritic antioxidants, we performed computational studies using the Density Functional Theory (DFT) calculations for a series of small phenol-based dendritic antioxidants (in methanol), each containing either an EDG or an EWG that is *ortho* to the phenolic OH group.

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

All calculations were performed using Gaussian 16 [16]. Geometry optimizations as well as frequency calculations were obtained using the B3LYP functional and the 6-311++G\*\* basis set. Several conformations were used as initial structures to find the minimum energy conformation due to the flexibility of the studied compounds. All structures have been relaxed without symmetry constraints until maximum forces and displacements were smaller than 4.5 <sup>×</sup> <sup>10</sup>−<sup>4</sup> Hartree/Bohr and 1.8 <sup>×</sup> <sup>10</sup>−<sup>3</sup> Bohr, respectively. The requested convergence of the root mean square (RMS) variation of the density matrix elements and the total energy per cell in the self-consistent field procedure is 10−<sup>8</sup> a.u. and 10−<sup>6</sup> Hartree, respectively. To model the solvent effect of methanol, the Polarizable Continuum Model/Solvation Model Density (PCM/SMD) from Truhlar et al. was used in all calculations, including geometry optimizations [17].

For the different mechanisms studied here, the enthalpies are defined as follows:


From these definitions, it becomes clear that in order to find the total enthalpies of each mechanism we need the enthalpies for the electron, proton, and the H atom in methanol. As reported before, the solvation enthalpy of the H atom in most organic solvents is about 5 kJ/mol and we therefore adopted this value for methanol [17]. Methanol solvation enthalpies of the proton (−1071 kJ/mol) and the electron (−77 kJ/mol) have been calculated here at the same level of theory as the main compounds (B3LYP/6-311++G\*\*). These values are obtained by performing the calculation of a methanol molecule in methanol (PCM/SMD model) and then the methanol molecule after addition of a proton (or addition of an electron) in methanol (PCM/SMD model) as indicated by:


where the enthalpy of a proton in the gas phase is taken as <sup>5</sup> <sup>2</sup>*RT*.

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

#### *3.1. Geometry Optimization*

For each compound in Figure 1, we performed geometry optimizations. Structural relaxation of all compounds was carried out to find their minimum energy conformation as detailed under Materials and Methods. Our calculations indicate that in the optimized lowest energy conformation, the three phenol rings of each dendritic antioxidant were almost perpendicular to each other, indicating the absence of π-π stacking interaction between the phenol rings and no steric hindrance between the *ortho*-substituents.

The *o*-OCH3 group of the common phenol ring (in the reference and compounds **1**–**5**) was oriented in the same plane as the aromatic ring. It has been reported that in such an orientation, the 2p-type lone pair of electrons on the oxygen is parallel to the aromatic p-orbitals, thereby efficiently overlapping the aromatic π-electron cloud of PhOH [18]. This orientation stabilizes the incipient phenoxy radical (PhO•) and weakens the OH bond of PhOH [18]. The methyl of the OCH3 groups orients away from the OH group and the H of the phenol-OH points towards the oxygen of *o*-OCH3, forming an intramolecular hydrogen bond. If there are two OCH3 groups *ortho* to the phenolic OH (such as in compound **2**), both groups are in the same plane as the phenol ring with the methyl groups orienting away from the OH. It was also observed that *o*-NO2, *o*-Br, and *o*-Cl groups formed an H-bond with the H of phenolic OH in the energy-minimized conformation.

**Figure 1.** Structures of reference and target compounds **1–5**.

#### *3.2. Enthalpy Calculations*

To the best of our knowledge, this is the first computational study on the mechanism of action of dendritic antioxidants with multiple free radical scavenging sites.

We determined the enthalpy for each individual ring present on the dendritic antioxidants. To distinguish the enthalpy of each phenol O–H in the target compounds, X, X' and Y notations were used (Figure 2). X and X' represent equivalent phenol rings with either an EDG or EWG *ortho* to the phenolic OH. Y denotes the common ring with one *o*-OCH3 group.

**Figure 2.** General structure of the compounds and notation (X, X' and Y) for the phenol rings. R = H, electron-donating or electron-withdrawing group.

In this study, only compound **2** contains PhOH with two *ortho* substituents (OCH3). The enthalpy of **2** can be affected by not only electronic but also steric effects. The additional OCH3 group might help stabilize the incipient phenoxy radical electronically but might also confer a slight negative impact via steric effect. Since all other compounds have only one *ortho* substituent on each ring, a direct comparison of compound **2** with others might not be fair, but to get further insights into the effect of multiple electron-donating groups on antioxidant activity, we decided to design the compound and calculate its enthalpy nonetheless.

In most computational studies involving the electronic effects of substituents on antioxidant activities, the 1st step of each proposed mechanism, BDE (HAT), IP (1st step of SET-PT), and PA (1st step of SPLET), was reported to be the thermodynamically significant [9–11]. In some cases, the 2nd step of SET-PT and SPLET is not even calculated. To understand the antioxidant mechanisms (in methanol) more thoroughly, we calculated the enthalpy of the energy minimized structure of each dendritic molecule for all steps in the three proposed mechanisms (HAT, SET-PT, and SPLET) at the B3LYP/6-311G\*\* level of theory. For the HAT mechanism (Table 1), we calculated the enthalpy for each ring independently as if only one ring in the molecule reacted with the radical: three values in the first column, one for the common phenol ring with *o*-OCH3 (Y) and the other two for the two equivalent phenol rings (X, X') where R is either an *o*-EDG or *o*-EWG. In the second column of each compound, the enthalpy values for the two equivalent phenol rings (X, X') were averaged. In the case of SET-PT, only one value is given for each step for each compound (Table 2) since the loss of an electron is a global property and will most likely occur at the phenol ring with the highest electron density in an antioxidant with multiple phenol OH groups. For SPLET (Tables 3 and 4), we present three values in the first column and two in the second column as we did for HAT (calculated raw data is shown in Table S1 in Supplementary Information).

**Table 1.** Calculated O–H bond dissociation enthalpy (BDE) values (kJ/mol) of the phenol rings (X, X', and Y) in methanol.


**Table 2.** Calculated ionization potential (IP) and proton dissociation enthalpy (PDE) values (kJ/mol) in methanol.


<sup>a</sup> In parenthesis, we indicated the ring where the proton dissociation occurs.

**Table 3.** Calculated proton affinity (PA) values (kJ/mol) of the phenol rings (X, X' and Y) in methanol.


**Table 4.** Calculated electron transfer enthalpy (ETE) values (kJ/mol) of the phenol rings (X, X' and Y) in methanol.


#### 3.2.1. HAT Mechanism

In MeOH, *o*-OCH3 (EDG) decreased BDEO–H whereas *o*-NO2 (EWG) increased BDEO–H. *Ortho*-halogens (Cl and Br) showed similar BDEO–H compared to the unsubstituted reference (Table 1).

#### 3.2.2. SET-PT Mechanism

The IP values of all compounds (except for **2**) are most likely derived from their common phenol ring (Y) with one *o*-OCH3 because the ring has the highest electron density. In the case of compound **2**, its IP value could be derived from either of the two phenol rings (with two *o*-OCH3) since they are electron richer than the phenol ring with one *o*-OCH3.

Unexpectedly, compound **2** showed higher IP and lower PDE values than the reference and compound **1** (Table 2). Since **2** is the only one with two *o*-substituents, direct comparison with others might be misleading. Without considering compound **2**, *o*-OCH3 (EDG) decreased IP whereas EWGs (*o*-NO2, *o*-Cl, and *o*-Br) increased IP. The *ortho*-OCH3 group increased PDE whereas the *o*-NO2, *o*-Cl, and *o*-Br groups decreased PDE compared to the reference. It should be noted that the differences in IP and PDE values between the compounds are very small, implying they might be derived from the same ring, probably Y.

#### 3.2.3. SPLET Mechanism

Our results show that *o*-OCH3 (EDG) slightly increased PA whereas *o*-NO2 (EWG) significantly decreased PA (Table 3). *Ortho*-OCH3 decreased ETE considerably while *o*-NO2 increased ETE substantially compared to the reference (Table 4). *Ortho*-halogens (*o*-Cl and *o*-Br) exhibited slightly lower PA and slightly higher ETE values compared to the unsubstituted reference.

#### *3.3. Thermodynamically Favorable Antioxidant Mechanism for Dendritic Antioxidants*

Thermodynamically favorable processes should be determined by calculating the change in the free energy of a reaction (ΔG). The entropy component (−TΔS) in the proposed mechanisms is negligible, meaning that ΔG largely depends on the enthalpy (ΔH). Therefore, it is reasonable to determine the most preferred mechanism from enthalpy values [9,11].

In methanol, the SET-PT is clearly the least favored mechanism due to the large enthalpy of the first step. The SPLET mechanism has an overall low enthalpy in the first step but the enthalpy of the second step is comparable to the BDE in the HAT mechanism. However, the second step in SPLET is not isolated and occurs simultaneously with the reduction of the radical (in this case DPPH•) (reaction 2a + 2b in Scheme 1) as stated in the original papers that introduced the SPLET mechanism [13,19]. The authors state that based on studies between phenolic antioxidant and the DPPH radical in hydroxylic solvents like methanol or ethanol, the deprotonation step (1st step of SPLET) is equilibrium (reaction 1 in Scheme 1) and electron transfer from PhO− to the DPPH radical is fast (reaction 2a + 2b in Scheme 1) [13,14,19]. Many studies consider the first step as thermodynamically significant, perhaps based on these reports, and thus the enthalpy of only the 1st step is used to determine the major operating mechanism or in assessing the antioxidant potential. However, we wish to include the role of the DPPH radical in determining the major antioxidant mechanism. If we consider the electron transfer from PhO− (reaction 2a, Scheme 1) to the DPPH radical (reaction 2b, Scheme 1) to be concurrently occurring in the second step, the overall enthalpy of the second step can be reduced by −303 kJ/mol (reaction 2b), thus significantly favoring the SPLET mechanism over HAT in methanol. We note that the enthalpy for DPPH reduction was obtained at the same level of theory as all other calculations following reaction 2b (Scheme 1).

$$\begin{array}{ll} \text{PhOH} \xrightarrow{\text{-} \longrightarrow \text{-} \text{PhO}^{-} + \text{H}^{+} \text{(PA, 14 steps)} \longrightarrow \text{-} \text{(1)} \\\\ \text{PhO}^{-} \xrightarrow{\text{-} \text{-} \text{-} \text{-} \text{(ETE, 24 steps)} \longrightarrow \text{-} \text{(2a)} \\\\ \text{DPPH}^{+} + \text{e}^{-} \xrightarrow{\text{-} \text{-} \text{-} \text{-} \text{-} \text{DPH-} \text{(A} \text{H} \text{o} \text{e} \text{e}\_{\text{i}} \text{s} = -303 \text{ kJ/mol} \longrightarrow \text{-} \text{(2b)} \\\\ \text{PhO}^{-} + \text{DPPH} \bullet \xrightarrow{\text{Fast}} \text{PhO}^{-} + \text{DPPH-} \text{(net ETE)} \longrightarrow \text{-} \text{(2a+2b)} \end{array}$$

**Scheme 1.** Reaction between antioxidant (PhOH) and the DPPH radical via the SPLET mechanism.

Our computational study indicates that although both HAT and SPLET mechanisms are possible, SPLET is the more prevalent mechanism for dendritic antioxidants in the presence of the DPPH radical in methanol. This result is consistent with previous studies reporting that SPLET is the major

operating mechanism in polar ionizing solvents, while HAT is more favorable in nonpolar solvents or in gas [9–11,13,14]. However, it should be cautioned that the SPLET process between dendritic antioxidants and DPPH• in methanol should not be extrapolated to other radicals or solvents.

Values of pKa depend on the PA of phenol O–H: the higher the PA, the more difficult it is to deprotonate, meaning a higher pKa. Based on the pKa values calculated using the ChemAxon pKa calculator (Table S2 in Supplementary Information), the phenol rings in the dendritic antioxidants have pKa values of less than 11.0, indicating that they are weakly acidic. It was shown that phenols with pKa values below 12.5 can undergo the SPLET process with radicals derived from molecules with low pKa values, e.g., DPPH• and ROO• [19].

#### *3.4. Prediction of Overall Antioxidant Activity of Dendritic Antioxidants*

Since our dendritic antioxidants have three potential radical scavenging sites, it is important to know whether all three phenol rings can undergo the SPLET mechanism and contribute to the antioxidant activity or whether their enthalpy becomes too high after one or two radical scavenging and stops. Hence, we determined the cumulative total enthalpy (ΔHcum) using SPLET as the major operating mechanism in methanol to estimate the overall antioxidant activity of our dendritic compounds. Starting from the phenol ring which had the lowest PA value based on the PA order determined in the SPLET mechanism (Table 3), H<sup>+</sup> was removed and then an electron was removed immediately after to form a PhO• radical. The process was repeated based on the PA enthalpy order of the PhOH rings and continued cumulatively until all three PhOH rings became PhO• radicals. Figure 3 shows a scheme of the changes in the reference compound after each step. The ΔHcum of each compound was determined by adding the PA values of steps 1–3 and the ETE values of steps 1–3. These cumulative enthalpy calculations enable us to determine potential enthalpy cooperative effects of the rings in the dendritic antioxidants.

**Figure 3.** Scheme of the reference compound undergoing consecutive sequential proton loss electron transfer (SPLET) mechanisms. The order was determined based on the computed PA values shown in Table 3. At each step, charge and spin multiplicity of the resulting molecule are indicated in parenthesis (charge, spin multiplicity).

In the SPLET mechanism, the 1st step corresponds to the PA. EWGs decrease PA whereas EDGs increase PA. This means PhO− formation occurs more readily with PhOH containing an EWG and the resulting PhO− will undergo fast e-transfer to the radical. This may lead to the misconception that a PhOH with an EWG is a better antioxidant than the one with an EDG (higher PA). Based on our experience, antioxidants containing EDGs are better antioxidants than those containing EWGs. In our laboratory, syringaldehyde (4-hydroxy-3,5-dimethoxybenzaldehyde) and vanillin (4-hydroxy-3-methoxybenzaldehyde), each containing an aldehyde group, were used to synthesize dendritic antioxidants via reductive amination. Syringaldehyde and vanillin are very weak antioxidants by themselves, but dendrimers derived from them showed significantly higher antioxidant activities (as determined by the DPPH assay [20]). For example, antioxidant dendrimers containing eight syringaldehyde or vanillin derivatized units showed over 170- and 70-fold increase in DPPH radical scavenging activity rather than 8-fold compared to syringaldehyde and vanillin starting material, respectively [3]. The dramatic increase was caused in part by the replacement of the electron-withdrawing aldehyde (in the starting material) with an electron-donating benzyl group (in the dendrimer). Based on this observation, EDGs are more beneficial than EWGs for antioxidant activity. Although EDGs contribute to increasing PA in polar solvents, they will decrease ETE, thus helping electron transfer from the antioxidant to the radical occur more efficiently. Hence, we argue that the 2nd step (ETE) needs to be considered in determining the full antioxidant potential of our dendritic antioxidants since EDGs reduce ETE considerably. In addition, to emulate each phenol ring reacting with a radical (DPPH• was used in this study), the enthalpy change of the DPPH radical that is undergoing reduction by an electron from PhOH ring was considered. Therefore, the net enthalpy of the 2nd step of each PhOH ring was determined by combining ETE of PhOH (reaction 2a in Scheme 1) with ΔH of DPPH•, which is −303 kJ/mol (reaction 2b in Scheme 1).

In the reference compound, the phenol rings with no substituent (X and X') had a lower PA than the common ring (Y) and thereby are expected to lose H<sup>+</sup> before the common ring. Thus, the H<sup>+</sup> was removed from one of the phenol rings with no substituent (X) and formed a PhO− ion. The 1st PA enthalpy was 150 kJ/mol. In turn, an electron was removed to form a PhO• radical. The 1st ETE was 355 kJ/mol to which the ETE of DPPH• (−303 kJ/mol) was added, giving a net ETE of 52 kJ/mol. The process was repeated for the 2nd phenol ring (X') to determine the 2nd PA and ETE, which were 147 kJ/mol and 56 kJ/mol, respectively. The 3rd PA and ETE, which originate in the common ring, were 128 kJ/mol and 61 kJ/mol, respectively.

Compound **1** has the same three phenol rings with one *o*-EDG (OCH3) on each ring. Each phenol ring underwent the SPLET process one by one. The 1st, 2nd, and 3rd PA/ETE were 154/30, 150/37, 147/41 kJ/mol, respectively. Our results show a slight decrease in PA values as more H<sup>+</sup> is removed. In contrast, ETE increases as the reaction progresses.

In the case of compound **2**, the first H<sup>+</sup> was removed from one of the rings with two *o*-OCH3 (X or X'), because it had a lower PA than the common ring (Y). The 1st, 2nd, and 3rd PA/ETE were 151/16, 149/19, 148/40 kJ/mol, respectively. Its ETE values are substantially lower than those of the reference compound and **1**. By looking at the order of overall ETE values (compound **2** < compound **1** < reference compound), we can see the importance of EDG on the electron-donating potential of the antioxidants.

In compound **3**, the two phenol rings containing *o*-NO2 had the lowest PA values. Therefore, the first proton was removed from one of the PhOH rings with NO2. The 1st PA was determined to be 116 kJ/mol. Then, an electron was removed to form a PhO• radical (1st ETE = 119 kJ/mol). The same process was repeated for the 2nd phenol ring with the NO2 group. The 2nd PA and ETE were 116 kJ/mol, and 116 kJ/mol, respectively. The common ring underwent SPLET last. Its PA and ETE were 69 kJ/mol and 119 kJ/mol, respectively. Calculations were also done for compounds **4** and **5** to determine their PA and ETE in the same manner. The 1st, 2nd, and 3rd PA/ETE of compound **4** were 132/66, 130/70, 114/73 kJ/mol, respectively. The 1st, 2nd, and 3rd PA/ETE of compound **5** were 136/61, 133/66, 118/70 kJ/mol, respectively (calculated raw data is shown in Table S3 in Supplementary Information).

Figure 4 summarizes how each step affects the enthalpy of the following step, according to DFT. Overall, PA values gradually decrease while ETE values increase as more and more phenol rings undergo radical scavenging. This means that the enthalpy values in the later steps are affected by the phenol ring(s), which underwent deprotonation and ETE earlier. None of the steps showed a

significant increase in the corresponding enthalpy values, suggesting that the multiple SPLET process can occur and all three PhOH rings in our dendritic antioxidants are able to scavenge free radicals.

**Figure 4.** Enthalpy value of each step in a multiple SPLET process for the compounds studied. Each ETE value has been decreased by 303 kJ/mol (ΔHDPPH• = −303 kJ/mol) to account for the reaction with the DPPH radical.

The ΔHcum of each compound, determined by cumulatively adding PA and ETE of all three rings with consideration of the ΔHDPPH• is shown in Figure 5. The overall enthalpy order is compound **2** (523 kJ/mol) < compound **1** (559 kJ/mol) < compound **5** (584 kJ/mol) ≈ compound **4** (585 kJ/mol) < reference compound (594 kJ/mol) < compound **3** (655 kJ/mol).

**Figure 5.** The cumulative total enthalpy (ΔHcum) of consecutive SPLET mechanisms.

If the ΔHDPPH• is not considered, the overall ΔHcum order is compound **2** (1432 kJ/mol) < compound **1** (1468 kJ/mol) < compound **5** (1493 kJ/mol) ≈ compound **4** (1494 kJ/mol) < reference compound (1503 kJ/mol) < compound **3** (1564 kJ/mol). It is worth mentioning that the cumulative total enthalpy values (ΔHcum) are slightly higher than the summed-up enthalpy (ΔHtot = PA in Table 3 + ETE in Table 4) of the three rings in each compound: compound **2** (1428 kJ/mol) < compound **1** (1462 kJ/mol) < compound **5** (1490 kJ/mol) = compound **4** (1490 kJ/mol) < reference compound (1496 kJ/mol) < compound **3** (1561 kJ/mol). Nonetheless, the order remains the same. The difference in the ΔHcum and ΔHtot indicates that the phenol rings in the dendritic antioxidants are mutually dependent on each other.

Based on these results, compound **2** (two *o*-OCH3) had the lowest total enthalpy, followed by **1** (one *o*-OCH3), suggesting that PhOH with EDG will have higher antioxidant activity compared to the unsubstituted PhOH; the more EDG on PhOH, the better the antioxidant. Compound **3** (containing *o*-NO2) showed a higher total enthalpy compared to the reference compound. This suggests that PhOH with EWG will have poorer antioxidant activity than the unsubstituted PhOH. Compounds containing *o*-halogens (Cl and Br) had slightly lower enthalpy than the reference. However, the enthalpy values are very close to each other, suggesting that halogens do not have a significant effect on the overall antioxidant activity.

In order to determine if the dendritic antioxidants composed of 3 phenol rings are more active radical scavengers than their starting materials and fragments, we determined PA, ETE and ΔH of vanillin and syringaldehyde (starting materials) as well as the fragments of compound **1**, 4-(aminomethyl)-2-methoxyphenol (Y') and 4-methyl-2-methoxyphenol (Y") and fragments of compound **2**, 4-(aminomethyl)-2,6-dimethoxyphenol (X') and 4-methyl-2,6-dimethoxyphenol (X") (Table 5).


**Table 5.** PA, ETE and ΔH of starting materials and fragments of compounds **1** and **2**.

<sup>a</sup> ΔH = 3 equiv. PA (Y) + 3 equiv. ETE (Y), <sup>b</sup> ΔH = 3 equiv. PA (Y') + 3 equiv. ETE (Y'), <sup>c</sup> ΔH = 3 equiv. PA (Y") + 3 equiv. ETE (Y"), <sup>d</sup> ΔH = 2 equiv. PA (X) + 1 equiv. PA (Y) + 2 equiv. ETE (X) + 1 equiv. ETE (Y), <sup>e</sup> ΔH = 2 equiv. PA (X') + 1 equiv. PA (Y') + 2 equiv. ETE (X') + 1 equiv. ETE (Y'), <sup>f</sup> ΔH = 2 equiv. PA (X") + 1 equiv. PA (Y") + 2 equiv. ETE (X") + 1 equiv. ETE (Y").

Compound **1** has three rings derived from vanillin. Therefore, the PA and ETE of vanillin were tripled (3 equivalents) and then added together to determine the total enthalpy (ΔH). ΔH of vanillin (1518 kJ/mol) was higher than that (1468 kJ/mol) of compound **1**. In the case of compound **2**, it can be considered to be derived from 2 equivalents of syringaldehyde (X) and 1 equivalent of vanillin (Y). Therefore, both PA and ETE were calculated by using the formula, 2X+Y, giving a PA of 356 kJ/mol and an ETE of 1128 kJ/mol. The total enthalpy (ΔH) of the starting material (1484 kJ/mol) was higher than that of compound **2** (1432 kJ/mol). The higher enthalpy values of the starting materials are likely due to the presence of the electron-withdrawing aldehyde group on both molecules.

Both compounds **1** and **2** had lower PA but higher ETE and ΔH, compared to their respective fragments. This trend suggests that radical scavenging of antioxidants with multiple radical scavenging sites is a multistep cumulative process, which results in a gradual decrease in PA and increase in ETE as more radical scavenging sites are used up. The higher ΔH of dendritic antioxidants is understandable because their radical scavenging is a cumulative process. By the time the 2nd and 3rd phenol rings undergo the SPLET process, the antioxidant already has one phenoxy and two phenoxy radicals, respectively, which are higher in energy than the ground state (neutral phenol).

#### *3.5. Correlation of Enthalpy Values with DPPH IC50*

It is important to know how well theoretically computed enthalpy values of dendritic antioxidants correlate with their experimental DPPH radical scavenging activities [21]. We first determined the cumulative total PA (PAcum) and ETE (ETEcum) by adding up the PA and ETE of steps 1–3 shown in Figure 4, respectively. Then, PAcum, and ETEcum were added to determine ΔHcum (Table 6). To see the enhancing/decreasing effect, the enthalpy differences between each compound and the reference were determined (Δ(ΔHcum), ΔPAcum, and ΔETEcum) and ΔIC50 values were determined the same way. These differences are depicted in Figure 6. Based on these graphs, we observe that the trend of ΔIC50 resembles that of Δ(ΔHcum) most closely, followed by ΔETEcum.

**Table 6.** The cumulative enthalpy values, ΔHcum, PAcum, and ETEcum, for dendritic antioxidants and the experimental DPPH IC50 values.


<sup>a</sup> Data from [21].

**Figure 6.** Graphical representation of the enthalpy deviation of each compound from the value of the reference compound: (**a**) Δ(ΔHcum), (**b**) ΔPAcum, (**c**) ΔETEcum and (**d**) ΔIC50.

We also determined the Pearson correlation coefficients (r) between the IC50 and ΔHcum/ PAcum/ ETEcum values to determine which parameter of the SPLET mechanism better correlates with the IC50 values. Although the IC50 order looks slightly different from the order of ΔHcum (Table 6), their correlation was quite impressive: r = 0.9596 with a 95% confidence interval (0.6692, 0.9957) based on Fischer's Z-transformation (Figure 7a). The PAcum showed a strong negative correlation with IC50: r =

−0.9189 with a 95% confidence interval (−0.9912, −0.4222) (Figure 7b). The r of IC50 to ETEcum was 0.9623 with a 95% confidence interval (0.6882, 0.9960) (Figure 7c).

**Figure 7.** Correlation between IC50 and ΔHcum (**a**), IC50 and PAcum (**b**), and IC50 and ETEcum (**c**).

These r values, based on the collected data, indicate that the IC50 correlates better with ΔHcum and ETEcum as opposed to PAcum. These results suggest that between ETE and PA, the ETE might be a more significant measure in estimating the antioxidant potential than PA. From our study, it is clear that *o*-EDGs like OCH3 increase antioxidant activity whereas *o*-EWGs such as NO2 decrease it and the trend is more consistent with ETE. Some studies reported compounds containing EWGs to be better antioxidants in the SPLET mechanism than their counterparts with EDGs because of the lower PA values obtained from EWG containing compounds [11]. Their interpretation is reasonable if we consider the findings in the original papers that proposed the SPLET mechanism: the 1st step (PA) of SPLET is equilibrium and PhOH is converted to PhO− with the help of ionizing solvents. Once, the PhO− is formed in the 1st step, the 2nd step (ETE) occurs rapidly in the presence of DPPH radicals [13,14]. It was also reported that rate constants of PhOHs/DPPH•reactions in alcohols were increased by the addition of a base, indicating that PhO− formation is important in SPLET [19]. The formation of PhO− is directly related to the acidity of the PhOH: the more acidic PhOH is, the faster PhO− forms. Based on these reports, phenols containing EWGs seem to be better antioxidants than unsubstituted ones and those containing EDGs. This is because phenols with EWGs can form PhO− more readily and enter the SPLET mechanism faster. The comparison of IC50 values between our *o*-NO2 (55 μM) and *o*-OCH3 (8 μM) containing antioxidants (both having similar intramolecular and intermolecular H-bonding properties in ionizing solvents like methanol) suggest that faster formation of PhO<sup>−</sup> (lower PA of *o*-NO2) does not necessarily result in better antioxidant activities. The negative effect of *o*-NO2 on the antioxidant activity was shown not only by our study but also other experimental studies [22]. Thus, we propose that the 2nd step (ETE) of SPLET should be considered in assessing the full antioxidant potential of phenolic antioxidants. In agreement with our proposal, Li et al. showed that the introduction of *o*-NH2 will be more beneficial for the antioxidant activity than *o*-NO2 although their *o*-NO2 containing compound had lower PA value than the *o*-NH2 counterpart [10].

#### **4. Conclusions**

Enthalpy values for model dendritic antioxidants with either an *o*-EDG or *o*-EWG were determined for each step of the three proposed mechanisms (HAT, SET-PT and SPLET) in methanol using DFT calculations. Based on our results, *o*-OCH3 (*o*-EDG) decreases BDE, IP (1st step of SET-PT), and ETE (2nd step of SPLET) and increases PDE (2nd step of SET-PT) and PA (1st step of SPLET) while *o*-NO2 (EWG) showed the opposite effects. SPLET was found to be the preferred mechanism in methanol when DPPH• was used as the radical.

To compare relative antioxidant activity between the dendritic compounds, the total enthalpy of each compound was determined by cumulatively running SPLET for all three PhOH rings and adding all the enthalpy values. The results showed that *o*-OCH3 (EDG) increases the antioxidant activity whereas *o*-NO2 (EWG) decreases it. In comparison, *ortho*-halogens (Br and Cl) showed negligibly lower enthalpy values compared to the unsubstituted PhOH. Correlation of the DPPH radical scavenging activities with each step in SPLET revealed that the experimental IC50 values better correlate with ETE (the 2nd step) rather than PA (the 1st step), implying that ETE is a more important factor in estimating the role of substituents on antioxidant activity than PA.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-3921/9/3/189/s1. Table S1. DFT calculations in methanol; Table S2. Values of pKa obtained using ChemAxon pKa calculator; Table S3. Multistep DFT calculations in methanol.

**Author Contributions:** Conceptualization, C.Y.L.; calculations, V.B.; data curation, J.S., C.A., K.N.C. and V.B.; formal analysis, C.Y.L., A.S. and V.B.; writing—original draft, C.Y.L.; writing—review and editing, C.Y.L. and A.S. 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*
