**Preface to "Dietary Polyphenols and Neuroprotection"**

As editors of this book, our aim was to collect new data from experienced authors in order to further advance the knowledge on the protective effects of polyphenols' intake, e.g., when included in the human diet, to modulate cellular functions and pathways associated with neurodegenerative diseases.

Most, if not all, edible vegetables included in human diet are rich sources of phenolic compounds. Such compounds are commonly used by the plants for several reasons, e.g., as metabolic intermediates, protective substances, reproductive attractants, etc. Many of those phytochemicals have been recognized as potent antioxidants, also sharing other important actions that can impact human health like their anti-inflammatory properties and their proposed capability to modulate several cell-signalling pathways and mediators.

Neurodegenerative diseases are among the main causes of death worldwide and, in most of them like Alzheimer's or Parkinson's, neurodegeneration occurs long before the onset of first symptoms, where a large population of brain neurons are already lost. Besides neurons, glial cells like astrocytes and microglia, are involved in oxidative and neuroinflammatory pathological pathways, making them interesting targets for neuroprotective strategies. Polyphenols are promising candidates for those strategies, either as prophylactic substances or as therapeutic molecules.

The proposed benefits of polyphenols, either as protective/prophylactic substances or as therapeutic molecules, may be achieved by the consumption of a natural polyphenol-enriched diet, by the use as food supplements or formulation as pharmaceutical drugs/nutraceuticals. It was also proved that the health effects of polyphenols depend on the amount consumed as well as on their bioavailability.

With this collection of original data and literature reviews, we hope to raise the interest of scientist, researchers, medical doctors and students in the promising field of polyphenols neuroprotection as a relevant approach to prevent or treat mental disorders, dementia and neurodegenerative diseases.

> **Rui F. M. Silva, Lea Pogaˇcnik** *Editors*

## *Editorial* **Neuroprotective Properties of Food-Borne Polyphenols in Neurodegenerative Diseases**

**Rui F. M. Silva 1,\* and Lea Pogaˇcnik <sup>2</sup>**


Fruits and vegetables are the richest source of polyphenols in the regular human diet. These substances belong to plants' secondary metabolites and can have several roles, such as being metabolic intermediates, reproductive attractants, and protective agents. Most of these molecules possess a high antioxidant capacity, as well as several other important activities that can affect human health, among which anti-inflammatory properties and the potential ability to modulate different cell signaling pathways seem to be the most important.

Taking into account these significant properties of polyphenols, together with their abundance in various food products that are a part of a healthy human diet, a wide range of different approaches, both in vitro and in vivo, address their potential role in the prevention and treatment of different pathological conditions associated with oxidative stress and/or inflammation.

The significance of food-borne polyphenols on human health has considerably increased the number of studies dedicated to their various proposed actions in many different pathologies such as cancer, and cardiovascular and neurodegenerative diseases, which has also resulted in a growing number of clinical trials on the acute or chronic use of dietary polyphenols. Studies in the field of cancer and neurodegenerative disorders are particularly important since an effective treatment is still not available. It was shown that food-borne polyphenols can be used either as protective/prophylactic molecules or as therapeutic substances. They can be consumed as part of a natural polyphenol-enriched diet, with the use of food supplements or as pharmaceutical drugs/nutraceuticals.

In this Special Issue of *Antioxidants*, several research papers and two reviews explore the chemical properties of naturally occurring polyphenols and some new possibilities for the therapeutic and/or prophylactic roles of these molecules in neurodegeneration and neurodegenerative diseases.

Ribeiro et al. [1] proposed the use of a new enzymatic biosensor to determine the antioxidant activity of commercially available teas, as well as the level of ascorbic acid in effervescent products, which can also be used to detect the real level of antioxidants in samples and depict the validity of "antioxidant" labeling in the product information. Using flavone derivatives, Sakalauskas et al. [2] further explored the potential of polyphenols to prevent amyloid aggregation, an important hallmark of Alzheimer's disease (AD) as well as other amyloid-related disorders. They demonstrated that after oxidation, flavones, particularly oxidized 6,2′ ,3′ -trihydroxyflavone, not only keep but even increase their anti-amyloid properties, which might be a relevant addition to the discussion relating to whether, after physiologic metabolization, polyphenol derivatives still keep their beneficial properties.

Continuing with the study of the antioxidant properties of polyphenol-rich vegetables, Jug et al. [3] evaluated extracts of Japanese knotweed rhizomes produced by several extraction solvents. Knotweed is an invasive botanical species, and it is important to find

**Citation:** Silva, R.F.M.; Pogaˇcnik, L. Neuroprotective Properties of Food-Borne Polyphenols in Neurodegenerative Diseases. *Antioxidants* **2021**, *10*, 1810. https:// doi.org/10.3390/antiox10111810

Received: 9 November 2021 Accepted: 11 November 2021 Published: 15 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

novel economically viable applications for this destructive botanical species. Using size exclusion–high-performance liquid chromatography (SEC-HPLC)-UV and reversed-phase HPLC-UV coupled with multistage mass spectrometry to fractionate the extracts, they identified (−)-epicatechin as a potent and stable antioxidant.

The well-proven antioxidant properties of polyphenols can be explored in the prevention of or reduction in oxidative stress-induced injury. Accordingly, the results from Bobadilla et al. continue to reinforce the potential of natural food polyphenols to reduce neuronal demise by several mechanisms, as in the case of aluminum maltolate neurotoxicity, where several commercially available natural food supplements decreased neuronal cell death in vitro by reducing ROS levels and caspase-3 activity, and also increased the antioxidant enzyme activity in mice, preventing the formation of lipid peroxidation products in the brain [4].

The multiple targets for polyphenols and their beneficial effects on several human chronic diseases were reviewed by Bucciantini et al. [5]. These authors explore, in particular, the actions from polyphenols present in extra virgin olive oil, expanding from the antioxidant to the anti-inflammatory properties. They advance to the use of olive oil polyphenols in human chronic diseases that involve inflammation due to their inhibitory effects on oxidative stress-induced signaling pathways and minimal secondary effects.

The second review from this Special Issue reinforces polyphenols' multiple mechanisms of action, focusing on amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), two of the neurodegenerative disorders still without effective treatments or cures. For that, Novak et al. [6] analyze the current therapeutical options for ALS and FTD in parallel with several of the more prominent polyphenols such as resveratrol, curcumin and green tea catechins, emphasizing the therapeutic potential of polyphenols.

In fact, according to the results from Zheng et al. [7], resveratrol might also influence the clearance of beta-amyloid peptide-42 (Aβ42), related to AD neurodegeneration, by modulating the insulin-degrading enzyme that also has a strong ability to degrade Aβ4.

Moving to in vivo studies, a crucial pre-clinical step, the findings from Hong et al. [8] show very important amelioration effects on cognitive and memory functions in models of neurodegeneration. Interestingly, ampelopsin A from Vitis vinifera was shown to have neuroprotective properties, increasing both cognitive and memory functions by, in part, elevating the BDNF/CREB-related signaling, in a mice model as well as in hippocampal brain slices (CA3-CA1 synapses), where neurodegeneration was induced by scopolamine.

Finally, in a pilot study performed in older human volunteers with memory complaints, but not AD, Robinson at al. [9] evaluated the effects of the administration of a whole coffee cherry extract nutraceutical, rich in polyphenols, using MRI and determination of BDNF blood levels. In summary, they found significant improvements in cognition that may be related to the increase in exosomal BDNF.

In conclusion, polyphenols seem to be effective molecules for preventive and therapeutic strategies in a wide range of pathological conditions. However, it will be important to take into account the possible issues raised by their dosage and toxicity and monitoring of their safe usage.

**Funding:** This work was supported by iMed.ULisboa, Fundação para a Ciência e Tecnologia (FCT), Portugal (UID/DTP/04138/2013), and the Slovene Research Agency (P4-0121).

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

#### **References**


## *Article* **Determination of the Antioxidant Activity of Samples of Tea and Commercial Sources of Vitamin C, Using an Enzymatic Biosensor**

**Danilo Braga Ribeiro <sup>1</sup> , Gabriela Santos Silva <sup>1</sup> , Djanira Rubim dos Santos <sup>1</sup> , Andressa Rose Castro Costa <sup>1</sup> , Eliane Braga Ribeiro <sup>1</sup> , Mihaela Badea 2,\* and Gilvanda Silva Nunes 1,\***



**Citation:** Ribeiro, D.B.; Santos Silva, G.; dos Santos, D.R.; Castro Costa, A.R.; Braga Ribeiro, E.; Badea, M.; Nunes, G.S. Determination of the Antioxidant Activity of Samples of Tea and Commercial Sources of Vitamin C, Using an Enzymatic Biosensor. *Antioxidants* **2021**, *10*, 324. https://doi.org/10.3390/antiox10020324

Academic Editors: David Arráez-Román, Rui F. M. Silva and Lea Pogaˇcnik

Received: 28 December 2020 Accepted: 17 February 2021 Published: 22 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

**Abstract:** Antioxidants are synthetic or natural compounds capable of preventing or delaying oxidative damage caused by chemical species that can oxidize cell biomolecules, such as proteins, membranes, and DNA, leading to the development of various pathologies, such as cancer, atherosclerosis, Parkinson, Alzheimer, and other diseases serious. In this study, an amperometric biosensor was used to determine the antioxidant activity of teas and effervescent products based on vitamin C, available on the market. A sensor composed of three electrodes was used. The performance of the following electrochemical mediators was evaluated: meldola blue combined with Reineck salt (MBRS), Prussian blue (PB), and cobalt phthalocyanine (CoPC), as well as the time of polymerization in the enzymatic immobilization process and the agitation process during chronoamperometric measurements. Prussian blue proved to be more efficient as a mediator for the desired purposes. After optimizing the construction stages of the biosensor, as well as the operational parameters, it presented stability for a period of 7 months. The results clearly indicate that the biosensor can be successfully used to detect fraud in products called "antioxidants" or even in drugs containing less ascorbic acid than indicated on the labels. The detection limit was set at 4.93 µmol·L −1 .

**Keywords:** antioxidants; biosensors; xanthine oxidase; teas; drugs; vitamin C

#### **1. Introduction**

Due to redox reactions that provide cells with the energy necessary for their functioning, external factors such as pollution, bad habits (smoking, alcohol consumption), and inadequate nutrition cause an increase in formation of free radicals in the human body [1].

Diabetes, cirrhosis, cardiovascular diseases, some types of cancer, and neurological disorders are examples of diseases often associated with irregular and uncontrolled processes in the production of these radicals. In excess, they can cause oxidative damage to cells, forming advanced glycation products, inactivating proteins (enzymes), and attacking membrane lipids, carbohydrates, and DNA [2–4].

The search for new methods of evaluating the antioxidant activity of several compounds potentially capable of inhibiting such damage in biological systems or even in food has increased in the last years, as shown by a search in the Web of Science database on 4 February 2021 (Supplementary Data)—1917 references since 1997 for antioxidant capacity detection, most of them from the last five years. From all the indicated references, 1801 (93.95%) were articles, 86 (4.46%) were proceedings papers, and 81 (4.23%) reviews. Considering the journals that published these papers, most of them are dealing with applications in food chemistry. Considering the entire group of proposed articles, an H-index—112

was generated, indicating in this way the importance of this topic for scientific media. We believe that also our study will be an important one due to the applications proposed in the sample of tea and commercial sources of vitamin C, using a novel methodology developed recently by our group and previously successfully applied for testing antioxidant capacity of fresh and frozen fruit. Of natural or synthetic origin, antioxidants are able to prevent or delay oxidative damage generated by oxidizing sources, even when in lower concentration compared to the oxidizable substrate [5]. The human body's antioxidant defense system consists of a range of bioactive compounds capable of neutralizing the action of free radicals, such as vitamins (A, C, E, K), glutathione, mineral salts, metalloproteins, enzymes (SOD—superoxide dismutase) and polyphenols [6,7]. Antioxidants of exogenous origin come from plant sources, such as fruits and teas, and are also found in commercially available supplements [8–10].

Over the years, the use of plants for medicinal purposes has grown considerably, either empirically or in complex compositions of the pharmaceutical industry and consumption in the form of teas has been quite expressive due mainly to the incentive to use natural products.

Different methodologies based on electrochemical, spectrophotometric, and chromatographic techniques have been described in the literature in order to assess the antioxidant capacity in different matrices. In general, they differ in terms of the mechanisms for obtaining oxidizing species and in the way the final products are measured. However, they have the drawback of being relatively time consuming and employing expensive bioreagents [11–13]. In addition, due to the presence of unsaturation in its chemical structure, antioxidants from natural products can present stability problems that make them sensitive to exposure to heat, light, and the presence of oxygen [14].

On the other hand, in recent years, electrochemical biosensors have been considered as a promising tool in determining the antioxidant potential, due to their characteristics as selectivity, low cost of obtaining, ease of storage, miniaturization capacity, easy automation, and portability, which combined make possible in situ analysis, reducing the risk of interference resulting from the destabilization of compounds [15,16].

Therefore, this work aims to improve and apply an analytical device in the determination of the antioxidant capacity in samples of teas and commercial sources of vitamin C, which offers the additional advantages of facilitated construction, high precision, and sensitivity of detection and that allows use both in the laboratory and in situ. In this paper, electrochemical evaluations will be performed using the cyclic voltammetry (CV) technique and chronoamperometric measurements, comparing with the previous paper [12] by our group where amperometry was used.

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

#### *2.1. Reagents and Solutions*

All reagents used were of analytical grade and the water used was deionized (Milli-Q Millipore 18.2 MΩ cm−<sup>1</sup> ). Prussian blue or ferric ferrocyanide (PB) was obtained from Gwent Group (Torfaen, United Kingdom). Water-soluble polyvinyl alcohol photopolymer (PVA-AWP) was purchased from Toyo Kogyo Corporation (Chiba, Japan). Monobasic potassium phosphate (KH2PO4), dibasic potassium phosphate (K2HPO4), potassium chloride (KCl), hypoxanthine (HX), bovine milk xanthine oxidase enzyme (XOD) were all purchased from Sigma Aldrich Corporation (Nasdaq-Sial, Darmstarm, Darmst, Germany). Ascorbic acid (C6H8O6) was purchased from Merck (Seelze, Germany). The 50 mmol·L −1 K-PBS buffer solution (K2HPO<sup>4</sup> 33.33 mmol·L −1 , KH2PO<sup>4</sup> 16.67 mmol·L −1 ) containing 10 mmol·L <sup>−</sup><sup>1</sup> KCl (pH 7.5) was used in the preparation of the enzymatic solutions (stock and work), solutions of the substrate hypoxanthine (HX) 5 mmol·L <sup>−</sup><sup>1</sup> and as the electrolyte in electrochemical measurements. It was also used as a solvent in tea infusions and dissolution of effervescent vitamins C.

#### *2.2. Instrumentation*

The electrochemical measurements using the biosensors were made in a a Ivium-nstat potentiostat/galvanostat controlled by the IviumSoft software (Ivium Technologies, Eindhoven, Netherland). The working, reference, and auxiliary electrodes were printed on a thin transparent polyvinyl chloride (PVC) plate, which constituted the electrochemical sensor. The reference pseudo electrode was constituted by a straight line 5 × 1.5 mm in diameter, and is formed by a mixture (paste) of Ag/AgCl. The working electrode consisted of a 4 mm diameter disk, formed by a commercial graphite paste containing Prussian blue salt (PB) or Meldola blue with Reinecke salt (MBRS) or cobalt phthalocyanine (CoPC) as a modifier. The auxiliary electrode, formed by a 16 × 1.5 mm curved line, contained only the commercial graphite paste. Screen-printed electrodes (SPE) were produced in laboratory of University of Perpignan via Domitia, using a DEK 248 printing machine, and offered for these studies by Prof. Dr. Jean-Louis Marty.

#### *2.3. Electrochemical Characterization*

Electrochemical evaluations of sensors and biosensors were performed using the cyclic voltammetry (CV) technique. The biosensors were initially prepared according to the methodology developed in our research group [12]. An enzymatic charge of 8 mU XOD was immobilized under the surface of the modified working electrode, from the deposition of 3 µL of a homogeneous mixture of the enzymatic solution and PVA- AWP in the proportion 1:2 and later polymerization under neon light at 4 ◦C for 30 min. The effect of pH on the biosensor response was also evaluated.

#### *2.4. Chronoamperometric Measurements and Parameter Optimization*

All chronoamperometric measurements were performed at room temperature using a 10 mL dark electrochemical cell. The biosensor was previously subjected to 10 voltametric cycles and the current generated was then measured at a fixed working potential of −100 mV vs Ag/AgCl, where there is a reduction in H2O2.The intensity of the initial current was recorded after swelling of the PVA-AWP, followed by signal stabilization, in a total time of approximately 20 min. Then, an analytical curve was constructed by successively adding aliquots of the HX 5 mmol·L −1 solution under constant stirring. Then, the polymerization time (30 and 60 min) and the stirring conditions during measurements were optimized, as well as the current measurement time.

#### *2.5. Determination of the Antioxidant Capacity of Real Samples*

As a negative control, the production of reactive oxygen species (ROS) was used without neutralizing them by antioxidants. An analytical curve of current intensity (at the end of 75 s) as a function of the concentration of the hypoxanthine substrate (HX) was constructed, and the angular coefficient recorded (ma). Then, a new analytical curve was built, but in the presence of the antioxidant solution (standards or samples), and the antioxidant potential of the solution or sample was determined. The antioxidant capacity was expressed by the percentage of ROS inhibition by comparing the slope obtained in the curves constructed in the absence (ma) and the presence of antioxidants (mb), according to the Equation (1):

$$\text{Antioxidant capacity } \%= 100 \ast \left[1 - \left(\frac{mb}{ma}\right)\right]. \tag{1}$$

Sample Preparation and Analysis

Samples of teas (fennel, chamomile, mint, cimegripe tea) and effervescent vitamin C were obtained from pharmacies in the city of São Luís, Maranhão, Brazil.

In assessing the antioxidant capacity of the tea, a volume of 50 mL of an infusion in K-PBS buffer was prepared by heating for 10 min on a magnetic stirrer with heating. A 10 mL volume of the infusion was transferred to the electrochemical cell. Then, analytical curves were constructed in the presence of HX in different concentrations, and the slopes of the curves constructed in the presence and absence of the samples were compared.

The samples of effervescent vitamin C were prepared by dissolving the mass of the tablet corresponding to 500 mg of ascorbic acid in 20 mL of K-PBS buffer. An aliquot of 20 µL of this solution was then transferred to the electrochemical cell, and the volume was made up to 10 mL with K-PBS. The curves were constructed, according to the procedure described above, and the antioxidant capacity determined.

The experiments were carried out using an infrared spectrometer from Shimadzu, model IR-Prestige-21 with an extended KBr (potassium bromide) beam splitter. The data were collected in a range of 500 to 4000 cm−<sup>1</sup> , at a resolution of 4 cm−<sup>1</sup> using a spectral medium of 40 scans in potassium bromide.

The content of ascobic acid in the drugs was determined by the standard addition method. Then, 100 µl aliquots of the samples were transferred to a 50 mL volumetric flasks and volumes (500, 1000, 1500, 2000, and 2500 µL) of a 1 mg·L −1 solution of pure ascorbic acid was added to them, completing them with water. The absorbance was measured at a wavelength of 264 nm, using a UV-VIS spectrophotometer (Themoscientific, Orion AquaMate 8000), quartz cuvette 1 cm of optical path.

#### **3. Results and Discussions**

#### *3.1. Electrochemical Characterization*

The sensors used were built on a flexible and chemically inert base (PVC), where the three electrodes were printed, using a simple methodology based on semi-automatic screen printing. Such technology enables the manufacture of economic, portable, quick-response electrodes, with high sensitivity, low power required, disposable, and with the ability to operate at room temperature, thus enabling the performance of in situ analyses [17].

The O<sup>2</sup> and H2O<sup>2</sup> molecules monitored in the system proposed here are electroactive species that undergo oxidation and/or reduction when subjected to high work potentials, generating an electrical signal. Uric acid, the product of the enzymatic reaction, as well as several antioxidant compounds (ascorbic acid, for example), are also oxidized when they occur at high potentials, which can generate interference in the measured current [18]. Thus, the use of electrochemical mediators aims to annul or reduce such interference since it allows working with lower potentials [19]. Mediators are chemical species capable of donating or receiving electrons, thus helping to regenerate the oxidation state of the enzyme and its active center in an enzymatic reaction. The modifying agent has the function of increasing the sensitivity of the electrodes and can be incorporated into the carbon paste by directly adding a certain mass of the modifier in a mixture of graphite powder and binder [20].

The generated H2O<sup>2</sup> is reduced on the polarized (−100 mV vs. Ag/AgCl) WE surface, in presence of PB mediator, which has a specific catalytic effect for the H2O<sup>2</sup> reduction due to its structure [21]. The O2•- radicals and/or H2O<sup>2</sup> are scavenged with a decrease of the cathodic current which permits the quantification of the antioxidant capacity of different samples. In the Figure 1, the principle of detection using XOD based biosensor, using PB as mediator, is shown. *Antioxidants* **2021**, *10*, x FOR PEER REVIEW 5 of 14

**Figure 1.** Principle of detection using XOD based biosensor, using Prussian Blue (PB), as mediator. **Figure 1.** Principle of detection using XOD based biosensor, using Prussian Blue (PB), as mediator.

quite desirable, when the focus of the electrochemical system is the determination of antioxidant capacity since at potentials below −200 mV there is a reduction in molecular oxygen, that at potentials above 0 mV oxidation of antioxidant compounds occurs [12,22]. Figure 2 shows the cyclic voltammetry of printed electrodes of carbon paste, bare and modified, as well as the electrochemical response after immobilization of the enzyme xan-

thine oxidase.

It is important to note that a range of working potential between 0 and −200 mV is quite desirable, when the focus of the electrochemical system is the determination of antioxidant capacity since at potentials below −200 mV there is a reduction in molecular oxygen, that at potentials above 0 mV oxidation of antioxidant compounds occurs [12,22]. *Antioxidants* **2021**, *10*, x FOR PEER REVIEW 5 of 14

Figure 2 shows the cyclic voltammetry of printed electrodes of carbon paste, bare and modified, as well as the electrochemical response after immobilization of the enzyme xanthine oxidase.

There is a limitation for the printed carbon paste electrode in the cathodic region at potentials below −0.5 V and anodic above 0.5 V vs. Ag/AgCl, regions in which the supporting electrolyte is discharged. However, such behavior did not produce interference, as it is outside the desired potential window.

The use of Prussian blue (PB) as a mediator proved to be feasible for application in determining the antioxidant capacity, as it has a cathodic peak at −133 mV, within the desired potential window (0 to −200 mV). An increase in cathodic and anodic current was also observed with the immobilization of the enzyme, indicating that the electronic transfer process was favored, demonstrating an electrochemical affinity between the enzyme and the mediator. In addition, the incorporation of the PB mediator into the working electrode has been described as a simple, cost-effective, and highly stable process in acid and neutral media. **Figure 1.** Principle of detection using XOD based biosensor, using Prussian Blue (PB), as mediator. It is important to note that a range of working potential between 0 and −200 mV is quite desirable, when the focus of the electrochemical system is the determination of an-

The intensity of cathodic and anodic current obtained with the biosensors when MBRS was the mediator was greater, demonstrating that there is an electrochemical affinity between the enzyme and this mediator as well. In the case of CoPC, in addition to not having seen such an increase, there was also a negative shift in the cathodic peak potential. It is also noted that, in the region of interest, there is no electrocatalytic activity, making its use unfeasible in this study. tioxidant capacity since at potentials below −200 mV there is a reduction in molecular oxygen, that at potentials above 0 mV oxidation of antioxidant compounds occurs [12,22]. Figure 2 shows the cyclic voltammetry of printed electrodes of carbon paste, bare and modified, as well as the electrochemical response after immobilization of the enzyme xanthine oxidase.

**Figure 2.** *Cont*.

**Figure 2.** Cyclic voltamogram of (**a**) carbon paste electrode (SPE), (**b**) SPE modified with meldola blue with Reinecke salt (MBRS), (**c**) SPE modified with Cobalt phthalocyanine (CoPC), (**d**) SPE modified with PB and biosensors (**e**) without mediator and mediated with (**f**) MBRS, (**g**) CoPC, and (**h**) (PB) in K-PBS 50 mmol·L −1 pH 7.5. Scan rate 50 mV·s **Figure 2.** Cyclic voltamogram of (**a**) carbon paste electrode (SPE), (**b**) SPE modified with meldola blue with Reinecke salt (MBRS), (**c**) SPE modified with Cobalt phthalocyanine (CoPC), (**d**) SPE modified with PB and biosensors (**e**) without mediator and mediated with (**f**) MBRS, (**g**) CoPC, and (**h**) (PB) in K-PBS 50 mmol·L <sup>−</sup><sup>1</sup> pH 7.5. Scan rate 50 mV·s −1 .

−1 .

There is a limitation for the printed carbon paste electrode in the cathodic region at potentials below −0.5 V and anodic above 0.5 V vs. Ag/AgCl, regions in which the supporting electrolyte is discharged. However, such behavior did not produce interference, as it is outside the desired potential window. The use of Prussian blue (PB) as a mediator proved to be feasible for application in determining the antioxidant capacity, as it has a cathodic peak at −133 mV, within the In Figure 3, it can be seen that the electrochemical activity of PB is even more favored in an acidic environment. Due to the affinity of the enzyme with MBRS, its behavior at other pH was investigated, envisioning a possible application. However, considering the preservation of the enzyme activity, the medium with a pH close to neutrality was chosen as the working medium. *Antioxidants* **2021**, *10*, x FOR PEER REVIEW 7 of 14

Oxidation processes can be favored or compromised at specific pH levels, generating

It is also noteworthy that the catalytic properties of Prussian blue on the reduction of hydrogen peroxide are well known and have been discussed previously by several re-

The detection principle explored in the present work was based on the measurement of the H2O2 reduction current generated as a final product of the hypoxanthine (HX) oxidation reaction to uric acid, catalyzed by the XOD enzyme. The current was proportional to its concentration. Antioxidants inhibit such ROS, causing a decrease in the cathodic

In accordance with the methodology described in the experimental part, analytical curves were constructed (Figure 4), evaluating the signal obtained when using different polymerization times in the enzymatic immobilization step, and a calibration curve was

ity (higher peak currents). For the MBRS mediator, despite its high efficiency in the electronic transfer process, proven in the present study, the negative shift in the cathodic peak potential at pH 3.5 and pH 6.5 disadvantaged its use. In a more alkaline environment, there is a decrease in the cathodic current and the formation of ill-defined peaks, which

desired potential window (0 to −200 mV). An increase in cathodic and anodic current was

other pH was investigated, envisioning a possible application. However, considering the preservation of the enzyme activity, the medium with a pH close to neutrality was chosen as the working medium. **Figure 3.** Evaluation of the effect of pH on the electrochemical behavior of the biosensor modified with (**a**) MBRS and (**b**) PB in K-PBS buffer as a function of the mediator's immediate behavior. Scan rate: 50 mV·s −1 . **Figure 3.** Evaluation of the effect of pH on the electrochemical behavior of the biosensor modified with (**a**) MBRS and (**b**) PB in K-PBS buffer as a function of the mediator's immediate behavior. Scan rate: 50 mV·s −1 .

may be possibly caused by problems in electronic kinetics.

current, thus evaluating the antioxidant capacity.

*3.2. Biochemical Principles and Electrochemical Characterization of the Biosensor* 

searchers [23–25].

subsequently constructed.

Oxidation processes can be favored or compromised at specific pH levels, generating different oxidation and reduction potentials (greater or lesser), as well as greater sensitivity (higher peak currents). For the MBRS mediator, despite its high efficiency in the electronic transfer process, proven in the present study, the negative shift in the cathodic peak potential at pH 3.5 and pH 6.5 disadvantaged its use. In a more alkaline environment, there is a decrease in the cathodic current and the formation of ill-defined peaks, which may be possibly caused by problems in electronic kinetics. there is a decrease in the cathodic current and the formation of ill-defined peaks, which may be possibly caused by problems in electronic kinetics. It is also noteworthy that the catalytic properties of Prussian blue on the reduction of hydrogen peroxide are well known and have been discussed previously by several researchers [23–25].

−1 .

Oxidation processes can be favored or compromised at specific pH levels, generating

different oxidation and reduction potentials (greater or lesser), as well as greater sensitivity (higher peak currents). For the MBRS mediator, despite its high efficiency in the electronic transfer process, proven in the present study, the negative shift in the cathodic peak potential at pH 3.5 and pH 6.5 disadvantaged its use. In a more alkaline environment,

It is also noteworthy that the catalytic properties of Prussian blue on the reduction of hydrogen peroxide are well known and have been discussed previously by several researchers [23–25]. *3.2. Biochemical Principles and Electrochemical Characterization of the Biosensor*  The detection principle explored in the present work was based on the measurement

#### *3.2. Biochemical Principles and Electrochemical Characterization of the Biosensor* of the H2O2 reduction current generated as a final product of the hypoxanthine (HX) oxidation reaction to uric acid, catalyzed by the XOD enzyme. The current was proportional

*Antioxidants* **2021**, *10*, x FOR PEER REVIEW 7 of 14

**Figure 3.** Evaluation of the effect of pH on the electrochemical behavior of the biosensor modified with (**a**) MBRS and (**b**)

PB in K-PBS buffer as a function of the mediator's immediate behavior. Scan rate: 50 mV·s

The detection principle explored in the present work was based on the measurement of the H2O<sup>2</sup> reduction current generated as a final product of the hypoxanthine (HX) oxidation reaction to uric acid, catalyzed by the XOD enzyme. The current was proportional to its concentration. Antioxidants inhibit such ROS, causing a decrease in the cathodic current, thus evaluating the antioxidant capacity. to its concentration. Antioxidants inhibit such ROS, causing a decrease in the cathodic current, thus evaluating the antioxidant capacity. In accordance with the methodology described in the experimental part, analytical curves were constructed (Figure 4), evaluating the signal obtained when using different

In accordance with the methodology described in the experimental part, analytical curves were constructed (Figure 4), evaluating the signal obtained when using different polymerization times in the enzymatic immobilization step, and a calibration curve was subsequently constructed. polymerization times in the enzymatic immobilization step, and a calibration curve was subsequently constructed.

**Figure 4.** (**a**) Response of the amperometric biosensor with the successive addition of HX under constant agitation of 300 rpm, with the PB as a mediator and 8 mU of immobilized enzyme load, with polymerization times of 30 and 60 min. E = −100 mV. (**b**) Calibration curve of the amperometric biosensor modified with PB, when polymerization time of 60 min is used.

The reduction of hydrogen peroxide was more favored when a 60 min enzyme polymerization time was used in the graphite network, under neon radiation, becoming evident that the degree of polymerization depends on the time of exposure to neon light and interferes with enzyme retention in polymer and permeability of substrate and enzyme reaction products.

The agitation conditions also interfered with the biosensor response (Figure 5); therefore, the best conditions for carrying out the measurements were determined.

is used.

*Antioxidants* **2021**, *10*, x FOR PEER REVIEW 8 of 14

**Figure 4.** (**a**) Response of the amperometric biosensor with the successive addition of HX under constant agitation of 300 rpm, with the PB as a mediator and 8 mU of immobilized enzyme load, with polymerization times of 30 and 60 minutes. E = − 100 mV. (**b**) Calibration curve of the amperometric biosensor modified with PB, when polymerization time of 60 min

zyme reaction products.

The reduction of hydrogen peroxide was more favored when a 60 min enzyme polymerization time was used in the graphite network, under neon radiation, becoming evident that the degree of polymerization depends on the time of exposure to neon light and interferes with enzyme retention in polymer and permeability of substrate and en-

The agitation conditions also interfered with the biosensor response (Figure 5); there-

fore, the best conditions for carrying out the measurements were determined.

**Figure 5.** Curves obtained under different conditions: (**a**) 15 s of agitation, 15 s of rest, and 75 s measuring the current, fixed the time of 60 minutes of polymerization, (**b**) 30 s of agitation, 15 s of rest, and 60 s measuring the current, fixed the time of 60 minutes of polymerization, and (**c**) 15 s agitation, 15 s rest, 60 s current measurement, fixed the time of 60 minutes of polymerization. **Figure 5.** Curves obtained under different conditions: (**a**) 15 s of agitation, 15 s of rest, and 75 s measuring the current, fixed the time of 60 min of polymerization, (**b**) 30 s of agitation, 15 s of rest, and 60 s measuring the current, fixed the time of 60 min of polymerization, and (**c**) 15 s agitation, 15 s rest, 60 s current measurement, fixed the time of 60 min of polymerization.

**Equation R<sup>2</sup>**

**Table 1.** shows the analytical efficiency of the biosensor in terms of R<sup>2</sup> , sensitivity, and linear range, under constant or controlled agitation.  **Sensitivity Linear Range**  From an analytical point of view, a greater linear range of work allows numerous alternatives for using the prototype, due to the variability of antioxidant capacities Table 1. Linear range and similar sensitivities, in amperometric biosensors, using printed electrodes modified with PB were achieved, fixing the work potential at −100 mV [25,26].

**(µmol·L −1)**  I 1=−137.91+(−9.98) × [HX] 0.994 −9.98 0–50 **Table 1.** Shows the analytical efficiency of the biosensor in terms of R<sup>2</sup> , sensitivity, and linear range, under constant or controlled agitation.


<sup>1</sup> Constant agitation (300 rpm); <sup>2</sup> 15 s agitation, 15 s rest, 75 s current measurement; <sup>3</sup> 30 s agitation, 15 s rest, 60 s current measurement; <sup>4</sup> 15 s agitation, 15 s rest, 60 s current measurement. A new chronoamperometric run was performed at each substrate concentration.

The operating conditions were fixed at 15 s of agitation, 15 s of rest, and 60 s of current measurement and detection and quantification limits were set at 4.93 µmol·L −1 and 16.43 µmol·L −1 , respectively. They were calculated by the ratio between the average of

10 blanks measurements and an inclination of the analytical curve, multiplied by a factor of 3 and 10, respectively [12]. *3.3. Determination of the Antioxidant Capacity of Commercial Samples* 

From an analytical point of view, a greater linear range of work allows numerous alternatives for using the prototype, due to the variability of antioxidant capacities Table 1. Linear range and similar sensitivities, in amperometric biosensors, using printed electrodes modified with PB were achieved, fixing the work potential at −100 mV [25,26].

The operating conditions were fixed at 15 s of agitation, 15 s of rest, and 60 s of current measurement and detection and quantification limits were set at 4.93 µmol·L−1 and 16.43 µmol·L−1, respectively. They were calculated by the ratio between the average of 10 blanks measurements and an inclination of the analytical curve, multiplied by a factor of 3 and

#### *3.3. Determination of the Antioxidant Capacity of Commercial Samples* 3.3.1. Tea Samples

*Antioxidants* **2021**, *10*, x FOR PEER REVIEW 9 of 14

#### 3.3.1. Tea Samples As shown in Figure 6a, the initial currents recorded for the samples of chamomile tea

10, respectively [12].

As shown in Figure 6a, the initial currents recorded for the samples of chamomile tea and cimegripre® tea, were discrepant in relation to that obtained in the absence of antioxidant and other samples. This may be the result of the different resistivity of the medium depending on the composition of the samples. However, it is noticeable in all cases the decrease in the cathodic current resulting from the reduction of H2O<sup>2</sup> when compared to that obtained in the absence of the samples, resulting in smaller angular coefficients. Based on this difference, the total antioxidant capacity of the teas was calculated and the result shown in Figure 5b. and cimegripre® tea, were discrepant in relation to that obtained in the absence of antioxidant and other samples. This may be the result of the different resistivity of the medium depending on the composition of the samples. However, it is noticeable in all cases the decrease in the cathodic current resulting from the reduction of H2O2 when compared to that obtained in the absence of the samples, resulting in smaller angular coefficients. Based on this difference, the total antioxidant capacity of the teas was calculated and the result shown in Figure 5b.

**Figure 6.** (**a**) Response of the biosensor in the presence of the antioxidant samples. (**b**)Antioxidant capacity. **Figure 6.** (**a**) Response of the biosensor in the presence of the antioxidant samples. (**b**)Antioxidant capacity.

The wide variety of antioxidants, which may respond differently to the different oxidizing sources, makes it difficult to develop a single, simple, and universal method to assess antioxidant capacity, which is why different methods are generally used to charac-The wide variety of antioxidants, which may respond differently to the different oxidizing sources, makes it difficult to develop a single, simple, and universal method to assess antioxidant capacity, which is why different methods are generally used to characterize a sample [27].

terize a sample [27]. The antioxidant capacity in the analyzed ones followed the order: chamomile > mint > fennel > cimegripe. This result is in agreement with that obtained by Nakamura et al. [28] when investigating the total antioxidant capacity of the infusions of chamomile, mint, and fennel teas by the CUPRAC (Cupric Reducing Antioxidant Capacity) method, based on the reduction of Cu2+ to Cu1+ when certain reducing agents are present in the medium, The antioxidant capacity in the analyzed ones followed the order: chamomile > mint > fennel > cimegripe. This result is in agreement with that obtained by Nakamura et al. [28] when investigating the total antioxidant capacity of the infusions of chamomile, mint, and fennel teas by the CUPRAC (Cupric Reducing Antioxidant Capacity) method, based on the reduction of Cu2+ to Cu1+ when certain reducing agents are present in the medium, forming a complex of Cu1+/NC of intense color with maximum absorption at 454 nm, using the chromogenic reagent. This corroborates the applicability of the biosensor.

forming a complex of Cu1+/NC of intense color with maximum absorption at 454 nm, using the chromogenic reagent. This corroborates the applicability of the biosensor. The importance of the study could be related also to the connection of these plants The importance of the study could be related also to the connection of these plants with their neuroprotective properties. In many countries, traditional herbal medicines are used to prevent or treat neurodegenerative disorders, and some have been developed as nutraceuticals or functional foods [29,30].

with their neuroprotective properties. In many countries, traditional herbal medicines are used to prevent or treat neurodegenerative disorders, and some have been developed as nutraceuticals or functional foods [29,30]. Fennel (*Foeniculum vulgare Mill*.) is a herbal that has antioxidant properties, with effects for prevention and treatment of stress-induced neurological disorders [31]. Fennel Fennel (*Foeniculum vulgare Mill*.) is a herbal that has antioxidant properties, with effects for prevention and treatment of stress-induced neurological disorders [31]. Fennel oil and trans-anethole, the main component of fennel oil, significantly inhibit SOCE-induced [Ca2+] increase in vascular endothelial cells and that these reactions may be mediated by NSC, IP3-dependent Ca2+ mobilization, and PLC activation [32]. Several clinical studies indicated fennel for its therapeutic potential to minimize neuronal toxicity by normalizing the expression levels of APP isoforms and oxidative stress markers [33]. Efficacy of oral fennel oil in the management of dysmenorrhea, premenstrual syndrome, amenorrhea,

menopause, lactation, and polycystic ovary syndrome were confirmed according to results of clinical studies [34].

*Matricaria chamomilla* L. (chamomile) extract may produce clinically meaningful antidepressant effects in addition to its anxiolytic activity in subjects with a generalized anxiety disorder (GAD) and comorbid depression [35]. Chamomile had moderate antioxidant and antimicrobial activities, and significant antiplatelet activity in vitro. Animal model studies indicate potent anti-inflammatory action, some antimutagenic and cholesterol-lowering activities, as well as anti-spasmotic and anxiolytic effects [36].

Mints are aromatic plants traditionally used as a remedy and as culinary herbs. Methanolic extracts of Mentha x piperita and Mentha aquatica produced significant (*p* < 0.05) protection of the PC12 cells against oxidative stress. There were observed antioxidant and MAO-A inhibitory properties, M. x piperita being the most active. M. aquatica showed the highest affinity to the GABA(A)-receptor assay [37]. Its beneficial effect on the central nervous system as a neuroprotective potential, for example, has been explored. In addition, it targets multiple Alzheimer's disease events [38].

Cimegripe® is a mixture of paracetamol, clorfeniramina, and fenilefrina, with oral adult usage, for the relief of nasal congestion, runny nose, fever, and body pain present in flu-like states [39]. In the form of tea, its main ingredient is paracetamol. This medicine induces drowsiness, so it should not be used by vehicle drivers, machine operators, or those whose attention depends on the safety of others.

#### 3.3.2. Commercial Sources of Vitamin C

The enzymatic biosensor was used to determine the antioxidant capacity, not only of ascorbic acid (pure standard), but also of commercial effervescent vitamin C formulations.

Vitamin C provides protection against uncontrolled oxidation in the aqueous medium of the cell, due to its high reducing power, being a water-soluble and thermolabile vitamin. Humans and other primates, as well as the guinea pig, are the only mammals unable to synthesize it, requiring its administration through feeding or artificial supplementation [40].

Figure 7a shows the analytical curves constructed in the presence of different concentrations of ascorbic acid, in increasing concentrations of the HX substrate. By fixing the ascorbic acid concentration at 50 µmol·L −1 , greater sensitivity was obtained and, in higher concentrations, a considerable loss of it (Figure 7b). A similar behavior was observed when samples of effervescent vitamin C were used as antioxidants in the system. For comparative purposes, the concentration of 0.53 mmol·L <sup>−</sup><sup>1</sup> of vitamin C was fixed, according to manufacturers' specifications, and the antioxidant capacity of the effervescent vitamin C samples was determined (Figure 7c,d).

According to the manufacturers, each tablet of the effervescent contained 1 g of pure ascorbic acid. Thus, by weighing the equivalent quantity of each product, in order to make a final concentration of ascorbic acid equal to the three brands, it was expected that the results found for the antioxidant capacity were the same or very close. However, these were quite different from each other. The actual vitamin C content in the formulations may be the cause of the discrepancies found.

The composition of the samples was investigated by infrared spectroscopy and their spectra were similar, demonstrating that there are no significant differences between them. The content of ascorbic acid in the samples was determined by the standard addition method. Although samples 1 and 3 exhibited different antioxidant capacities, both contained the same content of ascorbic acid. This shows that the antioxidant capacity of the drugs is influenced by other compounds present in them. Some studies have shown a negative correlation between vitamin C content and antioxidant capacity [41,42]. Therefore, the individual and combined contribution of each component of the sample can be studied in the future.

The results of this study are in accordance with other published data using differential pulse voltammetry for vitamin C detection in pharmaceutical samples [43].

*Antioxidants* **2021**, *10*, x FOR PEER REVIEW 11 of 14

**Figure 7.** (**a**) Evaluation of the antioxidant capacity of ascorbic acid (standard) using the amperometric biosensor. (**b**) Sensitivity curve, in terms of the slopes of the curves showed in (**a**). (**c**) Response of the biosensor in the presence of the antioxidant samples. (**d**) Antioxidant capacity of drugs containing ascorbic acid against the biosensor. **Figure 7.** (**a**) Evaluation of the antioxidant capacity of ascorbic acid (standard) using the amperometric biosensor. (**b**) Sensitivity curve, in terms of the slopes of the curves showed in (**a**). (**c**) Response of the biosensor in the presence of the antioxidant samples. (**d**) Antioxidant capacity of drugs containing ascorbic acid against the biosensor.

#### *3.4. Analytical Stability of the Amperometric Biosensor*

According to the manufacturers, each tablet of the effervescent contained 1 g of pure ascorbic acid. Thus, by weighing the equivalent quantity of each product, in order to make a final concentration of ascorbic acid equal to the three brands, it was expected that the results found for the antioxidant capacity were the same or very close. However, these Adequate quality control is necessary to obtain reliable results. The use of control charts can be an effective strategy to ensure that there has been no change in a particular process over time. It helps to detect variations outside a statistically acceptable standard, making it possible to correct them.

were quite different from each other. The actual vitamin C content in the formulations may be the cause of the discrepancies found. The composition of the samples was investigated by infrared spectroscopy and their Figure 8 shows the statistical control chart, built from chronoamperometric measurements, obtained with the same biosensor, in a medium containing 9.9 µmol·L <sup>−</sup><sup>1</sup> HX, in the absence of antioxidants, over the course of 7 months. The biosensor proved to be statistically stable in this period, with the control lines not being exceeded once.

tial pulse voltammetry for vitamin C detection in pharmaceutical samples [43].

Adequate quality control is necessary to obtain reliable results. The use of control

Figure 8 shows the statistical control chart, built from chronoamperometric measurements, obtained with the same biosensor, in a medium containing 9.9 µmol·L−1 HX, in the

charts can be an effective strategy to ensure that there has been no change in a particular process over time. It helps to detect variations outside a statistically acceptable standard,

spectra were similar, demonstrating that there are no significant differences between them. The content of ascorbic acid in the samples was determined by the standard addition method. Although samples 1 and 3 exhibited different antioxidant capacities, both contained the same content of ascorbic acid. This shows that the antioxidant capacity of the drugs is influenced by other compounds present in them. Some studies have shown a negative correlation between vitamin C content and antioxidant capacity [41,42]. Therefore, the individual and combined contribution of each component of the sample

*3.4. Analytical Stability of the Amperometric Biosensor* 

can be studied in the future.

making it possible to correct them.

*Antioxidants* **2021**, *10*, x FOR PEER REVIEW 12 of 14

absence of antioxidants, over the course of 7 months. The biosensor proved to be statisti-

cally stable in this period, with the control lines not being exceeded once.

**Figure 8**. Statistical control chart for monitoring the biosensor stability. **Figure 8.** Statistical control chart for monitoring the biosensor stability.

#### **4. Conclusions 4. Conclusions**

The use of silk-screened sensors, whose working electrode contains graphite and the Prussian blue mediator (PB) proved to be efficient in determining the antioxidant capacity when the target molecule is hydrogen peroxide. This was possible due to the catalytic properties of this modifier, at the time of its reduction. The use of Meldola blue combined with Reinecke salt (MBRS) was not effective in the studied conditions. However, the possibility of future applications should not be ruled out, due to its high potential in electron transfer processes. The use of silk-screened sensors, whose working electrode contains graphite and the Prussian blue mediator (PB) proved to be efficient in determining the antioxidant capacity when the target molecule is hydrogen peroxide. This was possible due to the catalytic properties of this modifier, at the time of its reduction. The use of Meldola blue combined with Reinecke salt (MBRS) was not effective in the studied conditions. However, the possibility of future applications should not be ruled out, due to its high potential in electron transfer processes.

The method of immobilization by occlusion/entrapment of the enzyme xanthine oxidase (XOD) in polymeric PVA-AWE film on the surface of the carbon paste electrode modified with PB proved to be simple and efficient. However, the polymerization conditions must be properly controlled because a small variation in the polymerization time has considerably affected the analytical response of the biosensor The method of immobilization by occlusion/entrapment of the enzyme xanthine oxidase (XOD) in polymeric PVA-AWE film on the surface of the carbon paste electrode modified with PB proved to be simple and efficient. However, the polymerization conditions must be properly controlled because a small variation in the polymerization time has considerably affected the analytical response of the biosensor

The elimination of the constant agitation process, during chronoamperometric measurements, for example, resulted in an increase in the linear range reached by the device. If higher levels of enzyme activity and substrate content are used, but always below the level of kinetic saturation, it is believed that the linear region can be further expanded. The elimination of the constant agitation process, during chronoamperometric measurements, for example, resulted in an increase in the linear range reached by the device. If higher levels of enzyme activity and substrate content are used, but always below the level of kinetic saturation, it is believed that the linear region can be further expanded.

The biosensor showed high stability and showed promise in determining the antioxidant capacity of teas and/or drugs that are sources of vitamin C and can also be used to detect fraud. Its use can be expanded to assess the antioxidant potential of fresh or processed foods and also to control different products used for their neuroprotective effects. The developed biosensors could be used as possible tools to the monitoring of reac-The biosensor showed high stability and showed promise in determining the antioxidant capacity of teas and/or drugs that are sources of vitamin C and can also be used to detect fraud. Its use can be expanded to assess the antioxidant potential of fresh or processed foods and also to control different products used for their neuroprotective effects.

tive oxygen species, the free radicals in different samples, as plants extracts, drugs, or The developed biosensors could be used as possible tools to the monitoring of reactive oxygen species, the free radicals in different samples, as plants extracts, drugs, or other biological liquids, aiming to obtain a correlation between the index obtained from these indicators with the oxidative stress levels in the samples.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2076-392 1/10/2/324/s1, Figure S1: Distribution of the references concerning antioxidant capacity detection, according with publication years—since 1997; Figure S2: Distribution of the references concerning antioxidant capacity detection, according with document type—since 1997; Figure S3: Distribution of

the references concerning antioxidant capacity detection, according with source titles—since 1997; Figure S4: Distribution of the references concerning antioxidant capacity detection, according research area—since 1997; Figure S5: H-index related to the ISI WOS references dealing with antioxidant capacity detection—since 1997.

**Author Contributions:** Conceptualization, D.B.R. and G.S.N.; Data collection and experimentation, D.B.R., G.S.S. and A.R.C.C.; software and data curation, D.B.R., D.R.d.S. and E.B.R.; writing—original draft preparation, D.B.R.; writing—review and editing, M.B. and G.S.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES)-Finance Code 001.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

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

#### **References**


## *Article* **Autoxidation Enhances Anti-Amyloid Potential of Flavone Derivatives**

**Andrius Sakalauskas , Mantas Ziaunys, Ruta Snieckute and Vytautas Smirnovas \***

Life Sciences Center, Institute of Biotechnology, Vilnius University, LT-10257 Vilnius, Lithuania; andrius.sakalauskas@gmc.vu.lt (A.S.); mantas.ziaunys@gmail.com (M.Z.); r.snieckute@gmail.com (R.S.) **\*** Correspondence: vytautas.smirnovas@bti.vu.lt

**Abstract:** The increasing prevalence of amyloid-related disorders, such as Alzheimer's or Parkinson's disease, raises the need for effective anti-amyloid drugs. It has been shown on numerous occasions that flavones, a group of naturally occurring anti-oxidants, can impact the aggregation process of several amyloidogenic proteins and peptides, including amyloid-beta. Due to flavone autoxidation at neutral pH, it is uncertain if the effective inhibitor is the initial molecule or a product of this reaction, as many anti-amyloid assays attempt to mimic physiological conditions. In this work, we examine the aggregation-inhibiting properties of flavones before and after they are oxidized. The oxidation of flavones was monitored by measuring the UV-vis absorbance spectrum change over time. The protein aggregation kinetics were followed by measuring the amyloidophilic dye thioflavin-T (ThT) fluorescence intensity change. Atomic force microscopy was employed to image the aggregates formed with the most prominent inhibitors. We demonstrate that flavones, which undergo autoxidation, have a far greater potency at inhibiting the aggregation of both the disease-related amyloid-beta, as well as a model amyloidogenic protein—insulin. Oxidized 6,2′ ,3′ -trihydroxyflavone was the most potent inhibitor affecting both insulin (7-fold inhibition) and amyloid-beta (2-fold inhibition). We also show that this tendency to autoxidize is related to the positions of the flavone hydroxyl groups.

**Keywords:** aggregation; amyloid-beta; insulin; flavones; inhibition; autoxidation

#### **1. Introduction**

Protein aggregation into highly structured aggregates is associated with various amyloidoses, such as Alzheimer's disease (AD) and Parkinson's disease (PD) [1]. AD alone is recognized to be the most common cause of dementia (60–80%) [2] that affects more than 50 million people worldwide and, according to the World Alzheimer's Report, is set to increase up to 152 million by 2050. The cause of this forecast is that the onset of AD mostly occurs after 60 years of age, and the increasing life expectancy leads to more people suffering from dementia. The pathological hallmark of this disease is the increased concentration of the 42 amino acid peptide—amyloid-beta (Aβ42) that prompts the formation of its oligomeric and fibrillar species [3].

The increasing focus on anti-amyloid-β compounds has led to many different in vitro studies showing positive effects against protein aggregation [4]. Despite this fact, many suggested disease-modifying compounds, ranging from small organic molecules to large monoclonal antibodies, have not led to an effective cure, leaving 99.5% of clinical trials unsuccessful [5,6]. Several potential problems with the very low clinical trial success rate are linked to targeting the wrong pathological substrates, concerns with drug development, and problems with methodologies [7,8]. Subsequently, it is of utmost importance to take into consideration the gap between the initial drug screening and human physiology [4,9].

The aggregation process of the Aβ<sup>42</sup> peptide is exceptionally complicated; however, the mechanism is rather well described [10,11]. The process of several steps involves

**Citation:** Sakalauskas, A.; Ziaunys, M.; Snieckute, R.; Smirnovas, V. Autoxidation Enhances Anti-Amyloid Potential of Flavone Derivatives. *Antioxidants* **2021**, *10*, 1428. https:// doi.org/10.3390/antiox10091428

Academic Editors: Rui F. M. Silva and Lea Pogaˇcnik

Received: 9 August 2021 Accepted: 1 September 2021 Published: 7 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

primary nucleation, elongation, fibril surface-catalyzed nucleation (often referred to as secondary nucleation), and fragmentation [12]. While primary nucleation causes the formation of nuclei that eventually grow into fibril aggregates, secondary nucleation is shown to be the main source of more cytotoxic oligomeric species that cause direct neurotoxicity [13–15]. For that reason, it is beneficial to find an anti-amyloidogenic compound that prevents primary and secondary nucleation as well as elongation processes [16].

Flavones are abundant in nature and found in a variety of herbs, fruits, vegetables, and spices [17]. This group of natural anti-oxidants has been reported to possess antiamyloid characteristics, exhibit neuroprotective, anti-inflammatory, and anti-microbial properties [16,18]. In addition, flavone derivatives have shown positive effects when treating diabetes, cancer, malaria, asthma, and cardiovascular system diseases [19]. Studies have also shown that a variety of flavonoids function as acetylcholinesterase inhibitors (AChEI) [20,21]. AChEI is currently one of the most prominent options for symptomatic treatment of AD, mostly by increasing neurotransmitter acetylcholine concentrations in synaptic gaps of the nervous system [22,23]. If the same compound would also inhibit amyloid formation, it could be an ultimate anti-amyloid drug. Moreover, the small molecular weight and widely abundant flavonoids could be a better option for drug development. Compared to the large monoclonal antibody-based drugs, such molecules do pass Lipinski's rule of 5, have high availability and stability, and could potentially be used for less expensive prevention against the onset of neurodegenerative diseases [24].

Studies with flavones demonstrated properties against Aβ<sup>42</sup> aggregation in vitro [25,26]. In many cases, the anti-aggregation potential is evaluated via measurement of amyloidophilic dye thioflavin-T (ThT) fluorescence intensity [27,28], assuming that relatively lower fluorescence intensity correlates with fewer fibrils formed. While this hypothesis is quite prominent, various counterfactors exist. Typically, Aβ<sup>42</sup> aggregation is examined at neutral pH without evaluating the characteristics of the potential inhibitor in question. Numerous flavones have light absorbance properties in the same range as typically used fluorescent amyloid-dyes [29]. In addition, flavones could potentially bind to either the dye molecule itself or the formed aggregates, preventing its interaction with the fibril [30].

Many polyphenolic compounds, including flavones, are reported to undergo autoxidation at neutral or higher pH [31,32]. One particular study shows the oxidation mechanism of quercetin, suggesting that the process involves the breakdown of the flavone C ring, enabling different structure formations [32]. In another report, the Aβ<sup>42</sup> inhibitory effect is based on the autoxidation of (+)-taxifolin [28]. This leads to an assumption that molecule autoxidation could be the main cause of the inhibitory effect in vitro. Furthermore, several reports demonstrate low mono- and polyhydroxylated flavone oral bioavailability due to direct metabolism [33]. In addition, human cytochrome P450 enzymes oxidize the 5 hydroxyflavone to specific di- or trihydroxyflavones [34]. These aforementioned aspects raise questions about whether the tested molecule or its oxidized species inhibit amyloid formation in vitro.

In this work, we examined the oxidation potential of 64 mono- and polyhydroxylated flavones and tested their inhibitory effect on the aggregation of amyloid-beta and a commonly used model amyloid protein—insulin. We show that the positions of flavone hydroxy groups have a remarkably high impact on autoxidation which enables the inhibitory effect on both proteins under the tested conditions.

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

#### *2.1. Flavone Solution Preparation*

Each non-oxidized flavone stock solution was prepared by dissolving the flavones (Indofine Chemical Company, Inc., Hillsborough, NJ, USA) in dimethylsulfoxide (DMSO, Carl Roth, Karlsruhe, Germany) to a final concentration of 10 mM. The oxidation solution of each flavone was prepared by diluting 10 mM flavone stock solution with 10 mM sodium phosphate buffer (pH 8.0) and DMSO to yield a final flavone concentration of 0.2 mM in

9 mM sodium phosphate buffer solution containing 10% DMSO. The 10% DMSO buffer solution was used to increase the solubility of flavones.

#### *2.2. Absorbance Measurements*

The autoxidation of flavones was monitored by measuring UV-Vis absorbance spectrum changes over time using a ClarioStar Plus plate reader (BMG Labtech, Ortenberg, Germany). Each flavone oxidation solution was stored as 100 µL samples in a UV-clear 96-well plate (Thermo Fisher Scientific, Inc., Waltham, MA, USA, cat. No. 11670352) and incubated at 37 ◦C, while the measuring absorbance spectra were in the range from 240 nm to 800 nm. Data were collected each hour for a total of 100 h. Spectra was baseline corrected at 800 nm. The resulting samples, which are later referred to as "incubated" or "oxidized" flavones, were then used in aggregation kinetic experiments.

Samples for the measurement of ThT and flavone interaction were prepared by mixing 0.5 mM incubated flavone, 10 mM ThT stock solution, and 20 mM phosphate buffer solution (pH 7.0), yielding either separate 50 µM flavone and 20 µM ThT or combined 50 µM flavone and 20 µM ThT solutions in 20 mM phosphate buffer (pH 7.0). Samples were scanned using a Shimadzu UV-1800 spectrophotometer (1 nm steps). Separate 50 µM flavone and 20 µM ThT spectra were added together for comparison with their mixture. Each sample was scanned three times and averaged; the baseline was corrected at 800 nm.

#### *2.3. Fluorescence Measurements*

Samples for the fluorescence measurements were prepared by mixing 0.5 mM incubated flavone, 10 mM ThT stock solution, 2 µM of Aβ<sup>42</sup> aggregates, and 20 mM phosphate buffer solution (pH 7.0), yielding 1 µM of Aβ<sup>42</sup> fibril samples with either 20 µM ThT or 50 µM flavone samples with both ThT and flavone. The fluorescence intensity was scanned using a Varian Cary Eclipse fluorescence spectrophotometer, with excitation and emission wavelengths being 440 nm and 480 nm, respectively (5 nm excitation and 2.5 nm emission slit widths). The intrinsic fluorescence emission intensity, occurring from non-fibril-bound ThT or flavones, was subtracted from their respective fibril-compound sample intensities. This was done by acquiring fluorescence emission intensity values of ThT or flavone samples in the absence of Aβ<sup>42</sup> aggregates.

## *2.4. Purification of Recombinant Aβ<sup>42</sup>*

The expression vector of Aβ<sup>42</sup> was described previously [35]. The peptide was expressed in *E.coli* BL-21StarTM (DE3) (Invitrogen, Carlsbad, CA, USA) and purified as described previously [36]. In brief, the transformed cells were incubated on LB agar plates containing ampicillin (100 µg/mL) overnight at 37 ◦C. The next day, the overnight cultures were prepared from single colonies and grown in LB medium with ampicillin (100 µg/mL). The 1 mL of the culture was transferred to 400 mL of auto-inductive ZYM-5052 medium [37] containing ampicillin (100 µg/mL) and grown for 15 h. The collected cell pellet was washed 3 times to remove all soluble proteins. The procedure involves pellet homogenization, sonication, and centrifugation. After removing soluble proteins, the cell pellet was resuspended in 50 mL of 20 mM Tris/HCL pH 8.0 buffer solution containing 8 M urea and 1 mM EDTA, homogenized, and centrifuged as in the previous steps. The collected supernatant was diluted with 150 mL of 20 mM Tris/HCL (pH 8.0) buffer containing 1 mM EDTA, mixed with 60 mL DEAE-sepharose and agitated at 80 rpm for 30 min at 4 ◦C. The chromatography procedure was performed using a Buchner funnel with Fisherbrand glass microfiber paper on a vacuum glass bottle. The resin with bound proteins was washed with 20 mM Tris/HCL pH 8.0 buffer containing 1 mM EDTA in increasing NaCl concentrations in a step-gradient (0, 20, 150, 500 mM). The target protein fractions were collected by washing the resin with a 50 mL buffer solution (containing 150 mM NaCl) four times. Collected fractions were mixed together, lyophilized, and stored at −20 ◦C.

The Aβ<sup>42</sup> peptide powder was dissolved in a 20 mM sodium phosphate buffer solution (pH 8.0) containing 5 M guanidine thiocyanate (GuSCN, Carl Roth). The dissolved sample was loaded on a Tricorn 10/300 column (packed with Superdex 75) and eluted at 1 mL/min using a 20 mM sodium phosphate buffer solution (pH 8.0) containing 0.2 mM EDTA and 0.02% NaN3. Collected fractions were mixed together, lyophilized, and stored at −20 ◦C. Before aggregation experiments, the purification procedure was repeated, but this time the collected fraction (0.75 mL) was purified Aβ<sup>42</sup> was stored on ice for 5 min. The concentration was determined by calculating the integrated chromatographic UV absorbance peak (ε<sup>280</sup> = 1 490 M−<sup>1</sup> cm−<sup>1</sup> ). Afterward, it was diluted and immediately used for aggregation experiments.

#### *2.5. Aggregation Kinetics of Aβ<sup>42</sup> Peptide*

The purified peptide fraction (1.5 mL, pH 8.0) was mixed with 3 mL of 20 mM sodium phosphate buffer solution (pH 6.33) to yield a 3-fold diluted peptide solution (pH 7.0). The peptide and each oxidized or incubated flavone solution was mixed together with 20 mM sodium phosphate buffer solution (pH 7.0), 10 mM ThT stock solution, and DMSO to a final reaction mixture, containing 1 µM Aβ42, 20 µM ThT, 50 µM of selected flavone compound and 1% DMSO. The kinetic aggregation measurements were performed in nonbinding 96-well plates (Fisher, Waltham, MA, USA, cat. No. 10438082) (sample volume was 80 µL) at 37 ◦C by measuring ThT fluorescence using 440 nm excitation and 480 emission wavelengths in a ClarioStar Plus (BMG Labtech, Ortenberg, Germany).

#### *2.6. Aggregation Kinetics of Insulin*

Human recombinant insulin powder (Sigma-Aldrich, St. Louis, MO, USA, cat. No. 91077C) was dissolved in a 20% acetic acid solution (prepared from 100% acetic acid; Carl-Roth) containing 100 mM NaCl (Fisher) to a protein concentration of 400 µM. This insulin stock solution was mixed with non-oxidized/incubated or oxidized/incubated flavone solutions and 10 mM ThT stock solution to a final insulin concentration of 200 µM, 100 µM ThT, and 20 µM of each flavone. The aggregation kinetic measurements were performed similarly as in the case of Aβ42, but at 60 ◦C.

#### *2.7. Kinetic Data Analysis*

After reaching the plateau, kinetic aggregation curves were fit using Boltzmann's sigmoidal equation:

$$y = \frac{(A\_1 - A\_2)}{\left(1 + e^{\frac{x - x\_0}{dx}}\right)} + A\_2 \tag{1}$$

where, *A*<sup>1</sup> is the starting fluorescence intensity, *A*2—final fluorescence intensity, *x*0 aggregation halftime. The relative halftime and relative ThT fluorescence intensity values were calculated based on the control sample in their specific microplate. These values were calculated by dividing each sample's average value by the average control value. Data were processed using Origin software (OriginLab, Northampton, MA, USA).

#### *2.8. Atomic Force Microscopy (AFM)*

The samples for AFM images were collected after kinetic measurements and scanned similarly as previously described [31,38]. In short, 40 µL of 1% (*v/v*) APTES (Sigma-Aldrich, cat. No. 440140) in MilliQ water was deposited on freshly cleaved mica and incubated for 5 min. Then, mica was rinsed with 2 mL of MilliQ water and dried under gentle airflow. Each sample was deposited (40 µL) on the functionalized surface and incubated for another 5 min. Prepared samples were rinsed with 2 mL of MilliQ water and dried under gentle airflow. AFM imaging was performed using a Dimension Icon (Bruker, Billerica, MA, USA) atomic force microscope. Images were 1024 × 1024 pixel resolution and were analyzed using Gwyddion 2.5.5 software. Fibril heights were determined by tracing perpendicular to each fibril's axis.

#### *2.9. FTIR*

Aβ<sup>42</sup> fibrils were separated from the buffer solution by placing the mixture in the 0.5 mL 10 kDa concentrators (Fisher, cat. No. 88513) and spinning at 10,000 g for 10 min. Then 0.5 mL of D2O was added, and the process of buffer exchange to D2O was repeated 3 times. After the last spinning step, fibrils were resuspended in 0.1 mL of D2O. FTIR spectra were recorded using an Invenio S IR spectrophotometer equipped with an MCT detector. The sample was placed in the CaF<sup>2</sup> transmission windows with 0.05 mm Teflon spacers, 256 interferograms of 2 cm−<sup>1</sup> resolution were averaged per spectrum. All spectra were normalized in the 1705–1595 cm−<sup>1</sup> region, and baseline corrected after subtracting the D2O and water vapor spectrums. The data were processed using GRAMS software (Thermo Fisher Scientific, Inc., Waltham, MA, USA).

#### **3. Results**

We first incubated flavones at 37 ◦C in order to evaluate potential structural transitions that occur due to autoxidation. The time-dependent changes in the UV-vis spectra of flavones were recorded over a period of 100 h, comparing the absorbance in the 240–800 nm region. At the start of the experiment, each flavone spectrum (Figure 1) exhibited two characteristic maxima that are associated with the π → π\* transitions within rings A and C, referred to as benzoyl system, band II (~240–290 nm), and ring B that is conjugated with the carbonyl of ring C, referred to as cinnamoyl system, band I (~300–415 nm) [39] (Figure S1). A decrease in the magnitude of these bands was observed in all displayed spectra that led to no characteristic maxima (No. 11, 22, 31, 38, 44, 46, 48, 51–52, 57, 59, 64) or appearance of new maxima peaks in other cases. The absorbance spectra changes and reduced characteristics of the band I indicate structural changes, loss of conjugation in a chromophore, and development of different intra- and intermolecular interactions [40]. A few trihydroxyflavones (THF) (No. 38, 46), tetrahydroxyflavones (TeHF) (51–52), and most of penta- and hexahydroxyflavones (PHF and HHF) (No. 59, 61, 63, 64) had major spectrum changes within the first 5 h. Most of the other flavones, including dihydroxyflavones (DHF), THF, and TeHF (No.10, 11, 22, 31, 38, 42, 44, 46, 48, 53, 55, 57), had significant absorbance changes within a 5–40 h period, while only a few (No. 1, 32, 37, 58) exhibited most of their spectrum transitions only after > 40 h of incubation. The rest of the flavones had minor spectra changes during incubation that are reflected in slight transitions of the maxima positions (No. 30, 45, 54, 56, 60, 62) or a decrease in the magnitude of the maxima in the 380–420 nm region (No. 21, 43).

Examining the effect of non-oxidized flavones reveals that only the presence of luteolin (Figure 2A,B No. 56) slightly increased the aggregation halftime of insulin (Table S1) while not affecting the fluorescence intensity. Other flavone relative halftime and ThT fluorescence intensity did not change, except for a few cases, where they even decreased the aggregation halftime (Figure 2A No. 10, 21–22, 48, 53, 54, 59, 61–63). However, once flavones were oxidized, many of them displayed substantial inhibitory potential. Some flavones (Figure 2A,B No. 31, 59, 63) increased the aggregation halftime more than fivefold, which correlates with the ten-fold elevated fluorescence intensity (compared to the control sample). In most cases, oxidized flavones inhibited insulin aggregation, except for a few (Figure 2A,B No. 1, 30, 32, 37, 46, 54–55, 58) that did not possess such properties, as neither ThT fluorescence intensity nor halftime changed compared to the previously tested non-oxidized forms. A completely different effect was seen on Aβ<sup>42</sup> aggregation. Here, the fluorescence intensity (Figure 2D) was diminished in all cases, except for four flavones (Figure 2D No. 1, 30, 32, 37) which seem to have had no impact on either protein aggregation process, while several oxidized compounds (Figure 2D No. 22, 31, 52, 59) showed reduced intensity values ranging from 93% to 98%, which also reduced the aggregation rate. Despite the fact that most oxidized flavones inhibited insulin aggregation, only thirteen (Figure 2C No. 22, 31, 38, 43, 46, 48, 51, 52, 56–57, 59–60, 63) appeared to increase Aβ<sup>42</sup> relative halftime and only three (Figure 2C No. 22, 31, 52) slowed the aggregation by at least 50%.

**Figure 1.** UV-visible absorbance spectra of flavones, recorded at 0 h (black), 5 h (red), 40 h (blue), and 100 h (green). Spectra were baseline corrected at 800 nm. Most of the flavone spectra experienced a significant change in the 250–450 nm region. In contrast, 21, 30, 43, 45, 54, 56, 60, and 62 experienced only a slight transition of maxima or decrease in the magnitude of the initial absorbance spectrum.

**Figure 2.** Effects of non-oxidized and oxidized flavones on insulin aggregation kinetics (**A**) and relative ThT fluorescence intensity (**B)**. Effect of oxidized flavones on Aβ<sup>42</sup> aggregation kinetics (**C**) and relative ThT fluorescence intensity (**D**). Error bars are for one standard deviation (*n* = 4). None of the non-oxidized flavones, except 56, inhibited insulin aggregation; after the oxidation, more than half of the flavones showed an inhibitory effect, with 31, 59, and 63 having the most significant impact. Oxidized flavones 22, 31, 52, and 59 increased the relative halftime of Aβ<sup>42</sup> the most, while 1, 30, 32, 37 did not affect the relative halftime nor the relative ThT fluorescence intensity.

The flavone autoxidation experiment described above allowed us to evaluate the effect of oxidized flavones on protein aggregation. Nevertheless, not all compounds may undergo structural changes in the reaction mixture; thus, an additional number of flavones were incubated at the experimental conditions to evaluate whether UV-vis spectrum changes occur. Every tested flavone maintained the absorbance of Band I and Band II, with no major changes in the tested region (Figure 3). However, spectra of many compounds exhibited intensity changes with no shape or maximum transitions (No. 3, 6, 8, 9, 12, 13, 14, 19, 23, 26, 33, 47) that may be related to the solubility of each molecule, especially when the change occurred between the first two scans.

**Figure 3.** UV-visible absorbance spectra of flavones, recorded at 0 h (black), 5 h (red), 40 h (blue), and 100 h (green). Spectra were baseline corrected at 800 nm. Numbers 3, 6, 13, 14, 19, 23, 33 experienced the most significant decrease in the magnitude of the spectrum, while 4, 18, 27, 34, 36 had no notable change over the course of the experiment.

An identical experiment was conducted with the second set of flavones to evaluate their influence on insulin (Figure 4A,B) and Aβ<sup>42</sup> (Figure 4C,D) aggregation processes. Here, similar results were observed, where most of the non-incubated and incubated flavones did not inhibit insulin aggregation, yet some increased its rate (Figure 4A No. 2, 3, 7, 8, 9, 12, 13, 15, 16, 18, 19, 20, 49, 50). The majority of flavones did not affect Aβ<sup>42</sup> aggregation as well. However, a significant decrease in ThT fluorescence intensity was mostly evident for flavones with a higher number (Figure 4D No. 34–35, 41, 49–50), which represents THF and TeHF. In addition, dihydroxyflavones did not reduce the intensity value, except for no. 15. Three flavones (Figure 4C,D No. 5, 14, 16) that stand out appear to have altered the aggregation process by increasing the ThT fluorescence intensity and decreasing Aβ<sup>42</sup> aggregation halftime.

**Figure 4.** Effects of non-incubated and incubated flavones on insulin aggregation kinetics (**A**) and relative ThT fluorescence intensity (**B**). Effect of incubated flavones on Aβ<sup>42</sup> aggregation kinetics (**C**) and relative ThT fluorescence intensity (**D**). Error bars are for one standard deviation (*n* = 4). The non-incubated and incubated flavones did not impact insulin and Aβ<sup>42</sup> relative halftime, while incubated flavones 34, 35, 41, and 50 had the most significant impact on the relative ThT fluorescence intensity of Aβ42.

Atomic force microscopy imaging was employed to observe whether fibrils were formed at the end of the Aβ<sup>42</sup> aggregation experiment (when plateau was reached). Five samples were tested that represented the control sample (Figure 5A,B) and Aβ<sup>42</sup> with incubated 2′ ,3′ -DHF (Figure 5C,D), 6,2′ ,3′ -THF (Figure 5E,F), 3,6,2′ ,3′ -TeHF (Figure 5G,H), 3,6,3′ ,4′ -TeHF (Figure 5I,J), 5,7,3′ ,4′ ,5′ -PHF (Figure 5K,L). These particular compounds were selected due to their high impact on Aβ<sup>42</sup> aggregation rate and bound-ThT. All samples with flavones revealed Aβ<sup>42</sup> fibrillar aggregates on the mica, despite the fact that the surface was mostly covered by round-shaped oligomeric, very short fibrillar structures. Samples with 2′ ,3′ -DHF, 6,2′ ,3′ -THF, 3,6,2′ ,3′ -TeHF, and 5,7,3′ ,4′ ,5′ -PHF (Figure 5C,F,G,K) appeared to have clumps of fibrils with round-shaped oligomeric structures attached to them, leaving the area empty around this structure. This suggests that inhibition requires the binding of an active molecule to the protein or its oligomeric/fibrillar species. In order to further analyze AFM images, we measured the height of a hundred oligomeric structures or fibrils and compared their height distribution (Figure 5M). Structures formed with inhibitors

had a dispersed height distribution, revealing that oligomeric structures may resemble clumped protofibrils. To understand this aspect more, the FTIR spectra of control Aβ<sup>42</sup> fibrils and the sample with 2′ ,3′ -DHF (when both samples reached a plateau in the ThT intensity) were recorded (Figure 5N). Samples for this experiment were prepared by using 10 kDa concentrator tubes that aided in changing the reaction solution to D2O. This method also eliminated monomeric species of amyloid-β. Notably, the FTIR spectrum of control fibrils exhibited the only major maximum at 1630 cm−<sup>1</sup> , typical for β-sheet structures, commonly found in amyloid fibrils, while the spectrum of Aβ<sup>42</sup> + 2′ ,3′ -DHF sample, had a less expressed β-sheet-related band at 1629 cm−<sup>1</sup> , and another broad peak at 1675 cm−<sup>1</sup> , which can mean the presence of substantial amounts of turns or different types of β-sheets. Unfortunately, the FTIR spectra could not be analyzed deeper; due to very low signal intensity, the signal-to-noise ratio was too high. It is necessary to note that, before spectra were normalized, the area of the amide I band of the sample with inhibitor was almost twice as small as the area of the amide I band of the control sample, leading to an assumption that less oligomeric and fibrillar species were present.

**Figure 5.** Atomic force microscopy images of Aβ<sup>42</sup> formed without (**A**,**B**) and with 50 µM of oxidized 2 ′ ,3′ -DHF (**C**,**D**), 6,2′ ,3′ -THF (**E**,**F**), 3,6,2′ ,3′ -TeHF (**G**,**H**), 3,6,3′ ,4′ -TeHF (**I**,**J**) and 5,7,3′ ,4′ ,5′ -PHF (**K**,**L**) flavones. Fibril and oligomeric species height distribution (**M**), where box plots indicate mean ± SD and error bars are in the 5%–95% range (*n* = 100). FTIR spectra (**N**) of Aβ<sup>42</sup> fibrils formed alone and with 50 µM of 2′ ,3′ -DHF. The AFM images of Aβ<sup>42</sup> aggregates formed with all inhibitors showed a similar distribution in height and revealed round shape structures that were not present in the image of the control sample. The FTIR spectrum of the sample with 2′ ,3′ -DHF had less expressed β-sheet-related band at 1629 cm−<sup>1</sup> than the control sample.

#### **4. Discussion**

The characteristics of insulin aggregation kinetic data show that 63 out of 64 tested non-oxidized flavones possess no anti-amyloid properties under the tested conditions (Table S1), while most flavones that undergo the autoxidation process slow down insulin fibril formation. This is expressed in altered relative aggregation halftime. However, compounds also change the ThT fluorescence intensity (Figure 2A,B), which can be explained based on our previous report, where we show that insulin is capable of forming distinct fibril conformations in 20% acetic acid solution, with one exhibiting ~10-fold higher bound-ThT intensity values [41]. Increased fluorescence intensity is also observed using oxidized gallic acid [38], leading to a hypothesis that oxidized flavones redirect insulin amyloid formation.

Contrary results are seen during the Aβ<sup>42</sup> aggregation process. Here, oxidized flavones led to a reduced ThT fluorescence intensity (Figures 2D and 4D), and only 14 oxidized flavones (Figures 2C and 4C) affected the aggregation rate. These diverse results introduce several potential explanations which may act simultaneously during the kinetic experiment. First, molecules that act as inhibiting agents should bind to monomers, intermediate oligomeric species, or aggregation nuclei to prevent the aggregation process [42]. Matos et al. revealed that quercetin, luteolin, and (+)-dihydroxyquercetin non-covalently bind to Aβ<sup>42</sup> lysine residues [27] and Sato et al. displayed the mechanism where catecholtype flavonoids, namely (+)-taxifolin, autoxidize forming an o-quinone on the B-ring that covalently binds to the amino group of lysine [28]. Second, the fluorescence quenching is unavoidable when using ThT as the excitation and emission wavelengths overlap with the majority of oxidized flavones absorbance region (Figures 1 and 3) and appear to form oxidized flavone-ThT interactions (as seen from differences in absorbance spectra, when the compounds are separate or together, Figure S2) that may lead to less bound-ThT on the fibril surface, reducing the fluorescence intensity even further (Figure S3). Therefore, most of the oxidized flavones (especially with more OH groups) suppress the fluorescence intensity in Aβ<sup>42</sup> aggregation experiments. This effect has been observed when two dye molecules interact alone or in the presence of fibrils [30].

Taking a deeper look into the AFM images, we see a tendency for the formation of major clumps when Aβ<sup>42</sup> aggregates with oxidized flavones, especially with 2′ ,3′ -DHF (Figure 5C) and 3,6,2′ ,3′ -TeHF (Figure 5G). This indicates that flavone derivatives bind to the surface of higher-level oligomeric particles as well as fibrils. While most of the mica is covered by oligomeric species, the AFM images may be analyzed, and it can be concluded that inhibitors redirect the aggregation pathway towards the arrangement of different structures. However, this explanation is just the tip of the iceberg, and a more revealing image is seen after a larger-scale analysis. The fibrillar clumps, which appear to be a combination of oligomeric structures and fibrils, consist mostly of aggregates present on the mica that is hardly found. Despite this, some oligomeric species that are found around these clusters led to the assumption that aggregation was partially stopped.

The main objective of this work was to understand the variety of flavones that may act as inhibiting molecules. There is a distinct correlation between the positions of hydroxyl groups, flavone oxidation, and inhibition of the insulin and the Aβ<sup>42</sup> aggregation process. Adjacent OH groups have a tendency to increase the solubility compared to other flavones and enable the autoxidation process, which was seen via UV-vis absorbance spectral data (Figures 1 and 3). Taking into consideration dihydroxyflavones, only four (5,6-DHF, 7,8-DHF, 2′ ,3′ -DHF and 3′ ,4′ -DHF) had an influence on the protein aggregation process. Surprisingly, 6,7-DHF does not autoxidize or affect protein aggregation. Despite this, the majority of hydroxyflavones that have neighboring hydroxy groups undergo oxidation leading to an enhanced inhibitory potential. This structural aspect is similar for 6,7,3′ -THF and 5,6,7-THF, while 5,6,7,4′ -TeHF and 6,7,3′ ,4′ -TeHF tend to oxidize, potentially due to the additional hydroxy groups on the flavone B ring. 2′ ,3′ -DHF appears to have the highest inhibition potential out of all tested flavones, which is then followed by 6,2′ ,3′ -THF. This may resemble a close connection between structures and the autoxidation end products.

Surprisingly, some of the flavonol derivates that do not have neighboring hydroxyl groups (3-hydroxyflavone, 3,5,7-THF, 3,7,3′ -THF, 3,7,4′ -THF, 3,5,7,4′ -TeHF and 3,5,7,2′ ,4′ -PHF) undergo autoxidation; however, these autoxidized molecules do not increase the Aβ<sup>42</sup> aggregation time. This finding suggests a distinct autoxidation mechanism as well as different cinnamoyl system characteristics that are decisive for the developed anti-amyloid properties. Further, flavones with a higher number of hydroxyl groups that contain the aforementioned neighboring OH groups do autoxidize and inhibit insulin aggregation, but only some extend the Aβ<sup>42</sup> aggregation time. These flavones can be categorized into two groups: 7,8-DHF derivatives (7,8,2′ -THF, 7,8,3′ -THF, and 7,8,3′ ,4′ -TeHF) and flavones that have at least two hydroxyl groups on ring B (3,6,3′ ,4′ -TeHF, 5,7,3′ ,4′ -TeHF, 5,7,3′ ,4′ ,5′ -PHF, 3,6,2′ ,4′ ,5′ -PHF and 3,5,7,3′ ,4′ ,5′ -PHF). Even though the number of effective inhibitors directly correlates with the number of OH groups on the molecule, the pentaand hexahydroxyflavone groups are far more complex. One probable scenario is that the flavone inhibitory effect is enabled by the appearance of particular molecular structures that form during the autoxidation process. These molecules should be structurally related, as the positions of OH groups on the molecule repeat, potentially leading to similar autoxidation mechanisms and products. While this study shows that the autoxidation of flavones leads to the formation of different structures, it is essential to note that due to this process, flavones may lose their initial characteristics, such as being inhibitors of AChE or anti-oxidants.

#### **5. Conclusions**

Taking everything into account, non-oxidized flavones do not inhibit the aggregation process of insulin or amyloid-beta, while their oxidized forms show potential against fibril formation. We also show that flavone autoxidation and inhibition are strictly related to the structure of the molecule and depend highly on the position of hydroxyl groups.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/antiox10091428/s1, Figure S1: Flavone structure; Figure S2: Absorbance spectra of incubated flavone mixed with ThT (red) and combined spectra of incubated flavone and ThT when they are scanned separately (black); Figure S3: Fluorescence intensity values of preformed Aβ<sup>42</sup> aggregates mixed with incubated flavone (red), ThT (blue) and incubated flavone with ThT (green); Table S1: Relative halftime and ThT fluorescence intensity values of Insulin and Amyloid-β aggregation.

**Author Contributions:** A.S. and V.S. designed the experiments; A.S., M.Z., and R.S. performed the experiments; A.S., M.Z., and V.S. analyzed the data and prepared the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by grant no. S-SEN-20-3 from the Research Council of Lithuania.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available in this manuscript.

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

#### **References**


## *Article* **(**−**)-Epicatechin—An Important Contributor to the Antioxidant Activity of Japanese Knotweed Rhizome Bark Extract as Determined by Antioxidant Activity-Guided Fractionation**

**Urška Jug 1,2, Katerina Naumoska 1,\* and Irena Vovk 1,\***


**Abstract:** The antioxidant activities of Japanese knotweed rhizome bark extracts, prepared with eight different solvents or solvent mixtures (water, methanol, 80% methanol(aq), acetone, 70% acetone(aq), ethanol, 70% ethanol(aq), and 90% ethyl acetate(aq)), were determined using a 2,2 diphenyl-1-picrylhydrazyl (DPPH) free radical-scavenging assay. Low half maximal inhibitory concentration (IC50) values (2.632–3.720 µg mL−<sup>1</sup> ) for all the extracts were in the range of the IC<sup>50</sup> value of the known antioxidant ascorbic acid at t<sup>0</sup> (3.115 µg mL−<sup>1</sup> ). Due to the highest extraction yield (~44%), 70% ethanol(aq) was selected for the preparation of the extract for further investigations. The IC<sup>50</sup> value calculated for its antioxidant activity remained stable for at least 14 days, while the IC<sup>50</sup> of ascorbic acid increased over time. The stability study showed that the container material was of great importance for the light-protected storage of the ascorbic acid(aq) solution in a refrigerator. Size exclusion–high-performance liquid chromatography (SEC-HPLC)–UV and reversed phase (RP)-HPLC-UV coupled with multistage mass spectrometry (MS<sup>n</sup> ) were developed for fractionation of the 70% ethanol(aq) extract and for further compound identification, respectively. In the most potent antioxidant SEC fraction, determined using an on-line post-column SEC-HPLC-DPPH assay, epicatechin, resveratrol malonyl hexoside, and its in-source fragments (resveratrol and resveratrol acetyl hexoside) were tentatively identified by RP-HPLC-MS<sup>n</sup> . Moreover, epicatechin was additionally confirmed by two orthogonal methods, SEC-HPLC-UV and high-performance thin-layer chromatography (HPTLC) coupled with densitometry. Finally, the latter technique enabled the identification of (−)-epicatechin. (−)-Epicatechin demonstrated potent and stable time-dependent antioxidant activity (IC<sup>50</sup> value ~1.5 µg mL−<sup>1</sup> ) for at least 14 days.

**Keywords:** *Polygonum cuspidatum*; Reynoutria; invasive species; phenolic compounds; flavan-3-ols; stilbenes; vitamin C; size-exclusion chromatography; DPPH test; DPPH derivatization

## **1. Introduction**

Japanese knotweed (*Fallopia japonica* Houtt.; synonyms: *Polygonum cuspidatum* Siebold & Zucc., *Reynoutria japonica* Houtt., *Polygonum reynoutria* Houtt., *Pleuropterus cuspidatus* (Siebold and Zucc.) H. Gross, *Tiniaria japonica* (Houtt.) Hedberg), which is native to East Asia, is an invasive plant species in Europe and North America [1]. The Japanese knotweed rhizome has already been tested in various biological studies [2], and its extract or existing compounds showed antioxidant [3–11], estrogenic [12], antiproliferative [3], antibacterial [13], antiviral (anti-human immunodeficiency virus) [14], anti-inflammatory [15], antiatherosclerotic [16] activities, etc. A lot of health benefits of Japanese knotweed rhizome extract were correlated with the content of some antioxidant compounds [4].

Mechanisms of antioxidant activity, such as free radical scavenging, singlet oxygen quenching, transition metal chelation, enzyme mimetic activity, and enzyme inhibition, have

**Citation:** Jug, U.; Naumoska, K.; Vovk, I. (−)-Epicatechin—An Important Contributor to the Antioxidant Activity of Japanese Knotweed Rhizome Bark Extract as Determined by Antioxidant Activity-Guided Fractionation. *Antioxidants* **2021**, *10*, 133. https:// doi.org/10.3390/antiox10010133

Received: 27 December 2020 Accepted: 14 January 2021 Published: 18 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

been described [17]. There are several methods for evaluating antioxidant activity [18,19] that are based on different mechanisms and can give results that are not comparable. A universal test does not exist; therefore, the use of at least two different methods is strongly recommended [18]. The methods can be classified according to their performance ("in vitro" and "in vivo") [19], the type of the measurement (e.g., spectrophotometric [3,5–11], electrochemical [20,21], chromatographic (gas chromatography [22], HPLC [20,21], HPTLC [23–26])), and the type of the reaction used for the assay (hydrogen atom transfer (HAT)-based assays and electron transfer (ET)-based assays) [18,21].

Among the free radical-scavenging methods, the DPPH assay is the fastest, the most straightforward, relatively inexpensive, efficient, and, therefore, the most frequently employed. Many studies on DPPH assay-guided fractionation of various plant materials have already been performed [27–40], including *on-line* methods with pre-column [27] or post-column [28–33,41] DPPH reactions. *On-line* methods for measurement of the free radical-scavenging activity indicate the antioxidant fractions/compounds in a fast and inexpensive way, without the need to isolate and test them *off-line*, which is time consuming, as described in [28].

The antioxidant activity of the Japanese knotweed rhizome has been tested and confirmed using various assays: DPPH radical-scavenging capacity [3–5,7–11]; superoxidescavenging (nitroblue tetrazolium (NBT) reduction) capacity [4]; 2′ -azinobis-[3 ethylbenzthiazolin-6-sulfonic acid] (ABTS) radical-scavenging capacity [5,6]; electron spin resonance spectrometry (ESR) [3]; oxygen radical absorption capacity (ORAC) [6]; ferricreducing antioxidant power (FRAP) [9]; chemiluminescence [5]; phosphomolybdenum reduction [10]; lipid peroxidation inhibition performed on linoleic acid [10] and on mouse brain tissue [4]; DNA strand scission assay [3,4]; and superoxide dismutase (SOD) inhibition assay–water-soluble tetrazolium salt-1 (WST-1) [8].

Tests for determining the total polyphenol content, such as the Folin–Ciocalteu assay [4,5,9,10] have also been frequently used to estimate the antioxidant capacity of Japanese knotweed rhizomes, as the polyphenol content is generally significantly correlated to the sample's total antioxidant activity [9,18,42,43]. Phenolic compounds act as reducing agents, hydrogen donors, singlet oxygen quenchers, and metal chelators [44].

The antioxidant activities of the extracts obtained from the rhizome of Japanese knotweed and from two other knotweed species, giant knotweed (*Fallopia sachalinensis* Schm.) and their hybrid Bohemian knotweed (*Fallopia*×*bohemica* Chrtek & Chrtková), using different solvents, have already been compared [10]. The choice of solvent was shown to be of great importance for the extraction of antioxidants [10]. The relationship between the antioxidant activity and the chemical content was determined using principal component analysis (PCA) [10], showing that proanthocyanidins are the most important contributors to the total antioxidant capacity [10].

Japanese knotweed rhizome extract is already commercially available as food supplements, marketed as a source of resveratrol as an antioxidant from the stilbenes. Analyses of the bioactive compounds of Japanese knotweed rhizome extract were predominantly performed by (ultra)high-performance liquid chromatography coupled with a UV detector and (multistage) mass spectrometry ((U)HPLC-UV-MS(n)) [6,8,10,45–51] using RP stationary phase, although HPLC-UV [11,12], HPTLC [52–55], HPTLC-MS<sup>n</sup> [52,53], and capillary electrophoresis [56,57] were also used.

The objectives of our work were: (1) to select the most suitable solvent or solvent mixture for the extraction of antioxidants from Japanese knotweed rhizome bark; (2) to determine the antioxidant activity of Japanese knotweed rhizome bark extract; (3) to determine the stability of the antioxidant activity of the selected extract over time; (4) to fractionate the extract by a new SEC-HPLC method and to determine its most potent antioxidant fraction by an *on-line* post-column reaction with DPPH; and (5) to further identify the compounds present in the isolated antioxidant SEC fraction(s) by RP-HPLC-MS and HPTLC.

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

#### *2.1. Chemicals and Materials*

All solvents were at least of analytical grade. Methanol (HPLC and LC-MS grade), acetone, and acetonitrile (LC-MS grade) were obtained from Honeywell Reagents (Seelze, Germany). Ethanol (absolute anhydrous) was purchased from Carlo Erba Reagents (Val de Reuil, France). Ethyl acetate, acetic acid (glacial (100%) and glacial (100%) LC-MS grade), concentrated hydrochloric acid (37%), and 4-(dimethylamino)cinnamaldehyde (DMACA) were acquired from Merck (Darmstadt, Germany). Ammonium acetate, 2,2-diphenyl-1 picrylhydrazyl (DPPH), (−)-epicatechin (90%), and (−)-catechin (98%) were acquired from Sigma-Aldrich (Steinheim, Germany). Ascorbic acid was obtained from Fluka, Sigma-Aldrich (Steinheim, Germany). (−)-Epicatechin (of high purity) was purchased from Fluka Chemie (Buchs, Switzerland), while (+)-catechin (98%) was obtained from Carl Roth (Karlsruhe, Germany). A Milli-Q water purification system (18 MΩ cm−<sup>1</sup> ; Millipore, Bedford, MA, USA) was used to obtain ultrapure water. Disposable plastic cuvettes were purchased from Brand (Wertheim, Germany).

#### *2.2. The Preparation, Extraction Yield, and Antioxidant Activity of Various Extracts*

Japanese knotweed (*Fallopia japonica* Houtt.) rhizomes were harvested in Ljubljana, Slovenia (Vrhovci, by a bridge over the Mali Graben, N 46◦02′33.9′′; E 14◦27′00.9′′). A voucher specimen was deposited in the Herbarium LJU (LJU10143477). After the rhizomes were cleaned with tap water, the bark was peeled and lyophilized at −50 ◦C for 24 h (Micro Modulyo, IMAEdwards, Bologna, Italy). The obtained dry material was frozen using liquid N<sup>2</sup> and pulverized by a Mikro-Dismembrator S (Sartorius, Göttingen, Germany) for 1 min at a frequency of 1700 min−<sup>1</sup> . The lyophilized and pulverized rhizome bark (200 mg; eight replicates) was extracted with 2 mL of the following solvents or solvent mixtures: water, methanol, 80% methanol(aq), acetone, 70% acetone(aq), ethanol, 70% ethanol(aq), and 90% ethyl acetate(aq), followed by 5 min vortexing, 15 min ultrasonication, and 5 min centrifugation at 6700× *g*.

The supernatants were transferred into pre-weighted glass storage vials, where the solvents were evaporated under N<sup>2</sup> flow. The vials with obtained dry extracts of Japanese knotweed rhizome bark (JKRB) were weighed to calculate the extraction yield. The dry extracts were further dissolved in methanol (stock solutions, which also served as first working solutions: 400 µg mL−<sup>1</sup> ) and diluted with the same solvent to obtain additional working solutions with the following concentrations (µg mL−<sup>1</sup> ): 200, 100, 50, 25, 12.5, 6.25, 3.125, 1.563, 0.781, 0.391, and 0.195. Immediately after dilution, they were tested using the DPPH assay described in [58]. The DPPH reagent (1 mL of 200 µM methanolic solution of DPPH) was added to 3 mL of each working solution in triplicate (solution A) [59].

To prepare the sample blanks, 1 mL of methanol was added to 3 mL of separate working solutions (solution B) [59]. A control sample (for DPPH) was prepared by the addition of 1 mL of DPPH reagent to 3 mL of methanol in triplicate (solution **C**) [59]. All prepared solutions were vortexed for 5 s and stored in amber glass storage vials for 30 min in the dark at room temperature. Spectrophotometric measurements of the absorbances of solutions A, B, and C (named AA, AB, and AC, respectively; Equation (1)) were performed at 517 nm using a Lambda 45 UV/Vis spectrometer (Perkin Elmer, Waltham, MA, USA) with methanol as a blank solvent for the instrument. The IC<sup>50</sup> values were calculated and the curves were plotted in GraphPad Prism 7 [60].

Calculation of the DPPH scavenging effect [59]:

DPPH scavenging effect (%) = 100 − ((A<sup>A</sup> − AB) × 100/AC) (1)

in which A<sup>B</sup> is included in the case of yellow-colored working solutions to exclude their absorbance contributions [59].

#### *2.3. Comparison between the Antioxidant Activities of the 70% Ethanol(aq) Extract of Japanese Knotweed Rhizome Bark and Ascorbic Acid over Time*

A DPPH assay of the selected dry 70% ethanol(aq) extract, re-dissolved in methanol (400 µg mL−<sup>1</sup> ) and diluted to the concentrations (µg mL−<sup>1</sup> ): 200, 100, 50, 25, 12.5, 6.25, 3.125, 1.563, 0.781, 0.391, and 0.195, and a DPPH assay of ascorbic acid dissolved in methanol (1000 µM) and diluted to the concentrations (µM): 500, 250, 100, 50, 40, 30, 20, 10, and 1, were performed at t = 0, 2, 4, 6, 8, 10, 24, and 50 h, 7 and 14 days (at T = 25 ◦C) after the preparation of solutions.

The influence of glass vs. plastic storage containers on the stability of ascorbic acid was studied by a 24 h aging of 50 µM aqueous ascorbic acid solutions stored in the refrigerator (T = 4 ◦C) or at room temperature (T = 25 ◦C) and: (i) protected from light in plastic centrifuge tubes (T = 4 ◦C), (ii) protected from light in glass storage vials (T = 4 ◦C), (iii) exposed to daylight in plastic centrifuge tubes (T = 25 ◦C), and (iv) exposed to daylight in glass flasks (T = 25 ◦C). HPLC-UV analyses of ascorbic acid solutions were performed at t = 0 h and at t = 24 h after solution preparation using an in-house HPLC method (confidential) at 254 nm. As ascorbic acid degrades very quickly, three fresh ascorbic acid solutions were prepared at t = 24 h to confirm the intermediate precision of the method (*n* = 6; *t*<sup>R</sup> = 3.1 min).

## *2.4. SEC-HPLC-UV Fractionation of the 70% Ethanol(aq) JKRB Extract Guided by an On-Line Post-Column Reaction with DPPH*

The SEC-HPLC-UV method was developed for the fractionation of the 70% ethanol(aq) extract of JKRB using an Agilent Bio SEC-3 column (150 mm × 4.6; 3 µm, 100 Å) on an HPLC-PDA Agilent Technologies 1260 Infinity system (Santa Clara, California, USA), equipped with a fraction collector (Agilent 1260 Infinity II). OpenLAB CDS ChemStation software (Agilent) was used for data collection and analysis. A pre-mixed mobile phase was prepared with 150 mM ammonium acetate buffer and ethanol in the ratio 75:25 (*v/v*).

The ammonium acetate buffer was prepared by dissolving 5.778 g of ammonium acetate in 500 mL ultrapure water, and acetic acid was used to adjust the pH value to 4.8. An isocratic elution was performed with a flow rate of 0.325 mL min−<sup>1</sup> and a run time of 40 min. The temperatures of the column and autosampler were set to 40 ◦C and 25 ◦C, respectively. The dry 70% ethanol(aq) extract of JKRB was re-dissolved in the mobile phase to achieve a concentration of 0.5 mg mL−<sup>1</sup> and was filtered through a 0.45 µm polyvinylidene fluoride (PVDF) membrane filter before injection (5 µL). Chromatograms were recorded at different wavelengths (280, 300, and 360 nm), and absorption spectra were acquired as well.

To determine the antioxidant fractions, an *on-line* post-column reaction was performed using DPPH solution (400 µM in 80% methanol(aq)) delivered at a flow rate of 5 µL min−<sup>1</sup> through a syringe pump, leading to one inlet of a T-unit. The second inlet of the T-unit was connected to the column effluent capillary, while the outlet led to a 3.5 m long reaction coil (0.13 mm internal diameter (I.D.)) and later to the photodiode array (PDA) detector. The chromatographic conditions were as explained above. The reaction coil allowed a longer contact time between the eluting fractions' compounds and the DPPH reagent, thus enabling radical scavenging reactions before reaching the PDA detector. UV/Vis spectra were acquired, and the chromatograms were recorded at 280 and 517 nm.

The decrease in absorbance at 517 nm indicated the antioxidant activity of the fractions, visible as negative peaks on the chromatogram. The *on-line* post-column reaction of the SEC fractions was performed in triplicate. Blank and control analyses were executed as follows: (i) injection of the sample extract and post-column introduction of 80% methanol(aq); (ii) injection of the procedural blank (the mobile phase filtered through a 0.45 µm PVDF membrane filter) and post-column reaction with DPPH; and (iii) injection of the procedural blank and post-column introduction of 80% methanol(aq).

As expected, the reaction coil led to a shift of the retention times (*t*Rs) to higher values. Therefore, it was used for all analyses, including fraction collection, although the reaction with DPPH was not applied during this step.

Fourteen fractions, detected at 280 nm, were selected for retention time-based collection (Section 3.4). The temperature of the fraction collector was maintained at 4 ◦C. The collected fractions were pooled, the solvent was evaporated under N<sup>2</sup> flow, and the solid residues were stored in a freezer at −20 ◦C.

#### *2.5. Analyses of SEC Fractions and Determination of the Strongest Antioxidant by RP-HPLC-MS*

The compounds of the isolated SEC fractions were analyzed using a UHPLC-UV-MS system (Accela 1250, coupled to an LTQ Velos MS, Thermo Fisher Scientific, Waltham, MA, USA). A new HPLC-UV-MS method was developed for the separation and characterization of the compounds from the 70% ethanol(aq) extract of JKRB using a Hypersil ODS column (150 × 4.6 mm; 5 µm I.D., Thermo A). SEC fractions (FRs) obtained from 122 (FRs 1–7, 9, and 14), 96 (FRs 10–13), and 77 (FR 8) runs were dried under a N<sup>2</sup> flow, dissolved in 150 (FRs 2, 5, and 7), 200 (FRs 1, 3, and 4), 250 (FRs 6 and 9), 300 (FR 8), 500 (FR 14), and 1000 µL (FRs 10–13) of solvent (water:ethanol, 3:1, *v/v*) and injected in different volumes (5 µL: FRs 10–13; 10 µL: FRs 1, 6, 9, and 14; 15 µL: FRs 2–5, 7, and 8).

These values were adapted to the peak heights and widths of the SEC fractions. The mobile phase, consisting of 0.1% acetic acid(aq) (A) and acetonitrile (B), and a linear gradient elution with a flow rate of 0.7 mL min−<sup>1</sup> of 10–100% B (0–30 min), were used. The column and autosampler temperatures were maintained at 25 ◦C and 10 ◦C, respectively. Chromatograms were recorded at 280, 300, and 360 nm, and absorption spectra were acquired as well. For the ionization of compounds, heated electrospray ionization (HESI) in negative ion mode was used, and the MS parameters were as follows: heater and capillary temperatures of 400 and 350 ◦C, respectively, sheath gas 30 arbitrary units (a.u.), auxiliary gas 5 a.u., sweep gas 0 a.u., spray voltage 4 kV, and S-Lens RF level 69.0%.

To optimize the MS parameters, a methanolic standard solution of (−)-epicatechin (0.1 mg mL−<sup>1</sup> , 10 µL min−<sup>1</sup> ) was combined with the column effluent (55% B, 0.7 mL min−<sup>1</sup> ) using a T-unit, thus directing the combined flow into the MS source. The MS spectra were recorded in the *m/z* range of 50–2000, while the precursor ions of interest were fragmented in MS<sup>n</sup> using a collision energy of 35%. Xcalibur software (version 2.1.0, Thermo Fisher Scientific) was used to evaluate the collected chromatograms and spectra.

#### *2.6. Identification of the Compounds in the Antioxidant Fraction by Orthogonal Methods and Confirmation of Their Antioxidant Activity by DPPH Assay*

To confirm the presence of (−)-epicatechin in the isolated antioxidant fraction, commercially available standards of flavan-3-ols were used. Standards of (+)-catechin and (−)-epicatechin were used for the SEC-HPLC-UV and RP-HPLC-UV-MS analyses, while for the HPTLC analysis performed on an HPTLC cellulose stationary phase, which enables separation of enantiomers, a (−)-catechin standard was also applied. Standards were prepared in concentrations of 0.01 mg mL−<sup>1</sup> (in water:ethanol (3:1, *v/v*) for SEC-HPLC-UV) and 0.1 mg mL−<sup>1</sup> (in methanol for RP-HPLC-UV-MS and HPTLC). The injection volume was 5 µL for both HPLC methods.

The antioxidant fraction FR 8, collected from 77 runs (by SEC-HPLC method; Section 2.4), was dissolved in 200 µL of methanol. All standards (4 µL, 0.4 µg) and the antioxidant fraction (40 µL) were applied on an HPTLC cellulose plate (Merck, Art. No. 1.05786.0001, cut to 10 cm × 10 cm) as 8 mm bands, 8 mm from the bottom of the plate using a Linomat 5 (Camag, Muttenz, Switzerland). The plate was developed up to 90 mm (45 min) in a normal developing chamber (for 10 cm × 10 cm plates, Camag) using water as a developing solvent [23,61,62] and dried with a stream of warm air for 3 min after development.

Post-chromatographic derivatization was performed by immersing the plate for 1 s into DMACA detection reagent, prepared by dissolving 60 mg of DMACA in 13 mL of concentrated hydrochloric acid (37%) and diluted with ethanol to make up a total volume of 200 mL [61]. The plate was then dried with warm air for 2 min. The DigiStore 2 documentation system in conjunction with Reprostar 3 (Camag) was used for the documentation of

the chromatograms at 254 nm, 366 nm, and white light illumination after development and 10 min after post-chromatographic derivatization with DMACA reagent.

After derivatization, the plate was also scanned with a slit-scanning densitometer (TLC Scanner 3, Camag) set in absorption/reflectance mode at 655 nm. The selected wavelength was derived from our previously published studies [61,63]. The other settings were as follows: slit length 6 mm, slit width 0.30 mm, and scanning speed 20 mm s−<sup>1</sup> . Both instruments were controlled using winCATS software (version 1.4.9.2001).

As in Sections 2.2 and 2.3, the spectrophotometric DPPH assay of methanolic solutions of the (−)-epicatechin standard (Sigma-Aldrich; 1000, 500, 250, 100, 50, 40, 30, 20, 10, 1, and 0.1 µM) was performed at t = 0, 2, 4, 6, 8, 10, 24, and 50 h, 7 and 14 days (storage at T = 25 ◦C) after solution preparation to determine its IC<sup>50</sup> value for the radical scavenging activity, as well as the stability of its antioxidant activity.

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

#### *3.1. Extraction Yields and Antioxidant Activity of Various Extracts*

The extraction of JKRB was performed with water, polar organic solvents (methanol, acetone, and ethanol), and aqueous solutions thereof (80% methanol(aq), 70% acetone(aq), 70% ethanol(aq), and 90% ethyl acetate(aq)). The highest extraction yield was achieved by 70% ethanol(aq), and a slightly lower yield was achieved by 70% acetone(aq) (Table 1). Significantly lower extraction yields were obtained with pure ethanol and acetone. The difference between the extraction yields obtained with methanol and 80% methanol(aq) was not significant. Water gave a higher extraction yield than did pure acetone. The lowest extraction yield was obtained using 90% ethyl acetate(aq) (Table 1).

**Table 1.** The extraction yields and the calculated values of the half maximal inhibitory concentrations (IC50) of antioxidant activity (GraphPad Prism 7 [60]) of extracts prepared with different solvents or solvent mixtures tested in the concentration range of 0.195–400 µg mL−<sup>1</sup> .


The antioxidant activities of all dry extracts, re-dissolved in methanol, were tested using a DPPH assay. Re-dissolving all dry extracts in the same solvent (methanol) was preferred (providing equal polarity and pH of the reaction medium for all extracts) to enable comparison of the obtained DPPH assay results, as discussed in [58]. As all dry extracts and DPPH were soluble in methanol, this was selected as a reaction medium.

The obtained results of the DPPH radical scavenging assay are expressed as IC<sup>50</sup> values, which represent the concentration of the antioxidant required to scavenge 50% of the DPPH free radicals and consequently lead to a 50% decrease in the DPPH absorption [64–66].

As different protocols of the same antioxidant assay may lead to incomparable results, a known antioxidant, ascorbic acid, was used as a reference. The IC<sup>50</sup> values of all JKRB extracts (Figure 1) prepared by different extraction solvents and solvent mixtures were very low (2.632–3.715 µg mL−<sup>1</sup> ; Table 1) and in the range of the IC<sup>50</sup> value of ascorbic acid at t<sup>0</sup> (3.115 µg mL−<sup>1</sup> ; Table 2). This indicates the high antioxidant potential of JKRB extracts, which may be attributed to the activity of the various phenolic compounds present in JKRB [67]. A JKRB extract prepared with 70% ethanol(aq) was used for further analyses

due to the highest extraction yield (only 70% acetone(aq) resulted in a comparable yield) (Table 1). Additional reasons for the selection of this green extraction solvent include that ethanol is considered less harmful than other solvents when present as a residual solvent in pharmaceutical formulations [68], 70% ethanol(aq) is suitable for the preparation of tinctures, and ethanol is commercially available as a "food grade" solvent.

**Figure 1.** Logarithmic curves of the antioxidant activities of extracts of Japanese knotweed rhizome bark (JKRB) (*n* = 3) prepared with the following solvents and solvent mixtures: water (**A**), methanol (**B**), 80% methanol(aq) (**C**), acetone (**D**), 70% acetone(aq) (**E**), ethanol (**F**), 70% ethanol(aq) (**G**), and 90% ethyl acetate(aq) (**H**). The calculated values of IC<sup>50</sup> are 3.561 (**A**), 3.715 (**B**), 3.469 (**C**), 2.632 (**D**), 3.350 (**E**), 2.893 (**F**), 3.503 (**G**), and 2.786 (**H**) µg mL−<sup>1</sup> (obtained by GraphPad Prism 7 [60]).


**Table 2.** The calculated IC<sup>50</sup> values of the antioxidant activity of ascorbic acid (AA, in the range 1–1000 µM or 0.176–176.12 µg mL−<sup>1</sup> ) and JKRB 70% ethanol(aq) extract (in the range 0.195–400 µg mL−<sup>1</sup> ) over time.

The logarithmic curves representing the radical scavenging activity of the extracts of JKRB with different concentrations are shown in Figure 1.

A time-dependent decrease in the antioxidant activity (increase in the IC<sup>50</sup> value) of the ascorbic acid solutions was observed, which is most likely a consequence of its oxidation, which can particularly be promoted by light, heat, and heavy metal cations [69]. Therefore, the preparation of the ascorbic acid solutions was carried out very quickly, and the time from their preparation to exposure to DPPH was kept as short as possible (10 min in the worst-case scenario). The addition of the chelating agent ethylenediaminetetraacetic acid (EDTA) to ascorbic acid solution was previously found to have an indirect stabilizing effect on the ascorbic acid molecule through the chelation of traces of heavy metals residing on the surface of glass containers [70]. To examine the influence of glass vs. plastic storage containers on the stability of ascorbic acid in solution, its content after storage in different containers was determined by the use of the HPLC-UV method (Section 3.2).

## *3.2. Antioxidant Activity over Time—Ascorbic Acid Compared to the JKRB 70% Ethanol(aq) Extract*

The antioxidant activity of ascorbic acid continuously decreased over time (IC<sup>50</sup> value increased from 3.115 up to 62.787 µg mL−<sup>1</sup> , Table 2, Figure 2A), while the antioxidant activity of the JKRB 70% ethanol(aq) extract remained constant during the same time interval (0 h to 14 days) (Table 2, Figure 2B). These results suggest a potential use of the JKRB 70% ethanol(aq) extract as a strong antioxidant material. The potential applications might include the formulation of food supplements (e.g., tincture, powder, and solid dosage forms) or its utilization as a food antioxidant. On the other hand, the stability of ascorbic acid (and its antioxidant effect) in various beverages (bottled and left standing) rich in or enriched with ascorbic acid remains questionable.

**Figure 2.** Logarithmic curves plotting the 2,2-diphenyl-1-picrylhydrazyl (DPPH) scavenging effect (%) of ascorbic acid (**A**) and JKRB 70% ethanol(aq) extract (**B**) against the concentration, measured over time.

As the mobile phase used for the HPLC quantification of ascorbic acid was aqueous based, ascorbic acid for the HPLC analyses was dissolved in water. After 24 h of aging in daylight and at room temperature, practically all ascorbic acid was lost (ascorbic acid <1%), regardless of the container material used for storage. On the other hand, the container material was of great importance for the light-protected storage of the ascorbic acid(aq) solution in the refrigerator. After 24 h of aging (dark, refrigerator) in a plastic container, the content of ascorbic acid(aq) was 65.19% of the initial concentration, while storage in a glass container resulted in a loss (<1% of the initial concentration) comparable to that reported for the room conditions (room temperature and daylight).

Based on these results, the combination of light and temperature, as well as trace metals on the glass surface, influence the stability of ascorbic acid in aqueous solution (Figure 3). On the other hand, the JKRB extract showed stable antioxidant activity for at least 14 days in the worst-case scenario conditions for ascorbic acid (light, room temperature, glass container). The flavan-3-ols, proanthocyanidins, and anthraquinones, which represent major groups of compounds in the JKRB extract [67], are proven chelating agents of glass surface ions [71,72], acting through their hydroxyl or both carbonyl and hydroxyl groups, located on the vicinal or *peri* positions [71]. This supports our findings regarding the stability of the measured antioxidant activity of the JKRB extract.

**Figure 3.** Ascorbic acid(aq) (50 µM) analyzed immediately after preparation (**A**) was subjected to 24 h of aging (**B**–**E**), stored in the refrigerator and protected from light (**B**,**C**) in plastic (**B**) and glass containers (**C**) or stored in daylight at room temperature (**D**,**E**) in plastic (**D**) and glass containers (**E**). The peak areas corresponding to ascorbic acid (*t<sup>R</sup>* 3.1 min, 254 nm) in aged solutions were compared to the peak area of ascorbic acid in the fresh solution. The intermediate precision of the method was 3% (*n* = 6).

## *3.3. SEC-HPLC Fractionation of the JKRB 70% Ethanol(aq) Extract, On-Line Post-Column Reaction of the SEC Fractions with DPPH and Determination of the Antioxidant Fractions*

A SEC-HPLC-UV method was developed for the first time to separate the compounds from the Japanese knotweed rhizome extract. A SEC column with a pore size of 100 Å was used to enable better separation of the smaller molecules from the extract. A high concentration of the buffer (150 mM) was used to reduce the secondary interactions on the column. Ethanol as a co-solvent, mixed with the buffer in a ratio of 25:75 (*v/v*), improved the solubility of JKRB compounds in the mobile phase. A higher percentage of ethanol in the mobile phase causes precipitation of the ammonium acetate buffer. The chromatograms were recorded at 280 nm, where the highest sensitivity for most of the compounds was achieved. Fourteen of the most abundant fractions (FR 1–FR 14) were selected for isolation.

The antioxidant potential of the Japanese knotweed rhizome bark was tested for the first time using the SEC-HPLC-UV/Vis method with an *on-line* post-column DPPH assay. Finding the right concentration, flow rate, and solvent for the DPPH reagent (insoluble in water and soluble in methanol) to be introduced into the mobile phase (buffer insoluble in methanol) was challenging. However, a 400 µM DPPH solution in 80% methanol(aq), delivered at a flow rate of 5 µL min−<sup>1</sup> , proved to be a good choice as it did not cause precipitations in the system upon contact with the mobile phase. The isocratic elution of the SEC method enabled the constant solubility of the DPPH reagent in the mobile phase and equal chemical reaction conditions throughout the whole run, thus ensuring more relevant results related to the antioxidant activity in comparison to gradient mode chromatography (e.g., RP in the gradient mode).

The noisy baseline of the chromatogram at 517 nm (Figure 4) was most probably due to the imperfections of the in-house built equipment for *on-line* post-column derivatization. Therefore, some antioxidant fractions might have been overlooked, due to a potentially too low decrease in the baseline at 517 nm. However, FR 8, eluting at *t*<sup>R</sup> 16.8–18 min (Figure 4), was undoubtedly determined as the most potent antioxidant (a clear baseline drop at 517 nm). Although only FR 8 showed antioxidant activity, all fractions (FR 1–FR 14, Figure 4) were collected and screened using RP-HPLC.

**Figure 4.** SEC-HPLC-UV/Vis chromatogram at 280 nm (without post-column reaction) and at 517 nm (after post-column reaction with DPPH). The fractions and time intervals selected for fraction collection are marked. Fraction 8 was determined to be the strongest antioxidant due to the decrease in the absorbance at 517 nm.

#### *3.4. Characterization of the Compounds in the Isolated SEC Fractions, Identification of the Antioxidant Fraction Compounds by Orthogonal Methods, and their Antioxidant Activity over Time*

An RP-HPLC-MS<sup>n</sup> method was developed to analyze the compounds in the isolated SEC fractions. The compounds were tentatively identified by comparing the obtained and literature MS and MS<sup>2</sup> data (Table 3). For the antioxidant fraction FR 8, MS<sup>3</sup> was also performed. Although expected, the size distribution of the SEC eluting compounds (from larger to smaller molecular masses) was not obvious (Table 3). One of the possible explanations relates to the content of the organic solvent (25%) in the mobile phase, which might promote secondary interactions and might subsequently impact the distribution of the compounds.

The presence of flavan-3-ol monomer, as the main representative in the antioxidant fraction FR 8, was suspected based on the mass spectra and fragmentation patterns, which were compared to those of the (−)-epicatechin standard (Figure 5) and to the literature data [52,53,67]. MS signals of resveratrol malonyl hexoside, resveratrol, and resveratrol acetyl hexoside were also observed in FR 8, where the last two most likely corresponded to in-source fragments of resveratrol malonyl hexoside [67]. Additional MS signals (Table 3) were not identified due to their low abundance.

**Figure 5.** The flavan-3-ol monomer identified by (−)ESI-MS based on the mass spectra and fragmentation patterns obtained for the signal at *t*<sup>R</sup> 6.4 min in the antioxidant fraction (FR 8) (**A**) and confirmed by (−)-epicatechin standard (**B**). Figure abbreviations: selected ion monitoring (SIM), and total ion current (TIC).

In our previous study [67], (+)-catechin and (−)-epicatechin were identified as the two main flavan-3-ol monomers in JKRB. Therefore, both standards were analyzed using the developed RP-HPLC-MS method, which resulted in the separation of (−)-epicatechin (*t*<sup>R</sup> 6.4 min) and (+)-catechin (*t*<sup>R</sup> 5.6 min) (Figure 6). The presence of epicatechin (Figure 6) was thus confirmed in FR 8 (Figures 5 and 6).

**Figure 6.** RP-HPLC-MS chromatograms of the antioxidant fraction (FR 8) in SIM mode (*m/z* 289), (−)-epicatechin and (+)-catechin standards (both in TIC mode; *m/z* 50–2000).

The (−)-epicatechin and (+)-catechin standards were also analyzed using the SEC-HPLC method and the *t*<sup>R</sup> of (−)-epicatechin (17.2 min) matched the *t*<sup>R</sup> range of the antioxidant fraction FR 8 (16.8–18.0 min) (Figure 7), while (+)-catechin (*t*<sup>R</sup> 19.8 min) eluted at the *t*<sup>R</sup> range of FR 9 (19.4–19.8 min). MS signals of flavan-3-ols at *m/z* 289 were observed by RP-HPLC-MS in both fractions (Table 3). Unexpectedly, catechin and epicatechin diastereoisomers were separated by SEC-HPLC (Figure 7), likely due to their conformational differences or as a consequence of secondary interactions in the column. However, C18- RP-HPLC and SEC-HPLC methods do not enable distinguishing between the enantiomers ((+)-catechin and (−)-catechin; (+)-epicatechin and (−)-epicatechin). According to our previous study [67], the presence of diastereoisomer (−)-epicatechin in FR 8 and (+)-catechin in FR 9, was suspected.

Matching UV spectra of the isolated epicatechin in FR 8 and (−)-epicatechin standard obtained by RP-HPLC and SEC-HPLC methods (λmax at 230 and 280 nm—data not shown) showed that epicatechin is the main compound of the most potent antioxidant fraction, FR 8.

To confirm the presence of (−)-epicatechin, HPTLC analysis of FR 8 and three standards, (−)-epicatechin, (+)-catechin, and (−)-catechin, was performed on the cellulose stationary phase, which acts as a chiral selector. (+)-Epicatechin was not applied on the plate as it is not commercially available. Derivatization of the chromatograms with the DMACA detection reagent (flavan-3-ol-specific reagent) confirmed the presence of (−) epicatechin in FR 8 (matching *R*<sup>F</sup> values of the bands of FR 8 and the (−)-epicatechin standard; Figure 8). In addition to the band for (−)-epicatechin, another poorly resolved band appeared in FR 8 (Figure 8, track 1), which also showed a positive reaction with DMACA. The presence of the two peaks was also confirmed densitometrically (Figure 9).

Resveratrol malonyl hexoside, which was also detected in the antioxidant fraction, was reported for the first time in the Japanese knotweed rhizome in our previous study [67]. Unfortunately, the standard of this compound is not commercially available, thus, an additional confirmation of its presence in the antioxidant fraction was not possible (too low an amount for nuclear magnetic resonance (NMR) spectroscopy).

The antioxidant potential of resveratrol, an aglycone moiety of resveratrol malonyl hexoside, is already well known, and resveratrol's presence has also been linked to the antioxidant potential of the Japanese knotweed rhizome [8,73]. Resveratrol may cause synergistic or additive antioxidant effects in combination with epicatechin or other extract constituents (Table 3). Recently, an important contribution to the high antioxidant potential of this plant material was attributed to flavan-3-ols and proanthocyanidins [10]. Epicatechin was confirmed in the antioxidant fraction by three orthogonal methods, SEC-HPLC, RP-HPLC-MS, and HPTLC, among which the latter enabled the identification of (−)-epicatechin, which was already recognized as an antioxidant [18,29,42].

We also compared the antioxidant activities of (−)-epicatechin and *trans*-resveratrol standards by DPPH assay. The results showed that (−)-epicatechin is a stronger DPPH radical scavenger than *trans*-resveratrol (higher IC50; 7.08 µg/mL or 31.02 µM). Moreover, (−)-epicatechin was shown to be present in higher quantities in Japanese knotweed rhizome bark in comparison to *trans*-resveratrol [11].

**Table 3.** RP-HPLC-MS analysis of the compounds in the SEC-HPLC-UV fractions (FR 1–FR 14), corresponding to the SEC *t*R ranges (FR 1: 7.7–8.0, FR 2: 8.3–8.6, FR 3: 9.3–9.8, FR 4: 11.0–11.7, FR 5: 13.0–13.4, FR 6: 13.6–15.0, FR 7: 16.1–16.6, FR 8: 16.8–18.0, FR 9: 19.4–19.8, FR 10: 20.8–22.9, FR 11: 24.0–25.5, FR 12: 26.4–27.8, FR 13: 28.7–30.2, and FR 14: 32.0–32.9). Different numbers of collection runs were performed to isolate the SEC fractions: 122 (FRs 1–7, 9, 14), 96 (FRs 10–13), and 77 (FR 8). Fractions were dissolved in different volumes of water:ethanol (3:1, *v/v*): 150 (FRs 2, 5, 7), 200 (FRs 1, 3, 4), 250 (FRs 6, 9), 300 (FR 8), 500 (FR 14), and 1000 µL (FRs 10–13) and injected in different volumes (5 (FRs 10–13), 10 (FRs 1, 6, 9, 14), and 15µL (FRs 2–5, 7, 8)) into the RP-HPLC-MS system.


**Table 3.** *Cont.*


<sup>a</sup>*t*Robtained by RP-HPLC-ESI-MS; b MS3; c Not identified; d [M-H+(2)H2O].

**Figure 7.** Matching the *t*Rs of the antioxidant fraction FR 8 (**A**) and (−)-epicatechin (**B**), and the *t*Rs of FR 9 (**A**) and (+)-catechin (**B**). Chromatograms were acquired at 280 nm using the SEC-HPLC method.

**Figure 8.** Chromatogram of FR 8 (**track 1**, 40 µL), (+)-catechin (**track 2**, 400 ng), (−)-catechin (**track 3**, 400 ng), and (−)-epicatechin (**track 4**, 400 ng) applied as 8 mm bands on the HPTLC cellulose plate developed with water, derivatized with DMACA reagent, and documented with illumination with white light.

**Figure 9.** Densitograms of FR 8 (40 µL) and monomeric flavan-3-ol standards (400 ng) scanned at 655 nm in the absorption/reflectance mode on the HPTLC cellulose plate developed with water and derivatized with DMACA.

As in previous experiments with 70% ethanolic(aq) JKRB extract and ascorbic acid methanolic solutions (Sections 3.1 and 3.2), a spectrophotometric DPPH assay was performed to determine the IC<sup>50</sup> value of the free radical scavenging potential of the methanolic solutions of (−)-epicatechin and to test the stability of its antioxidant activity over time (t = 0, 2, 4, 6, 8, 24, and 50 h, and 7 and 14 d). The IC<sup>50</sup> value of the radical scavenging of (−)-epicatechin was ~1.56 µg mL−<sup>1</sup> , which indicated a higher antioxidant activity compared to that of ascorbic acid. Surprisingly, the antioxidant activity of the (−)-epicatechin standard remained constant over time (0 h to 14 days; ~1.56 µg mL−<sup>1</sup> ) (Figure 10, Table 4).

The low IC<sup>50</sup> value of (−)-epicatechin's radical scavenging potential and the stability of its antioxidant activity for at least 14 days indicated that (−)-epicatechin could represent one of the most important contributors to the antioxidant activity of the JKRB extract. The reaction of antioxidants with DPPH is influenced by steric hindrance, with a preference for small antioxidant molecules [18]. Therefore, the antioxidant potential of fractions composed of different molecules is only indicative [18]. The results of the radical scavenging activity of JKRB extract, ascorbic acid, and (−)-epicatechin could not be directly compared to the results of other antioxidant assays.

The antioxidant potential of the whole extract may be the result of a synergistic or additive effect of different matrix compounds, which may be even more potent compared to the isolated single compounds' effect either in the human body [18] or as food antioxidants [76,77]. In the current study, a high time-dependent stability (up to 14 days) of the antioxidant activities of the JKRB extract and (−)-epicatechin (standard solution) was observed.

**Figure 10.** Logarithmic curves plotting the DPPH scavenging effect (%) of (−)-epicatechin against the concentration, measured over time.

**Table 4.** The calculated IC<sup>50</sup> values of the antioxidant activity of (−)-epicatechin tested in the range of 0.1–1000 µM or 0.029–290 µg mL−<sup>1</sup> over time.


#### **4. Conclusions**

Antioxidant activities of Japanese knotweed rhizome bark extracts prepared with water, methanol, 80% methanol(aq), acetone, 70% acetone(aq), ethanol, 70% ethanol(aq), and 90% ethyl acetate(aq) were measured using a DPPH free radical-scavenging assay (IC<sup>50</sup> = 3.561, 3.715, 3.469, 2.632, 3.350, 2.893, 3.503, and 2.786 µg mL−<sup>1</sup> , respectively). Due to the highest extraction yield, the 70% ethanol(aq) extract was selected for further fractionation.

A SEC method was developed for the first time for fractionation of the Japanese knotweed rhizome (bark) extract. Its antioxidant potential was tested for the first time using the SEC-HPLC-UV/Vis method with an on-line post-column DPPH assay. This approach can also be used for the isolation of other plant bioactive constituents. The compounds in the isolated SEC fractions were determined with a new RP-HPLC-UV-MS<sup>n</sup> method. Epicatechin was confirmed in the antioxidant fraction by three orthogonal methods, SEC-HPLC-UV, RP-HPLC-MS, and HPTLC, among which the latter enabled the identification of (−)-epicatechin. The antioxidant activity of the (−)-epicatechin standard was additionally proven with a DPPH free radical-scavenging assay.

The IC<sup>50</sup> values of the antioxidant activity of the selected extract (~3.7 µg mL−<sup>1</sup> ) and of (−)-epicatechin (~1.6 µg mL−<sup>1</sup> ) remained constant for 14 days, while the IC<sup>50</sup> values of ascorbic acid increased over time (3.115–62.787 µg mL−<sup>1</sup> ). The antioxidant activity of the extract was comparable to that of ascorbic acid at t0, while the antioxidant activity of (−)-epicatechin was even higher.

**Author Contributions:** Conceptualization, U.J. and K.N.; methodology, U.J. and K.N.; validation, U.J. and K.N.; formal analysis, U.J.; investigation, U.J. and K.N.; resources, I.V.; data curation, U.J.; writing—original draft preparation, U.J.; writing—review and editing, K.N. and I.V.; visualization, U.J.; supervision, K.N. and I.V.; project administration, I.V.; funding acquisition, I.V. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Slovenian Research Agency (research core funding No. P1-0005 and "Young Researchers" program.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available in this article.

**Acknowledgments:** The authors would like to thank the Slovenian Research Agency (research core funding No. P1-0005 and "Young Researchers" program), Kaja Loboda Bergant for the help with the preparation of graphics in GraphPad Prism 7, Andreja Krušiˇc and Tinka Palkoviˇc for their help with the experimental work, and Jure Zekiˇc for the discussions regarding the protocol for the analysis of ascorbic acid.

**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**

