**Antioxidant, Anti-Inflammatory, and Antibacterial Properties of an** *Achillea millefolium* **L. Extract and Its Fractions Obtained by Supercritical Anti-Solvent Fractionation against** *Helicobacter pylori*

**Marisol Villalva 1, Jose Manuel Silvan 1, Teresa Alarcón-Cavero 2,3, David Villanueva-Bermejo 4, Laura Jaime 4, Susana Santoyo <sup>4</sup> and Adolfo J. Martinez-Rodriguez 1,\***


**Abstract:** The main objective of this work is to evaluate the potential utility of an *Achillea millefolium* extract (yarrow extract, YE) in the control of *H. pylori* infection. The supercritical anti-solvent fractionation (SAF) process of YE allowed the obtaining of two different fractions: yarrow's precipitated fraction (YPF), enriched in most polar phenolic compounds (luteolin-7-*O*-glucoside, luteolin, and 3,5-dicaffeoylquinic acid), and yarrow's separator fraction (YSF), enriched in monoterpenes and sesquiterpenes, mainly containing camphor, artemisia ketone, and borneol. YE was effective in reducing reactive oxygen species (ROS) production in human gastric AGS cells by 16% to 29%, depending on the *H. pylori* strain. YPF had the highest inhibitory activity (38–40%) for ROS production. YE modulated the inflammatory response in AGS gastric cells, decreasing IL-8 production by 53% to 64%. This IL-8 inhibition also showed a strain-dependent character. YPF and YSF exhibited similar behavior, reducing IL-8 production, suggesting that both phenolic compounds and essential oils could contribute to IL-8 inhibition. YSF showed the highest antibacterial activity against *H. pylori* (6.3–7.1 log CFU reduction, depending on the strain) and lower MIC (0.08 mg/mL). Results obtained have shown that YE and SAF fractions (YPF and YSF) were effective as antioxidant, anti-inflammatory, and antibacterial agents regardless of the *H. pylori* strain characteristics.

**Keywords:** *Achillea millefolium*; yarrow extract; *H. pylori*; supercritical anti-solvent fractionation; anti-inflammatory activity; antioxidant activity; antibacterial activity

### **1. Introduction**

*Helicobacter pylori* (*H. pylori*) is one of the most prevalent human pathogens, as over half of the world's population is colonized with this Gram-negative bacterium [1]. The gastric colonization by *H. pylori* occurs asymptomatically in most individuals, although most people infected with *H. pylori* usually have histological changes in gastric mucosa consistent with the presence of gastritis. However, long-term infection with the pathogen can cause a wide range of clinical manifestations associated with several diseases, including gastric inflammation, peptic ulcer, gastric cancer, gastric mucosa-associated lymphoidtissue lymphoma, and other extra-gastric pathologies [2]. Due to the high correlation between *H. pylori* infection and gastric cancer, most therapeutic guidelines aim to eradicate

**Citation:** Villalva, M.; Silvan, J.M.; Alarcón-Cavero, T.; Villanueva-Bermejo, D.; Jaime, L.; Santoyo, S.; Martinez-Rodriguez, A.J. Antioxidant, Anti-Inflammatory, and Antibacterial Properties of an *Achillea millefolium* L. Extract and Its Fractions Obtained by Supercritical Anti-Solvent

Fractionation against *Helicobacter pylori*. *Antioxidants* **2022**, *11*, 1849. https://doi.org/10.3390/ antiox11101849

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 17 August 2022 Accepted: 16 September 2022 Published: 20 September 2022

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

**Copyright:** © 2022 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/).

this pathogen using a combination of antibiotics with a proton pump inhibitor in triple or quadruple therapy [3]. However, there are a number of concerns related to the use of eradicative therapies, especially in asymptomatic individuals. First, the global increase in antibiotic resistance [4] and the significant distress that antibiotic therapy causes in the microbiota [5]; and second, the relationship that has been found between the use of eradicative therapies and the emergence or worsening of other pathologies, such as esophageal reflux [6]. This situation has led to increased interest in bioactive compounds obtained from natural sources for the treatment of *H. pylori* infection [7]. Natural extracts not only with antibacterial activity against *H. pylori* but also with anti-inflammatory and antioxidant properties could be potentially interesting in *H. pylori* treatment [8–11]. This is because the immune response to *H. pylori* is a combination of events involving both protective and damaging responses to the host. In fact, it has been described that much of the pathological evidence related to *H. pylori* infection may be due more to the effects of the host's immune system than to the bacterial infection itself [12].

*Achillea millefolium* L.—traditionally known as yarrow—is a flowering plant commonly used in folk medicine not only in Europe but also in Asia, Africa, and America [13]. Due to the widely known benefits of this plant, the study of its composition and biological properties has awakened a constant interest in developing pharmaceutical, nutraceutical, and food products [14–16]. Dried and fresh upper parts from yarrow have been used to prepare aqueous and alcoholic extracts for the treatment of several health problems, such as diabetes and cardiovascular, respiratory, hepatobiliary, spasmodic, and gastrointestinal disorders [17]. In addition, yarrow has been used externally for the treatment of skin and mucous membrane inflammation [18]. The main bioactive compounds present in different yarrow extracts have been associated to health benefits. The presence of phenolic compounds, specifically chlorogenic and dicaffeoylquinic acids, luteolin, apigenin, and quercetin, as well as volatile fraction constituents, predominantly terpenes, such as borneol, camphor, 1,8-cineole, and chamazulene, have been related to antioxidant, antiinflammatory, antibacterial, antitumor, and antidiabetic properties [19–22]. The antioxidant effect of yarrow extracts has been extensively studied in both in vitro and in vivo models; likewise, the radical scavenging capacity, intracellular oxidative damage, and reduction in lipid peroxidation in rats have been reported [13,23–26]. The anti-inflammatory properties of yarrow ethanolic extracts have shown their role in the suppression of pro-inflammatory cytokines [26,27]. Regarding antibacterial activity, aqueous and ethanolic extracts of yarrow have been effective against different microorganisms, including those causing skin infections, such as *Staphylococcus aureus*, *Staphylococcus epidermidis*, and *Pseudomonas aeruginosa*, and others related with gastrointestinal diseases, such as *Salmonella thypi* and *Escherichia coli* [19,28,29]. However, the effect of yarrow on *H. pylori* is scarcely known despite it being one of the main human pathogens. Only two previous studies screening different extracts obtained from plants used in traditional medicine have shown an antibacterial effect of yarrow against *H. pylori* [30,31]. There are no previous reports on the antioxidant and anti-inflammatory effect of yarrow on human gastric cells infected with *H. pylori*.

Since continuously increasing research on bioactive components and rising interest in high-quality ingredients is evident, manufacturers are motivated to use enriched extracts, fractions, or purified components instead of crude extracts. Furthermore, the use of clean and sustainable extract processes is an essential requirement nowadays. For that purpose, different approaches have been explored to obtain fractions enriched in bioactive molecules from plant extracts, mainly phenolic compounds, such as the use of membrane technology [32], solid-phase extraction with reusable macroporous resins [33], and supercritical anti-solvent fractionation (SAF) with CO2 at supercritical conditions as a solvent [34,35]. The SAF technique has gained interest as a fractionation or purification process with the potential to reduce the number of steps, since as well as separation of compound(s) occurring in the precipitate, a dried enriched-precipitate is produced [36]. Another advantage of SAF is the low use of chemicals and the reduction in waste that is due to CO2 being recycled for further extractions. Recently, we have demonstrated that SAF resulted in an adequate

method to improve the antioxidant and anti-inflammatory properties of a yarrow ethanolic extract [37], although its impact against *H. pylori* is unknown. For this reason, in this study, we have evaluated the antioxidant, anti-inflammatory, and antibacterial properties of a yarrow extract and its fractions obtained by SAF against three different *H. pylori* strains.

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

#### *2.1. Sample Material and Ultrasound-Assisted Extraction of Yarrow*

Inflorescences and upper dried leaves from yarrow (*Achillea millefollium* L.) were purchased from a local herbalist (Murciana Herbolisteria, Murcia, Spain). The sun-dried plant from a Bulgarian variety was ground in a hammer mill (Premill 250, Lleal S.A., Granollers, Spain) and sieved to reduce its particle size (<500 μm). Then, the UAE extraction was carried out by using an ultrasonic device (Branson Digital Sonifier 250, Danbury, CT, USA) with a power of 200 W and frequencies of 60 kHz. For this purpose, 40 g of ground and sieved yarrow plant were added to 400 mL of pure ethanol (Panreac Madrid, Spain) for 30 min at 40 ◦C. An output of 70% with respect to the nominal amplitude was applied during extraction. Finally, the obtained yarrow extract (YE) was concentrated to a final concentration of 17.9 mg/mL by rotary evaporation at 35 ◦C (IKA RV-10 control, VWR, Madrid, Spain) and stored at −20 ◦C.

#### *2.2. Supercritical Anti-Solvent Fractionation (SAF) of Yarrow Extract*

Fractionation of YE was performed by means of a piece of supercritical technology equipment (Thar SF2000, Thar Technology, Pittsburgh, PA, USA) with two pumps for the separate supply of supercritical CO2 (SC-CO2) and YE solution, and a precipitation vessel and two separators' vessels (0.5 L each), with independent control of temperature and pressure as described by Villanueva-Bermejo et al. [35]. Briefly, SC-CO2 was pumped into the precipitation vessel until 15 MPa of pressure and 40 ◦C were attained. Then, the solution of YE (17.9 mg/mL concentration) was pumped into the precipitator while maintaining the SC-CO2 flow. A CO2/extract flow ratio of 31.3 g/g (50 g/min for CO2 and 1.6 g/min for YE) was employed. During the process, both separators' vessels were kept at ambient pressure. After system depressurization, two fractions were collected, one corresponding to the YE components that were not soluble in the SC-CO2+ethanol mixture and precipitated in the precipitation vessel (yarrow's precipitated fraction, YPF). The second fraction corresponded to the YE components soluble in the SC-CO2+ethanol recovered in the separators (yarrow's separator fraction, YSF) with an oleoresin appearance. To obtain a dried YSF fraction, the samples of both separator vessels were recovered with ethanol and combined in a single fraction to finally remove the solvent by rotary evaporation under vacuum. The YPF and YSF fractions were kept at −20 ◦C in darkness until analysis.

#### *2.3. Chemical Characterization of YE and Its Fractions by HPLC-PAD-ESI-QTOF-MS and GC-MS Analyses*

The phenolic composition was determined by HPLC using an Agilent HPLC 1260 Infinity series system (Agilent Technologies Inc., Santa Clara, CA, USA) according to the Villalva et al. [37] methodology. Chromatographic separation was carried out by using a reverse phase ACE Excell 3 Super C18 column (150 mm × 4.6 mm, 3 μm particle size) from Advanced Chromatography Technologies (Aberdeen, Scotland), thermostated at 35 ◦C and protected by an ACE 3 C18-AR (10 mm, ×3 mm) guard column. Dry samples were dissolved in DMSO (HPLC grade, ≥99.7%) (Merck, Madrid, Spain) to allow a final concentration of 5 mg/mL and filtered by a PVDF filter (0.45 μm) before injection (20 μL). For identification purposes, the retention time (Rt) and UV–Vis spectrum of each chromatographic peak were compared with the analytical standards (Phytolab, Madrid, Spain); additionally, the accurate mass from HPLC-ESI-QTOF-MS in negative mode analysis was used for compounds assignment, as previously described in Villalva et al. [37]. For quantification, standard calibration curves were built for each pure compound, namely, caffeic acid, caftaric acid, chlorogenic acid, cryptochlorogenic acid, 1,5-dicaffeoylquinic

acid (DCQA), 3,4-DCQA, 3,5-DCQA, 4,5-DCQA, ferulic acid, neochlorogenic acid, rosmarinic acid, apigenin, apigenin-7-*O*-glucoside, diosmetin, homoorientin, luteolin-*β*-7- O-glucuronide, luteolin-7-*O*-glucoside, schaftoside, vicenin 2, casticin, quercetin, rutin, and vitexin. Moreover, luteolin-6,8-di-*C*-glucoside and 6-hydroxyluteolin-7-*O*-glucoside were quantified by the calibration curve of orientin and luteolin-7-*O*-glucoside. In addition, vicenin 2 and schaftoside calibration curves were used for apigenin-*C*-hexoside-*C*pentoside and schaftoside isomer quantification; as well, quercetin and casticin were used for methoxyquercetin isomer and centaureidin, respectively.

Volatile compounds from yarrow extracts were characterized by GC-MS using an Agilent 7890A system (Agilent Technologies, Santa Clara, CA, USA) equipped with a split/splitless auto-injector (G4513A), a flame ionization detector, a triple-axis mass spectrometer detector (5975C), and GC/MS Solution software. Extracts were dissolved in ethanol (5 mg/mL final concentration), filtered (0.45 μm), and injected (1 μL) in splitless mode. Then, the chromatographic analysis was carried out as described by Villalva et al. [37]. Briefly, the mass spectrometer operated under electron impact mode (70 eV) and it was used in total ion current (TIC) mode (mass range from 40 to 500 *m*/*z*). The analysis was performed using an Agilent HP-5MS capillary column (30 m × 0.25 mm i.d., 0.25 μm phase thickness) and the following chromatographic method: 40 ◦C initial temperature, from 40 ◦C to 150 ◦C at 3 ◦C min−1, isothermal at 150 ◦C for 10 min, then increased from 150 to 300 ◦C at 6 ◦C min−1, and finally isothermal at 300 ◦C for 1 min. Helium (99.99%) was employed as the carrier gas (1 mL/min flow rate). The temperature used for the injector was 250 ◦C. For the identification of volatile compounds, the obtained mass spectral fragmentation patterns were contrasted with those of the Wiley 229 mass spectral library. In addition, their corresponding retention indices were calculated and compared to the information reported in the literature [38–41] and contained in the NIST database.

#### *2.4. Helicobacter pylori, Growth Media, and Culture Conditions*

*H. pylori* strains (Hp48, Hp53, and Hp59) were isolated from gastric mucosal biopsies obtained from symptomatic patients from the Microbiology Department, Hospital Universitario La Princesa (Madrid, Spain). Biopsies were cultured in selective (Pylori agar, BioMerieux, Madrid, Spain) and non-selective media (blood-supplemented Columbia Agar, BioMerieux, Madrid, Spain). Hp48 and Hp59 strains are resistant to metronidazole, while Hp53 is a multi-resistant strain with resistance to amoxicillin, clarithromycin, and rifampicin. *H. pylori* strains were stored at −80 ◦C in Brucella broth (BB) (Becton, Dickinson, & Co., Madrid, Spain) with 20% glycerol. The agar-plating medium consisted of Müeller– Hinton agar supplemented with 5% defibrinated sheep blood (MHB) (Becton, Dickinson, & Co.), and the liquid growth medium consisted of BB supplemented with 10% horse serum (HS) (Biowest, Barcelona, Spain). *H. pylori* inoculum strains were prepared as follows: the frozen stored strains were reactivated by inoculation (200 μL) in a MHB plate and incubation in a microaerophilic atmosphere using a variable atmosphere incubator (VAIN) (85% N2, 10% CO2, 5% O2) (MACS-VA500, Don Whitley Scientific, Bingley, UK) at 37 ◦C for 72 h. Bacterial biomass grown in one MHB plate was collected with a sterile cotton swab and suspended in 2 mL of BB supplemented with 10% HS (BB-HS) or a culture medium cell (~1 × <sup>10</sup><sup>8</sup> colony forming units/mL (CFU/mL)), and was used as an experimental bacterial inoculum in the different experimental assays.

#### *2.5. Human Gastric Epithelial Cell Cultures*

The human gastric epithelial cell line AGS was obtained from the American Type Culture Collection (ATCC, Barcelona, Spain). Cells were grown in Dulbecco's Modified Eagle's Medium/F12 (DMEM/F12) (Lonza, Madrid, Spain) supplemented with 10% fetal bovine serum (FBS) of South American origin (Hyclone, GE Healthcare, Logan, UK) and 1% penicillin/streptomycin (5000 U/mL) (Lonza). Cells were plated at densities of ~1 × 106 cells in 75 cm2 culture flasks (Sarstedt, Barcelona, Spain) and incubated at 37 ◦<sup>C</sup> under 5% CO2 in a humidified incubator until 90% confluence was reached. The culture

cell medium was changed every two days. Before a confluent monolayer appeared, a cell sub-culturing process was carried out. All experiments were performed between passage 5 and 15 to ensure cell uniformity and reproducibility.

#### *2.6. Cell Viability*

Before conducting experiments on antioxidant and anti-inflammatory activity, it was necessary to evaluate the potential cytotoxicity of YE and its fractions (YPF and YSF) against the AGS cell line at different concentrations. For this purpose, cell viability was determined by the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Merck) reduction assay, as was previously described by Silvan et al. [10]. Confluent cell cultures (~90%) were trypsinized (Trypsin/EDTA solution 170,000 U/L) (Lonza) and cells were seeded (~5 × <sup>10</sup><sup>4</sup> cells per well) in 96-well plates (Sarstedt) and incubated in cell culture medium at 37 ◦C under 5% CO2 in a humidity incubator for 24 h. Cell culture medium was replaced with a serum-free cell culture medium containing YE and its fractions at 0.4, 0.2, and 0.08 mg/mL (final concentration), and cells were incubated at 37 ◦C under 5% CO2 in a humidity incubator for 24 h. Viability control cells (non-treated) were incubated in a serum-free cell culture medium without samples. Thereafter, cells were washed twice with phosphate-buffered saline (PBS) (Lonza), and the medium was replaced with 200 μL of serum-free cell culture medium plus 20 μL of MTT solution in PBS (5 mg/mL) that were added to each well for the quantification of the living, metabolically active cells after 1 h incubation at 37 ◦C under 5% CO2 in a humidity incubator. MTT is reduced to purple formazan in the mitochondria of living cells. Formazan crystals in the wells were solubilized in 200 μL of DMSO. After incubation, cell concentration was estimated as ranging from ~5 × <sup>10</sup><sup>4</sup> to 5.5 × <sup>10</sup><sup>4</sup> cells per well. Finally, absorbance was measured at 570 nm wavelengths, employing a microplate reader Synergy HT (BioTek Instruments Inc., Winooski, VT, USA). Cell viability was calculated considering controls containing the serum-free medium as 100% viable cells, and using the following formula:

Cell viability (%) = (absorbance of sample)/(absorbance of control) × 100

Data represent the mean and standard deviation (SD) of triplicates of three independent experiments (*n* = 9).

#### *2.7. Antioxidant Activity of YE and Its Fractions against Intracellular Reactive Oxygen Species (ROS) Production on H. pylori-Infected Gastric Cells*

Intracellular ROS were measured by the DCFH-DA (carboxy-2- ,7- -dichloro-dihydrofluorescein diacetate) (Merck) assay, as previously reported by Silvan et al. [10]. Cells were seeded (~5 × <sup>10</sup><sup>4</sup> cells per well in 500 <sup>μ</sup>L) in 24-well plates (Sarstedt) and incubated at 37 ◦<sup>C</sup> under 5% CO2 in a humidity incubator until a monolayer was formed. Cells were incubated with YE and its fractions (YPF and YSF) (0.08 mg/mL) dissolved in a serum-free cell culture medium for 24 h. After that, cells were washed twice with PBS and incubated with 20 mM DCFH-DA (Merck) at 37 ◦C for 30 min. Next, cells were washed twice with PBS to remove the unabsorbed probe and were then infected with *H*. *pylori* inoculum strains (500 μL) suspended in a serum/antibiotics-free cell culture medium (~1 × <sup>10</sup><sup>8</sup> CFU/mL). ROS production was immediately monitored for 180 min in a Synergy HT (BioTek Instruments Inc.) fluorescent microplate reader using λex 485 nm and λem 530 nm. After incubation, cell concentration was estimated as ranging from ~5 × <sup>10</sup><sup>5</sup> to 5.5 × <sup>10</sup><sup>5</sup> cells per well. After being oxidized by intracellular oxidants, DCFH-DA changes to dichloro-fluorescein (DCF) and emits fluorescence. Cells incubated only with the *H. pylori* inoculum were used as an oxidation control (100% of intracellular ROS production). All samples were analyzed in triplicate in three independent experiments (*n* = 9).

#### *2.8. Anti-Inflammatory Activity of YE and Its Fractions on H. pylori-Infected Gastric Cells*

The inflammatory response was evaluated as IL-8 production in AGS cells after being infected with different *H. pylori* strains following the procedure described by Silvan et al. [9]. Briefly, human gastric AGS cells were seeded (~5 × <sup>10</sup><sup>4</sup> cells/well) in 24-well plates (Sarstedt) and incubated in a cell culture medium at 37 ◦C under 5% CO2 in a humidity incubator until a monolayer was formed. Cells were incubated with YE and its fractions (YPF and YSF) (0.08 mg/mL) at 37 ◦C in a 5% CO2 humidified atmosphere for 2 h. Cells were washed twice with PBS and infected with 0.5 mL of *H. pylori* inoculum prepared in a serum/antibiotics-free cell culture medium (~1 × <sup>10</sup><sup>8</sup> CFU/mL for all tested strains). The infected cells were incubated at 37 ◦C under 5% CO2 for 24 h to allow the bacteria to adhere and invade the cells. Uninfected and nontreated cells were included in the experiment as a negative and positive control of IL-8 production, respectively. At the end of incubation, cell supernatants were collected, particulate material was removed by centrifugation (10 min at 12,000 rpm), and samples were stored at −20 ◦C until analyses were performed. The amounts of secreted interleukin IL-8 in the collected supernatant from gastric epithelial cell samples were determined by an ELISA assay. A commercially available ELISA kit (Diaclone, Besancon, France) for the quantitation of IL-8 cytokine was used as described by the manufacturer's instructions. Absorbance was measured at 450 nm using a microplate reader Synergy HT (BioTek Instruments Inc.). Since, in the absence of bacteria, gastric AGS cells release small amounts of IL-8 [42], titers of cytokine released by AGS cells (pg/mL) were determined experimentally. The data represent the mean and SD of triplicates of three independent experiments (*n* = 9).

#### *2.9. Antibacterial Activity of YE and Its Fractions against H. pylori Strains*

The antibacterial activity of YE and its fractions (YPF and YSF) against the *H. pylori* strains was tested following the procedure described by Silvan et al. [10]. Briefly, 1 mL of the sample at 0.4, 0.2, 0.14, and 0.08 mg/mL (final concentration) was transferred into different flasks containing 4 mL of BB-HS. Bacterial inoculum (100 <sup>μ</sup>L of ~1 × 108 CFU/mL) was then inoculated into the flasks under aseptic conditions. The culture was incubated in the VAIN in the conditions described above. *H. pylori* growth controls were prepared by transferring 100 <sup>μ</sup>L of bacterial inoculum (~1 × 108 CFU/mL) to 5 mL of BB-HS. After 24 h incubation, serial decimal dilutions of cultures were prepared in 0.9% saline solution (NaCl). Then, they were plated onto fresh MHB agar and incubated at 37 ◦C under microareophillic conditions in the VAIN. After 72 h of incubation, the CFU were assessed. Results were expressed as CFU/mL.

#### *2.10. Statistical Analysis*

Results were reported as means ± SD. Significant differences among the data were estimated by applying analysis of variance (ANOVA). Tukey's least significant differences (LSD) test was used to evaluate the significance of the analysis. Differences were considered significant at *p* < 0.05. All statistical tests were performed with IBM SPSS Statistics for Windows, Version 27.0 (IBM Corp., Armonk, NY, USA).

#### **3. Results**

#### *3.1. Characterization of YE and Its Fractions*

Phenolic composition of YE and its fractions and the details of HPLC-ESI-QTOF-MS of the identified phenolic compounds are shown in Table 1 and Table S1 (supplementary material). Phenolic compounds from two different families (phenolic acids and flavonoids) were identified in the extract and fractions. Flavonoids were the major family within YE (2924.4 mg/100 g), constituting 77% of the total phenolic compounds. Among the flavonoids, flavones were the prevalent group (2018.1 mg/100 g), representing 69% of flavonoids. Luteolin-7-*O*-glucoside (768.7 mg/100 g; 38% of flavones) and luteolin (447.4 mg/100 g; 22.2% of flavones) were the main compounds identified within YE and in the flavones group.


**Table 1.** Phenolic composition and quantification of YE and its fractions (mg/100 g dry sample).

YE: yarrow extract. YPF: yarrow's precipitator fraction. YSF: yarrow's separator fraction. <L.Q.: below limit of quantification. <sup>1</sup> Comparison with authentic standard. a,b,c Values in the same row marked with different superscript letters indicates statistical differences (*p* < 0.05).

In the flavonols group (906.3 mg/100 g; 31% of flavonoids), centaureidin and methoxyquercetin isomer were the major compounds accounting for 43% and 41% of total flavonols, respectively. With regards to phenolic acids (22% of total phenolic compounds), chlorogenic acid and its derivatives (1,5-DCQA, 3,4-DCQA, 3,5-DCQA, and 4,5-DCQA) were predominant (75% of total phenolic acids), 3,5-DCQA (361.7 mg/100 g) being the most abundant phenolic acid in YE (43% of total phenolic acids). Rosmarinic acid content was also relevant in YE (22% of total phenolic acids).

Concerning SAF fractions, YPF showed a similar phenolic composition to YE, but it was enriched 2.4 times in total phenolic compounds (9060 mg/100 g) in comparison with YE (3768 mg/100 g). Flavonoid content increases up to 7093.3 (2.4 times more than YE), representing 78% of total phenolic compounds in YPF, similar to that obtained in YE (77%). Within the flavonoids compounds, and as was observed in YE, flavones were the prevalent group (5932.1 mg/100 g), increasing its content up to 83% of total flavonoids, luteolin-7- *O*-glucoside (40% of total flavones) and luteolin (22% of total flavones) being the major compounds in this class of compounds. Phenolic acids concentration increases 2.3 times in YPF compared to YE, chlorogenic acid derivatives being the major compounds in this fraction (98% of total phenolic acids), outstanding the 3,5-DCQA as the most abundant phenolic acid (1163.4 mg/100 g).

On the other hand, only some low-polarity phenolic compounds were recovered as part of YSF (1325.7 mg/100 g), mainly flavonoid compounds (1296.8 mg/100 g) representing 97% of total phenolic compounds identified. Mostly, aglycones of flavonoids, the lesser polar compounds originally described in YE, were found in this fraction. Among them, the biflavonoid amentoflavone (62.2 mg/100 g) and methoxylated flavonols casticin (61.8 mg/100 g) and centaureidin (669.6 mg/100 g) were in significantly (*p* < 0.05) higher concentrations in YSF than in YE. Due to the oleoresin appearance of YSF, it was expected that it could contain volatile oil components. That hypothesis was confirmed with a GC-MS analysis and the results are presented in Table 2.


**Table 2.** Volatile compounds identified by GC-MS in YE and YSF represented as peak area contributions (normalized percentage of area).

Rt: retention time. YE: yarrow extract. YSF: yarrow's separator fraction. N.i.: non-identified compound. AUC: area under curve.

As shown, a great abundance of monoterpenes and sesquiterpernes was found for both YE and YSF. In particular, four monoterpenes, camphor, artemisia ketone, borneol, and 2,6-dimethyl-1,7-octadiene-3,6-diol, were the most abundant compounds in both extracts. When comparing the total peak area contribution, it can be observed that YSF (43.8 × <sup>10</sup><sup>6</sup> AUC) represented a double richness of volatile compounds with respect to YE (23.6 × <sup>10</sup><sup>6</sup> AUC). The fraction obtained in the precipitator vessel (YPF) was also analyzed; however, as expected, it lacks volatile components (data not shown).

#### *3.2. Antioxidant Activity of YE and Its Fractions against Intracellular ROS Production in H. pylori-Infected AGS Cells*

Before the antioxidant activity experiments, the viability of the AGS cells was evaluated in the presence of YE and its corresponding fractions (YPF and YSF). For this purpose, AGS cells were placed in contact with variable concentrations of YE and its fractions (0.08 to 0.40 mg/mL), and the MTT assay was performed. The data obtained demonstrated

that the maximum concentration that was non-cytotoxic resulted in 0.08 mg/mL for all tested samples (>95% cell viability) (data not shown). Higher concentrations of YE and its fractions significantly reduced cell viability compared to the control of untreated cells (viability lower than 80%). Infection of gastric cells with *H. pylori* strains (Hp48, Hp53, and Hp59) induced ROS production in AGS cells (data not shown), as has been demonstrated in previous works using this cell model [10]. As shown in Figure 1, in all cases, YE and YPF significantly (*p* < 0.05) reduced intracellular ROS production in AGS-infected cells in comparison with the control group (untreated infected cells). However, YSF had a strain-dependent behavior and only significantly inhibited (*p* < 0.05) ROS production when AGS cells were infected with the Hp48 strain.

**Figure 1.** Inhibition effect of yarrow extract (YE) and its fractions (YPF and YSF) (0.08 mg/mL) on ROS production by human gastric epithelial AGS cells after *H. pylori* strains infection. Values are the mean ± SD (*n* = 3). \* Asterisk indicates significant differences compared to the untreated infected control (no inhibition) (*p* < 0.05). a,b,c,d,e Different letters indicate statistical difference between samples and *H. pylori* strains (*p* < 0.05).

The inhibition effect of YE on ROS production ranged from 16% to 29% depending on the *H. pylori* strain. YPF, the fraction enriched in phenolic compounds, was the most active fraction regardless of the strain used. It provoked the inhibition of intracellular ROS production of about 40%. YSF, which contained only the most non-polar phenolic compounds, showed a lower antioxidant activity (3–14%) than the YE, which is also coherent with the presence of lower amounts of phenolic compounds in this fraction.

#### *3.3. Effect of the YE and Its Fractions on the Inflammatory Response Induced by H. pylori in AGS Cells*

Previously, we evaluated in vitro the secretion of different pro-inflammatory cytokines produced in *H. pylori*-infected AGS cells, IL-8 being the most secreted cytokine, similarly to that described by others [2]. For this reason, we selected IL-8 as a biomarker to evaluate the anti-inflammatory effect of YE and its fractions on AGS cells infected by *H. pylori* strains. As can be observed in Figure 2, the background level of IL-8 production in uninfected AGS cells was 105.0 ± 12.0 pg/mL (Ctrl. AGS; untreated and uninfected cells). Infection with *H. pylori* strains effectively stimulated the secretion of IL-8 pro-inflammatory cytokine (Ctrl. Hp; untreated infected control) in AGS cells (413 to 521 pg/mL). Furthermore, IL-8 production showed a strain-dependent character, since statistical differences between strains were found (*p* < 0.05).

**Figure 2.** Inhibition effect of yarrow extract (YE) and its SAF fractions (YPF, yarrow's precipitator fraction and YSF, and yarrow's separator fraction) (0.08 mg/mL) on pro-inflammatory cytokine IL-8 production (pg/mL) by human gastric epithelial AGS cells infected by *H. pylori* strains. Control Hp (Ctrl Hp) represents the values obtained from untreated cells infected with *H. pylori* strains. Control AGS (Ctrl AGS) represents the values obtained from untreated and uninfected AGS cells. Values are the mean <sup>±</sup> SD (*<sup>n</sup>* = 3). a–g Different letters indicate statistical differences between treatments for each *H. pylori* strain (*p* < 0.05).

For all strains, YE significantly (*p* < 0.05) decreased IL-8 production by 53% to 64% when compared to its respective control Hp. Unlike the antioxidant activity, it was more difficult in this case to evaluate the impact of each fraction on the observed behavior. For two of the strains (Hp48 and Hp59), both the fraction enriched in phenolic compounds (YPF) and the fraction containing essential oils (YSF) showed similar behaviors, reducing IL-8 production, suggesting that the two types of compounds could contribute to IL-8 inhibition. In contrast, for the Hp53 strain, YPF presented a greater contribution to IL-8 inhibition than YSF (*p* < 0.05).

#### *3.4. Antibacterial Activity of YE and Its Fractions against H. pylori Strains*

The antibacterial effect of YE, YPF, and YSF against *H. pylori* growth is presented in Table 3. YE was significantly (*p* < 0.05) effective as an antibacterial agent against all *H. pylori* strains tested, although the effect was greater or lesser depending on the strain and varied in a range of CFU reduction between 4.8 and 7.1 log. However, MIC was the same for all strains (0.14 mg/mL). Analysis of the contribution of each fraction to the antibacterial effect showed that YSF, the fraction enriched in volatile compounds, had a significantly (*p* < 0.05) greater antibacterial effect (6.3–7.1 log CFU reduction, depending on the strain) and lower MIC (0.08 mg/mL) than YE. On the other hand, phenolic-enriched YPF also significantly (*p* < 0.05) reduced bacterial growth of all strains and this reduction was independent of the strain used.

**Table 3.** Antibacterial activity of YE and its fractions (YPF and YSF) at 0.4 mg/mL against *H. pylori* strains. Results represent the mean ± standard deviation of colony forming units (CFU)/mL (*n* = 3).


CFU detection limit was 1.00 × <sup>10</sup><sup>2</sup> CFU/mL. MIC: minimal inhibitory concentration (mg/mL). YE: yarrow extract. YPF: yarrow's precipitator fraction. YSF: yarrow´s separator fraction. a,b,c,d Different lowercase letters denote significant differences within a line (*p* < 0.05). A,B Different uppercase letters denote significant differences within a column (*p* < 0.05).

#### **4. Discussion**

The phenolic composition of the YE obtained by ethanolic extraction was similar to that reported in previous works for this same yarrow variety [26,35]. The use of ethanol or ethanol mixtures as extraction solvents has been described as a useful method to obtain extracts rich in bioactive phenolic compounds and volatile essential oils from yarrow [26,43]. Because of the well-known bioactivities of phenolic compounds contained in yarrow, the SAF technique was employed to selectively obtain enriched fractions from YE, according to its greater or lesser affinity to the SC-CO2 and ethanol mixture performing as solvents. YPF was enriched in phenolic compounds, while YSF was enriched in monoterpenes and sesquiterpenes, which are very abundant compounds in yarrow's essential oil [13,19]. It has been described that these fractions represent an advantage in the recovering of the extract with high purity and free of solvent, contributing to producing high-quality products [36]. YE and its fractions (YPF and YSF) demonstrated their potential utility for use in both the control of *H. pylori* growth and the modulation of the oxidative and inflammatory response of the human gastric cells associated with *H. pylori* infection. Modulation of the oxidative and inflammatory response in the gastric epithelium has been shown to be particularly relevant in preventing tissue damage and the progression of pathologies associated with *H. pylori* infection [2]. YPF, which presents phenolic compounds 2.4 times more concentrated than YE, had the highest inhibitory activity for ROS production. This behavior seems consistent with the potent antioxidant activity described for many of the major phenolic compounds identified in this fraction. For example, the flavones luteolin-7-*O*-glucoside and luteolin, the most predominant phenolic compounds in the YPF fraction, have been described as potent antioxidant agents, since their molecular structure, formed by a 2–3 carbon double bond of C ring (C2=C3) conjugated with a carbonyl group in C4, confers them with the capacity to react and neutralize ROS, behaving as scavengers in the cellular processes that generate this type of molecules [44]. Other major compounds in YE concentrated in the YPF, such as 3,5-DCQA, have also been shown to have a relevant capacity to scavenge intracellular ROS [45]. In general, since in YPF most of the phenolic compounds present are in a higher concentration than in YE, it is expected that many of them, whose antioxidant properties have been described [46–48], may contribute to a higher inhibition of ROS production found for YPF. On the other hand, the scarce presence of phenolic compounds in the YSF fraction was correlated with low antioxidant activity. The high antioxidant capacity of phenolic-compound-enriched YPF also led to a decrease in IL-8 production. In the case of YE and YSF, not only phenolic compounds but also some essential oils seemed to be involved in their anti-inflammatory capacity. Similar results have recently been obtained evaluating the effect of a yarrow extract and its fractions on differentiated human macrophages, observing that the inhibition in the secretion of some pro-inflammatory cytokines (IL-6, IL-1*β*, and TNF-α) could be related to the presence of essential oils such as camphor, borneol, or artemisia ketone, which constituted approximately 30% of the fraction studied [37]. Likewise, luteolin-7-*O*glucoside and luteolin, predominant phenolic compounds in the YE and YPF, have been shown to be able to downregulate IL-1*β*, IL-6, and TNF-*α* production acting on NF-κB, MAPK, and JAK/STAT inflammatory pathways by reducing inflammation in cellular models [44,49]. It has also been reported in experiments carried out with a hydro-alcoholic extract of thistle that 3,5-dicaffeoylquinic acid, another major phenolic compound in YE and YPF, was primarily responsible for inhibiting the secretion of IL-8 and NF-κB pathways in human gastric epithelial AGS cells [50].

Although YE and its two fractions were effective as inhibitors of *H. pylori* growth, the contribution of YSF was higher in the antibacterial activity of the extracts. Numerous essential oils are known to have significant antibacterial activity against *H. pylori* [51]. Particularly in yarrow, the major volatile compounds identified (camphor, borneol, and artemisia ketone) have also been shown to be effective as inhibitors of *H. pylori* growth [48–54]. Although the *H. pylori* strain may influence the intensity of the bioactive response obtained, the present work showed that YE and its fractions were effective as antioxidant, anti-inflammatory, and antibacterial agents regardless of the characteristics of the used strain.

#### **5. Conclusions**

Among other uses, yarrow has been widely utilized as a part of folk medicine to alleviate symptoms related to gastrointestinal discomfort, many of them similar to those associated with *H. pylori* infection. The historical background of its efficacy in the treatment of these pathologies is complemented in this work by more scientifically based evidence to support the pharmacological effects of various compounds present in YE against *H*. *pylori*. YE may be potentially effective in combating oxidative stress and modulating the inflammatory response associated with gastric *H. pylori* infection. In addition, YE exhibits strong antibacterial activity against *H. pylori*. Both the phenolic compounds and essential oils present in the extract appear to contribute to the bioactive properties of the extract, although the degree of contribution varies depending on each property (antioxidant, antiinflammatory, or antibacterial). The SAF technique allows the obtaining of YE fractions enriched in phenolic compounds or essential oils, on the basis of the concept of green extraction, and may be useful in the design of bioactive extracts against *H. pylori* in which it is desirable to enhance specific bioactivity. This approach is attractive in terms of cost, tolerability, and cultural acceptability and can be especially useful in those countries where modern health facilities and access to certain pharmacological substances are not always adequate or available.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antiox11101849/s1, Table S1: Phenolic compounds identified in yarrow samples by using HPLC-ESI-QTOF-MS.

**Author Contributions:** Conceptualization, A.J.M.-R., L.J., S.S. and M.V.; methodology, A.J.M.-R. and J.M.S.; validation, A.J.M.-R., J.M.S. and M.V.; formal analysis, D.V.-B., J.M.S., M.V.; investigation, J.M.S., D.V.-B. and M.V.; resources, A.J.M.-R., T.A.-C., L.J. and S.S.; data curation, A.J.M.-R., J.M.S. and M.V.; writing—original draft preparation, A.J.M.-R., J.M.S. and M.V.; writing—review and editing, A.J.M.-R., J.M.S. and M.V.; visualization, A.J.M.-R. and J.M.S.; supervision, J.M.S. and M.V.; project administration, A.J.M.-R.; funding acquisition, A.J.M.-R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Projects AGL2017-89566-R (HELIFOOD) (MCIN/AEI/10.13039/ 501100011033/ (Spanish Ministry of Science and Innovation) and Fondo Europeo de Desarrollo Regional (FEDER) "Una manera de hacer Europa") and ALIBIRD-CM2020 P2018/BAA-4343 (Comunidad de Madrid, Spain).

**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.

**Acknowledgments:** The authors thank Soledad Diaz Palero for experimental support (Garantía Juvenil CAM 2020, Ref. 37722).

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

#### **References**


#### *Comment*

## **Comment on Villalva et al. Antioxidant, Anti-Inflammatory, and Antibacterial Properties of an** *Achillea millefolium* **L. Extract and Its Fractions Obtained by Supercritical Anti-Solvent Fractionation against** *Helicobacter pylori***.** *Antioxidants* **2022,** *11***, 1849**

**Rafał Fra ´nski 1,\* and Monika Beszterda-Buszczak <sup>2</sup>**


**Abstract:** Villalva et al. evaluated the potential utility of an *Achillea millefolium* (yarrow) extract in the control of *H. pylori* infection. The agar-well diffusions bioassay was applied to determine the antimicrobial activity of yarrow extracts. The supercritical anti-solvent fractionation process of yarrow extract was made to give two different fractions with polar phenolic compounds and monoterpenes and sesquiterpenes, respectively. Phenolic compounds were identified by HPLC-ESIMS by using the accurate masses of [M−H]<sup>−</sup> ions and the characteristic product ions. However, some of the reported product ions seem to be disputable, as described below.

**Keywords:** yarrow; flavonoids; fragmentation pathway; mass spectrometry; electrospray ionization

**Citation:** Fra ´nski, R.; Beszterda-Buszczak, M. Comment on Villalva et al. Antioxidant, Anti-Inflammatory, and Antibacterial Properties of an *Achillea millefolium* L. Extract and Its Fractions Obtained by Supercritical Anti-Solvent Fractionation against *Helicobacter pylori*. *Antioxidants* 2022, *11*, 1849. *Antioxidants* **2023**, *12*, 1226. https://doi.org/10.3390/ antiox12061226

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 8 November 2022 Revised: 21 April 2023 Accepted: 1 June 2023 Published: 7 June 2023

**Copyright:** © 2023 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/).

Eradication of *Helicobacter pylori* has become a serious challenge due to increasing antimicrobial resistance. That Gram-negative, microaerophilic bacterium that is known to affect over 50% of the worldwide population occurs initially during childhood but, if left untreated, could persist for life, resulting in a broad spectrum of gastropathies. It is particularly important because *H. pylori* is involved in the development of 80% of gastric cancers and 5.5% of all malignant conditions worldwide [1–3].

The most effective conventional therapies for *H. pylori* infection require a minimum of two antibiotics (commonly, amoxicillin and clarithromycin) in combination with a gastric acid inhibitor or bismuth to guarantee high eradication rates. That is why nowadays, much research work is concentrated on the search for new potential anti-*H. pylori* candidates, among others, vegetables and plant extracts showing antibacterial properties [4,5]. They act by inhibiting bacterial enzymes, suppressing nuclear factor-κB, adhesions with gastric mucosa, and by inhibiting oxidative stress [6,7]. Organic extracts (ethanol, methanol, acetone, chloroform, petroleum ether, and mixtures of mentioned) of *Acacia nilotica*, *Alchornea triplinervia*, *Arrabidaea chica*, *Bridelia micrantha*, *Calotropis procera*, *Camellia sinensis*, *Carum carvi*, *Cocculus hirsutus*, *Derris trifoliate*, *Geranium wilfordii*, *Hydrastis canadensis*, *Myristica fragrans*, *Xanthium brasilicum*, and *Trachyspermum copticum* and many others have demonstrated antibacterial activity against clinical isolates of *H. pylori* [8–15]. In the study by Villalva and coworkers, the potential utility of an *Achillea millefolium* (yarrow) extract in the control of *H. pylori* infection was evaluated [16]. The supercritical anti-solvent fractionation process of yarrow extract was made to give two different fractions with polar phenolic compounds and monoterpenes and sesquiterpenes.

Yarrow (*Achillea millefolium* L.) is one of the most commonly used medicinal herbs, in both folk and conventional medicine, for over 3000 years, growing wild and as a cultivated plant in the region of Eurasia and North America [17]. According to the conducted studies, *Achillea millefolium* L. is a biologically active plant that demonstrates multiple beneficial effects, including antioxidant, anti-inflammatory, spasmolytic, diaphoretic, hepatoprotective, choleretic, antipyretic, analgesic, antimicrobial and anticancer properties [18–20]. *Achillea millefolium* L. helps eliminate toxins from the body, controls bleeding, lowers blood pressure, relieves menstrual pain, and is used in the treatment of various diseases such as diabetes, tuberculosis, Alzheimer's, and Parkinson's disease [21–23]. Moreover, as indicated by Tilwani and coworkers, the yarrow-treated SARS-nCoV-2 cell exhibits the disintegration of the virus membrane [24].

Different extracts (hexane, petroleum ether, and methanol) of *A. millefolium* aerial parts were found to be active toward the following pathogens: *Bacillus cereus*, *Staphylococcus aureus*, *Escherichia coli*, *Klebsiella pneumoniae*, *Pseudomonas aeruginosa*, *Salmonella enteritidis*, *Yersinia enterocolitica*, *Aspergillus niger*, and *Candida albicans* [17,25,26]. As the literature indicates, the Gram-positive bacteria were more susceptible than the Gram-negative ones, whereas *S. aureus* and *B. cereus* were the most susceptible Gram-positive bacteria [25,27].

In the study by Villalva et al., the agar-well diffusions bioassay was applied to determine the antimicrobial activity of yarrow extracts (YE). YE was significantly (*p* < 0.05) effective as an antibacterial agent against all *H. pylori* strains tested (Hp48, Hp53, and Hp59) in the range of CFU reduction between 4.8 and 7.1 log [16]. Moreover, even better results were obtained for the fraction enriched in volatile compounds. In turn, in the experiment with the inflammatory response induced by *H. pylori* in AGS cells, interleukin 8 (IL-8) has been used as a biomarker. It is a well-documented fact that *H. pylori* infection is associated with an increase in IL-8 concentration, both in vitro and in vivo; it was among the first cytokines described to be produced by infected gastric epithelium; and leads to the recruitment of leukocytes in the gastric mucosa, representing a major step in the regulation of immune-inflammatory responses [2]. In the study by Villalva et al., the application of YE decreased IL-8 synthesis by 53% to 64% in human gastric epithelial cells, with a suggestion that the two types of compounds could contribute to this inhibition—not only phenolic compounds but also some essential oils. In addition, the authors analyzed the antioxidant activity of YE and its fractions against intracellular reactive oxygen species (ROS) synthesis in *H. pylori*-infected AGS cells. The inhibition effect of YE on ROS production depended on the examined *H. pylori* strain and ranged from 16% to 29%, while the fraction enriched in phenolic compounds was more active, regardless of the strain used, with about 40% intracellular ROS inhibition.

In the study by Villalva et al., the volatile compounds have been identified by GC-MS, and only retention times have been provided for them (Table 2 of [16]). Phenolic compounds have been identified by HPLC-ESIMS in the negative ion mode, and aside from the retention times, the accurate masses of [M−H]<sup>−</sup> ions and the characteristic product ions (MS/MS ions) have also been provided (Supplementary Material, Table S1 of [16]). However, some of the product ions seem to be disputable, as described below.

The most disputable is the product ion at *m*/*z* 112 reported for apigenin and diosmetin. This is the only product ion reported for these two compounds and has 100% ri (Supplementary Material, Table S1 of [16]). According to the published data, apigenin should yield characteristic product ions at *m*/*z* 225, 151, 149, 117, and 107, although the relative abundances of these ions may be different, depending on the instrumental conditions used [28–31]. Diosmetin should yield an abundant product ion at *m*/*z* 284 as a result of methyl radical loss—a characteristic feature of methoxylated flavonoids [32,33]. The other characteristic diosmetin product ions should be at *m*/*z* 256, 227, 151, and 107 [34]. Besides diosmetin, Villalva et al. have found one other methoxylated flavonoid, named 'methoxyquercetin isomer' (the third most abundant flavonoid in yarrow's precipitated fraction, Table 1 of [16]); however, taking into account the *m*/*z* of [M−H]<sup>−</sup> at 315, it should be *O*-methyl quercetin. Villalva et al. claim only the detection of a product ion at *m*/*z* 301, thus the loss of mass 14 (elimination of a CH2 moiety). As mentioned above, the characteristic feature of the fragmentation of [M−H]<sup>−</sup> ions of methoxylated flavonoids is the loss of mass 15 (loss of a methyl radical). Furthermore, other product ions should be observed as

well, enabling at least tentative identification of this compound (isorhamnetin glycosides have already been found in the *Achillea millefolium* L. [35,36]). It has to be stressed that the two other methoxylated flavonoids, centaureidin and casticin, have been observed to have methyl radical losses by Villalva et al. [16]. The authors claim the detection of one biflavonoid, namely amentoflavone ([M−H]<sup>−</sup> at *m*/*z* 537), for which they have reported two product ions at *m*/*z* 519 and 495. However, the characteristic amentoflavone product ions are at *m*/*z* 443, 417, 375 (the most abundant), and 331 [37,38].

Villalva et al. have also detected a number of flavone C-glycosides. Among them are three isomers, namely apigenin-C-hexoside-C-pentoside, schaftoside, and schaftoside isomer ([M−H]<sup>−</sup> at *m*/*z* 563). For all these three compounds, Villalva et al. have obtained two identical product ions with identical relative abundances (*m*/*z* 473 and 443, 10 and 20% ri, respectively). The loss of mass 90 and 120 is a characteristic feature of flavone C-glycosides fragmentation [39]; thus, the product ions at *m*/*z* 473 and 443 are typical of these compounds. However, other product ions should also be detected, e.g., at *m*/*z* 383 and 353, and at least minor differences in relative ion abundances should be observed for these three isomers detected [40,41]. Villalva et al. claim the detection of luteolin-6,8-di-C-glucoside ([M−H]<sup>−</sup> at *m*/*z* 609), for which they have detected two product ions at *m*/*z* 489 and 325 [16]. The first one (loss of mass 120) is a characteristic product ion of luteolin-6,8-di-C-glucoside, often having 100% ri; however, at *m*/*z* 325 it is not the characteristic one. Other characteristic product ions which should be observed for this compound are at *m*/*z* 591, 399, 369, and 327 [41,42]. Villalva et al. claim the detection of vicenin 2 (apigenin 6,8-di-C-glucoside); however, no product ions have been reported for this compound [16]. For the two last flavone C-glycosides, namely homoorientin (luteolin 6-C-glucoside) and vitexin (apigenin 8-C-glucoside), the reported by Villalva et al. product ions match perfectly with those, the most abundant ones, reported elsewhere (although vitexin cannot be classified as flavonols) [43,44].

The product ions detected by Villalva et al. for other flavonoids are in agreement with those reported in the literature, at least as concerns the most characteristic ones. For example, for quercetin, the authors have detected only the product ion at *m*/*z* 151 [16]. Although deprotonated quercetin molecule yields a few other product ions (Figure 1), that at *m*/*z* 151 (formally 1,3A<sup>−</sup> product ion) is the most abundant, formed through the retro-Diels–Alder reaction, and can be regarded as a diagnostic ion for 5,7-dihydroxyflavonoids [28,45,46].

**Figure 1.** ESI mass spectrum (CID in-source) of quercetin obtained upon HPLC-MS(−) analysis of *Prunus persica* bark extract [46].

Villalva et al. have also detected a number of hydroxycinnamic acids and their conjugates, and the detection of most of them does not raise any doubts. The only disputable ones are the product ions reported for chlorogenic acid (*trans*-5-*O*-caffeoylquinic acid) and its isomer, cryptochlorogenic acid (*trans*-4-*O*-caffeoylquinic acid). For chlorogenic acid and cryptochlorogenic acid, Villalva et al. have obtained two identical product ions of almost identical relative abundances, namely at *m*/*z* 191 (100% ri) and 161 (10–11% ri) [16]. However, the fragmentation patterns of these compounds should be significantly different. Chlorogenic acid yields abundant product ions at *m*/*z* 191, and other product ions have

very low abundances (practically are not detectable), whereas cryptochlorogenic acid, except the product ion at *m*/*z* 191, yields abundant product ions at *m*/*z* 179, 173, and 135 [47–50]. On the other hand, it has to be stressed that the product ions detected by Villalva et al. for neochlorogenic acid (*trans*-3-*O*-caffeoylquinic acid) perfectly match those reported elsewhere with respect to their *m*/*z* and ri values [47–49]. The detection of other hydroxycinnamic acids and their conjugates also does not raise any doubts, although the accurate *m*/*z* of ferulic acid [M−H]<sup>−</sup> ion should be 193.0504, not 103.0504 (minor typos error). It is also worth adding that the elution order of the mono-*O*-caffeoylquinic acid isomers ([M−H]<sup>−</sup> at *m*/*z* 353) and di-*O*-caffeoylquinic acid isomers ([M−H]<sup>−</sup> at *m*/*z* 515) reported by Villalva et al. perfectly matches that obtained elsewhere during reversed-phase liquid chromatographic analysis [51].

It should be emphasized that our very specific comments concerning the product ions do not affect the paper by Villalva et al., the paper is characterized by a high scientific level, and the authors finding may imply the next practical application of *Achillea millefolium* L. extracts. On the other hand, the correction of Table S1 is desirable, maybe as a reply to our comment or as a corrigendum to their paper since it surely would improve the quality of the paper.

**Author Contributions:** Conceptualization, R.F. and M.B.-B.; writing—original draft preparation, R.F. and M.B.-B.; writing—review and editing, R.F. and M.B.-B. All authors have read and agreed to the published version of the manuscript.

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

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Reply*

**Reply to Fra ´nski, R.; Beszterda-Buszczak, M. Comment on "Villalva et al. Antioxidant, Anti-Inflammatory, and Antibacterial Properties of an** *Achillea millefolium* **L. Extract and Its Fractions Obtained by Supercritical Anti-Solvent Fractionation against** *Helicobacter pylori***.** *Antioxidants* **2022,** *11***, 1849"**

**Marisol Villalva 1, Jose Manuel Silvan 1, Teresa Alarcón-Cavero 2,3, David Villanueva-Bermejo 4, Laura Jaime 4, Susana Santoyo <sup>4</sup> and Adolfo J. Martinez-Rodriguez 1,\***


Franski and Beszterda-Buszczak [1] report some errors made in the identification of compounds in *Achillea millefolium* extract by MS/MS analysis included in the supplementary material of our published article [2]. We thank them for the observations, and we are pleased to be able to clarify the doubts from these authors. The following response offers an analysis of the comments made, compound by compound.

Regarding apigenin identification, as shown in Figure 1, spectrum 269 remains as the main product after MS/MS analysis along with other product ions such as 151 or 117, which have been reported as characteristics of apigenin fragmentation elsewhere [3–6]. In addition, *m/z* at 112.9858 was found for apigenin after MS/MS analysis. According to this finding, it was the only fragment included in Table S1, although 151 and 117 could also be included. 113 ion is in accordance with Taamalli et al. [3], who found it to be one of the product ions for apigenin-*O*-glucuronide. Therefore, a typographical error would be attributable, in this case, by reflecting 112 instead of 113. Moreover, based on its accurate mass, C15H10O5 (corresponding to the molecular formula of apigenin) was proposed for this product with an error of 3.2 ppm. In addition, the UV-Vis spectrum (data not included in this manuscript) and retention time matched those corresponding to the authentic apigenin standard.

As can be seen in Figure 2, diosmetin yielded the ions 299 (100), 284 (55), and 256 (12) as the main ion products, corresponding to a characteristic fragmentation ion from diosmetin [7]. No other ions were found in this analysis (e.g., 227, 151, or 107). However, based on the accurate mass [M−H]<sup>−</sup> at 299.0554, the molecular formula C16H12O6 (error 1.2 ppm) was obtained, which corresponded to diosmetin. Moreover, further identification was performed according to the UV-Vis spectrum and the retention time compared to diosmetin's authentic standard. Therefore, 112 ion, currently registered in Table S1, is a typographical error, and the omitted product ions (284 and 256) should be included in Table S1.

Alarcón-Cavero, T.; Villanueva-Bermejo, D.; Jaime, L.; Santoyo, S.; Martinez-Rodriguez, A.J. Reply to Fra ´nski, R.; Beszterda-Buszczak, M. Comment on "Villalva et al. Antioxidant, Anti-Inflammatory, and Antibacterial Properties of an *Achillea millefolium* L. Extract and Its Fractions Obtained by Supercritical Anti-Solvent Fractionation against *Helicobacter pylori*. *Antioxidants* 2022, *11*, 1849". *Antioxidants* **2023**, *12*, 1384. https:// doi.org/10.3390/antiox12071384

**Citation:** Villalva, M.; Silvan, J.M.;

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 3 February 2023 Revised: 29 April 2023 Accepted: 1 June 2023 Published: 4 July 2023

**Copyright:** © 2023 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/).

**Figure 1.** MS/MS spectrum peak identified as apigenin.

**Figure 2.** MS/MS spectrum peak identified as diosmetin.

Franski and Beszterda-Buszczak noted an incorrect assignment of [M−H]<sup>−</sup> at 315 as methoxyquercetin isomer. In the present study, the molecular formula C16H12O7 resulted with the *m/z* of [M−H]<sup>−</sup> at 315.0508 with an error of −3.3 ppm. The fragmentation pattern resulted in 300 ion (100) as the main product. Accordingly, a loss of CH3 (15 Da) could suggest the loss of [M−H]2<sup>−</sup> instead of 14 Da, and therefore the loss of [M−H], corresponding to 301 as the fragmentation ion. Franksi and Beszterda-Buszczak also suggest the possibility of examining other product ions, such as isorhamnetin glycoside, which was reported by Dias et al. [8] in *Achillea millefolium* L. However, the mass [M−H]<sup>−</sup> at 477 was not detected in our case. Certainly, other product ions were considered for preliminary analysis. In this regard, we considered the authentic standard of isorhamnetin, which corresponds to 3- -*O*-methylquercetin (also known as 3- -methoxyquercetin in the literature), but its retention time did not correspond with any of the identified compounds for *Achillea millefolium* L. Therefore, we have used methoxyquercetin, as its generic name, instead of *O*-methylquercetin.

Regarding amentoflavone identification, there was an error about the fragmentation ions of this compound shown in the HPLC-MS/MS spectra in Table S1. As can be seen from the fragmentation pattern shown in Figure 3, the characteristic amentoflavone product ions were detected but not reported properly according to the *m/z* at 375 (100), 443 (10), and 417 (20). This is in accordance with the literature for amentoflavone product ion mass spectra [9,10]. In addition, amentoflavone was designated by comparing its UV-Vis spectra and retention time using an authentic standard. Hence, the fragmentation pattern for amentoflavone in Table S1 should be modified.

Franski and Beszterda-Buszczak also mentioned the product ions of three isomers of flavone *C*-glycosides: apigenin-*C*-hexoside-*C*-pentoside, schaftoside, and schaftoside isomer. They claimed that the product ions—and their abundance—are similar for the three reported compounds. Certainly, no other product ions were detected for these isomers, although the accuracy of the *m/z* product ions varied slightly (with an accuracy within

four decimal places), along with their relative abundance. Hence, the relative abundance displayed in Table S1 for apigenin-*C*-hexoside-*C*-pentoside, schaftoside, and schaftoside isomer should be updated.

**Figure 3.** MS/MS spectrum peak identified as amentoflavone.

For luteolin-6,8-di-*C*-glucoside with an [M−H]<sup>−</sup> ion at *m/z* 609, the molecular formula C27H30O16 was found for the most likely compound with an error of 1.6 ppm. The most reported MS/MS fragmentation pattern included a fragmentation ion of [M−H−120]<sup>−</sup> from *m/z* 609, corresponding to a neutral loss of sugar residue, which yields the main ion fragment at *m/z* 489 [3]. The second fragment ion yielded a mass (*m/z*) of 325, which Franski and Beszterda-Buszczak did not recognize as a characteristic product ion. Surely, common fragmentation patterns in a negative ion mode include an ion at *m/z* 327 for this compound [11]. We decided to include the questioned ion fragment (*m/z* at 325) since luteolin possesses four -OH radicals, and it is possible that further ionization can occur, resulting in three hydrolyzed -OH radicals by means of a loss of [M−3H]3<sup>−</sup>.

For the product ions of cryptochlorogenic and chlorogenic acid, an inaccuracy exists in the reported fragmentation pattern in Table S1. The correct *m/z* of the main product ions correspond to 191 (100) for chlorogenic acid and 179 (67) and 173 (100) for cryptochlorogenic acid. Therefore, Table S1 should be updated. It is worth mentioning that these three chlorogenic acid isomers were also identified via a comparison with their authentic standards, as was already indicated in Table S1.

The identification of Vicenin 2 product ions was omitted by error but has now been updated and shown in Table S1. As can be observed in Table S1, the [M−H]<sup>−</sup> at *m/z* 593.1513, its corresponding molecular formula C27H30O15 (with an error of −0.2 ppm), and its main ion product *m/z* at 473 (100) are suggested to be the correct identification of Vicenin 2. In addition, an authentic standard was used to elucidate the proper identification. Vitexin was misclassified in Table S1 since it appeared in the flavonols section. Now, vitexin can be found in the flavones section since it is a flavone glycoside derivative of apigenin. The accurate mass *m/z* of ferulic acid was reported with a typographical error; the correct mass [M−H]<sup>−</sup> corresponds to *m/z* 193.0504. This mass has been corrected in Table S1.

Thus, after the revisions and modifications reflected in this letter, we would like to state that we have clarified all doubts. In addition, the results already published have full rigor and quality according to the standards of the scientific community and the procedures of the journal itself.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/antiox12071384/s1, Table S1: Phenolic compounds identified in yarrow samples by using HPLC-ESI-QTOF-MS.

**Author Contributions:** Conceptualization, A.J.M.-R., L.J., S.S. and M.V.; methodology, A.J.M.-R. and J.M.S.; validation, A.J.M.-R., J.M.S. and M.V.; formal analysis, D.V.-B., J.M.S. and M.V.; investigation, J.M.S., D.V.-B. and M.V.; resources, A.J.M.-R., T.A.-C., L.J. and S.S.; data curation, A.J.M.-R., J.M.S. and M.V.; writing—original draft preparation, A.J.M.-R., J.M.S. and M.V.; writing—review and editing, A.J.M.-R., J.M.S. and M.V.; visualization, A.J.M.-R. and J.M.S.; supervision, J.M.S. and M.V.; project administration, A.J.M.-R.; funding acquisition, A.J.M.-R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Projects AGL2017-89566-R (HELIFOOD) (MCIN/AEI/10.13039/ 501100011033/ (SpanishMinistry of Science and Innovation) and Fondo Europeo de Desarrollo Regional (FEDER) "Una manera de hacer Europa") and ALIBIRD-CM2020 P2018/BAA-4343 (Comunidad de Madrid, Spain).

**Data Availability Statement:** The data presented in this study are available in the article and supplementary materials

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

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Polyphenols from Thinned Young Apples: HPLC-HRMS Profile and Evaluation of Their Anti-Oxidant and Anti-Inflammatory Activities by Proteomic Studies**

**Giulio Ferrario 1,†, Giovanna Baron 1,†, Francesca Gado 1, Larissa Della Vedova 1, Ezio Bombardelli 2, Marina Carini 1, Alfonsina D'Amato 1, Giancarlo Aldini <sup>1</sup> and Alessandra Altomare 1,\***


**Abstract:** The qualitative profile of thinned apple polyphenols (TAP) fraction (≈24% of polyphenols) obtained by purification through absorbent resin was fully investigated by LC-HRMS in positive and negative ion mode and using ESI source. A total of 68 polyphenols were identified belonging to six different classes: flavanols, flavonols, dihydrochalchones, flavanones, flavones and organic and phenolic acids. The antioxidant and anti-inflammatory activities were then investigated in cell models with gene reporter for NRF2 and NF-κB and by quantitative proteomic (label-free and SILAC) approaches. TAP dose-dependently activated NRF2 and in the same concentration range (10–250 μg/mL) inhibited NF-κB nuclear translocation induced by TNF-α and IL-1α as pro-inflammatory promoters. Proteomic studies elucidated the molecular pathways evoked by TAP treatment: activation of the NRF2 signaling pathway, which in turn up-regulates protective oxidoreductases and their nucleophilic substrates such as GSH and NADPH, the latter resulting from the up-regulation of the pentose phosphate pathway. The increase in the enzymatic antioxidant cellular activity together with the up-regulation of the heme-oxygenase would explain the anti-inflammatory effect of TAP. The results suggest that thinned apples can be considered as a valuable source of apple polyphenols to be used in health care products to prevent/treat oxidative and inflammatory chronic conditions.

**Keywords:** thinned apples; polyphenols; anti-oxidant; anti-inflammatory; NRF2; NF-κB; proteomics

#### **1. Introduction**

Epidemiological studies indicate that consumption of apples and derivatives, such as apple juice, are beneficial in the treatment of some human diseases including CVD and related events, cancer and diabetes [1]. Beneficial effects have also been confirmed by intervention studies. Vallée Marcotte et al. [2] recently reviewed 20 intervention studies using apple juice, concluding that its consumption could exert some benefits on a variety of markers associated with the risk of developing chronic diseases including cardiovascular, cancer, and neurodegenerative diseases.

Recently, a regular consumption of 2–3 apples per day was associated to beneficial effects. In a randomized, controlled, crossover intervention study, in healthy subjects with mildly raised serum cholesterol concentrations, consumption of two Renetta Canada apples for eight weeks improved CVD risk factors, by reducing total and LDL cholesterol and ICAM-1 and increasing microvascular vasodilation [3]. Liddle et al. recently reported that consumption of three whole Gala apples per day for 6 weeks may be an effective strategy to mitigate inflammation in overweight and obese subjects, by reducing circulating biomarkers of inflammation and endotoxin exposure, including CRP, IL-6 and LBP and increasing the plasma antioxidant capacity [4].

**Citation:** Ferrario, G.; Baron, G.; Gado, F.; Della Vedova, L.; Bombardelli, E.; Carini, M.; D'Amato, A.; Aldini, G.; Altomare, A. Polyphenols from Thinned Young Apples: HPLC-HRMS Profile and Evaluation of Their Anti-Oxidant and Anti-Inflammatory Activities by Proteomic Studies. *Antioxidants* **2022**, *11*, 1577. https://doi.org/10.3390/ antiox11081577

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 5 July 2022 Accepted: 10 August 2022 Published: 15 August 2022

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

**Copyright:** © 2022 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/).

Most of the documented beneficial effects of apples are attributed to the fraction of polyphenols represented by five main groups, namely, flavanols (catechins, epicatechin and procyanidins), flavonols (quercetin glycosides), phenolic acids (chlorogenic, gallic and coumaric acids), dihydrochalcones (phloretin glycosides) and anthocyanins (cyanidin) [5–7]. Apple polyphenols exert antioxidant, anti-inflammatory and lipid lowering effects as also confirmed by pre-clinical studies. In particular, isolated apple polyphenols (AP) have been proven to be effective in preventing/treating hypercholesterolemia [8,9], atherosclerosis in ApoE-deficient mice [10], colorectal cancer [11], non-alcoholic hepatitis [12], ulcerative colitis [13] and indomethacin-induced gastric damage [14].

At a molecular level, the antioxidant and anti-inflammatory activities of apple polyphenols can be partially explained by considering the role of polyphenols as activators of the NRF2 pathway [15]. NRF2 is a transcriptional factor associated with antioxidant enzymes playing a master role in redox homeostasis in cells. NRF2 and its principal negative regulator, KEAP1, play a central role in the maintenance of intracellular redox homeostasis and regulation of inflammation. NRF2 is proved to contribute to the regulation of the heme oxygenase-1 (HMOX1) axis, which is a potent anti-inflammatory target [16], and to inhibit oxidative stress by up-regulating antioxidant enzymes and co-factors [17]. Recently, there is an increase in the research literature regarding the regulation of NRF2 signaling pathways in different aspects of inflammation such as cytokines, chemokine releasing factors, MMPs and other inflammatory mediators affecting the NF-κB and MAPK networks to control inflammation, as recently reviewed by Saha et al. [18]. Activation of NRF2 by polyphenol compounds is mainly mediated by those compounds bearing an ortho-diphenol moiety which is oxidized to the corresponding quinone, which, being an electrophilic compound, reacts with the thiols of the KEAP1, thus releasing NRF2, which then translocates into the nucleus [19].

Hence, the phenolic enriched fraction from apples represents a valuable source of natural compounds with a beneficial effect against inflammation and oxidative stress, and they may be applied as a food supplement and/or functional ingredient for the treatment of chronic inflammatory diseases.

There is nowadays great interest in the bioactive compounds obtained from the waste products deriving from agriculture and the food industry (circular economy) [20]. Thinning young apples (around one month after blossom), which constitute a massive waste product of the apple production chain, is carried out in order to guarantee the output and to increase the quality of the harvested apples and are usually discarded in the orchard soil [21]. However, this has the negative effect of increasing the soil acidity and thus disturbing the microbial community, which in turn affects the growth of fruit trees [22]. Thinned young apples are particularly rich in polyphenols, more than 10-fold with respect to harvested apples. The total polyphenols and total antioxidant activity show a rapid decrease after 45 days from blossoming, stabilizing after 85 days, as found in the Fuji apples [23]. Recently, some papers have reported methods for purifying polyphenols from thinned apples by successive use of polyethylene and polyamide resins [24]. Moreover, studies have reported the beneficial effects of polyphenols from thinned apples which have a significant antibacterial activity and were found to be effective in inhibiting halitosisrelated bacteria through damage to the cell membrane and hence should be a valuable waste material as a source of bioactive compounds [25]. Polyphenols from unripe apples were also found to exert anti-obesity activity in rats through the modulation of the fatty acid metabolism in the liver and the inhibition of the absorption of carbohydrates and fat [26].

Considering, on the one hand, the growing scientific interest in the apple and particularly in the apple polyphenol fraction as a source of bioactive compounds effective against inflammation and oxidative stress, and, on the other hand, taking into account that thinned apples are a waste product particularly rich in polyphenols, this work is aimed at fully characterizing the qualitative profile of polyphenols purified from thinned apples by a dual LC-HRMS approach (targeted and non-targeted), and at investigating their antiinflammatory and antioxidant activities. Cell lines with gene reporters for NRF2 and NF-κB nuclear translocation were first used to assess the dose-dependent antioxidant and antiinflammatory activities, and then to outline the molecular pathways involved by means of two integrated proteomic approaches, based on label-free and SILAC methodologies.

#### **2. Material and Methods**

#### *2.1. Chemicals*

Thinned Golden, Fuji, Bella del Bosco and Rosa Mantovana apples were sourced from farms located in Trentino-Alto Adige, Italy. Ultrapure water was prepared by a Milli-Q purification system (Millipore, Bedford, MA, USA). Protocatechuic acid, caffeic acid, *p*-coumaric acid, (+)-catechin, (−)-epicatechin, prunin (naringenin glucoside), phlorizin, phloretin, luteolin, cysteine (Cys), iodoacetamide (IAA), tris(2-carboxyethyl)phosphine (TCEP), tetraethylammonium bromide (TEAB), 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT), IL-1α, TNF-α, 6-hydroxy-2,5,7,8-tetramethyl-3,4 dihydrochromene-2-carboxylic acid (Trolox), ascorbic acid, catechin, sodium carbonate, Folin–Ciocalteu reagent, ferulic acid, naringenin-7-*O*-glucoside, quercetin, ethanol, methanol, phlorizin dihydrate, sodium acetate, acetic acid, 2,2-diphenyl-1-picrylhydrazyl (DPPH), CDDO-Me, rosiglitazone, formic acid (FA), trifluoroacetic acid (TFA), acetonitrile (ACN) and all ultra-pure-grade (99.5%) solvents used in LC-MS analysis were obtained from Merck KGaA, Darmstadt, Germany. Kaempferol, quercetin, quercetin-3-*O*-rhamnoside and quercetin-3-*O*-galactoside were purchased from Extrasynthese (Genay Cedex, France). S-TRAP™ columns were provided by Protifi (Huntington, NY, USA).

#### *2.2. Isolation of Thinned Apple Polyphenols (TAP) Fraction*

A total of 500 kg of thinned Golden, Fuji, Bella del Bosco and Rosa Mantovana apples were harvested 1 month after blossoming and stored at 2 ◦C for 1 month. The apples were thoroughly washed in a mixer with an aqueous solution containing 0.05% citric acid and then coarsely ground in a hammer mill. A 0.1% solution of pectinase was added to the resulting mush, which was then heated in a linear tunnel at 25 ◦C for 20 min.

The mass was then continuously forced through a filter with pressure of 200 bar and the squeezed juice was enzymatically treated with pectinase and centrifuged until a colorless solution was obtained. The liquid was then eluted through 20 L of XAD7 absorbent resin and washed with demineralized water until elimination of all the substances not retained by the resin. At the end of the elution with water, the retained substances (polyphenols) were eluted with 95% ethanol until all the bound components were recovered. Elution of polyphenols was checked by TLC. The hydro-alcoholic solution was concentrated under vacuum and then micronized.

#### *2.3. Qualitative Analysis by LC-HRMS*

The phytochemical profile of TAP fraction was performed by LC-HRMS as described by Baron et al. [27]. The extract was dissolved in methanol and diluted with mobile phase A to a final concentration of 2 mg/mL, added with Trolox (50 μM, final concentration) as internal standard (IS). The mixture was separated on a RP Agilent Zorbax SB-C18 column (150 × 2.1 mm i.d., 3.5 μm, CPS analitica, Milan, Italy) by an Ultimate 3000 system (Dionex, Sunnyvale, CA, USA) with a multistep program (80 min) of mobile phase A (H2O/HCOOH, 100/0.1, %*v/v*) and B (CH3CN/HCOOH, 100/0.1, %*v/v*). An LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) was set to perform the analysis in data-dependent scan mode, and acquired the spectra in positive and negative ion mode. Full MS spectra were acquired by the FT analyzer (resolution 30,000 FWHM at *m/z* 400) in profile mode and in the range of *m/z* 120–1800. The MS/MS spectra of the 3 most intense ions exceeding 1 × <sup>10</sup><sup>4</sup> counts of the full MS scan were acquired by the linear ion trap (LTQ) by using a normalized collision energy (CID) of 40 eV. A database containing the known components of apples and derivatives was built for the targeted data analysis [28–44]: the putative identification was obtained by comparing the accurate mass (5 ppm mass tolerance), the isotopic pattern and the fragmentation pattern with the compounds in the database. The identity of some molecules was confirmed by means of pure standards available in our laboratory. An untargeted data analysis was performed of the most intense ions not identified in the targeted analysis, as described by Baron et al. [27].

#### *2.4. Quantitative Analysis of Total Polyphenol Content*

#### 2.4.1. Colorimetric Analysis

The total polyphenol content was determined by the Folin–Ciocalteu colorimetric method, as reported by Baron et al. [45]; the calibration curve was built using catechin as a standard in a 1–1000 μg/mL range.

#### 2.4.2. HPLC Analysis

High-performance liquid chromatography (HPLC) coupled with a PDA detector was performed to evaluate the total polyphenol content. For the analysis, a methanolic solution of thinned apple polyphenols (TAP) fraction was diluted 1:4 in H2O/HCOOH, 100/0.1, %*v/v* (mobile phase A) to obtain a final concentration of 1 mg/mL.

The sample (injection volume, 10 μL) was analyzed in triplicate with a HPLC system (Surveyor, ThermoFinnigan Italy, Milan, Italy), equipped with a PDA detector (Surveyor, ThermoFinnigan Italy, Milan, Italy) and an RP Agilent Zorbax SB-C18 column (150 × 2.1 mm i.d., 3.5 μm, CPS analitica, Milan, Italy). The same gradient program described in Section 2.3 was used for the separation and quantification of TAP polyphenols, setting the PDA detector in a 200–600 nm range.

For the quantification, five calibration curves were built using a standard for each class of polyphenols characterizing the TAP extract: catechin for flavanols (10–100 μg/mL; λmax 278 nm), ferulic acid for phenolic acids (1–10 μg/mL; λmax 323 nm), naringenin-7 glucoside for flavanones (10–100 μg/mL; λmax 283 nm), phlorizin for dihydrochalchones (10–100 μg/mL; λmax 284 nm) and quercetin for flavonols (10–100 μg/mL; λmax 371 nm).

Each chromatographic peak of the sample was assigned to the corresponding polyphenolic class on the basis of the UV spectrum. For the quantification, the AUC was interpolated using the calibration curve of the relative standard and the sum of the concentrations of all the compounds belonging to the same polyphenol class was calculated. The total polyphenol content was then expressed as a percentage (%), that is, mg of polyphenols present in 100 mg of extract.

#### *2.5. Direct Radical Scavenging Activity*

The antioxidant activity was evaluated with the DPPH assay. A solution of thinned apple polyphenols (TAP) fraction was prepared with H2O:EtOH 50:50 (%*v/v*) to obtain final concentrations in the range 1–20 μg/mL. An aliquot of TAP extract solution (500 μL) was mixed with 1 mL of acetate buffer (pH 5.5, 100 mM) and 1 mL of EtOH. Then, 500 μL of DPPH (500 μM, ethanolic solution) was added and samples were kept in the dark for 90 min. A Shimadzu UV 1900 spectrophotometer (Shimadzu, Milan, Italy) was used for reading the absorbance at 517 nm. Trolox and ascorbic acid were used as reference antioxidant compounds. The percentage of inhibition (I%) was calculated as expressed by Equation (1) and the results are reported as mean ± SD.

$$I\% = \frac{Abs\,\,(blank\,sample) - Abs\,\,(sample)}{Abs\,\,(blank\,\,sample)} \times 100\tag{1}$$

#### *2.6. Cell-Based Assays*

#### 2.6.1. MTT Assay

The effect of thinned apple polyphenols (TAP) fraction on the cell viability for all the concentrations tested in the anti-inflammatory and antioxidant assays was verified by MTT assay on R3/1-NF-κB cells and HEK293 cells in transparent 96-well plates seeded with 4000 cells/well and 10,000 cells/well, respectively. Cells were treated with different concentration of the extract (1 μg/mL–250 μg/mL) for 18 h in complete medium (DMEM 10% FBS, 1% penicillin/streptomycin). Subsequently, media were removed and only for R3/1-NF-κB cell line, one wash with 100 μL PBS occurred. Then, 100 μL of DMEM, not supplemented with FBS and penicillin/streptomycin, was added to each well and the 4 h incubation started after the addition of 11 μL, 5 mg/mL MTT reagent. After medium removal, cells were lysed using 100 μL of a solution composed of DMSO, 8 mM HCl and 5% TWEEN20. The 96-well plate was shaken for 15 min in a plate shaker in the dark and the absorbance at 575 nm and 630 was measured using a plate reader (BioTek's PowerWave HT, Winooski, VT, USA). Cells incubated with DMSO (0.1%) were used as a control for 100% cell proliferation, while cells incubated with DMSO (3%) were used as a negative control.

#### 2.6.2. NRF2 Gene Reporter Cell Model

Thinned apple polyphenols (TAP) extract was evaluated for its ability to modulate the antioxidant response mediated by NRF2 activation using NRF2/ARE Responsive Luciferase Reporter HEK293 stable cell line (Signosis, Santa Clara, CA, USA) as previously described [46]. Briefly, HEK293 cells (10,000 cells/well) were treated with the extract (concentrations between 1 and 250 μg/mL). CDDO-Me 75 (bardoxolone methyl) 75 nM was used as a positive control [47]. After adding ONE-Glo™ Luciferase Assay Substrate (purchased from Promega Corporation, Madison, WI, USA) (100 μL/well), luciferase measurement was performed with a luminometer (Wallac Victor2 1420, Perkin-Elmer™ Life Science, Monza, Italy). Experiments were carried out with biological and technical replicates; values are reported as mean ± SD compared to untreated control cells. One-way ANOVA with Bonferroni's multiple comparisons test (*p* < 0.05 was considered significant) was used for the statistical analysis.

#### 2.6.3. NF-κB Gene Reporter Cell Model

The in vitro anti-inflammatory activity of the TAP extract was evaluated by using a cell model previously described [27]. Briefly, R3/1 NF-κB cells (5000 cells/well) were pre-treated for 18 h with different concentrations of the extract (1–250 μg/mL) in complete medium (DMEM 10% FBS, 1% L-glutamine, 1% Penicillin/Streptomycin). Rosiglitazone 10 μM was used as a positive control [27]. Then, cells were stimulated for 6 h with 10 ng/mL IL-1α, and for 6 and 24 h with 10 ng/mL TNF-α. Luciferase measurements was performed with a luminometer (Wallac Victor2 1420, Perkin-Elmer™ Life Science, Monza, Italy) after adding 100 μL of ONE-Glo™ Luciferase Assay Substrate (purchased from Promega Corporation, Madison, WI, USA). Experiments were carried out with biological and technical replicates; values are reported as mean ± SD compared to untreated control cells. One-way ANOVA with Bonferroni's multiple comparisons test (*p* < 0.05 was considered significant) was used for the statistical analysis.

#### *2.7. Quantitative Proteomic Studies*

#### 2.7.1. SILAC Culture

The R3/1 NF-κB reporter cell line used for SILAC experiments was cultured in DMEM for SILAC supplemented with 10% FBS, 1% pen/strep and 1% sodium pyruvate; the medium was completed by adding 0.5 mL of heavy or light L-lysine and L-arginine 1000X stock solution diluted in PBS. The final concentration of the light amino acids was 84 mg/mL for arginine and 146 mg/mL for lysine. For the heavy condition, both amino acids (13C6 15N2 lysine and 13C6 15N4 arginine) were diluted in the medium up to the working concentration of 88 mg/mL and 151.3 mg/mL for the heavy arginine and lysine, respectively; this difference in concentration is due to the different molecular weights of the amino acids [48,49]. The cell line was cultivated for at least 9 passages and the incorporation rate was checked: more than 95% of all peptides are required to be labeled.

#### SILAC Experimental Design

Each condition tested was cultivated in biological triplicate, and once 70% confluence was reached in a T75 flask, the cells were treated according to the planned experimental design, as shown in Figure 1. Overall, the conditions chosen were as follows: (i) *Control (CTR)*, i.e., untreated cells; (ii) *Thinned apple polyphenols (TAP) extract treatment (CTR-TAP),* i.e., cells that underwent a double 24 h treatment with the extract at a 200 μg/mL concentration; (iii) *Inflammation* (CTR-TNF), condition achieved treating cells for 24 h with TNF-α 0.01 μg/mL; (iv) *Thinned apple polyphenols (TAP) extract treatment of inflamed cells (CTR-TNF-TAP)*, consisting of the 24 h pre-treatment with the extract at 200 μg/mL and then with TNF-α at 0.01 μg/mL for another 24 h.

**Figure 1.** Experimental design of the quantitative proteomic analysis based on the two orthogonal approaches applied, namely, SILAC (**left side**) and LFQ (**right side**), interspersed by a timeline describing the three basic treatment steps. Below is a schematic of the sample processing protocol.

The metabolic labeling strategy was planned so that all conditions are compared with each other in order to have a comprehensive and meaningful pathway modulation overview (Figure 1). All the experiments were planned to finish in the same day.

After light or heavy amino acid incorporation, the flasks were washed twice with warmed PBS and the cell was detached using 2 mL of warmed trypsin incubated at 37 ◦C for 5 min. Cells were then recovered. After diluting the trypsin with 8 mL of complete medium, the suspension was centrifuged for 5 min at 300× *g* at room temperature and the pellet was then washed 5 times with 5 mL of 4 ◦C PBS. The cell pellets were then lysed using a suitable S-trap (Protifi) buffer composed of SDS 5%, TEAB 50 mM, MgCl 2 mM, one complete™ EDTA-free protease inhibitor cocktail tablet (Sigma-Aldrich, Milan, Italy) and 100 Unit of benzonase to reduce nucleic acids content that could occlude the S-trap column. The lysates completely solubilized after 30 min of mechanic resuspension in a tube rotator at 4 ◦C and were centrifuged at 10,000× *g* for 15 min at 4 ◦C in a refrigerated centrifuge. Once the lysates were ready, the protein concentration of differentially labeled samples was assessed using the BCA kit and the conditions were mixed in a 1:1 ratio to obtain 100 μg of proteins before any further processing.

Each mixture was prepared by mixing one specific biological replicate with another at random.

#### S-Trap Digestion

Samples collected from all the prepared incubation mixtures were then processed according to the bottom-up proteomics procedure. Sample preparation begins with the solubilization of samples in 5% SDS followed by further denaturation by acidification and subsequent exposure to a high concentration of methanol. The collected incubation mixtures were then solubilized 1:1 in lysis buffer (10% SDS, 100 mM TEAB). The reduction of disulfide bridges was performed by adding 5 μL of the reducing solution of tris(2 carboxyethyl)phosphine (c.f. 5 mM TCEP in 50 mM AMBIC) and incubating the mixtures under slow shaking in a Thermomixer for 10 min at 95 ◦C. Next, a volume of 5 μL of iodoacetamide solution (c.f. 20 mM IAA in 50 mM AMBIC) was added, with the aim of alkylating the free thiol residues; incubation in this case was carried out for 45 min at room temperature in the dark. Proteins were further denatured by acidification to pH < 1 by adding a 12% phosphoric acid solution in water (1:10 relative to sample volume). The next step consisted of the sample loading: 165 μL of the binding buffer (90% methanol, 10% TEAB 1 M) and 25 μL of sample were simultaneously added onto the spin columns, then centrifuged at a speed of 4000× *g* for 1 min at 15 ◦C; this step was repeated until the protein sample was fully loaded onto the columns.

After that, three washing steps by adding 150 μL of binding buffer onto the S-TRAP columns, followed by centrifugation (1 min, 4000× *g*, 15 ◦C), were performed to remove all the excess of unbound sample. At this point, the proteolytic digestion step started by adding a volume of 25 μL containing 1 μg of trypsin (sequencing-grade trypsin, Roche) diluted in 50 mM AMBIC. Upon addition of the protease, the physical confinement within the submicron pores of the trap forces substrate and protease interaction to yield rapid digestion; therefore, protein digestion requires much shorter incubation times, i.e., 1.5 h at 47 ◦C under slow stirring (400 rpm). The peptide mixture was recovered by loading two different solutions—40 μL of elution buffer 1 (10% H2O, 90% ACN, 0.2% FA) and 35 μL of elution buffer 1 (60% H2O, 40% ACN, 0.2% FA)—onto the columns (elution and 1- centrifugation, 4000× *g*, 15 ◦C). The collected peptide mixtures were dried in the SpeedVac (Martin Christ., Osterode am Harz, Germany) at 37 ◦C and stored at −80 ◦C until analysis.

#### High pH Fractionation

The digested peptide mixtures obtained by means of the S-Trap™ micro-spin column digestion strategy were fractionated using the Pierce™ High pH Reversed-Phase Peptide Fractionation Kit according to the manufacturer's instructions, with one additional fraction at 80% ACN.

#### nLC-HRMS Orbitrap Elite™ Mass Spectrometer Analysis

Tryptic peptides, resuspended in an appropriate volume (18 μL, sufficient for two technical replicates) of 0.1% TFA mobile phase, were analyzed using a Dionex Ultimate 3000 nano-LC system (Sunnyvale, CA, USA) connected to the Orbitrap Elite™ Mass Spectrometer (Thermo Scientific, Brema, Germania) equipped with an ionization source, the nanospray (Thermo Scientific Inc., Milano, Italia).

For each sample, 5 μL of solubilized peptides was injected in triplicate onto the Acclaim PepMap™ C18 column (75 μm × 25 cm, 100 Å pores, Thermo Scientific, Waltham, MA, USA), "protected" by a pre-column, the Acclaim PepMap™ (100 μm × 2 cm, 100 Å pores, Thermo Scientific, Waltham, MA, USA), thermostatically controlled at 40 ◦C. The chromatographic method used a binary pump system (LC/NC pumps) and started from sample loading onto the pre-column (3 min) using the loading pump with a flow rate of 5 μL/min of mobile phase consisting of 99% buffer A\_LC, 0.1% TFA/1% buffer B\_LC and 0.1% FA in ACN. After the loading valve switching, peptide separation was performed by the Nano Column Pump (NC\_pump) with a 117 min linear gradient (0.3 μL/min) of buffer B\_NC\_pump (0.1% FA in ACN) from 1% to 40%, and a further 8 min of linear gradient from 40% to 95% (Buffer B\_NC\_pump). Then, 5 min at 95% of buffer B\_NC\_pump to rinse the column followed the separative gradient, and the last 7 min served to re-equilibrate

the column to initial conditions. The total run time was 144 min. A washout injection with pure acetonitrile (5 μL) was performed between sample injections.

The nanospray ionization source was set as follows: positive ion mode, spray voltage at 1.7 kV; capillary temperature at 220 ◦C, capillary voltage at 35 V; tube lens offset at 120 V. The orbitrap mass spectrometer operating in data-dependent acquisition (DDA) mode was set to acquire full MS spectra in "profile" mode over a scan range of 250–1500 *m/z*, with the AGC target at 5 × 105, and resolution power at 120,000 (FWHM at 400 *m/z*). Tandem mass spectra were instead acquired by the linear ion trap (LTQ), set to automatically fragment in CID mode the ten most intense ions for each full MS spectra (over 1 × 104 counts) under the following conditions: centroid mode, normal mode, isolation width of the precursor ion of 2.5 *m/z*, AGC target 1 × <sup>10</sup><sup>4</sup> and normalized collision energy of 35 eV. Dynamic exclusion was enabled (exclusion dynamics for 45 s for those ions observed 2 times in 10 s). Charge state screening and monoisotopic precursor selection were enabled, and singly charged and unassigned charged ions were not fragmented. Xcalibur software (version 3.0.63, Thermo Scientific Inc., Milan, Italy) was used to control the mass spectrometer.

#### 2.7.2. LFQ Analysis

For the LFQ (Label-Free Quantitative Proteomics) experiment, the same cell line (R3/1 NF-κB reporter cell line) was cultivated in biological triplicate in T-25 flasks and the same experimental conditions were tested for SILAC ((i) *Control, CTR*; (ii) *Thinned apple extract treatment (CTR-TAP)*; (iii) *Inflammation (CTR-TNF)*; (iv) *Thinned apple extract treatment of inflamed cells (CTR-TNF-TAP),* as shown in Figure 1). Except for the labeling strategy, the lysate preparation procedure is the same, as is the proteolytic digestion performed exploiting the potential of the S-Trap™ Micro Spin Column Digestion Protocol.

#### nLC-HRMS Orbitrap Fusion™ Tribrid™ Mass Spectrometer Analysis

Tryptic peptides were analyzed using a Dionex Ultimate 3000 nano-LC system (Sunnyvale, CA, USA) connected to an Orbitrap Fusion Tribrid Mass Spectrometer (Thermo Scientific, Bremen, Germany) equipped with a nano-electrospray ion source according to the procedure previously described [50].

#### 2.7.3. Data Analysis

For both the SILAC and label-free analysis, the instrumental raw files were processed by MaxQuant software v.1.6.6.0 set on *Rattus\_Norvegicus* database (Uniprot taxonomy ID: 10116) against the Andromeda search engine. Protein quantification using SILAC (3 biological × 2 technical replicates for each condition) was based on the ratio of peptides' peak intensities in the mass spectrum reflecting the relative protein abundance, while for the label-free approach (3 biological × 3 technical replicates for each condition), the quantification was based on LFQ intensity. In both analyses, trypsin was specified as proteolytic enzyme, cleaving after lysine and arginine except when followed by proline, and up to two missed cleavages were allowed along with match between run option. The precursor ion tolerance was set to 5 ppm while the fragment tolerance was set to 0.5 Da. Carbamidomethylation of cysteine was defined as fixed modification, while oxidation of methionine and acetylation at the protein N-terminus were specified as variable modifications. For the SILAC experiment, only the multiplicity of the labels was set to 2, and Arg10 and Lys8 were selected as heavy aminoacidic residues. Interpretation of the results was performed using Perseus (v.1.6.1.43, Max Plank Institute of Biochemistry, Martinsried, Germany). For the SILAC analysis, the normalized ratio count was selected, transformed in log2, filtered for minimum of 3 valid values and then a one sample *t*-test was performed with Benjamini–Hochberg FDR for truncation with a threshold of 0.05. LFQ analysis was validated by applying a two-sample t-test of the log2 LFQ intensities. The network protein analyses related to significantly altered proteins were carried out by means of Cytoscape v.3.9.1 and the ingenuity pathways analysis (IPA) (QIAGEN Aarhus Prismet, Aarhus, Denmark, September 2021) licensed software based on the Gene Ontology database. All

statistical analyses of the inflammatory assays were conducted using GraphPad prism 8. The Venn diagrams with proportional areas were obtained using the bioVenn package for R [51].

#### Cytoscape Analysis

The protein–protein interaction network of the up-regulated proteins (log2 *Fold-Change* > 0.5) and down-regulated proteins (log2 *Fold-Change* < −0.5) was obtained by importing a list of Protein IDs into Cytoscape v.3.9.1 (http://cytoscape.org, accessed on 15 February 2022). We used the embedded STRING interaction database (http://apps. cytoscape.org/apps/stringApp, accessed on 15 February 2022) with a default confidence cut-off score of 0.4. *Rattus Norvegicus* was selected as a reference database with GO, KEGG, PFAM, WikiPathways, Reactome and InterPro cluster terms selected as functional annotations for the enrichment analyses.

#### Ingenuity Pathways Analysis (IPA)

The core analyses performed by IPA, using the differentially expressed proteins in the uploaded dataset, assess signaling pathways, molecular interaction network and biological functions that can likely be perturbed. The overall activation/inhibition states of canonical pathways are predicted through a z-score algorithm. This z-score is used to statistically compare the uploaded dataset with the pathway patterns. The pathways are colored to indicate their activation z-scores: orange predicts a gain of function, and blue a loss of function. The pathway is activated when molecules' causal relationships with each other (i.e., activation edge and the inhibition edge between the molecules based on literature findings) generate an activity pattern for the molecules and the end-point functions in the pathway.

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

#### *3.1. Qualitative Profile of Polyphenols Determined by Targeted LC-HRMS Analysis*

The qualitative profile of polyphenol components of TAP was firstly evaluated by a targeted and untargeted metabolomic approach by HPLC-HRMS in negative and positive ion mode. Figure 2 shows the chromatograms of the extract as total ion current (TIC) recorded in negative (A) and positive (B) ion mode where the identified peaks are numbered progressively (co-eluting peaks share the same number), according to the elution order. The peak of the internal standard (Trolox) is indicated by "IS". A total of 68 compounds were identified, 68 of which are polyphenols, 52 by the targeted approach and 16 by the untargeted approach.

The 52 compounds identified or putatively identified with the targeted analysis are listed in Table S1 of the Supplementary Materials; for each identified compound, the relative peak number (Figure 2), the retention time (RT), the experimental mass (as [M−H]−, [M+H]+, [M+2H]2+ or [M+Na]+), the mass accuracy (Δppm), MS/MS fragments and identification method are reported. Thirty-nine compounds were detected in both polarities, seven in negative ion mode and six in positive ion mode. Among the 52 components, 20 are phenolic and organic acids, 11 are flavanols, 10 are flavonols, 5 are flavanones, 4 are dihydrochalchones, 1 is a flavone (luteolin) and 1 is a triterpenoid (euscaphic acid). Thirteen compounds were confirmed by means of in-house available pure standards: three phenolic acids, protocatechuic acid, caffeic acid and *p*-coumaric acid; two flavanols, catechin and epicatechin; two flavanones, naringenin glucoside and naringin; four flavonols, quercetin-3-O-galactoside, quercetin-3-O-rhamnoside, quercetin and kaempferol; two dihydrochalcones, phlorizin and phloretin; one flavone, luteolin. For the compounds for which the standard was not available (39 molecules), a putative identification was carried out by matching the accurate mass, the isotopic pattern and the fragmentation pattern with data reported in the literature. Among phenolic acids, the glucoside forms of protocatechuic, ferulic, caffeic and coumaric acids, and also quinic acid esters of caffeic and coumaric acids were identified. Flavanols are represented in the monomeric forms (catechin and epicatechin) up to the nonamer oligomer, detected as [M+2H]2+. Naringenin and eriodictyol derivatives (and aglycones) were the flavanones identified in the extract. The untargeted analysis allowed the putative identification of 16 compounds (Table S2 of Supplementary Materials): nine flavonols, three phenolic acids, one flavanone, one dihydrochalchone and two lipids. As an example, 3-(benzoyloxy)-2-hydroxypropyl glucopyranosiduronic acid (compound **11**) was tentatively annotated, firstly by calculating the possible molecular formula with the QualBrowser tool of Xcalibur (mass tolerance of 5 ppm and using C, H, O, N, S as possible atoms). The resulting formulae were searched in databases and in the literature and we found a match with 3-(benzoyloxy)-2-hydroxypropyl glucopyranosiduronic acid (C16H20O10, 0.037 ppm, in negative ion mode) [52]. The MS/MS spectrum was also compared: the main fragment was at *m/z* 249 in negative ion mode deriving from the loss of benzoic acid (−122 Da). Glycosides of quercetin, methoxyquercetin (patuletin, detected with the untargeted approach) and kaempferol are the most representative flavonols identified, but they were also present as aglycones, and a coumaroylglucoside derivative of quercetin was annotated with the untargeted approach. Methoxyquercetin (patuletin, compound **36**) was putatively identified through HMDB database: the molecular formula C16H12O8 (0.107 ppm in negative ion mode; 1.862 ppm in positive ion mode) and MS/MS fragments at *m/z* 285, 209 and 181 (in negative ion mode) matched with those found in the database, and in addition, we detected an additional fragment at *m/z* 316, corresponding to the loss of a methyl group (−15 Da). We detected also five patuletin derivatives: three hexoside isomers (compounds **20**, **22** and **23**) at 20.8, 22.5 and 23.8 min, tentatively annotated with PubChem; one patuletin pentoside derivative (compound **27**), tentatively identified through the matching of the molecular formula (C21H20O12, −0.179 ppm and −1.591 ppm in negative and positive ion mode, respectively); and characteristic fragments of patuletin aglycone at *m/z* 331 and 316 due to a neutral loss of −132 Da, corresponding to a pentoside moiety and a subsequent loss of a methyl moiety. Similarly, patuletin rhamnoside (compound **29**) was annotated through the molecular formula (C22H22O12, 0.161 ppm and −2.066 ppm in negative and positive ion mode, respectively) and the MS/MS fragments (at *m/z* 331 and 316) deriving from the neutral loss of a rhamnoside moiety (−146 Da) and a subsequent loss of a methyl (−15 Da). Patuletin is a methoxy derivative of quercetin which, among the bioactivities reported in the literature, was found effective in the reduction of serum TNF-α in a rodent model of adjuvant-induced arthritis [53]. CFM-ID online software gave for compound **38** eriodictyol 7-(6-trans-*p*-coumaroylglucoside) as the best match, using the compound identification tool which compares the molecular ion (mass tolerance of 5 ppm) and the MS/MS fragments to data present in online databases. Coumaroyl glucosides seems to show a higher antioxidant effect with respect to the relative glucoside derivative as reported by Li X. et al. [54]: in TAP extract, we found eriodictyol 7-(6-trans*p*-coumaroylglucoside) and quercetin 3-(3-*p*-coumaroylglucoside) (putatively identified with HMDB). The phloretin derivatives typically identified in apples were also found, with a new putative annotation (through HMDB) of hydroxyphloretin glucoside (peak **25**) detected with the untargeted approach. The untargeted approach also revealed two lipids (peaks **41** and **42**), annotated as (10E,15Z)-9,12,13-trihydroxy-10,15-octadecadienoic acid and trihydroxy-octadecenoic acid.

#### *3.2. Quantitative Analyses (Total Polyphenols and HPLC)*

The total polyphenol content was determined both by spectrophotometry and HPLC analysis as reported in the Section 2. Overall, the results derived by these two methods are superimposable, being 24.14 ± 1.58 and 27.97 ± 0.68 mg/100 mg, as determined by the Folin–Ciocalteu colorimetric test and HPLC analysis, respectively.

**Figure 2.** Total ion currents (TICs) of PAT fraction acquired in (**A**) negative ion mode and (**B**) positive ion mode. Identified peaks are numbered progressively, according to the elution order, and their identity or putative identity is shown in Table S1 of the Supplementary Materials with the experimental mass, MS/MS fragments and identification method. IS: internal standard.

By considering the values reported by Sun et al., who found that the total extractable phenolic content from fresh apples is between 110 and 357 mg/100 g [55], we can say that the preparation of the TAP extract resulted in a significant increase in the recovery yield of the polyphenolic fraction.

#### *3.3. Direct Radical Scavenging Activity (DPPH)*

The direct radical scavenging activity was evaluated by the DPPH assay and the results are reported in Table 1 and expressed as IC50. The lower the value obtained, the higher the direct radical scavenging activity. The TAP extract was found to exert a significant radical scavenging activity, which was found to be significantly higher than the reference compounds when expressed on the basis of the polyphenol content as determined by HPLC.


**Table 1.** Direct radical scavenging activity of TAP extract, TAP expressed on the basis of polyphenol content, Trolox, and ascorbic acid. Results are reported as mean ± standard deviation.

#### *3.4. NRF2 Activation and Anti-Inflammatory Activity*

We first tested the effect of TAP on cell viability using the MTT assay up to a concentration of 250 μg/mL. Cell viability was found to be higher than 95% at all the tested doses (data not shown). Figure 3 reports the dose-dependent effect of TAP on NRF2 activation after 6 and 18 h of incubation in a concentration range between 1 and 250 μg/mL. After 6 h, the effect started to be significant at a concentration of 50 μg/mL and induced a 2.3-fold increase at a 250 μg/mL concentration. The fold increase was higher after an incubation time of 18 h to reach more than a 5-fold increase at the highest tested dose.

**Figure 3.** Dose-dependent effect of TAP on NRF2 nuclear translocation. NRF2 activation was tested using NRF2/ARE Responsive Luciferase Reporter HEK293 stable cell line incubated for 6 and 18 h with TAP in a concentration range between 1 and 250 μg/mL. CDDO-Me 75 nM was used as a positive control. Statistical analysis was calculated by two-way ANOVA with Dunnett's multiple comparison test, with individual variances computed for each comparison (\* *p* < 0.05, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, ns: not significant).

The anti-inflammatory activity of TAP extract was then tested in the same concentration range (1–250 μg/mL) using R3/1 NF-κB cells and two different inflammatory inducers: TNF-α and IL-1α. The results are summarized in Figure 4. (A) shows the NF-κB-dependent luciferase activity in cells incubated in the absence (black columns) and presence (gray columns) of IL-1α and treated with TAP. Results are reported as luciferase fold increase with respect to untreated cells. IL-1α induced more than a 5-fold increase in NF-κB-induced luciferase, which was dose-dependently reduced by TAP incubation. Luciferase activity as not affected by TAP in the absence of IL-1α. TAP was then tested using TNF-α as a stimulus (Figure 4B), and in this condition, it was found to dose-dependently reduce the NF-κB-dependent luciferase activity after both 6 and 24 h of incubation. The TAP anti-inflammatory activity was greater when IL-1α was used as an inflammatory stimulus with respect to TNF-α.

**Figure 4.** Dose-dependent anti-inflammatory activity of TAP. Activity was tested in R3/1 with the luciferase gene reported for NF-κB nuclear translocation. (**A**) The NF-κB-dependent luciferase activity in cells incubated in the absence (black columns) and presence (gray columns) of IL-1α and treated with TAP in a 1–250 μg/mL concentration range. Rosiglitazone 10 μM was used as a positive control. (**B**) The dose-dependent effect of TAP on the NF-κB-dependent luciferase activity of cells challenged for 6 h with IL1-α, and for 6 and 24 h with TNF-α, (\*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001—# *p* < 0.05, ## *p* < 0.005, ### *p* < 0.001).

#### *3.5. Quantitative Proteomic Studies*

The effect of TAP on the proteome of control cells and cells stimulated with TNF-α was then studied by using two different quantitative proteomic approaches, SILAC and labelfree, and the following four different experimental groups: control cells, cells stimulated with TNF-α, and cells incubated in the absence and presence of TAP.

#### 3.5.1. SILAC Proteomic Studies

Figure 5 shows the volcano plots obtained from the three comparative analyses: (i) cells incubated in the absence (CTR) and presence of TAP (CTR-TAP), (ii) cells incubated in the absence (CTR) and presence of TNF-α (CTR-TNF); (iii) cells stimulated with TNFα and incubated in the absence (CTR-TNF) and presence of TAP (CTR-TNF-TAP). The analyses were conducted through two orthogonal proteomic approaches, SILAC and the label-free method.

**Figure 5.** Volcano plot of the LFQ and SILAC analyses. The two columns represent the LFQ (left) or SILAC (right) analyses. In the first row, the two volcano plots (**A**,**B**) illustrate the analyses performed to understand the impact of TAP extract on the cell proteome. On the second row, (**C**,**D**) show the volcano plot for the comparative condition achieved by treating cells with TNFα. Lastly, (**E**,**F**) are the volcano plots showing the differential proteome expression for the inflamed cells treated with TAP extract. In each volcano plot, the most representative proteins identified in the various analyses are labeled in red if up-regulated (*p* value < 0.05, log2 *Fold-change* > 0.5), in blue if down-regulated (*p* value < 0.05, log2 *Fold-change* < −0.5) and in black when they do not fall into the two aforementioned conditions. A more comprehensive analysis of the log2 *Fold-change* values is available in Table 2.

**Table 2.** Key proteins identified by quantitative proteomics analysis; fold-change values calculated by each of the experimental approaches used are reported by gene product. The fold-change values are classified according to the comparative analysis of the conditions tested. ND indicates a notdetected protein identification with the corresponding proteomic approach. The value 0 indicates a fold-change not significantly different from zero. Each fold-change set, reported per gene product and per proteomic approach, is colored on a green (minimum value) to light orange (maximum value) color scale according to the conditional formatting function of excel. The gene products involved in the nuclear factor erythroid-derived 2 signaling pathway are highlighted in light violet and those involved in the pentose phosphate pathway in light blue.


\*\* *p* < 0.005.

The first experimental comparison analysis permits the evaluation of the general effect of TAP on the cell proteome in homeostatic conditions. From the SILAC experiment, 24 and 14 proteins were found to be up- and down-regulated, respectively, while 1235 were found to be unchanged. Ingenuity pathway analysis identified two main up-regulated pathways: the nuclear factor, erythroid-derived 2 signaling pathway with a z-Score of +2.9 (Figure 6A), and the pentose phosphate pathway with a z-Score of +2.0 (Figure 6B).

In particular, the up-regulated enzymes involved in the first pathway include HMOX1, CAT, GSTP1, AKR(1–7), NQO1 and SQSTM1, while for the second one, TKT, PGD, G6PDX and TALDO1 are identified. In addition to these proteins, PTGR1 and BLVRB were also found to be up-regulated, as reported in the volcano plots (Figure 5). PTGR1 is an NADPHdependent oxidoreductase which is found to be induced by NRF2 activators [56] and directly regulated by NRF2 [57]. BLVRB is found to be an NRF2 target gene and together with HMOX is critical in the heme metabolism [58] (Figure 11A).

The volcano plot relative to the second experimental comparison analysis (CTR-TNF vs. CTR, Figure 5C,D) mainly identifies two overexpressed proteins involved in inflammation, NF-KB2 and SOD2. As expected, the ingenuity pathway analysis identifies, with a z-Score of 1.8, the TNF-α as up-stream regulator (data not shown).

The volcano plot displaying the third comparison analysis (CTR-TNF-TAP vs. CTR-TNF, Figure 5E,F) reports that TAP treatment down-regulates the two proteins overexpressed in the inflammatory conditions (NF-KB2 and SOD2), while the two main pathways up-regulated by the TPA treatment were confirmed to be the NRF2 and the pentose phosphate pathway.

**Figure 6.** IPA analysis of the effect of the extract on the cell proteome. (**A**) illustrates that the differentially regulated proteins in the analysis are consistent with an NRF2 activation. (**B**) illustrates the Pentose Phosphate Pathway; in red are the proteins that were found to be up-regulated by TAP, resulting in a increment in the NADPH pool in the cell. Color legend: red represents the increased genes, green the decreased (not present in the figure). The intensity of the color is related to the intensity of up- or down-regulation. The orange line leads to activation and a blue line (not present in the figure) leads to inactivation. The yellow line indicates that findings that are not consistent with the proteomics results obtained.

Finally, protein–protein interaction network analysis using the Cytoscape interface in STRING confirmed the modulation of a protein set involved in the biological processes/pathways already highlighted in IPA, as a result of TAP treatment (Figure 7A,B). Assigned GO annotations lead firstly to confirmation of the activation of the NRF2 and pentose phosphate pathways and the involvement of several oxidoreductive enzymes as well as of proteins involved in the metabolism of glutathione (GSH). It should be taken into account that STRING–Cytoscape network analysis does not distinguish between up- and down-regulated proteins but only considers significant expression modulation.

**Figure 7.** Cytoscape and STRING analysis of the up-regulated gene products determined by SILAC. The panels illustrate the protein–protein interaction networks obtained by STRING in the CTR-TAP vs. CTR (**A**) and CTR-TNF-TAP vs. CTR (**B**) conditions. The color/s of each node (gene product) reflect/s the functional enrichment analysis performed; the color code describing the enriched biological processes with the corresponding FDR value is shown below each network. In both analyses, the singlets were left out.

#### 3.5.2. Label-Free Quantitative Proteomic Studies

The label-free quantitative proteomic study, through the use of a Fusion MS analyzer, identified a larger number of proteins with respect to those identified by the SILAC approach, which was based on an LTQ Orbitrap system. In particular, LFQ identified 2340 more proteins than SILAC in the CTR-TNF vs. CTR mixture, 2412 more in the CTR-TAP vs. CTR mixture and 2259 more in the CTR-TNF-TAP vs. CTR-TNF mixture. As shown by the volcano plots relative to the three experimental conditions' comparison, depicted in Figure 5 (lower panels), besides a larger number of identified proteins, the LFQ approach also permitted the identification of more up- and down-regulated proteins in all three conditions. Details about the comparison of the number of up- and down-regulated proteins identified by the two approaches in the three experimental groups are shown by Venn diagrams in Figure 8.

Pathway analysis (IPA) of the LFQ results, besides confirming the results achieved by the SILAC approach, identified further regulated pathways. As shown in Figure 9, a better description of the NF-κB activation was observed because two upstream regulators were found to be compatible with the over-expressed proteins, AKT (protein kinase B, Figure 9B) and IKBKB (inhibitor of nuclear factor kappa-B kinase subunit beta, Figure 9C). AKT regulates the transcriptional activity of NF-κB by inducing phosphorylation and the subsequent degradation of the inhibitor of κB (IκB). IKBKB phosphorylates the inhibitor in the inhibitor/NF-κB complex, causing dissociation of the inhibitor and activation of NF-κB. As a result, the number of proteins identified in the TNF-a pathway significantly increases. Besides AKT and IKBKB, another obvious upstream regulator is reported in Figure 9A, the TNF pathway, which simply confirms the effectiveness of the method used to induce inflammation.

**Figure 8.** Venn diagrams showing technical differences in terms of the number of significantly (log ratio cutoff ± 0.5) up- (first line diagrams) and down-regulated (second line diagrams) proteins identified by the two approaches in the three comparison analyses (CTR-TAP vs. CTR; CTR-TNF vs. CTR; CTR-TNF-TAP vs. CTR-TNF). Additionally, the percentage distribution of identifications in the two approaches and the percentage of common assignments are shown for each diagram.

**Figure 9.** Upstream regulators determined in the CTR-TNF vs. CTR condition by IPA analysis (LFQ). (**A**) The wheel of proteins involved in the TNFα stimuli. (**B**,**C**) The wheels of proteins associated with AKT and IKBKB as upstream regulators. Above is the prediction legend (IPA) essential for understanding networks. \* multiple isoforms.

Regarding the CTR-TAP vs. CTR comparison, the activation of the NRF2 pathway by TAP, as already evidenced by the SILAC approach, was confirmed and the number of proteins involved in the signaling was greatly extended. Furthermore, TAP treatment was associated with a reduction in ferroptosis, apoptosis and oxidative stress (data not shown).

Regarding the third group, CTR-TNF-TAP vs. CTR-TNF, the LFQ approach permitted a better description of the upstream regulators relative to a reduction in oxidative stress and NRF2 pathway due to the increased number of the identified proteins involved in the pathways (Figure 10).

**Figure 10.** IPA analysis in the CTR-TNF-TAP vs. CTR-TNF condition. (**A**) illustrates the different regulated proteins present in the analysis which correlate with NRF2 activation. (**B**) illustrates the effect that the differentially regulated proteins have on the quantity of reactive oxygen species, confirming a trend of reduction in oxidative stress. The prediction legend (IPA) is reported in Figure 9.

#### **4. Discussion**

Epidemiological and intervention studies indicate that consumption of apples and derivatives, such as apple juice, exert beneficial effects relating to some human diseases including CVD and related events, e.g., cancer and diabetes. Some of the health effects, and in particular, the antioxidant and anti-inflammatory activities, have been attributed to the polyphenol fraction, as also confirmed by pre-clinical studies. Hence, there is a growing interest in the use of apple polyphenols as health care products, as well as on their production with green methods. Apple polyphenols are contained in large quantities in thinned apples which represent a large waste material of the apple chain. Based on these premises, the aim of the present study was to evaluate thinned apples as a source for the isolation of bioactive apple polyphenols, effective as anti-inflammatory and antioxidant compounds to be used in health care products. To reach this goal, we first set up a scalable procedure for polyphenol isolation from thinned apple. The polyphenol content of the purified fraction obtained by absorption resin accounted for 24%, as determined by colorimetric analysis, and almost 28% by HPLC, indicating the suitability of the isolation process. The isolated polyphenols were then fully identified by a targeted and untargeted LC-HRMS approach which is, to our knowledge, the first qualitative profile of polyphenols from thinned apples.

The targeted and untargeted metabolomics approaches allowed us to identify a total of 68 polyphenols belonging to the six polyphenolic classes. Moreover, 16 compounds were found in the TAP fraction, though not yet reported in apple or derivatives. All the characteristic polyphenols identified in harvested apples as reported in the most recent literature were identified.

The biological evaluation of the extract followed the qualitative characterization; TAP was first evaluated in two cell models in order to evaluate the antioxidant and anti-inflammatory properties of the extract, the two main activities ascribed to apple polyphenols as reported by the most recent literature. The antioxidant activity was firstly studied by the DPPH assay, which demonstrated a greater radical scavenging activity of the polyphenol fraction with respect to reference compounds, i.e., Trolox and ascorbic acid. For several years, the in vivo antioxidant activity of plant polyphenols has been explained by considering their direct radical scavenging activity towards ROS and carbon centered radicals. However, as pointed out by Forman and Ursini [59], kinetic constraints indicate that in vivo scavenging of radicals is ineffective in antioxidant defense (except for vitamin E towards peroxyl radicals). Instead, enzymatic removal of non-radical electrophiles, such as hydroperoxides, in two-electron redox reactions, is the major antioxidant mechanism. Based on this overview, the major mechanism of action for nutritional antioxidants, and in particular, for plant polyphenols such as those contained in TAP, is the paradoxical oxidative activation of the NRF2 (NF-E2-related factor 2) signaling pathway, which maintains protective oxidoreductases and their nucleophilic substrates. As stated by Ursini, the maintenance of the "Nucleophilic Tone" by a mechanism called "Para-Hormesis" provides a means for regulating physiological non-toxic concentrations of the non-radical oxidant electrophiles that boost antioxidant enzymes, damage removal and repair systems (for proteins, lipids and DNA) at the optimal levels consistent with good health [59].

Hence, based on this new vision of the antioxidant activity of natural antioxidants, as a next step, we tested the ability of TAP to act as an indirect antioxidant by activating the NRF2 pathway.

Results well indicate that TAP dose-dependently activates the NRF2 pathway, as determined by measuring the activity of luciferase, which is the product of the gene reporter activated by the nuclear binding of NRF2. Activation of NRF2 by TAP can be firstly ascribed to the polyphenol compounds bearing an ortho-diphenol moiety which is oxidized to the corresponding quinone, which, being an electrophilic compound, reacts with the thiols of KEAP1, thus releasing NRF2, which then translocates into the nucleus. TAP contains a set of polyphenols containing the catechol moiety, including derivatives of caffeic acid, and quercetin. TAP also contains phenols or methoxy derivatives, such as derivatives of coumaric acid, phloretin and naringenin, which can also be NRF2 activators, but in this case, metabolic activation is required to form a di-phenol moiety. In particular, the metabolic activation of phenols requires the insertion of a hydroxyl group in ortho or para position mediated by the cytochrome enzymes, as found for resveratrol [60], while a CYP mediated O-demethylation activates methoxy derivatives, as reported for silybin [61]. Such metabolic reactions usually occur in the liver tissue and hence are unlikely to occur in the cell model used in the in vitro assay. Hence, we presume that the potency of NRF2 activation is underestimated and that presumably it is potentiated in vivo by metabolic activation occurring in the organism, as well as by the microbiota in the gastrointestinal tract.

NRF2 activation paralleled the dose-dependent anti-inflammatory activity, as demonstrated in the tested cell model (luciferase gene reporter for NF-κB nuclear translocation), and using two different pro-inflammatory agents, IL1-α and TNF-α. A strict interdependence between oxidative stress and inflammation and between the two transcriptional factors involved, NF-κB and NRF2, is well established [62,63]. When oxidative stress appears as a primary disorder, inflammation develops as a secondary disorder and further enhances oxidative stress. On the other hand, inflammation as a primary disorder can induce oxidative stress as a secondary disorder, which can further enhance inflammation [64]. Based on this evidence, a well-established strategy to block the inflammatory chronic condition consists of inhibiting oxidative stress, and in turn, the best approach to inhibit oxidative stress is to activate the nucleophilic tone by inducing NRF2 nuclear translocation.

Hence, the anti-inflammatory activity of TAP can be first of all ascribed to the NRF2- 2-dependent antioxidant effect. To gain a deeper insight into this mechanism, and in particular, to confirm the activation of antioxidant enzymes and to search for other enzymes involved in the anti-inflammatory activity, a proteomic investigation was then carried out.

Quantitative proteomic studies were carried out using the cell line with the gene reporter for NF-κB, used for the anti-inflammatory activity evaluation, and TNF-α as an inflammatory inducer. Two orthogonal proteomic approaches were applied, namely, a label-free method, using a Fusion MS analyzer, and a stable isotope in vivo labeling method, i.e., SILAC, coupled to a LTQ Orbitrap system. In general, the label-free approach identified many more proteins with respect to SILAC and was revealed to be more efficient in the identification of up- and down-regulated pathways. These differences could be easily explained by considering the intrinsic potential in terms of sensitivity and resolution characteristics of the mass analyzers used for each of the experiments. Besides the different performances, the two approaches identified the same modulated pathways, although with a different number of proteins, making the results more robust, as confirmed by two orthogonal approaches. This very interesting technical aspect is particularly evident in the following table (Table 2), which shows the experimentally calculated fold-change values for each key protein identified, whose expression is significantly modulated in the comparative analysis between the tested conditions.

The first relevant observation emerging from the comparison of the calculated foldchange values in the different conditions tested is the apparent shift in the expression of some of the proteins involved in the NRF2 and pentose phosphate pathways. A fitting example could be the HMOX1 expression modulation, having a negative fold-change (LFQ: −0.2424, SILAC: −0.1233) in the first comparison, meaning that inflammation causes its down-regulation, and showing a significant up-regulation when the cells are treated with TAP extract, in both conditions.

Targeting a more comprehensive biological significance, rather than investigating individual key proteins, the global effect of TAP treatment in physiological conditions (CTR), to identify the set of genes whose expression is variably modulated by treatment, was evaluated. As expected, many up-regulated genes involved in the NRF2 pathway activation were detected, particularly antioxidant enzymes, or those directly involved in the detoxification of peroxides and in two-electron redox reactions, e.g., SOD, CAT, GXP and redox proteins such as thioredoxins. Furthermore, a set of enzymes required for the synthesis of GSH and NADPH, co-factors for the enzymes involved in the antioxidant and redox regulation, were also identified. GCLC, GSTA3 and GSTP were the identified up-regulated enzymes involved in the GSH synthesis and metabolism while the enzymes belonging to the oxidative pentose-phosphate pathway are those up-regulated for NADPH synthesis, such as malic enzyme-1 (ME-1), isocitrate dehydrogenase-1 (IDH1), glucose-6 phosphate dehydrogenase (G6PDX) and 6-phosphogluoconate dehydrogenase (PGD).

The crucial finding is that TAP treatment over-expresses the inducible isoform of heme oxygenase (HMOX1), a well-established immunomodulator [65–67]; it has been proven that the induction of HMOX1 protects against the cytotoxicity caused by oxidative stress and apoptotic cell death, making HMOX1 an appealing target for the treatment of several chronic inflammatory diseases, including osteoporosis [68], cancer [69], acute kidney injury [70], retinal pigment epithelium degeneration [71] and Parkinson's disease [56,67].

In physiological conditions, HMOX1 is involved in rate-limiting heme degradation using NADPH–cytochrome P450 reductase (POR) and oxygen to generate linear tetrapyrrole biliverdin (further converted into linear tetrapyrrole bilirubin by the enzyme biliverdin reductase, BLVRB), ferrous iron (Fe2+) and carbon monoxide (CO) (Figure 11A). The enzymatic heme-degradation by-products are known to be the main cause of the beneficial protective effects promoted by HMOX1; bilirubin, in particular, is one of the most potent antioxidant and is particularly effective in protecting against lipid peroxidation.

**Figure 11.** Signaling pathways in which HMOX1 is involved. (**A**) Heme degradation pathway promoted by HMOX1 occurring in the cytosol of the cell. Heme degradation occurs in the cytosol by the enzyme HMOX1; the reaction involves cytochrome P450 reductase, NADPH as cofactor and O2, leading to the production of NADP+, and an equimolar amount of biliverdin, carbon monoxide and iron (Fe2+). Specifically, biliverdin can be further converted to bilirubin by the enzyme biliverdin reductase (BVR). (**B**) Regulation of HMOX1 expression under homeostatic conditions. NRF2 is bound to KEAP1, ubiquitinated and targeted for proteasomal degradation. BACH1 binds the antioxidant response element (ARE) at the promoter of HMOX1, inhibiting transcription of the gene. (**C**) Regulation of HMOX1 expression during inflammation and oxidative stress. Increased intracellular heme levels lead BACH1 to dissociate from ARE, and reactive oxygen species (ROS), interacting with cysteine residues on KEAP1, cause dissociation from NRF2, which in turn migrates into the nucleus, binds ARE and promotes transcription of HMOX1 (adapted from [67]).

HMOX1 can be up-regulated in response to numerous different stress stimuli, including various pro-oxidant and pro-inflammatory mediators; most of them do not interact directly with the transcription factors, but activate them via intermediate signaling pathways. HMOX1 expression is largely under the control of NRF2, a well-known binder of the ARE (antioxidant response element), aimed at promoting many antioxidant genes, including HMOX1. Under homeostatic conditions, NRF2 is bound to Kelch-like ECH-associated protein 1 (KEAP1), ubiquitylated and targeted by the proteosome for degradation (Figure 11B); meanwhile, BACH1 binds to ARE and inhibits gene transcription. An increased level of reactive oxygen species (ROS) during cell stress implies a concomitant KEAP1-NRF2 dissociation due to their reactivity against thiols in KEAP1; in turn, NRF2 translocates to the nucleus and binds the ARE to promote transcription of HMOX1 (Figure 11C). Basically, NRF2, KEAP1 and BACH1 constitute an intricate feedback system enabling cells to respond to oxidative stress ending at the up-regulation of HMOX1.

The pathway activation in this study is supported not only by the significant modulation of HMOX1 but also by the up-regulation of BLVRB and POR (NAPH-Cytochrome P450 reductase), which, although not significantly up-regulated (fold-change: 0.340528912, *p*-value n.s.), exactly matches the trend of the other associated gene products. In addition, activation of the pentose phosphate pathway leads to a plausible increase in circulating NADPH, a cofactor of both POR and BLVRB (Table 2).

The phenotypic effect described here, in terms of protein expression induced by TAP treatment, is completely concordant when comparing CTR-TNF-TAP to CTR-TNF conditions. The protective effect manifested, such as the triggering of the two pathways of interest, is perhaps even more evident looking at the volcano plots shown in Figure 5A,B: AKR1B1, AKR1B8, BLVRB, CAT, G6PDX, GSTP1, HMOX1, PGD, PTGR1, SQSTM1, TALDO1 and TKT (listed alphabetically) switch to a fold-change value indicating significant up-regulation, explaining the anti-inflammatory activity overall.

Besides the up-regulated set of NRF2-dependent pathways, another key point is the regulation of TAP towards the increased content of NF-κB in cells exposed to TNF-α with respect to control (Figure 5C,D), and the overexpression of a set of proteins which are known to be overexpressed by the TNF-signaling. Hence, NF-κB, in the cell model used, besides being activated and translocated at the nuclear level, as evidenced by the gene reporter, is also overexpressed, thus sustaining the inflammatory process. As stated above, TAP treatment of the inflamed cells not only inhibits NF-κB activation but also its overexpression: NF-κB2, together with SOD2, returns to homeostatic conditions, even if the fold-change was not significantly different from zero (Figure 5E,F). An interesting aspect to stress is that NF-κB, together with AP-1 (Figure 11C), can also bind the HMOX1 promoter to enhance its expression; indeed, some instances of HMOX1 up-regulation in vitro have been shown to be dependent on or correlated with NF-κB or AP-1 expression [72].

#### **5. Conclusions**

In conclusion, thinned apples represent a valuable source of apple polyphenols whose qualitative profile was fully elucidated by targeted and untargeted metabolomic approaches. The anti-inflammatory and antioxidant activities of TPA were first evaluated by cell-based assays using engineered cells with gene reporters for NRF2 and NF-κB nuclear translocation. Quantitative proteomic studies were then used to deeply investigate the molecular mechanisms involved in the proven biological activity. Two different proteomic approaches, i.e., label-free and SILAC, both confirmed TAP ability in improving the antioxidant cellular defense, by activating the NRF2 signaling pathway which up-regulates protective oxidoreductases and their nucleophilic substrates. A clear increase in enzymatic antioxidants together with the up-regulation of the heme-oxygenase explain the anti-inflammatory effect of TAP. Taken together, these results suggest that thinned apples can be effectively considered a valuable source of apple polyphenols to be used in health care products to prevent/treat oxidative and inflammatory chronic conditions.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antiox11081577/s1, Table S1: Compounds identified or putatively identified by LC-HRMS in negative and positive ion modes with the targeted method ordered on the basis of the retention time; Table S2: Compounds putatively identified by LC-HRMS in negative and positive ion modes with the untargeted method ordered on the basis of the retention time.

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

**Funding:** LDV and GF are supported as Ph.D student and temporary researcher (RTDA) by Ministero dell'Università e della Ricerca PON "Ricerca e Innovazione" 2014–2020, Azione IV.4—"Dottorati e contratti di ricerca su tematiche dell'innovazione" and Azione IV.6—"Contratti di ricerca su tematiche Green". This research is part of the project "MIND FoodS HUB (Milano Innovation District Food System Hub): Innovative concept for the eco-intensification of agricultural production and for the promotion of dietary patterns for human health and longevity through the creation in MIND of a digital Food System Hub", cofunded by POR FESR 2014–2020\_BANDO Call HUB Ricerca e Innovazione, Regione Lombardia. This work was supported by the University of Milan through the APC initiative.

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available.

**Acknowledgments:** The authors are grateful to AKHYNEX S.r.l. POLISTENA (RC) for the partial financial support.

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

#### **Abbreviations**



#### **References**


### *Review* **Pressurized Liquid Extraction for the Recovery of Bioactive Compounds from Seaweeds for Food Industry Application: A Review**

**Ana Perez-Vazquez 1,†, Maria Carpena 1,†, Paula Barciela 1, Lucia Cassani 1,2,\*, Jesus Simal-Gandara 1,\* and Miguel A. Prieto 1,2,\***


**Abstract:** Seaweeds are an underutilized food in the Western world, but they are widely consumed in Asia, with China being the world's larger producer. Seaweeds have gained attention in the food industry in recent years because of their composition, which includes polysaccharides, lipids, proteins, dietary fiber, and various bioactive compounds such as vitamins, essential minerals, phenolic compounds, and pigments. Extraction techniques, ranging from more traditional techniques such as maceration to novel technologies, are required to obtain these components. Pressurized liquid extraction (PLE) is a green technique that uses high temperatures and pressure applied in conjunction with a solvent to extract components from a solid matrix. To improve the efficiency of this technique, different parameters such as the solvent, temperature, pressure, extraction time and number of cycles should be carefully optimized. It is important to note that PLE conditions allow for the extraction of target analytes in a short-time period while using less solvent and maintaining a high yield. Moreover, the combination of PLE with other techniques has been already applied to extract compounds from different matrices, including seaweeds. In this way, the combination of PLE-SFE-CO2 seems to be the best option considering both the higher yields obtained and the economic feasibility of a scalingup approximation. In addition, the food industry is interested in incorporating the compounds extracted from edible seaweeds into food packaging (including edible coating, bioplastics and bionanocomposites incorporated into bioplastics), food products and animal feed to improve their nutritional profile and technological properties. This review attempts to compile and analyze the current data available regarding the application of PLE in seaweeds to determine the use of this extraction technique as a method to obtain active compounds of interest for food industry application.

**Keywords:** pressurized liquid extraction; seaweeds; green extraction technique; bioactive compounds; functional ingredients; food packaging; future trends

### **1. Introduction**

*1.1. Seaweeds as a Bioactive Compound Matrix*

Seaweeds, also known as macroalgae, are eukaryotic, photosynthetic, pluricellular organisms found in the marine environment. They are divided into three groups: green (Chlorophyta), brown (Phaeophyta) and red algae (Rhodophyta). They are widely consumed in Asia, with China being the world's largest producer [1]. In recent decades, Western countries have become interested in seaweeds due to their high nutritional value [2]. Seaweeds are distinguished by their high quality profile of lipids, proteins, essential minerals, phenolic compounds and pigments. Several species, for example, such as *Palmaria palmata*, *Vertebrata lanosa* and *Enteromorpha intestinalis*, have been reported to have high

**Citation:** Perez-Vazquez, A.; Carpena, M.; Barciela, P.; Cassani, L.; Simal-Gandara, J.; Prieto, M.A. Pressurized Liquid Extraction for the Recovery of Bioactive Compounds from Seaweeds for Food Industry Application: A Review. *Antioxidants* **2023**, *12*, 612. https://doi.org/ 10.3390/antiox12030612

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 31 December 2022 Revised: 14 February 2023 Accepted: 16 February 2023 Published: 1 March 2023

**Copyright:** © 2023 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/).

quality profiles of essential amino acids and lipids when compared to other food matrices such as rice, corn, or wheat. Seaweed are an interesting matrix for the industry due to the bioactive compounds and the hydrocolloids found in them [3]. It is important to note that the species, harvest season, and eco-habitat are all factors that influence the composition of seaweed [2]. Aside from their nutritional value, some compounds found in seaweeds have different technological and biochemical properties that can be used in the food industry, either to improve the food process or to increase the nutritional value of food products. The industrial functionality as well as the principal components and bioactive properties of seaweed compounds will be briefly described to demonstrate the potential benefits of their incorporation into food products.

Seaweed carbohydrates (CH) are classified in hydrocolloids (which include carrageenan, alginates, fucoidans and laminarin) and phycocolloids (agar being the most relevant example). These carbohydrates typically account for 4–76% of the dried weight (DW), with *Ulva lactuca* reporting an 65% of DW as one of the major CH contents reported in seaweeds [4]. The lipidic content of these organisms is typically less than 5% [3], but they have been reported to present a high-quality profile of fatty acids (FAs), which may vary between 1–5% [5] and 1–8% [6]. For instance, the FA content of *Laurencia filiformis*, *Cystoseira baccata* and *Fucus vesiculosus* is 6.2%, 6.7% and 6.6%, respectively, which proves that algae genera influence the FA content. Moreover, brown and red algae have higher lipid content than green algae [7]. Despite having a lower FA content than microalgae, seaweeds are interesting to the food industry for several reasons. For example, seaweed treatment is simpler than microalgae treatment, and seaweeds contain a significant amount of unsaturated fatty acids (USFA) [7]. Protein values range between 3 and 47% with brown algae having the lowest DW of this compound. The most important seaweed protein for the food industry is lectin. Because lectin is a glycoprotein with carbohydrate-binding properties, it can agglutinate yeasts, tumor cells and erythrocytes. Because lectin has antimicrobial, antitumor and antiviral activity, it could be used as a functional ingredient [4]. The main micronutrients in algae are inorganic minerals and vitamins. These organisms are high in potassium, sodium, magnesium and calcium, as well as vitamins A, B and E, and particularly vitamin B12.

Certain organisms, including algae, produce secondary metabolites as defensive and/or adaptive responses to environmental stresses. Among these compounds, pigments and phenolic compounds are the most investigated. Regarding pigments, chlorophylls, carotenoids and phycobilins are the three major classes of photosynthetic pigments in seaweeds. Carotenoids have higher industrial value due to their ability to provide color naturally as well as their bioactive function. Carotenoids (β-carotene, lutein and astaxanthin) have antioxidant activity, immune system effects and can help to prevent cardiovascular diseases and non-alcoholic fatty liver diseases [8]. Fucoxanthin, the pigment found primarily in brown algae has attracted industry attention in the recent years due to a wide range of biological properties that may be of interest to the food and nutraceutical industries [8]. Seaweed polyphenols have also been studied, with phlorotannins being the most extensively studied. Phlorotannins are polymers and oligomers composed of several phloroglucinol units linked in different ways [9,10]. These highly hydrophilic secondary metabolites are only produced by brown algae and range in molecular size from 162 Da to 650 kDa [11,12]. Phlorotannins have been linked to several biological functions including antioxidant, anticoagulant, antibacterial, anti-inflammatory and anti-diabetic activities [12]. Therefore, the incorporation of these compounds in food formulations may be noteworthy.

#### *1.2. Extraction Techniques Applied in Seaweeds Matrices*

The first step in extracting target compounds from any matrix is extraction. A solvent that can penetrate the solid matrix is required for this process so the target compound can be dissolved in and extracted, and the solute of interest can be separated from the raw material [13]. Different extraction techniques have been used to release the previously mentioned target compounds. The extraction methodologies can be classified in two groups: traditional extraction techniques and new extraction techniques. Traditional methods are maceration, percolation and reflux extraction. In these techniques, an organic solvent is usually used, and large volumes and a long time of extraction are needed. These drawbacks motivated the search for new alternatives to traditional extraction techniques, which is usually related to the green revolution concept. Food technologies for preservation, processing, extraction and analysis have evolved from those conventional procedures to more innovative and environmentally friendly processes by reducing fossil energy use and hazardous solvents while avoiding water loss and residues generation. Therefore, the design of green and sustainable processes, and particularly, green extraction processes, remains a hot topic in the food industry [14].

Maceration is a solid-extraction technique characterized by its low cost and simple equipment requirements. The solubility of the target compound is determined by agitation and temperature in this technique. Moreover, the protocols used in this process are easily adaptable to obtain a wide range of compounds of interest. This is possible because different solvents, temperatures and agitation conditions can be selected, increasing mass transfer selectivity and efficiency. Unfortunately, several cycles of filtration or centrifugation are required to separate the compounds from the matrix [15]. Percolation is an extraction technique that works continuously. As a result, the saturated solvent is continuously replaced by a fresh one, increasing efficiency when compared to maceration. The operating conditions are typically room temperature and atmospheric pressure, but heating can also be used. Finally, compared to maceration and percolation, reflux extraction has a higher efficiency, despite the fact that the time of extraction and the amount of solvent required are slightly lower. Moreover, the operational parameters are atmospheric pressure and heating [13].

New extraction techniques are characterized by shorter extraction times, lower operational temperatures, reduced solvent amount and process automation. Moreover, because of the previously mentioned benefits, these techniques are regarded as environmentally friendly. Different parameters should be optimized to obtain higher yields of the target analyte. Thus, in these new methodologies, common parameters to be optimized include solvent ratio, extraction solvent, extraction time, pH, temperature and particle size [16,17]. New methodologies for extracting seaweed compounds include ultrasound-assisted extraction (UAE), supercritical fluid extraction (SFE), microwave-assisted extraction (MAE), enzyme-assisted extraction (EAE) and pressurized liquid extraction (PLE).

UAE is based on the application of low-frequency (16–100 kHz) and high-power (8–20 W) waves to disrupt cells, releasing target analytes, accelerating the diffusion and increasing mass transfer [17]. This technique has been used successfully to extract pigments, phenolic compounds and carbohydrates from different seaweed species [18]. SFE is based on the use of solvents at pressures and temperatures above their critical points, so solvents are denser than gases but have comparable viscosity and intermediate diffusivities to liquids and gases. This method has been used to extract carotenoids, chlorophylls, PUFAs and polyphenols [17]. MAE is based on the heat produced by the direct interaction of electromagnetic waves (usually 2.45 GHz) with polar solvent molecules through dipole rotation and ionic conduction [17]. MAE has been used to extract several seaweed compounds, including carbohydrates and proteins [19]. EAE is a very selective and specific method since enzymes are used to degrade the cell wall of algae cells. For this process, the enzyme concentration and the optimal enzyme reaction conditions should be optimized to improve specificity and selectivity [17]. This extraction method has been used to extract phlorotannin, proteins or hydrocolloids from various seaweeds [20].

PLE is a green extraction technique that involves extracting analytes from a solid matrix using high temperature and pressure, typically between 50–20 ◦C and 3.5–20 MPa [21]. With these conditions, both solubility and mass transfer rates are increased, leading to a solvent diffusivity increment and, therefore, meliorating matrix kinetics [17]. In this way, an experimental design is needed so all the parameters can be selected to guarantee optimal operational conditions [22]. Moreover, PLE allows the use of several solvents, including green extraction solvents such as water, and a mixture of water with ionic liquids or eutectic solvents. When water is used as the solvent, this technique is also known as high-pressure solvent extraction

(HPSE), accelerated solvent extraction (ASE), enhanced solvent extraction, pressurized fluid extraction (PFE), pressurized hot solvent extraction (PHSE) or subcritical water extraction (SWE) [23]. When compared to the SFE technique, PLE operates without reaching the critical point of the liquid solvent and allows the use of broader range of solvents [21,22]. Moreover, PLE requires less extraction time than other traditional extraction techniques, such as Soxhlet extraction. In fact, the extraction time ranges from 5 to 20 min [24]. Furthermore, the use of PLE allows the achievement of higher yields, despite the fact that is not suitable for the extraction of thermolabile compounds and is not as selective as SFE [21,25]. However, PLE may be considered as a suitable green extraction technique to extract different bioactive compounds including polysaccharides, proteins and PUFAs since non-toxic solvents and high extraction yields are obtained. Considering the non-toxic solvents used, the extraction of these compounds with PLE may be an interesting pathway for the food industry.

The aim of this study was to compile, analyze and organize the available data on seaweeds components and the use of PLE as a potential extraction technique to obtain active compounds of interest to the food industry. Furthermore, the effects of combining PLE with other novel techniques for increasing extraction yield were revised. Finally, the use of different compounds extracted from edible seaweeds in the food industry was summarized to identify a potential new pathway in this production sector.

#### **2. Matrix Components**

This section compiles the main characteristics and interest of different compounds found in seaweeds. It is important to highlight that the information described is primarily aimed at the food industry, although pharmaceutical approximation was also considered. Moreover, the nutritional composition of different microalgae is compiled in Table 1 to provide comprehensive information.


**Table 1.** Bioactive compounds of microalgae, expressed as percentage dry weight, and their functional properties.


**Table 1.** *Cont.*

*Abbreviations:* EPA (eicosapentaenoic acid); DHA (docosahexaenoic acid); ARA (arachidonic acid).1\* is referred to *N. oculata* polysaccharides; 2\* is referred to *A. platensis* proteins; 3\* expressed as total fatty acids.

#### *2.1. Proteins*

Proteins are large molecules composed of smaller units known as amino acids that are linked together with aminoacidic bonds. Considering the amino acid profile of seaweeds, it is important to note that most of them contain all the essential amino acids, with aspartic and glutamic acid being particularly abundant. Figure 1 depicts the chemical structure of these amino acids found in edible seaweeds. *Ulva lactuca* (a green edible seaweed) has an amino acid profile that is like that recommended by the Food and Agricultural Organization (FAO) and the World Health Organization (WHO). Seaweed proteins are suitable for inclusion in food formulations due to their amino acid profile and agglutination properties [4]. Several authors have proposed producing antioxidant hydrolysates by hydrolyzing seaweed proteins. Researchers used commercial enzymes to hydrolyze *Ecklonia cava*, *Ishige okamurae*, *Sargassum fullvelum*, *Sargassum horneri*, *Sargassum corearum*, *Sargassum thunbergii* and *Pyropia columbina* to obtain bioactive peptides. *P. columbina* increased its antioxidant activity after a simulated gastrointestinal digestion [34]. Thus, since some studies have found drawbacks in the digestibility of proteins from some seaweeds, the use of cellulases, xylanases and β-glucanases has been studied to improve the digestibility of protein from

*Palmaria palmata* [15]. Furthermore, seaweed proteins such as phycobiliproteins have been linked with anti-inflammatory, hepatoprotective and antioxidant activities [46].

**Figure 1.** Chemical structure of the main amino acids, polysaccharides, lipids, and pigments extracted from edible seaweeds.

Moreover, a relationship has been discovered between seaweed bioactive peptides and metabolic syndrome. The metabolic syndrome is a collection of medical conditions that can lead to different cardiovascular diseases, and it can be avoided with functional foods. Indeed, NMEKGSSSVVSSRMKQ is the first antithrombotic peptide produced by hydrolyzing *Porphyra yezoensis* proteins with pepsin. This peptide binds to the coagulation pathway and inhibits it. Furthermore, seaweed-derived bioactive peptides inhibit some enzymes involved in the renin–angiotensin–aldosterone system (RAAS), which plays a key role in the hypertension treatment [4]. Microalgae proteins have been reported to be good gel and foam formers, with *Arthosphira platensis* being particularly notable. Moreover, microalgae are also an interesting source of proteins and bioactive peptides, as shown in Table 1. A study published in 2022 found that the protein composition of *Nannochloropsis granulata* and *Microchloropsis gaditana* was high, with 45.8 and 47% expressed as DW, respectively. Furthermore, studies revealed that microalgae peptides from different genera, including *Chlorella, Nitzschia* and *Bellerochae* are distinguished by their antihypertensive, antibiotic and antiviral activities [40,41]. Figure 2 depicts a graphical representation of different micro and macroalgae protein composition, except for *Nannochloropsis granulata, Nannochloropsis limnetica* and *Microchloropsis gaditana*, which had higher protein content. Based on the data presented, seaweeds are a suitable and complete source of proteins that can be incorporated into a variety food products.

**Figure 2.** Graphical comparison between different microalgae and macroalgae protein (**A**) and lipid (**B**) composition, expressed as dried weight (%). Green color refers to green seaweeds, brown color refers to brown seaweeds and purple color refers to microalgae species.

#### *2.2. Carbohydrates*

Seaweed contains monosaccharides, disaccharides and polysaccharides. Polysaccharides found in seaweeds are classified into two groups: sulphated and non-sulphated. As is shown in Figure 1, sulphated polysaccharides include fucoidans, carrageenan and laminarin, while non-sulphated polysaccharides are mainly alginates [47]. Seaweed polysaccharides and fiber are not digestible by humans. Moreover, in vitro studies have shown that polysaccharides from *Undaria pinnatifida*, *Laminaria japonica* and *Hizkia fusiformis* inhibit pepsin activity by 21%, 55% and 41%, respectively [15]. Despite their inability to be digested, these compounds serve as prebiotics in the human body because they can be degraded by intestinal bacteria [3]. However, it is important to highlight that some studies have linked prebiotic consumption to human flatulence. In fact, 14 women were studied for 4 weeks after consuming inulin, and 12% of the volunteers experienced severe flatulence [48]. Moreover, because seaweeds are composed of soluble fiber with high capacity to retain water (CRW), they can be used as hydrocolloids in food formulations. Some of the functions of hydrocolloids include thickening, stabilizing and emulsifying [3].

Carrageenan is a sulphated linear galactant found in 71% and 88% of *Chondrus crispus* and *Kappaphycus* spp., respectively. Carrageenan gel form from *Chondrus crispus* has been shown to have antiviral and anticoagulant properties against HIV and herpes simplex virus (HSV). The anticoagulant capacity of this compound has been related to the sulphate molecules in the polysaccharide chains [4].

Alginates are non-sulphated linear unbranched polysaccharides found in the intercellular spaces of brown algae [4,49]. *Laminaria*, *Saccharina*, *Lessonia*, *Macrocystis*, *Durvillaea*, *Eckonia* and *Ascophyllum* are the main seaweeds used to obtain this polysaccharide. Alginates are used in the food industry due to their technological properties, as these compounds are characterized by their gelation ability and become insoluble as a result of the formation of a cross-linked structure. Because of this, alginate is an excellent material for the active edible coating systems on foods such as fruits and vegetables [50]. Furthermore, the Food and Drug Administration (FDA) designated alginic acid and its salts as generally regarded as safe (GRAS) ingredients for oral administration [4].

Laminarins (Figure 1) are the main storage polysaccharides found in the cytoplasm of brown algae [40]. This small glucan is isolated from brown seaweeds and has a molecular weight that ranges from 1 to 10 kDa [51]. Laminarin can be found in a variety of seaweed species, including *Laminaria* spp., *Ascophyllum*, *Fucus* and *Undaria* [4]. This compound is well known for its gelling and emulsifying properties and is commonly used in the food industry as an additive [51]. Moreover, several studies have shown that laminarin has anti-apoptotic, anti-inflammatory, immunoregulatory, antitumor, anticoagulant and antioxidant activities [52]. These biomedical properties may be related to the sulphated composition of seaweed polysaccharides, which are not found in the terrestrial plants [4]. When the average molecular weight of laminarin was reduced to six kDa, its antioxidant activity increased from 7.5 to 79.7%. This could be explained by the fact that as the average molecular weight decreases, the number of carbonyl groups increases, interacting with transition metal ions and enhancing lipid oxidation protection [51]. Moreover, laminarin has a positive effect on the biochemistry and microbiology of the human gut microflora, modulating the intestinal metabolism in a positive way [4].

Fucoidans are sulphated polysaccharides (SP) found in brown seaweeds such as *Saccharina japonica*, *Laminaria ochroleuca* and *Himanthalia elongata*. The anti-inflammatory, anticoagulant, antitumoral, anti-thrombotic, antioxidant and antiviral activities of fucoidans have been associated with their sulphation level [34,40]. An in vitro and in vivo study of *Fucus evanescens* fucoidan's anticoagulant activity revealed that it has similar anticoagulant properties to heparin [4]. The anticoagulant activity has been related to the sulphation level, carbohydrate content and the position of sulphated groups on sugar residues [53]. In addition, fucoidans may alter the cellular surface properties effectively preventing virus penetration, and as a result, antiviral activity [53]. A study using *Laminaria japonica* as a source of fucoidans showed its ability to scavenge superoxide radicals and hypochlorous acid. The low molecular fraction of *L. japonica* fucoidans also had a significant inhibitory effect on low-density lipid (LDL) oxidation induced by Cu2+ [53]. Thus, fucoidans may be used to prevent free radical-mediated diseases. Finally, the antitumor activity of fucoidans

from brown seaweed was studied, and it was discovered that this sulphated polysaccharide could inhibit the proliferation of tumor cells by stimulating the apoptosis, blocking tumor cell metastasis, and enhancing immune response. These antitumor properties may lead to the use of fucoidans as functional ingredients or nutraceuticals [53]. Considering all the data presented and the increased interest of consumers in bioactive compounds in recent years, fucoidans may be considered as functional ingredients in the food industry [53].

The composition of microalgae polysaccharide (Table 1) has also been studied, showing immunomodulatory activity, low blood sugar and lipid levels in vitro and prebiotic activity [27,28]. Moreover, sulphated polysaccharides derived from *Arthosphira*, *Chlorella*, *Phaeodactylum*, *Schizochytrium* and *Thrautochytrium* showed antiviral, antitumor, and antiinflammatory activity [29,30].

#### *2.3. Lipids*

The lipidic fraction in seaweeds varies between 1–8%, with the most common longchain polyunsaturated fatty acids (PUFAs) found in seaweeds being γ-linolenic, α-linolenic, eicosapentaenoic and docosahexaenoic acid, whose chemical structure is shown in Figure 1. Because PUFAs have been linked to the prevention of cardiovascular, diabetes and hypertension diseases, their presence in seaweed has an interesting functional activity [3]. Moreover, the w-6/w-3 ratio is commonly used to define lipid quality [7]. According to the data, the w-6/w-3 ratio is 15/1 in Western diets, while the FAO recommends a ratio of less than 10. Because a high w-6/w-3 ratio is associated with the progression of various coronary diseases, the WHO recommends substituting saturated fatty acids for polyunsaturated fatty acids [6]. Furthermore, the atherogenic index (AI) and thrombotic index (TI) are parameters that indicate the lipid deposition in the artery wall as well as the thrombotic effect of saturated fatty acids (SFAs), respectively. Both indices are widely used to assess the quality of SFAs. In 2019, a study on the bioactive fatty acids extracted from *Laminaria ochroleuca* was conducted, with AI and TI results comparable to those obtained from some fish species [7]. It is important to note that the fatty acids in seaweeds are in a phospholipid and glycolipid form, which confer the cell wall membrane. This allows for very little degradation during digestion. Hence, a mechanical disruption of the cell wall is needed in order for the seaweed lipid content to be released and absorbed [3]. To demonstrate the differences between microalgae and macroalgae species, the lipid composition of each was compared. As is shown in Figure 2, the lipid composition of microalgae is richer than that of seaweed. In fact, the lipid composition of microalgae varies depending on the genus (Table 1), reaching 46.12% of DW in some species of *Nannochloropsis*. Moreover, the lipid composition of microalgae is rich in PUFAs such as EPA, ARA and DHA, which has been linked to different biological properties such as anti-inflammatory, anti-diabetes, antithrombotic and anticancer activities, as well as a high capacity to prevent cardiovascular diseases [31–33,38]. Although the lipid content of microalgae is higher than that of seaweed, based on the data presented, seaweed may be a valuable source of quality fatty acids for the food industry, capable of being used in vegetarian and vegan formulations.

#### *2.4. Pigments*

Chlorophyll (Figure 1) is one of the most abundant pigments on the planet, and it has mainly been studied in higher plants [54]. The significance of this phytochemical is due to its antimutagenic and antigenotoxic activity [54] as well as its potential use in the food industry as a natural pigment. A study conducted in 2017 identified the chlorophyll profile, distinguishing over 31 pigments. Although extraction protocols for these pigments in seaweeds are not well developed [55], the abundance of these phytochemicals in edible seaweeds makes this matrix a good source of them. Moreover, the stability of chlorophyll in fresh and cooked *Porphyra* seaweeds was studied during in vitro digestion. The bioaccessibility of cooked *Laminaria* chlorophyll seaweed improved, while processing decreased this parameter [54].

Fucoxanthin (FUCO) is a carotenoid found in the glycoglycerolipids of brown algae chloroplasts and is involved in the photosynthesis process [8,56]. FUCO accounts for 10% of total carotenoids in nature and has a market with a 2.47% annual growth rate. This pigment is present in several algae genera, with *Undaria pinnatifida* having the highest concentration. The main bioactive properties of FUCO are the antioxidant, anticancer and anti-inflammatory activity, as well as the cytoprotective and skin protective effects [8]. The antioxidant effect of this compound is explained by the presence of an allenic bond and an acetyl functional group in its structure (Figure 1), both of which can scavenge different free radicals. Thus, it has been demonstrated that FUCO reduces the production of intracellular ROS and DNA damage while increasing glutathione levels, which is a key molecule in oxidant defense and the maintenance of the redox cell homeostasis. All these actions contribute to the prevention of apoptotic processes [8]. The skin protective effect of FUCO has also been studied using oral administration. Results showed that this compound suppresses transcription of the melanogenesis factor by inhibiting mRNA expression. Therefore, FUCO could be used to prevent harmful effects of ultraviolet (UV) radiation, such as melanomas [8]. The ideal conditions for FUCO incorporation into food formulations have already been studied, with the conclusion that an encapsulation should be made with a solution with pH 5–7 and it should be stored at 4 ◦C [8].

#### *2.5. Metals*

As previously stated, seaweeds are a rich source of bioactive compounds that can be used in the food industry. However, seaweeds also contain significant amounts of metals. In fact, seaweed consumption has been considered as a high-risk route for heavy metals and metalloids due to their high capacity to bioaccumulate these compounds. Thus, the main metals found in these matrices are lead (Pb), cadmium (Cd), mercury (Hg), and arsenic (As) [57]. Red algae are high in selenium (Se), manganese (Mn), nickel (Ni) and silver (Ag), while brown seaweeds are high in copper (Cu), cobalt (Co), chromium (Cr), As and iron (Fe). Finally, zinc and Pb are commonly found in green seaweeds [58]. These metals are toxic and persistent, and their consumption may cause endocrine disruption and carcinogenic activity [59]. However, the potential risk of consuming seaweeds grown in Saint Martin's Island was studied, which is potentially a risk zone for heavy metal accumulation. A total of 21 heavy metals and metalloids were analyzed, and no health risk was found because bioaccumulation was below the established limits (Hazard Quantities < 1) [57]. Another study analyzed 11 species of seaweeds grown in South China Sea, showing a high degree of variability and complexity [58]. In this way, considering the importance of ensuring the consumers' health, heavy metal and metalloid analysis of the species used is required.

#### **3. Pressurized Liquid Extraction (PLE)**

#### *3.1. General Aspects of PLE*

PLE is an extraction technique that consists of the removal of analytes present in a solid matrix by applying high temperatures (*Textr*) and pressure (*Pextr*), usually up to 200 ◦C and over 200 bar, respectively according to Nieto et al. [22], without reaching the critical point using liquid solvents [23]. These conditions increase solubility and mass transfer rates, resulting in increased solvent diffusivity and, as a result, improved matrix kinetics [17].

Temperature, pressure, time, number of cycles, sample weight and solvent all influence extraction yield. To improve the efficacy of PLE, these parameters should be optimized by using a proper experimental design [22].

Figure 3 depicts a schematic representation of the PLE extraction equipment's operation. A high-pressure pump feeds the solvent into an extraction cell and the *Pextr* in the system is kept constant [24,60]. Because operational *Textr* and *Pextr* control is critical in this method, the extraction cell is kept in an oven with different valves and restrictors [60]. Moreover, an extract cooler system, a back pressure regulator and a vial to collect the extract are required [24]. Finally, it is important to keep in mind that the equipment must be resistant to corrosion and high pressure [24].

**Figure 3.** Pressurized liquid extraction equipment and schematic representation of its operation. First, the solvent (S) and the sample (A) are injected into the extraction cell. The extraction cell is composed of an oven (O) and a pressure valve (P) which together allow the achievement of the temperature and pressure selected to extract the compound present in the sample. Then, the extracted compound is cooled and collected in a carousel.

PLE can be used in three modes of action: static mode, dynamic mode and a combination of the two. The static mode is characterized by the use of constant temperature and pressure values, resulting in the sample being in contact with the solvent for a set period. On the contrary, in the dynamic mode, the solvent (usually water) flows continuously through the sample. As a result of the higher volume of the extract obtained, the analytes are diluted in the liquid extract. Analytes are typically pre-concentrated by liquid–liquid extraction or by solid-phase extraction to address this issue. Finally, a combination of both modes of action can be used, which may improve analyte extraction [23].

#### *3.2. Sample Pre-Treatments*

Before using PLE, samples must be pre-treated to increase the contact surface between the solvent and the matrix during the extraction [23]. Pre-treatment can be compiled into four steps, as explained below and in Figure 4:


**Figure 4.** Pre-treatment steps, extraction techniques and target compounds from seaweeds using PLE and SWE. In purple, a schematic representation of the steps that should be followed to prepare the sample before the extraction technique is applied. In green, PLE and SPE operational conditions considering water and CO2 as solvents, respectively. In red, a comparison between the compounds extracted using each extraction technique.

#### *3.3. Relevant Parameters in PLE*

#### 3.3.1. Solvent

One of the most important parameters to optimize is the extraction solvent [22]. The solvent's function is to solubilize the target analytes while minimizing the extraction of other components [23]. Therefore, it is important to choose a solvent that has the same polar behavior of the target analytes [22]. Non-polar and water immiscible solvents or a combination of nonpolar with medium-polarity solvents are used for non-polar or lipophilic compound extraction. Consequently, solvents with high polarity are used to extract polar and hydrophilic compounds. Finally, when extracting analytes with different polarities, a mixture of solvents with high and low polarity is commonly used. Indeed, some authors suggest following two PLE extractions when the target is for high and low polar analytes, so they can be removed in two steps [23]. As a result, because affinity and miscibility are the two parameters used in predictive approaches to determine the solubility of the target compound in green solvents at different temperatures, experimental trials may be limited [24].

Regarding the application of PLE on seaweeds, using water as a solvent is the most common green technique applied for the carbohydrates extraction since they are more soluble in water at 100–150 ◦C and the dielectric constant of water is reduced at this temperature. Moreover, subcritical water acts as an acid or an alkali, helping the polysaccharides extraction [47]. A study conducted in 2022 in which PLE was optimized by varying different parameters, showed that temperature was the most critical for the extraction of carbohydrates in microalgae, which can be extrapolated to seaweeds. Furthermore, bioactive polysaccharides extracted from seaweeds using PLE with water are not degraded because temperatures are kept below 200 ◦C, avoiding caramelization and other degradation reactions [47]. Although water is a good green solvent, it is important to consider that its use may result in unwanted

reactions or interference coextraction, affecting the procedure's selectivity [17]. Therefore, other eco-friendly solvent alternatives, such as deep eutectic solvents, are being considered for carbohydrate extraction. Deep eutectic solvents (DES) are eutectic mixtures composed of hydrogen bonding acceptors (HBAs) and hydrogen bonding donors (HBDs) [61]. Due to their stability, cost-competitiveness, and ease of synthesis, DES have been proposed to dissolve different polysaccharides such as cellulose, starch, chitin and lignin for biomass processing. Moreover, DES was recently used as a solvent to extract fucoidans and alginates from brown algae. The results of this study showed that DES functioned as a catalyst, yielding twice as much as acidified water extraction. Table 2 summarizes the information explained above by showing different solvents used in PLE extraction, and their target analytes.

**Table 2.** Solvents used in PLE for the extraction of different compounds.


When selecting a solvent, it is also important to consider subsequent steps of the process, such as the clean-up step or concentration step if necessary. Selectivity is the parameter that determines whether or not purification and concentration are required, and it is critical when developing a green technique process [69]. Finally, the solvent used must be both physically and chemically stable. Water, ethanol, organic esters such as ethyl acetate and ethyl lactate, (+)-limonene and their mixtures are the most commonly used solvents in PLE [24].

The operational conditions used to extract bioactive compounds from seaweeds using PLE are shown in Table 3. In 2017, for example, a study on the accuracy of some green solvents with PLE for the fucoxanthin extraction was conducted. Limonene, ethyl lactate and ethyl acetate were selected as green solvents and their ability to extract fucoxanthin was compared to that of ethanol [69,70]. The highest yields were obtained for each solvent when the operating temperature was set to 100 ◦C. None of the green solvents reached ethanol's yield, with ethyl lactate had the highest percentage. Despite the yield results, limonene had the highest selectivity (expressed as the ratio of total carotenoids to total chlorophylls), proving that limonene is a good alternative green solvent for fucoxanthin extraction.

#### 3.3.2. Temperature and Pressure

As previously stated, *Textr* and *Pextr* are two important parameters to be optimized when using PLE. Elevated temperatures are used to reduce the viscosity of the liquid solvent used, allowing it to a better wet the matrix, and solubilize the analytes of interest. In addition, diffusion of analytes in the matrix surface is facilitated because high temperatures aid in the breakdown of the analyte–matrix bonds [23]. *Textr* varies between 50 and 200 ◦C and is dependent on the target analyte. Thus, lower *Textr* are selected for extraction of certain bioactive compounds due to their thermolability. Because high temperatures above the atmospheric boiling point are required in order to keep the solvent liquid, a high operational *Pextr* is required [23]. Furthermore, high pressure increases the extraction yield by forcing the solvent to enter the matrix pores [60]. In PLE methodology, *Pextr* usually varies from 5.0 to 15 MPa. These high-pressure and temperature conditions allow for the extraction of the target analytes in a short period of time while using less solvent and showing a recovering ability in terms of extraction yield similar to other techniques [22].



(sodium hydroxide).

#### 3.3.3. Time and Number of Cycles

The time of extraction is defined as the duration of direct solvent contact with the sample for a given *Textr* and *Pextr* [24]. This value is determined by a variety of factors, including the mode of action. When using static mode d, extraction time is reduced (*textr* = 5–20 min) [24]. On the contrary, when the dynamic mode is established, the flow of the solvent must be determined to select *textr*. Furthermore, it is significant to notice that low flows cause PLE system blockages while high flows result in diluted extracts. Finally, it is known that several cycles with low volume lead to higher yields of the target analyte, while a single extraction with a large amount of solvent does not correspond with higher extraction yields [24].

#### *3.4. Post-Extraction Treatment (Clean-Up)*

During PLE, some compounds of the matrix could be co-extracted causing interferences, so a clean-up step could be necessary to decrease the limit of detection (LOD) value [22]. Extraction and clean-up steps can be carried out simultaneously, which leads to a reduction in time and quantity of solvents used, between 15% and 52% [78]. Different clean-up techniques can be used:


#### **4. Combinatorial Approaches of PLE with New Extraction Methodologies**

As shown in Table 3, there have been few studies on the use of PLE in seaweeds due to its novelty. However, following the green technology trends, PLE could be combined with other methodologies to improve bioactive compound extraction efficiency and reduce solvent and time consumption. Moreover, because combinatorial approaches of PLE applied in the extraction of bioactive compounds of seaweeds have not been thoroughly studied, results of trials where PLE and other techniques are applied in other matrices are shown in this section and compiled in Figure 5 as case studies for future applications of these techniques combined.

#### *4.1. PLE Combined with SPE*

The combination of PLE and SPE has been used for the separation of specific phenolic compounds [79]. The mode of action of this combination is based on the ability of PLE to extract bioactive compounds from the matrix and the ability of SPE to purify the extracted compounds [80]. Thus, SPE is mainly used as a post-extraction technique since PLE is a non-selective methodology.

There are no data on the use of PLE and SPE for the extraction of phenolic compounds from seaweeds, but it was applied in other matrices such as apple pomace, mate leaves and lemon peel. Higher yields of total flavonoids were obtained in all the three matrices, when compared to the extraction using PLE alone [79–81]. On the contrary, when lemon peel was used as the matrix, the yields of the polar compounds were lower, while total phenolic acids and flavonoids showed no statistical difference between PLE and PLE combined with SPE in mate leaves [80,81].

**Figure 5.** Improvements achieved in the recovery of compounds from different matrices when PLE is combined with other extraction techniques.

#### *4.2. PLE Combined with UAE*

Some studies have combined the use of PLE with UAE (UAPLE) in different matrices, including seaweeds. A recent study combined PLE and UAE to extract phenolic compounds from three brown and one red algae. The operational conditions (solvent 80% MeOH:H2O (*v*/*v*); 10 mL; 130 ◦C; 130 bar; two static cycles of 10 min) were able to increase the release of phenolic compounds from the matrix due to the high and stable pressure [82]. Nevertheless, further research is needed to know if scale-up is available from the economic point of view [83], since many companies have difficulties because of the high expense with facilities, extraction time and ultrasound power [84].

### *4.3. PLE Combined with SFE-CO2*

The combination of PLE with SFE-CO2 is a sequential process based on the ability of SFE-CO2 for the extraction of the lipophilic fraction of the matrix, and the ability of PLE for the extraction of the antioxidant or high polarity compounds [85]. Because there is no information available about this sequential process used in seaweeds, results from other studies were compared to determine if this methodology could be useful for extracting seaweed compounds. For the recovery of bioactive compounds from rowanberry pomace using SPE-CO2 and PLE consecutively, results showed that this is an effective method for the isolation of carotenoid-rich and antioxidant-rich fractions [86]. Same conclusions were achieved when this sequential process was applied in *Viburnum opulus* pomace and berries [85]. Moreover, an economic evaluation of this process applied in passion fruit by-products was carried out in Brazil. This study showed that the combination of these techniques is economically applicable in large-scale production since it increases process productivity and decreases the cost of manufacturing [87].

#### *4.4. PLE Combined with EAE*

The application of EAE as an extraction technique leads to some disadvantages, such as the high cost of enzymes, the limitation of cell disruption because of the specificity of the enzyme and the inactivation of enzymes with parameters such as temperature and pH change [88]. To solve these limitations, EAE studies combined with other new methodologies have been performed. For example, when EAE was combined with alkaline hydrolysis and PLE for the extraction of bioactive compounds from *Sargassum muticum*, the extraction yields were higher than when PLE was used alone [89]. This could be explained by the formation of a protein–polyphenol complex which results in decreased polyphenol recovery. This complex may be formed when the enzyme disrupts the cell of the seaweed, releasing proteins and other compounds. Thus, these compounds may form complexes with polyphenols, resulting in aggregation and precipitation and ultimately, lower yields [89]. Because no additional information was discovered, more research is required to determine whether combining EAE and SPE could result in higher yields.

Given the current data on PLE combined with various new methods, a sequential process using PLE and SFE-CO2 should be considered because extraction yields are increasing, and an industrial scale appears to be feasible. However, because there has been no research on the application of a sequential process involving PLE in seaweeds and different parameters affecting the percentage of recovery, including the matrix, more research is required. Furthermore, because UAPLE produces intriguing results, scale-up trials should be conducted to determine whether this sequential process is economically feasible.

#### **5. Evaluation of Pressurized Liquid Extraction (PLE) Applications**

As mentioned in previous sections, using PLE to recover seaweed compounds results in the extraction of various bioactive compounds. Because of their technological function, importance as functional ingredients, or application in innovative food packaging systems, these compounds can be used in the food industry.

#### *5.1. Technological Function and Functional Ingredients*

Different compounds extracted from seaweeds are already used in the food industry for the technological improvement of food products, such as carbohydrates from seaweeds which are mainly used for its functional properties. Thus, agar is applied in the confectionary industry for its hydration maintenance capacity. Moreover, the addition of agar in meat products allows the reduction of the fat content in the final product [90]. The use of seaweed extracts in meat emulsions is interesting from a technological point of view since a harder and chewier structure with better water and fat binding is achieved [91]. For example, due to the increase in vegan and vegetarian diets, meat analogues are increasingly in demand and carrageenan is used because of their stability properties [90]: it is already used in low-fat sausages, beef burgers and beef patties as a thickener and stabilizer agent [92]. In addition, since algae have essential micronutrients such as Mg, K and Fe, and low Na content, the addition of these extracts into meat products may be a good opportunity to increase the nutritional value of these products [93]. Carrageenan and alginates have been added in bakery products such as bread, being able to reduce the moisture loss during storage and the dehydration rate of the crumb. Additionally, alginate was able to retard the hardening of the crumb [94]. Fruit jellies, donuts and cakes are also examples of products where agar is added [90]. At last, the use of alginates has an antimicrobial growth activity

in the vegetable industry and is a good choice for encapsulation and delivery systems of probiotics, according to the bibliography [90].

On the other hand, proteins, peptides and amino acids are mainly used as stabilizers, thickener agents, protein replacements and gelling agents [95]. Peptides extracted from different seaweeds were incorporated in pasta products, showing superior pasta quality and antioxidant properties over the control [95]. Furthermore, the addition of *Palmaria palmata* hydrolysate in bread improved texture and sensory acceptability. In addition, several studies of peptides extracted from seaweeds have been carried out, and Wakame peptide jelly and Nori peptide S are two bioactive peptides included in some Japan foods due to their antihypertension activity [4]. Nutritional supplements are also considered in the scope of the food application of seaweed extracts. As seaweeds are a good source of proteins, not only because of the quantity but for the quality, the use of this extract as supplementation would be a good option for those athletes following a vegan diet [96]. Thus, there are already products on the market, such as Solaray, that contain extracts of *Rhodymenia palmata*, which helps in the maintenance of the immune system health [77].

The incorporation of phlorotannin into food formulations may be limited because of their astringency and bitter taste. Thus, these compounds were encapsulated into nanofibers made of sodium alginate and polyethylene oxide and successfully incorporated into chicken breasts. The chicken was stored, and thanks to the phlorotannin's encapsulation, *Salmonella* growth was prevented, while the sensorial characteristics of the product were unaffected [97]. Moreover, the preservation ability against polyphenol oxidase activity and melanosis formation was proved during white shrimps' ice storage when phlorotannin extracted from *S. tenerimum* were added. Furthermore, when shrimps were immersed in 5% phlorotannin solution, shelf life was extended by 4 days and higher scores on overall sensory acceptability when compared to control [97]. Finally, different studies proved that the addition of seaweeds to the food formulation of different products leads to a reduction in cooking loss and in an improvement in the texture, as it is shown in Table 4 [98].

Finally, due to the current trend of changing the soy- and animal-derived protein sources for animal feed, seaweed extracts have been also incorporated into these products [95]. According to one study, incorporating algae extracts into dairy cattle feed resulted in higher I and Se content [99]. In fact, it has been demonstrated that including seaweed extracts in animal feed is a good way to achieve the I daily intake for those people with I deficiency, since milk excretion meets the needs of this mineral [95]. The incorporation of red seaweed extracts to poultry feed was also studied. The results show that incorporating *Sarchodiotheca guadichaudii* and *Chondrus crispus* extracts into layer feed improves the growth of beneficial bacteria and reduces *Clostridium perifringens* proliferation in the gut, thus improving the safety of the products obtained. Moreover, the egg yolk and weight were increased by adding 1% of *Sarchodiotheca guadichaudii* into the feed without altering the color of the yolk and the shell thickness [100]. Extracts from seaweeds are also being studied for feeding fish, especially protein extracts, since they are the most expensive dietary requirement for fish and shellfish aquaculture. Moreover, seaweed is also a good source of highly unsaturated fatty acids. Considering the requirements for fish nutrition, studies suggest that seaweed extracts may be a good option for fish feed, allowing the substitution of animal meal by plant meal in these products [95,101].

Considering the advantages of using PLE alone or in combination with other green techniques to obtain previously exposed seaweed bioactive compounds, and how food products may improve with the addition of these biomolecules, the incorporation of seaweed extracts may be a good strategy to enhance their nutritional profile and the technological properties.


**Table 4.** Application of seaweed extracts in food products.

#### *5.2. Application as Innovative Food Packaging*

Some compounds extracted from seaweeds such as laminarin, phlorotannin, flavonoids, terpenes, lactones and proteins are active against bacteria and fungi cells and bacteria biofilm formation (which is more difficult to remove) [116]. The correct preservation of organoleptic characteristics, while avoiding microbial growth during the storage as well as the need to extend the shelf life of the products, are critical for the food industry because these are factors deeply involved with the increase in food waste. Therefore, the use of antimicrobial compounds extracted from seaweeds may be a good option for increasing the shelf life of food products.

Sensory analyses of different products with some of these antimicrobial compounds were carried out to identify their impact on different parameters such as flavor, taste, color and smell. The results showed that edible film made of chitosan and seaweed extracts from *H. longata* and *P. palmata* inhibit the growth of mesophilic and psychrophilic microorganisms by maintaining the pH and water activity without affecting the sensorial characteristics of fish burgers. On the contrary, the sensory evaluation of pork patties with fucoidan and laminarian extract in a ratio of 0.5 *w*/*w* proved an adverse impact on the product. To avoid a possible negative impact on the organoleptic characteristics of the products, adding the antimicrobial compounds onto the packaging instead of adding them directly in the product could be an option. In fact, considering that most of the spoilage and contamination of food occurs on its surface, adding these compounds in the packaging may extend the shelf life of the product without affecting its organoleptic characteristics [116].

In terms of packaging, seaweed derivatives were studied to determine their suitability for bioplastic production. Bioplastics are synthetic plastics derived from biodegradable sources and their main disadvantage is their hygroscopicity [117], which affects the mechanical and storage properties required for food packaging. Furthermore, edible coatings are thin membranes composed of GRAS such as polysaccharides, lipids and proteins. Moreover, edible coatings can act as carriers of different bioactive compounds useful in food preservation. Thus, these special coatings maintain firmness, inhibit microbial growth and prevent food weight loss during long-term storage [50]. Carrageenan is a polysaccharide that can be used as an additive combined with other compounds for bioplastic production since it has low water vapor permeability (WVP) [117]. A biodegradable film made of olive extract, glycerol and 1% of carrageenan (*w*/*v*) showed good mechanical and antimicrobial properties. Moreover, a bionanocomposite made of 10% of gelatin (*w*/*v*), 0.5% (*w*/*v*) of k-carrageenan and 1, 3 and 5% of nano-SiO2 showed a drop in the WVP from 100% to 68%. Finally, the bioplastic production using starch, glycerol and 5% of carrageenan showed that the addition of carrageenan enhanced the moisture resistance, brittleness and the tensile properties of the polymer. Thus, carrageenan can be used to produce edible food packaging [117].

Alginates are another type of polysaccharides that could be used in the formulation of biodegradable films because their main properties are tensile strength, elongation and WVP that are suitable for biodegradable packaging [117]. To determine how mechanical properties were affected, alginate biofilms were compared using hydrophilic and hydrophobic plasticizers. Tributyl citrate (TC) showed better results because TC and alginate secondary interactions improved mechanical resistance. Furthermore, it was proved that elongation at break can be increased by using hydrophilic plasticizers such as glycerol [117]. A study using alginate with aloe vera (AV) and garlic oil (GO) in different proportions to produce a brand-new edible coating was performed to determine their UV shielding, thermal and antimicrobial properties. The results showed that after 16 days of storage, tomatoes with the edible coating made of 33.3% alginate, 66.7% AV and 5% GO showed an 8% mass loss while tomatoes without edible coating showed a 47% of mass loss. In addition, tomatoes with 33.3% alginate, 66.7% AV and 5% GO as edible coating suffered less damaged in the UV light measurement and showed better elongation break properties. For the inhibition growth of *Staphylococcus aureus*, *Escherichia coli* and *Syncephalastrum racemosum*, better results were obtained when tomatoes were coated with 33.3% alginate, 66.7% AV and 5% GO. On the contrary, better tensile strength results were obtained in tomatoes with 33.3% alginate and 67.3% AV [50].

Based on the data presented, the incorporation of alginates, carrageenan, laminarin, phlorotannin, flavonoids, terpenes and proteins onto innovative food packaging is a viable option for this innovative pathway.

#### **6. Advantages and Drawbacks of the Application of PLE as an Extraction Technique of the Bioactive Compounds from Seaweeds**

Although PLE is a better option for extracting compounds than traditional extraction techniques, the main disadvantage of this methodology is the high cost of the equipment. This is mainly due to the high requirements in terms of temperature and pressure, as well as the fact that the equipment must be made from materials that can withstand these conditions while avoiding corrosion [24,118].

Despite the high cost of the equipment, PLE has many advantages. On the one hand, one of the primary benefits of using PLE is that it has a lower environmental impact than other conventional extraction methodologies such as maceration and Soxhlet. This improvement is primarily due to a reduction in extraction time and a lower amount of solvent required. As previously stated, high temperatures allow for a decrease in solvent viscosity, resulting in faster solubilization and diffusion of the target compound [17]. Generally, extraction takes about 15 min [118]. This reduces the amount of energy required in the extraction process, making this methodology greener than other conventional methods. Solvents, particularly organic solvents, have traditionally been a problem in conventional extraction methods from an environmental standpoint. This is due to their classification as

Volatile Organic Compounds (VOCs). VOCs are organic pollutants that contribute to the photochemical smog formation in the troposphere and ozone depletion in the stratosphere as a source of radical sources [119]. Furthermore, given that the food industry accounts for 2% of total global solvent consumption [119], primarily for extraction processes, it is necessary to develop techniques that reduce the amount of solvent used. Moreover, biobased solvents are commonly used in this technique. These are defined as solvents derived from biomass and characterized for their biodegradability, lower VOC content and near-zero carbon balance. The use of these solvents is even more critical because it is known that a portion of the solvent used remains in food, food additive excipients and packaging [119]. In PLE, the mainly biobased solvents used are alcohol, ethanol, ethyl acetate, methyl lactate, ethyl lactate and D-limonene. Then, PLE is considered as one of the novel techniques in which solvent consumption is not only low, but it is also better in terms of biodegradability and toxicity.

#### **7. Conclusions and Future Perspectives**

Nowadays, the scientific community is increasingly interested in obtaining bioactive compounds from novel matrices using less aggressive environmental methodologies. As explained in this review, PLE is a green extraction technique that allows the separation of target active compounds such as polysaccharides, lipids, proteins and bioactive compounds using short time cycles and low quantities of solvents due to the high-pressure and temperature operating conditions. In this way, since edible seaweeds are becoming more important in the Western world due to both their nutritional profile and their technological properties, using PLE as an extraction technique appears to be a viable option, according to the bibliography. Moreover, data from the combination of PLE with other extraction techniques were evaluated to determine if they were appropriate, and satisfactory results were obtained. There are little data available on the combination of PLE with other extraction techniques using edible seaweeds, but comparable results are expected based on the results obtained with other matrices.

However, even though the data available today are primarily focused on the pharmaceutical industry and PLE has not been applied to edible seaweeds, this review attempts to provide an approximation of the PLE technique applied to seaweeds to generate knowledge that could potentially be applied in the food industry. Thus, compounds derived from edible seaweeds using PLE appear to be suitable for the bioplastic production and edible coating required in packaging; the synthesis of bio-nanocomposites that can be incorporated into food packaging to improve the mechanical properties of bioplastics; the incorporation into nutritional supplements; and the improvement of the nutritional profiles of different food products and animal feed. More experimental approaches of the PLE use for the extraction of seaweed compounds used in food products are required to determine whether this technique is appropriate for this matrix and the final products.

**Author Contributions:** Conceptualization, A.P.-V., M.C. and M.A.P.; methodology, A.P.-V. and P.B.; software, A.P.-V. and P.B., M.C. and L.C.; formal analysis, M.C. and L.C.; investigation, A.P.-V., M.C. and L.C.; resources, M.A.P. and J.S.-G.; writing—original draft preparation, A.P.-V., M.C., P.B. and L.C.; writing—review and editing, M.A.P. and J.S.-G.; visualization, A.P.-V., M.C. and M.A.P.; supervision, L.C., M.A.P. and J.S.-G.; project administration, M.C. and M.A.P.; funding acquisition, M.A.P. and J.S.-G. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors are grateful to the Ibero–American Program on Science and Technology (CYTED—AQUA-CIBUS, P317RT0003) and the Bio Based Industries Joint Undertaking (JU) under grant agreement No. 888003 UP4HEALTH Project (H2020-BBI-JTI-2019). The JU receives support from the European Union's Horizon 2020 research and innovation program and the Bio Based Industries Consortium. The project SYSTEMIC Knowledge Hub on Nutrition and Food Security has received funding from national research funding parties in Belgium (FWO), France (INRA), Germany (BLE), Italy (MIPAAF), Latvia (IZM), Norway (RCN), Portugal (FCT) and Spain (AEI) in a joint action of JPI HDHL, JPI-OCEANS and FACCE-JPI launched in 2019 under the ERA-NET ERA-HDHL (n◦ 696295). **Acknowledgments:** The research leading to these results was supported by MICINN supporting the Ramón y Cajal grant for M.A. Prieto (RYC-2017-22891); by Xunta de Galicia for supporting the program EXCELENCIA-ED431F 2020/12, the post-doctoral grant of L. Cassani (ED481B-2021/152), and the pre-doctoral grant of M. Carpena (ED481A 2021/313).

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

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Shedding Light on the Hidden Benefit of** *Porphyridium cruentum* **Culture**

**Davide Liberti 1, Paola Imbimbo 1,\*, Enrica Giustino 1, Luigi D'Elia 1, Mélanie Silva 2, Luísa Barreira <sup>2</sup> and Daria Maria Monti 1,\***

<sup>1</sup> Department of Chemical Sciences, University of Naples Federico II, Via Cinthia 4, 80126 Naples, Italy

<sup>2</sup> Centre of Marine Sciences, University of Algarve, 8005-139 Faro, Portugal

**\*** Correspondence: paola.imbimbo@unina.it (P.I.); mdmonti@unina.it (D.M.M.)

**Abstract:** Microalgae can represent a reliable source of natural compounds with different activities. Here, we evaluated the antioxidant and anti-inflammatory activity of sulfated exopolysaccharides (s-EPSs) and phycoerythrin (PE), two molecules naturally produced by the red marine microalga *Porphyridium cruentum* (CCALA415). *In vitro* and cell-based assays were performed to assess the biological activities of these compounds. The s-EPSs, owing to the presence of sulfate groups, showed biocompatibility on immortalized eukaryotic cell lines and a high antioxidant activity on cell-based systems. PE showed powerful antioxidant activity both *in vitro* and on cell-based systems, but purification is mandatory for its safe use. Finally, both molecules showed anti-inflammatory activity comparable to that of ibuprofen and helped tissue regeneration. Thus, the isolated molecules from microalgae represent an excellent source of antioxidants to be used in different fields.

**Keywords:** microalgae; exopolysaccharides; phycoerythrin; antioxidant activity; anti-inflammatory activity; biocompatibility; wound healing

### **1. Introduction**

Microalgae are ubiquitous eukaryotic photosynthetic microorganisms that are able to live in different environments, in single colonies, chains, or groups; depending on the species, their size can vary from a few to hundreds of micrometers [1–3]. The biodiversity of microalgae is mainly due to their unique ability to adapt and grow even under unfavorable growth conditions (e.g., extreme temperatures, variable salinity, and low or high light intensity) and to produce a wide range of interesting chemical compounds with novel structures and biological activities [4,5]. Among the microalgae, the red marine microalga *Porphyridium cruentum* could be pointed to as a commercial source of various high-value bioproducts [1], to be recovered from the same culture, in order to make the whole process economically feasible [6–10]. In particular, *P. cruentum* produces sulfated exopolysaccharides (s-EPSs) that are accumulated in a layer surrounding the cytoplasmic membrane. These exopolysaccharides act as a mucilage, because *P. cruentum* is without a well-defined cell wall [11]. They are composed of glucuronic acid and several major neutral monosaccharides, such as D- and L-Gal, D-Glc, D-Xyl, D-GlcA, and sulfate groups. S-EPSs from *P. cruentum* have antioxidant [12], immunomodulatory, anti-inflammatory, hypocholesterolemic, antimicrobial, and antiviral activity [13,14]. S-EPSs from *P. cruentum* also exhibit specific rheological properties that can be exploited in food applications [12,15]. In addition to exopolysaccharides, *P. cruentum* produces a broad range of colored pigments, including chlorophylls, carotenoids, and phycobilins, which are commercially utilized in the food, pharmaceutical, and cosmetic industries [16]. Amongst them, phycoerythrin (PE) is a light-harvesting protein with a structure of (αβ)6γ complex and a MW ranging from 240 to 260 kDa. Due to its unique biological properties, PE has gained much attention from the food and pharmaceutical industries and in the molecular biology field [17–22]. Here, starting from our recent results [10], a comprehensive study on the biological activities

**Citation:** Liberti, D.; Imbimbo, P.; Giustino, E.; D'Elia, L.; Silva, M.; Barreira, L.; Monti, D.M. Shedding Light on the Hidden Benefit of *Porphyridium cruentum* Culture. *Antioxidants* **2023**, *12*, 337. https:// doi.org/10.3390/antiox12020337

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 9 January 2023 Revised: 25 January 2023 Accepted: 26 January 2023 Published: 31 January 2023

**Copyright:** © 2023 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/).

of s-EPSs and purified phycoerythrin was carried out in order to verify if the extraction techniques could affect their biological activities.

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

#### *2.1. Reagents*

All solvents, reagents, and chemicals were purchased from Sigma-Aldrich (St Louis, MO, USA).

#### *2.2. Biocompounds Isolation*

S-EPSs and PE were isolated and purified from the culture of *Porphyridium cruentum* (CCALA415) as previously described [10]. Briefly, at the end of cell growth, the culture was centrifuged to recover s-EPSs in the supernatant. The s-EPSs were precipitated by adding pure ethanol (1:2 *v*/*v*) and centrifuging the sample (12,000× *g*, 30 min, and 4 ◦C). The supernatant was discarded, and the precipitate was freeze-dried. S-EPSs yield was 300 ± 67 g/L, which corresponds to 0.53 g/gd.w. biomass. In the case of PE, a crude aqueous extract was obtained *via* sonication (40% amplitude; 20 min, 30 s on and 30 s off) from the harvested biomass. PE was then isolated *via* a one-step purification procedure as reported by Liberti, up to a purity grade of 4 [10].

#### *2.3. Eukaryotic Cell Culture and Biocompatibility Assay*

Immortalized human keratinocytes (HaCaT, Innoprot, Derio, Spain) and immortalized murine fibroblasts Balb/c-3T3 (ATCC, Virginia, USA) were cultured in 10% foetal bovine serum in Dulbecco's modified Eagle's medium, in the presence of 1% penicillin/streptomycin and 2 mM L-glutamine, in a 5% CO2 humidified atmosphere at 37 ◦C. To verify the biocompatibility of the crude extract of s-EPSs and of purified PE, cells were seeded in 96-well plates at a density of 2 × 103/well and, 24 h after seeding, were incubated with increasing concentrations of the extract/compounds (5 to 75 μg/mL for EPS, 5 to 500 μg/mL of total proteins for crude extracts, and 5 nM to 100 nM for purified PE) for 72 h. At the end of the incubation period, cell viability was assessed with the MTT assay. Cell survival is expressed as the percentage of viable cells in the presence of compounds compared with control cells (represented by the average obtained between untreated cells and cells supplemented with the highest concentration of buffer).

#### *2.4. In Vitro Antioxidant Assays*

The antioxidant activity of the extract/compounds was tested by measuring their ability to scavenge the free radicals 1,1-diphenyl-2-picrylhydrazyl radical and 2,2- -azinobis- [3-ethylbenzthiazoline-6-sulfonic acid] (DPPH and ABTS, respectively) and to reduce or chelate redox active iron and copper (ferric-reducing antioxidant power (FRAP); ironchelating activity (ICA), and copper-chelating activity (CCA), respectively). DPPH and FRAP assays were performed following the procedure reported by Rodrigues et al. [23], and ascorbic acid and butylhydroxytoluene (BHT), respectively, were used as positive controls at the same concentrations of the sample under test. The ability of the extract/compounds to scavenge the ABTS radical was assessed as previously reported [24]. The results were compared to a calibration curve obtained using Trolox (6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid) as the standard. ICA and CCA were determined by measuring the formation of the Fe2+-ferrozine complex and by using pyrocatechol violet, respectively, according to the method reported by Megias [25]. EDTA was used as a standard at a final concentration of 100 μg/mL. S-EPS or purified PE was tested between 0.05 and 120 μg/mL and 0.2 and 270 nM, respectively. The results are expressed as IC50, i.e., the concentration required to scavenge 50% of the free radical or as the highest percentage achieved.

#### *2.5. Determination of Intracellular ROS Levels on Eukaryotic Cell Lines by DCFDA Assay*

The protective effect of s-EPSs (from 5 to 75 μg/mL) or purified PE (10 nM) against oxidative stress was measured by determining the intracellular reactive oxygen species (ROS) levels, following the protocol used by Imbimbo [26].

#### *2.6. Determination of Intracellular Glutathione Levels (DTNB Assay) and Lipid Peroxidation Levels (TBARS Assay) on Eukaryotic Cell Lines*

Intracellular GSH levels and lipid peroxidation levels were measured by following the procedure described by Petruk [27] using 12 μg/mL of s-EPSs or 10 nM of purified PE.

#### *2.7. Anti-Inflammatory Activity*

The anti-inflammatory activity of the compounds was tested by their ability to inhibit cyclooxygenase-2 (COX-2). S-EPS or purified PE was tested at different concentrations (4 and 167 μg/mL for s-EPSs or 10 and 27 nM for purified PE) using a commercial inhibitory screening assay kit, Cayman test kit-560131 (Cayman Chemical Company, Ann Arbor, MI, USA). Ibuprofen was used as a positive control. Results are expressed as a percentage of inhibition of COX-2.

#### *2.8. Wound Healing Assay*

Wound healing was assessed with a scratch assay. HaCaT cells were seeded at a cell density of 3 × 105 cells/cm<sup>2</sup> for 24 h, to allow cells to reach about 95% of confluence. Then, cells were washed with PBS, scratched manually with a 200 μL pipet tip, and incubated with 12 μg/mL of s-EPSs or 10 nM of purified PE. The scratch size was monitored at 0 h and 24 h by acquiring images using optical microscopy (Zeiss LSM 710, Zeiss, Germany) at 10× magnification. The width of the wound was measured by using Zen Lite 2.3 software (Zeiss, Germany). Results are expressed as a reduction of the area (fold) compared with untreated cells.

#### *2.9. Statistical Analyses*

All the experiments were performed in triplicate. Results are presented as the mean of results obtained after three independent experiments (mean ±SD) and compared by one-way ANOVA according to Bonferroni's method (post hoc) using GraphPad Prism for Windows, version 6.01 (Dotmatics, California, USA).

#### **3. Results**

*3.1. s-Exopolysaccharides Characterization*

3.1.1. s-EPSs Biocompatibility on Cell-Based Model

s-EPSs were tested for their biocompatibility on two eukaryotic immortalized cell lines: HaCaT (human keratinocytes) and Balb/c-3T3 (murine fibroblasts). Twenty-four hours after seeding, cells were incubated with increasing amounts of s-EPSs (from 5 to 75 μg/mL). After 72 h of incubation, cell viability was assessed by the MTT assay; cell survival is expressed as the percentage of viable cells in the presence of s-EPSs compared with that of control samples (i.e., untreated cells). The results in Figure 1 show that, under all the experimental conditions, the s-EPSs were fully biocompatible on both the cell lines analyzed.

**Figure 1.** Effect of s-EPSs from *P. cruentum* on cell viability. Dose–response curves of HaCaT (black dots) and Balb/c-3T3 (empty squares) cells after 72 h of incubation with increasing concentrations of exopolysaccharides (5–75 μg/mL). Cell viability is reported as a function of s-EPS concentration.

#### 3.1.2. s-EPSs *In Vitro* Antioxidant Activity

The antioxidant activity of s-EPS was evaluated with different *in vitro* analyses: ABTS, DPPH, FRAP, and iron and copper chelating assays. As shown in Table 1, s-EPSs were not able to scavenge the ABTS and DPPH radicals, whereas a slight but significant activity was observed for the chelation of iron and for ferric ion reduction assays. Both tests are based on the ability to act on iron: the former measures the ability of the compounds under test to bind Fe2+, whereas the latter analyzes the ability to reduce Fe3+ to Fe2+. As for the copper chelating assay, the highest activity reached, at the highest concentration tested, was 9 ± 3%, a value much lower than the one obtained by testing the positive control molecule at the same concentration.

**Table 1.** *In vitro* antioxidant and chelating activity of s-EPSs. Results are expressed as percentage of inhibition. The concentration evaluated is referred to the final concentration of s-EPSs or positive control used in the well.


#### 3.1.3. s-EPSs Antioxidant Activity on a Cell-Based Model

The antioxidant activity of s-EPSs was also evaluated on HaCaT cells. For this purpose, cells were incubated with increasing concentrations of s-EPSs (from 5 to 50 μg/mL) for 2 h, and then oxidative stress was induced by UVA irradiation (100 J/cm2). Immediately after irradiation, the intracellular ROS levels were measured by using H2DCFDA as a probe. For each set of experiments, untreated cells were used as a control. As shown in Figure 2, UVA treatment significantly increased the DCF fluorescence (black bars, *p* <0.001). In the absence of stress, s-EPSs induced a slight but significant increase in the intracellular ROS level (Figure 2, white, dashed grey, and dark grey bars on the left part of the graph). Interestingly, when cells were preincubated with s-EPSs prior to being stressed, only 5 and 12 μg/mL were able to protect the cells from ROS formation (Figure 2, light grey and white bars on the right part of the graph), whereas the higher concentrations had no protective effect. This result is in agreement with those of Giordano et al. [28], as antioxidants act at low

concentrations, whereas, at high concentrations, they may work as pro-oxidants. Based on these results, s-EPSs were used at 12 μg/mL for further experiments.

**Figure 2.** Antioxidant activity of s-EPSs on UVA-stressed HaCaT cells. Intracellular ROS levels were determined with DCFDA assay. Cells were preincubated in the presence of increasing amounts (from 5 to 50 μg/mL) of s-EPSs for 2 h prior to UVA irradiation (100 J/cm2). Results are expressed as percentages compared with untreated cells. Black bars refer to untreated cells; light grey bars refer to cells incubated with 5 μg/mL of s-EPSs; white bars refer to cells incubated with 12 μg/mL; dashed bars refer to cells incubated with 25 μg/mL; dark grey bars refer to cells incubated with 50 μg/mL of s-EPSs in the absence (−) or presence (+) of UVA stress. Data shown are means ± S.D. of three independent experiments. \* indicates *p* < 0.05, \*\* indicates *p* < 0.01, and \*\*\*\* indicates *p* < 0.001.

To deeply analyze the protective effect of s-EPSs, the intracellular glutathione levels and lipid peroxidation levels were determined with DTNB and TBARS assays, respectively. In the absence of any treatment, a significant decrease (*p* < 0.01) in GSH levels was observed after UVA exposure (Figure 3A), and s-EPSs (grey bars) were able to inhibit GSH oxidation, thus confirming a protective effect against oxidative stress. As for the TBARS assay, a significant increase (*p* < 0.05) in lipid peroxidation levels was observed after UVA treatment (black bars, Figure 3B), but, notably, this effect was inhibited upon pretreatment with s-EPSs (grey bars). Treatment of the cells with exopolysaccharides did not significantly alter either glutathione or lipid peroxidation levels in the absence of UVA treatment (−). Taken together, the results clearly indicate that s-EPSs are able to protect cells from oxidative damage.

**Figure 3.** Protective effect of s-EPSs on HaCaT cells. Intracellular GSH levels were determined with a DTNB assay (**A**) and lipid peroxidation levels were determined with a TBARS assay (**B**). Cells were preincubated in the presence of 12 μg/mL of s-EPSs for 2 h prior to UVA irradiation (100 J/cm2). GSH and lipid peroxidation levels were measured 90 min after UVA irradiation. Black bars refer to untreated cells, and grey bars refer to cells incubated with s-EPSs, in the absence (−) or in the presence (+) of UVA stress. Values are expressed as percentages compared with untreated cells. Data shown are means ± S.D. of three independent experiments. \* Indicates *p* < 0.05; \*\* indicates *p* < 0.01.

#### 3.1.4. *In Vitro* Anti-Inflammatory Activity of s-EPSs

As inflammation is a condition strictly linked to oxidative stress, the anti-inflammatory activity of s-EPSs was measured by evaluating their capacity to inhibit the enzyme COX-2. When inflammation occurs, COX-2 is able to enhance the prostanoid production [29]. As reported in Table 2, surprisingly, s-EPSs showed no significant differences compared with ibuprofen used as positive control when tested at the same concentration, thus suggesting a new role of s-EPSs in inflammation control.

**Table 2.** *In vitro* s-EPSs anti-inflammatory activity.


#### *3.2. Phycoerythrin Characterization*

#### 3.2.1. Phycoerythrin Biocompatibility on Immortalized Eukaryotic Cells

Following biomass lysis, phycoerythrin (PE) had a purity grade of 1.5 [10]. This value is considered as reagent-grade, thus indicating that the protein can be used as it is for food applications [30]. In order to verify the safety of the protein on eukaryotic cells, an MTT assay was performed by comparing the crude extract with the purified protein (purity grade of four). The results of the MTT assay, reported in Figure 4, clearly show that only pure PE was fully biocompatible with both cell lines (Figure 4B), while the crude extract exerted a dose-dependent toxicity (Figure 4A). These results clearly indicate that PE needs to be purified to a higher purity grade before being used on cell-based models, or, at least, that it cannot be applied when present in the extract at concentrations higher than a certain threshold (100 μg/mL).

**Figure 4.** Biocompatibility of total extract (**A**) and purified PE (**B**) on eukaryotic cells. Dose–response curves of HaCaT (black dots) and Balb/c-3T3 cells (empty squares) after 72 h of incubation with increasing concentrations of total extract (**A**) and purified PE (**B**). Cell viability was assessed with an MTT assay and is reported as a function of extract/protein concentration.

#### 3.2.2. *In Vitro* Antioxidant Activity

*In vitro* analysis of the antioxidant activity of purified PE was carried out with the abovementioned experimental procedures. As reported in Table 3, purified PE was not able to scavenge the DPPH radical or chelate copper ions. However, it demonstrated a high capacity to scavenge the ABTS radical ion and to reduce ferric iron or chelate iron with considerably low IC50 values (0.072 ± 0.004 and 0.084 ± 0.012 μM, 0.084 ± 0.004 μM, respectively). Noteworthy, the purified PE IC50 values were about 160, 1000, and 600 times lower than the IC50 values obtained with the positive control molecules (Trolox, 12 ± 1 μM in the ABTS; BHT, 90 ± 4 μM in the FRAP; and EDTA, 51 ± 3 μM in the ICA).


**Table 3.** *In vitro* antioxidant and chelating activity of purified PE. Results are expressed as IC50 values, μM.

3.2.3. Cell-Based Antioxidant Activity of PE

Starting from the encouraging results obtained *in vitro*, purified PE was tested on the UVA-stressed HaCaT experimental system used for s-EPSs. Cells were treated with 2.5 μg/mL (10 nM) of purified PE for 2 h, and then oxidative stress was induced by UVA irradiation (100 J/cm2). At the end of irradiation, the intracellular ROS levels were evaluated. As shown in Figure 5, UVA induced a significant increase in intracellular ROS levels (black bars, 200%) compared with untreated cells (*p* < 0.001). When cells were treated with purified PE (grey bars), no increase in intracellular ROS levels was observed. Interestingly, when cells were incubated with purified PE prior to UVA exposure, an inhibition of the intracellular ROS production was observed.

**Figure 5.** Protective effect of purified PE on UVA-stressed HaCaT cells. Intracellular ROS levels were determined with DCFDA assay. Cells were preincubated in the presence of 10 nM of purified PE (grey bars) for 2 h prior to UVA irradiation (100 J/cm2). Black bars refer to untreated cells in the absence (−) or in the presence (+) of UVA stress. Values are expressed as percentages compared with untreated cells. Data shown are means ± S.D. of three independent experiments. \*\*\* indicates *p* < 0.005; \*\*\*\* indicates *p* < 0.001 with respect to UVA-treated cells.

The effect of purified PE on GSH and lipid peroxidation was also assessed. As shown in Figure 6, PE was able to fully protect cells from oxidative stress, as no alteration in either the GSH levels (Figure 6A) or in the lipid peroxidation levels (Figure 6B) was found when the cells were pretreated with purified PE prior to stress, thus confirming the protective effect of the protein against oxidative stress.

**Figure 6.** Analysis of intracellular GSH and lipid peroxidation levels on HaCaT cells. Cells were preincubated with 10 nM of purified PE for 2 h before UVA irradiation (100 J/cm2). (**A**) determination of intracellular GSH levels; (**B**) analysis of lipid peroxidation levels. In both experiments, measurements were recorded 90 min after UVA-induced stress. Values are expressed as a percentage compared with control (i.e., untreated) cells. Data shown are means ± S.D. of three independent experiments. \* indicates *p* < 0.05, \*\* indicates *p* < 0.01, and \*\*\* indicates *p* < 0.005.

#### 3.2.4. *In Vitro* PE Anti-Inflammatory Activity

Purified PE was also able to inhibit COX-2 (Table 4) by about 75%, although the level of inhibition attained with ibuprofen 24 nM could not be achieved at any of the analyzed concentrations.

**Table 4.** COX-2 inhibition by purified PE.


#### *3.3. Effect of s-EPSs and Purified PE on Wound Healing*

Finally, a scratch assay was carried out on HaCaT cells to test the ability of s-EPSs and purified PE to induce cell migration related to wound repairing. The results are reported in Figure 7 and in Table 5. In the absence of any treatment, the cells spontaneously migrated to induce the re-epithelialization. Interestingly, when the cells were treated with either s-EPSs or purified PE, a significant enhancement in the wound closure was observed after 24 h. Indeed, s-EPSs reduced the scratched area by 2.5 ± 0.17-fold and purified PE by 2.4 ± 0.1317-fold compared with untreated cells (1.80 ± 0.02-fold reduction).

**Figure 7.** Effect of s-EPSs and purified PE on wound healing. Confluent HaCaT cells were scratched and treated with either 12 μg/mL s-EPSs or 10 nM purified PE for 24 h. Optical microscopy images were acquired at 10× magnification at the beginning (t0) and end (24 h) of the incubation.

**Table 5.** Reduction of area (fold) of wound closure upon 24 h of incubation with either s-EPSs or purified PE. Data shown are means ± S.D. of three independent experiments. For each experiment, at least 10 images were acquired. \* indicates *p* < 0.05.


\* = *p <* 0.05.

#### **4. Discussion**

As the use of synthetic molecules is known to be harmful in the long run, the search for new natural compounds endowed with beneficial properties is urgent [31]. In this context, antioxidants from microalgae could represent an excellent alternative, but the costs of microalgae upstream and downstream processes are still too high [32].

We recently set up a cascade approach to recover four classes of molecules from *P. cruentum* culture: s-EPSs, PE, carotenoids, and saturated fatty acids. Among them, here, we evaluated the biological activity of s-EPSs and PE. S-EPSs were chosen as it is generally thought that polysaccharides with a high sulfated content have biological activities [33], such as antioxidant action [34]. It is known that antioxidant molecules can bind metal ions, forming metal–ion complexes. The presence of sulfate groups could increase the metal-binding capacity of the carbohydrates by donating an electron pair or by losing a proton, thus stabilizing the complex [35,36].

In agreement with the findings of Wang et al., we found that s-EPSs had no radical scavenging activity against DPPH, whereas they showed antioxidant activity in the ABTS assay, with IC50 values ranging from 6.59 to 8.92 mg/mL [37]. Interestingly, despite the low antioxidant activity observed *in vitro*, s-EPSs were active on a cell-based system at a concentration almost 600 times lower than that measured *in vitro*. This result is in agreement with literature, as it is well known that *in vitro* assays should not be compared with cellbased ones. Indeed, antioxidants provide their function by different mechanisms of action, so that bioavailability, stability, retention, or reactivity of the compound under test in a complex system, such as that of eukaryotic cells, cannot be either mimicked or evaluated *in vitro* [38]. Our results indicated that s-EPSs were able not only to inhibit the intracellular ROS production but also to prevent GSH depletion and lipid peroxidation.

Different is the case of PE, which was found to be a very powerful antioxidant agent *in vitro* and on a cell-based system. The ABTS assay was in line with that observed by Sonani on a PE from a different source (IC50 of 72 ± 4 nM vs. 101 nM, respectively) [39], whereas the PE prepared by this author had lower DPPH-scavenging (IC50 of 930 nM) and iron-chelating abilities (IC50 of 484 nM) than the purified PE prepared in this study. We hypothesize that the higher antioxidant activity measured in our experimental system may rely on the source or strain used. A different source may also affect the biocompatibility results: indeed, we found that only pure PE was biocompatible with eukaryotic cells, strongly suggesting the importance of purification of the protein for all the potential applications. Pure PE protected cells from UVA irradiation at a concentration in the low nanomolar range (10 nM). Generally, antioxidants prevent the generation of free radicals, which can significantly affect some physiological processes, including wound healing. In particular, ROS generation can damage tissues and slow down the regeneration process. The presence of antioxidants should counteract chronic inflammation and at the same time contribute to promoting tissue regeneration [40]. Considering that both s-EPSs and PE were able to inhibit one of the key enzymes in the inflammation process (COX-2) and to induce a significantly faster scratch closure compared with untreated cells, we can conclude that the bioproducts obtained by *P. cruentum* represent an excellent ingredient for new biomaterials, such as medical patches.

#### **5. Conclusions**

In this study, s-EPSs and PE, obtained from *P. cruentum* culture by a cascade approach described in a previous work [10], showed a remarkable antioxidant activity in a cell-based system, higher than that obtained by *in vitro* assays, thus suggesting that the reliability of *in vitro* assays has to be overhauled. Moreover, both molecules showed anti-inflammatory characteristics comparable with ibuprofen and a significant ability to promote cell proliferation.

**Author Contributions:** D.L., P.I. and D.M.M. designed the concept and supervised the experiments. D.L., P.I., E.G. and L.D. performed the experimental work with microalgae. D.L., P.I., E.G. and D.M.M. wrote the manuscript. L.B. supervised M.S. on the in vitro experiments. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was carried out with funding from the Portuguese Foundation for Science and Technology (FCT) through UIDB/04326/2020, UIDP/0436/2020, LA/P/0101/2020.

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

**Informed Consent Statement:** Not applicable.

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

**Acknowledgments:** P.I. would like to acknowledge the ALGAE4IBD project (From Natura to Bedside-Algae Based Bio Compound for Prevention) funded by the European Union's Horizon 2020 Research and Innovation program under grant agreement No. 101000501.

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

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **A Deadly Liaison between Oxidative Injury and p53 Drives Methyl-Gallate-Induced Autophagy and Apoptosis in HCT116 Colon Cancer Cells**

**Antonietta Notaro 1,†, Marianna Lauricella 2,†, Diana Di Liberto 2, Sonia Emanuele 2, Michela Giuliano 1, Alessandro Attanzio 1, Luisa Tesoriere 1, Daniela Carlisi 2, Mario Allegra 1, Anna De Blasio 1, Giuseppe Calvaruso <sup>1</sup> and Antonella D'Anneo 1,\***


**Abstract:** Methyl gallate (MG), which is a gallotannin widely found in plants, is a polyphenol used in traditional Chinese phytotherapy to alleviate several cancer symptoms. Our studies provided evidence that MG is capable of reducing the viability of HCT116 colon cancer cells, while it was found to be ineffective on differentiated Caco-2 cells, which is a model of polarized colon cells. In the first phase of treatment, MG promoted both early ROS generation and endoplasmic reticulum (ER) stress, sustained by elevated PERK, Grp78 and CHOP expression levels, as well as an upregulation in intracellular calcium content. Such events were accompanied by an autophagic process (16–24 h), where prolonging the time (48 h) of MG exposure led to cellular homeostasis collapse and apoptotic cell death with DNA fragmentation and p53 and γH2Ax activation. Our data demonstrated that a crucial role in the MG-induced mechanism is played by p53. Its level, which increased precociously (4 h) in MG-treated cells, was tightly intertwined with oxidative injury. Indeed, the addition of N-acetylcysteine (NAC), which is a ROS scavenger, counteracted the p53 increase, as well as the MG effect on cell viability. Moreover, MG promoted p53 accumulation into the nucleus and its inhibition by pifithrin-α (PFT-α), which is a negative modulator of p53 transcriptional activity, enhanced autophagy, increased the LC3-II level and inhibited apoptotic cell death. These findings provide new clues to the potential action of MG as a possible anti-tumor phytomolecule for colon cancer treatment.

**Keywords:** oxidative stress; phytocompounds; methyl gallate; autophagy; apoptosis; p53

#### **1. Introduction**

Nowadays, the identification of non-toxic drugs for normal cells capable of selectively targeting tumor systems represents one of the main challenges in the development of innovative and tailored therapies for cancer [1].

Cancer represents a global problem with a continued growing expansion, seriously affecting public health. For many decades, multimodal approaches based on surgery, radiation therapy and chemotherapy have been used. In recent years, a revolution in the development of tumor-targeting drugs has been based on the knowledge of neoplastic entities, which are hallmarks that have substantially improved the types of combinatorial strategies, thus opening the way for precision cancer medicine [2,3].

To achieve this goal, growing interest toward plant-derived chemicals has pushed cancer scientists to search for new unexplored molecules as preventative or anti-tumor

**Citation:** Notaro, A.; Lauricella, M.; Di Liberto, D.; Emanuele, S.; Giuliano, M.; Attanzio, A.; Tesoriere, L.; Carlisi, D.; Allegra, M.; De Blasio, A.; et al. A Deadly Liaison between Oxidative Injury and p53 Drives Methyl-Gallate-Induced Autophagy and Apoptosis in HCT116 Colon Cancer Cells. *Antioxidants* **2023**, *12*, 1292. https://doi.org/10.3390/ antiox12061292

Academic Editor: Stanley Omaye

Received: 19 May 2023 Revised: 9 June 2023 Accepted: 13 June 2023 Published: 16 June 2023

**Copyright:** © 2023 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/).

compounds. Theoretically, each plant can represent a significant reservoir of bioactive compounds, such as secondary metabolites, phytonutrients, nutraceuticals and supplements [4], that can be obtained from the vegetal world, tested in a laboratory, and applied to ameliorate human health or fight chronic diseases. Plants have developed the ability to synthetize secondary metabolites through highly controlled biochemical pathways that act in response to environmental insults related to abiotic or biotic stresses. In general, these metabolites can be classified as phenolic, terpenoid and alkaloid compounds [5], and due to their multifaceted activities, have stimulated increasing interest from industries for their application as dietary supplements, biocides, pharmaceuticals or potential medicinal drugs.

Due to their intrinsic properties based on the action on multiple targets, limited side effects and excellent efficacies, plant-derived natural products have been applied as healing agents to treat a very wide range of human diseases [6]. A lot of related examples stem from natural drugs used for the treatment of different human malignancies. The main features of these compounds, such as terpenoids (artesunate, atractylodes, andrographolide, etc.), phenols (resveratrol, quercetin, curcumin, capsaicin, etc.) and alkaloids (piperine, berberine, matrine, etc.), rely on their ability to inhibit tumor proliferation and angiogenesis, trigger apoptosis and modulate immune responses [7,8]. A compound that is widely spread in chemotherapeutic regimens is paclitaxel, which is a diterpenoid molecule originally isolated from *Taxus brevifolia* [9,10]. It represents a classical natural phytochemical that has attracted the attention of researchers for its therapeutical ability as an anti-neoplastic agent in the treatment of lung, breast and ovary cancers [10–12]. An analog successful chemotherapeutic agent is campthothecin, which is a bioactive compound isolated from *Camptotheca acuminate* [13], that has shown significant efficacy against many different solid tumors when used alone or in combination treatment with cisplatin [14].

In light of these considerations, our research has recently focused on the chemical characterization of plant and fruit extracts or essential oils to discover anticancer phytomolecules to apply alone or as adjuvants to conventional therapies [15–22]. Their identification could also offer the chance to chemically modify plant-derived molecules through the introduction of active pharmacophores, creating novel and powerful lead compounds [23]. In particular, our interest was recently directed toward phytochemicals that are widely found in *Mangifera indica* L., where the peel fraction of the fruit is rich in many bioactive compounds, such as mangiferin; citric acid; quinic acid; digallic acids; gallic acid; and its esters, such as methyl gallate and pentagalloyl glucose [15]. We demonstrated that methyl gallate and pentagalloyl glucose were the most common phytocostituents and were found to be particularly effective in reducing the cell viability of three different colon cancer cell lines [16].

Methyl gallate (MG) is a natural methyl ester of gallic acid and is endowed with many different biological activities ranging from anti-inflammatory to antioxidant and anti-microbial properties [24–27]. MG was demonstrated to harbor a clear selective antineoplastic action in many tumor systems. It plays a crucial inhibitory role in the tumor infiltration of CD4+ CD25+ regulatory T cells and its administration was demonstrated to delay tumor progression and survival in an EL-4 lymphoma model [28]. Other studies provided substantial evidence that this phytomolecule exerts anti-tumor activity on glioma cells, inhibiting proliferation and migratory cell ability via the suppression of the ERK1/2, Akt and paxillin phosphorylation signaling pathways [29]. Huang et al. reported that MG can also exhibit antitumor potential in different HCC cells (Hep3B, Mahlavu and HepJ5) triggering ROS-mediated and caspase-dependent apoptotic cell death [30]. Additionally, MG was also shown to exert a tumor inhibitory effect in in vitro and in vivo mouse models of hepatocellular carcinoma. Furthermore, it did not exert any cytotoxic effect in human normal hepatocytes, while it significantly suppressed the migration, invasion and epithelial– mesenchymal transition in tumor systems of HCC via the AMPK/NF-κB pathway [31].

In light of these observations, and since the anti-neoplastic potential of MG has not been described in colon cancer systems, we explored its possible impact on this tumor. Our investigations provided evidence that this phytochemical starkly reduces colon cancer

cell viability, sparing differentiated Caco-2 cells, which provide a model of polarized cells resembling enterocytes [32]. In addition, MG triggered both autophagy and apoptotic cell demise via intertwined crosstalk between oxidative stress and p53 activation as prime sources of the phytochemical action.

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

#### *2.1. Cell Cultures and Chemicals*

The Caco-2 and HCT116 colon cancer cells used in this study were obtained from Interlab Cell Line Collection (ICLC, Genoa, Italy) and cultured as monolayers in DMEM supplemented with 10% (*v*/*v*) heat-inactivated FCS and 2 mM glutamine and in the presence of a 1% penicillin/streptomycin solution. To obtain differentiated Caco-2 cells as enterocyte-like cells [33], Caco-2 cells were plated and cultured in a complete medium for 21 days as reported by Natoli et al. [34].

For the reported experiments, cells were seeded in a culture medium on 96-well microplates or 6-well plates as previously reported [16] and allowed to adhere overnight at 37 ◦C in a humidified atmosphere containing 5% CO2 followed by treatment with MG or a vehicle only. MG stock solution was prepared in DMSO and stored at −20 ◦C according to vendor specifications. In each experiment, MG working solutions were prepared in DMEM, never exceeding 0.01% (*v*/*v*) DMSO. The vehicle condition reported in each experiment as control was represented by untreated cells incubated in the presence of the corresponding DMSO volume. All cell culture media and culture reagents were provided from Euroclone SpA (Pero, Italy). All other reagents and chemicals, except where differently indicated, were purchased from Millipore Sigma (Milan, Italy).

#### *2.2. Cell Viability Assay*

Cell viability was assessed using an MTT assay as previously reported [35]. Cells were plated in 96-well plates, and after 24 h, they were incubated with compounds for indicated periods. Since the measure of cell viability using an MTT assay is based on the reduction of MTT to formazan and many polyphenols may interfere with formazan production, we considered this aspect in our experimental conditions. In particular, after incubation with the compounds, the plate was centrifuged and the medium was withdrawn and replaced with a fresh one before proceeding with the assay. Afterward, 20 μL of 5 mg/mL MTT was added to each well and the plate was incubated at 37 ◦C for 2 h. Therefore, the media was removed from each well and replaced with 100 μL lysis buffer (20% SDS and 10% dimethylformamide) before reading at 450 nm. For the determination of IC50 values nonlinear regression analysis with the equation of a sigmoidal dose response with a variable slope was performed using Graphpad Prism 7.0 software (San Diego, CA, USA).

The observation of cellular morphological changes was detected using a Leica DMR inverted microscope (Leica Microsystems, Wetzlar, Germany), while the pictures were taken using IM50 Leica software (Leica Microsystems, Wetzlar, Germany).

The cytotoxic action of MG was also evaluated using the LDH (lactate dehydrogenase) assay, which is a method based on the measure of the activity of a stable cytoplasmic enzyme commonly released upon cell damage. Cells were seeded in a 6-well plate at a density of <sup>2</sup> × 105 cells, and after incubation with the compound, they were collected and centrifuged at 120× *g* for 10 min. The supernatant medium of each sample was recovered and analyzed using ARCHITECT Lactate Dehydrogenase kit (Abbot Laboratories Diagnostics Division, IL, USA) according to vendor specifications. To detect the total LDH release, treated cells were compared with a positive control represented by cells incubated in the presence of 0.1% Triton 100×.

#### *2.3. Colony Formation Assay*

This assay measures cell proliferation in a cell-contact-independent way. Cells were plated in pre-tested appropriate densities yielding 500 cells per plate. The plates were cultured for 10 days in the presence or absence of different doses of MG. Then, the colony signals were measured after crystal violet staining as previously reported [15]. The clonogenic survival fraction was defined as the ratio of the signal intensity of the untreated group versus the MG-treated group. All assays were made in triplicate. The number of colonies for each experimental condition was determined using the "Colony Area" plugin for the open-source image analysis software ImageJ v 1.8.0 as reported by Guzman et al. [36].

#### *2.4. ROS Measurement*

To assess the intracellular generation of reactive oxygen species (ROS), cells were plated in 96-well plates and allowed to adhere overnight. Cells were treated with MG and incubation with 5-(and-6)-carboxy-2- ,7- -dichlorodihydrofluorescein diacetate (H2DCFDA) fluorochrome (Molecular Probe; Thermo Fisher Scientific, Inc., Life Technologies Italia, Monza, Italy) was performed as previously reported [37]. For this purpose, stock solutions of H2DCFDA were dissolved in DMSO and aliquots were stored at −20 ◦C until use. H2DCFDA working solution was prepared in a PBS solution containing 5 mM glucose and added to cells to the final concentration of 20 μM. Then, incubation was protracted for 30 min in the dark and ROS-positive cells were visualized using a fluorescein isothiocyanate (FITC) filter (excitation wavelength of 485 nm and emission wavelength of 530 nm) in a Leica inverted fluorescence microscope (Leica Microsystems S.r.l, Wetzlar, Germany) equipped with a DC300F camera. All pictures were captured using Leica Q Fluoro software (Leica Microsystems S.r.l, Wetzlar, Germany).

#### *2.5. Measurement of Intracellular Calcium Levels*

The intracellular calcium levels were assayed using the Ca2+-sensitive fluorescent dye Fluo 3-AM following vendor instructions (Thermo Fisher Scientific, Ferentino, Italy). After incubation with the compounds, cells (2 × <sup>10</sup>5/well) were collected, washed in calciumfree PBS and incubated in the presence of Fluo 3-AM for 1 h at 37 ◦C in the dark. Then, calcium generation was analyzed using flow cytometry on a FACSAria Cell Sorter (BD Biosciences Company, 283 Franklin Lakes, NJ, USA). At least 50,000 cells were analyzed for each experimental condition. The data analysis was then performed using FlowJo software workspace v10 (BD Biosciences).

#### *2.6. Analysis of Autophagic Vacuoles*

The generation of autophagic vacuoles was detected using monodansylcadaverine (MDC) staining according to Munafò et al.'s procedure [38]. Briefly, following to the treatment in the presence of MG, the medium was replaced and 50 mM MDC in PBS was added to cells. Then, cells were washed in PBS and the fluorescence was analyzed with a Leica fluorescence microscope (Leica Microsystems, Wetzlar, Germany) using a 4- ,6-diamidino-2-phenylindole dihydrochloride (DAPI) filter (excitation wavelength of 372 nm and emission wavelength of 456 nm). The analysis of autophagic vacuoles was also performed using acridine orange (AO) staining that specifically detects acidic vesicular organelles (AVOs) producing a bright red fluorescence, whereas it generates a bright green fluorescence for cytoplasm and nucleus [39]. For these analyses, cells were incubated for 15 min with 1 μg/mL AO prepared in PBS. Then, cells were analyzed under a Leica fluorescence microscope equipped with an image system (Leica Microsystems, Wetzlar, Germany) using Rhodamine (excitation wavelength of 596 nm and emission wavelength of 620 nm) and FITC (excitation wavelength of 485 nm and emission wavelength of 530 nm) filters. Merged images were obtained by combining pictures of both channels using Leica Q Fluoro software (Leica Microsystems, Wetzlar, Germany).

#### *2.7. Immunoblot Analyses*

Protein analysis was performed via a Western blotting procedure as previously reported [40]. For these analyses, 30 μg protein/lane were resolved using SDS-PAGE and then electroblotted on a nitrocellulose membrane filter (Bio-Rad Laboratories Srl, Segrate, Italy). All primary antibodies were purchased from Santa Cruz Biotechnology Inc. (Santa

Cruz, CA, USA), except for Protein kinase R-like endoplasmic reticulum kinase (PERK), phospho-PERK, eukaryotic initiation Factor 2α (eiF2α) and anti-caspase-3, which were from Cell Signaling Technology (Cell Signaling Technology Inc., Beverly, MA, USA). Anti-rabbit IgG (H + L) HRP conjugate and anti-mouse IgG (H + L) HRP conjugate (dilution 1:10,000) secondary antibodies were from Promega (Milan, Italy). In all experiments performed, γ-tubulin (diluted 1:1000, Sigma-Aldrich, Milan, Italy) was used as the loading control.

For all analyses, protein band detection was performed with an ECL™ Prime Western Blotting System (Cytiva, Merck KGaA, Milan, Italy) using a ChemiDoc XRS System equipped with the Quantity One software 4.6.6 (Bio-Rad Laboratories, Inc., Hercules, CA, USA).

#### *2.8. Analysis of Apoptotic Cell Death Using Hoechst and Annexin V/PI Staining*

To detect apoptotic cell death, cells were pre-incubated with Hoechst 33342 (Invitrogen; Thermo Fisher Scientific, Inc.) for 30 min before treatment with compounds. Next, blue nuclei showing condensed or fragmented chromatin were analyzed using fluorescence microscopy (Leica Microsystems, Wetzlar, Germany) as reported [15]. The quantification of apoptotic cell death percentage was determined via flow cytometry analysis using an Allophycocyanin (APC) Annexin V conjugate and propidium iodide (Annexin V-APC/PI) staining. For these experiments, HCT116 cells (2 × 105/2 mL medium) were seeded into 6-well plates and then subjected to treatments with MG. At the end of the treatment, the cells were taken via trypsinization, centrifuged at 120× *g* for 10 min, resuspended in PBS and counted. Next, 10<sup>5</sup> cells were incubated with Annexin V-APC (BD Pharmingen™ APC Annexin V kit, BD Biosciences, Milan, Italy) and PI (Sigma-Aldrich) in the dark according to the manufacturers' instructions. At the end of the incubation, the samples were analyzed using a FACSAria Cell Sorter flow cytometer (BD Biosciences Company, 283 Franklin Lakes, NJ, USA), acquiring at least 50,000 cells for each sample analyzed. The data obtained were then examined with FlowJo software (BD Biosciences).

#### *2.9. Preparation of Cytosolic and Nuclear Extracts*

For the isolation of nuclear and cytosolic fractions, 2 × <sup>10</sup><sup>6</sup> HCT116 cells were plated in 100 mm cell culture dishes and after incubation with MG, lysates were prepared as previously reported [41]. Briefly, cells were washed in PBS and scraped in a lysis solution containing 250 mM sucrose, 20 mM HEPES, 10 mM KCl, 1.5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM DTT and protease inhibitor cocktail, pH 7.4. Then, the homogenates were prepared by passing cells through a needle of 25 g on ice for 20 min (10 times) and centrifuging samples at 1000× *g* for 10 min at 4 ◦C. The pellets were recovered and resuspended in lysis solution, and homogenization was repeated by passing cells through a needle of 25 g for an additional 10 times. Therefore, samples were recentrifuged (1000× *g* for 10 min at 4 ◦C) and pellets representing the nuclear fraction were resuspended in RIPA buffer (1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, inhibitors of proteases: 25 μg/mL aprotinin, 1 mM PMSF, 25 μg/mL leupeptin and 0.2 mM sodium pyrophosphate) before proceeding with sonication. The supernatants obtained at the first centrifugation were recentrifuged (10,000× *g* for 30 min at 4 ◦C) and the supernatants obtained were used as the cytosolic fraction. Proteins from both fractions (nuclear and cytosolic) were quantified using a Bradford assay (Bio-Rad Laboratories, Inc.) and were resolved in a polyacrylamide gel to analyze the p53 cellular localization. To determine the purity of each cellular fraction obtained, GADPH and Lamin B were used as cytoplasmic and nuclear markers, respectively.

#### *2.10. Statistical Analyses*

The statistical analysis of the data was performed by using GraphPad PrismTM 7.0 software (Graph PadPrismTM Software Inc., San Diego, CA, USA) and data were reported as the mean ± S.E. The significant differences between the control (untreated) vs. treated samples were analyzed by applying Student's *t*-test, while the analysis of multiple groups of samples was conducted using the ANOVA test. The statistical significance threshold was considered to be *p* < 0.05.

#### **3. Results**

#### *3.1. MG Affected Colon Cancer Cell Viability in a Dose-Dependent Manner*

Studies on the cytotoxic effects of MG demonstrated that this phytochemical elicits remarkable cell viability inhibition in many tumor systems [29,31]. However, since no data are available for colon cancer, we undertook a study aimed at evaluating the MG antitumor potential on the viability of two colon cancer cell lines (HCT116 and Caco-2). Our results showed that the cytotoxic effects were visible after a lag phase of 24 h (not shown) and were clearly evident at 48 h. As can be observed from the response curves (Figure 1A) obtained using incremental doses of MG, the half-maximal inhibitory concentration of MG (IC50 value) was about 30 μg/mL in both cell lines after 48 h of incubation. A more consistent effect was observed in the presence of the 90 μg/mL dose, which caused a dramatic decrease in cell viability (about −80%). Differently, in comparison to the corresponding colon cancer Caco-2 cells, no cytotoxic effects were found when MG was administered to differentiated Caco-2 cells, which is a well-established model of polarized intestinal cells reproducing typical morphological and biochemical features of enterocytes [42].

The ability of MG in reducing colon cancer cell viability was also confirmed using both the lactate dehydrogenase (LDH) test and clonogenic assay, which is an in vitro survival test that estimates cell ability to maintain a reproductive potential over a prolonged period. As reported in panel B of Figure 1, MG enhanced the LDH release in HCT116 cells relative to the control. In addition, we also observed that the phytochemical reduced the colonyforming ability of colon cancer cells with doses spacing from 0.46 to 7.5 μg/mL range, while no colonies were found with higher doses. Based on cell viability tests, all further experiments were performed with HCT116 cells using those MG doses that caused cell reductions of about 50% (30 μg/mL) and 80% (90 μg/mL), respectively.

#### *3.2. MG Cytotoxicity Was Mediated by Oxidative Injury, ER Stress and Upregulation of Intracellular Calcium*

To clarify the underlying mechanism of MG cytotoxicity, we explored whether the observed cytotoxic effect could be ascribed to the induction of oxidative stress. For this purpose, we used NAC, which is a potent radical scavenger.

As reported in Figure 2A, when pre-incubating the cells in the presence of NAC for 2 h, the toxic effect of 90 μg/mL MG was consistently counteracted. Moreover, all morphological changes induced by MG, consisting of cell shrinkage and a reduction in cell number (Figure 2B), were also counteracted by NAC sustaining the induction of oxidative damage.

Such observations were confirmed by an evaluation of the ROS production assayed using H2DCFDA staining. Following MG exposure at different times, a dose-dependent ROS generation (green fluorescent cells) was observed. The effect, that was already visible after 2 h, reached a peak at 4 h to maintain lower levels for longer times of incubation in the presence of MG (Figure 2C).

Furthermore, we verified whether these effects were accompanied with the upregulation of stress-associated proteins at 24 h and 48 h. As it can be observed in Figure 3, the manganese superoxide dismutase (MnSOD) and catalase levels, which are two radical scavenger enzymes, increased at 24 h with the two doses of MG. Such an effect was counteracted by the pre-incubation of cells in the presence of NAC (Figure 3). No changes in the levels of these proteins were found at 48 h of treatment.

**Figure 1.** IC50 determination and colony formation assay of MG-treated colon cancer cells. (**A**) MTT assay of colon cancer cells (HCT116 and Caco-2) and differentiated Caco-2 cells incubated with incremental doses of MG. Cell viability was determined at 48 h as reported in the Materials and Methods section. IC50 values were assessed using GraphPad Prism 7 software. (**B**) LDH cytotoxicity test in MG-treated HCT116 cells. After incubation in the presence or absence of the phytocompound, cells were centrifuged and supernatants were used to assess the LDH content using a commercial kit. Data are reported as a percentage of the total LDH released from cells using as a positive control of cells incubated with 0.1% Triton 100×. (**C**) Clonogenic assay in MG-treated HCT116 cells. The colony formation inhibition was assessed by crystal violet staining after 10 days of exposure to MG. Representative images of colony formation are reported in the upper panel. The quantitative analysis of colonies (lower panel) was performed as reported in the Materials and Methods section. All experiments were performed in triplicate. (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with the untreated sample.

**Figure 2.** Oxidative stress is required for the cytotoxic efficacy of MG in colon cancer cells. (**A**) Effect of NAC pre-incubation on HCT116 cell viability of MG-treated cells for 48 h. Each value reported in the histogram represents the mean of three independent experiments ± SD. (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with the untreated sample. (#) *p* < 0.05 compared with the MG-treated sample. (**B**) Phase-contrast micrographs of morphological changes of HCT116 cells treated for 48 h with MG and the protective effect of NAC pre-incubation (original magnification 200×). (**C**) ROS generation induced by MG treatment in HCT116 cells. The ROS level was measured using H2-DCFDA, which is a redox-sensitive fluorescent probe, as reported in the Materials and Methods section. Original pictures were taken using a Leica fluorescence microscope equipped with a CCD camera and FITC filter (original magnification 200×).

**Figure 3.** Upregulation of the antioxidant enzymatic systems in HCT116 cells treated with MG. After treatment with the indicated doses of the phytocompound in the presence or absence of NAC, the cell lysates were prepared and the level of stress-associated proteins (MnSOD and catalase) was detected using Western blotting. Representative blots from three independent experiments were considered and a densitometry analysis histogram was normalized to γ-tubulin, which was used as a loading control. (\*\*) *p* < 0.01 and (\*\*\*) *p* < 0.001 compared with the untreated sample. (#) *p* < 0.05, and (##) *p* < 0.01 compared with the MG-treated condition.

In addition, we also investigated whether the oxidative injury triggered by MG exposure could be also associated with ER stress. With this in mind, we conducted Western blotting analyses to explore the status of key factors involved in ER stress. As reported in Figure 4, the higher dose of MG provoked a modest increase in the level of PERK, phospho-PERK and eiF2α. Such an effect was also accompanied at 24 and 48 h with an upregulation of Glucose-Regulated Protein 78 (Grp78), which is an ER chaperone acting as a key regulator of the unfolded protein response (UPR) [43], as well as that of C/EBP Homologous Protein (CHOP).

As it is well known, another event that can contribute to oxidative injury and lead to cell death is a calcium surge released from different cellular compartments, such the ER and mitochondria [44–46]. On the other hand, the calcium homeostasis that lies at the heart of many cell signaling processes is under redox control.

In accordance with these observations, flow cytometry analyses using a Fluo 3-AM probe provided evidence that the MG provoked a remarkable increase in the intracellular calcium content (Figure 5). Interestingly, this event, which had already occurred at 16–24 h of exposure to MG when cells were found to be still alive, also remained high in treated conditions up to 48 h, a time at which cell death took place, in correlation with the high levels of both Grp78 and CHOP.

**Figure 4.** MG exposure activated the ER-stress-associated protein levels. HCT116 cells were treated in the presence of 30 and 90 μg/mL MG for 24 h and 48 h. Western blot analysis was performed to evaluate the protein expression of the ER stress markers PERK, phospho-PERK, eiF2α, Grp78 and CHOP. The amount of analyzed proteins was assessed using γ-tubulin as the loading control protein and for band density normalization. The data are presented as the mean ± SD; (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with the untreated sample.

**Figure 5.** MG exposure increased the intracellular calcium level. Changes in the content of intracellular calcium analyzed at the indicated times via flow cytometry using Fluo 3-AM fluorochrome as reported in the Materials and Methods section.

#### *3.3. Autophagy Was Upregulated in MG-Treated Cells*

A growing number of studies indicated that many natural compounds, such as alkaloids, flavonoids, naphthoquinones, sequiterpene lactones and ginsenosides possess

an anti-cancer potential acting as autophagy modulators [47]. On the basis of these observations, to further dissect the underlying mechanism of MG, we determined whether this phytochemical might act through the induction of autophagy [48]. Moreover, calcium release from ER storages can also contribute to the generation of autophagosomes [49], which are typical spherical structures endowed with double-layer membranes participating in the autophagic process. Using the monodansylcadaverine (MDC)-based staining, which is a selective fluorescent probe that accumulates in autophagosomes, we observed that MG promoted the generation of autophagic vacuoles, which appeared as dot-like structures.

The event, which was already visible at 16 h of incubation, further increased at 24 h, when MDC-positive cells amounted to almost 90% with a 90 μg/mL dose (Figure 6A). Differently, no MDC-stained structures were highlighted at 48 h of treatment.

The induction of autophagy was also sustained by the conversion of microtubuleassociated protein 1A/1B-light chain 3 known as LC3-I to LC-3II, which is the phosphatidylethanolamine conjugated form that is recruited to autophagosomal membranes [50,51] and represents a crucial marker of the autophagic flux. An increasing trend of some autophagy-associated factors, such as p62, Beclin 1 and Atg7, was also observed in MGtreated conditions, while no changes were noticed for Atg1/Ulk1 (Figure 6B).

#### *3.4. MG Treatment Induced DNA Damage and p53-Mediated Apoptotic cell Death*

To evaluate whether the cytotoxic effect observed in the presence of MG could be ascribed to the induction of apoptotic cell death, we tested possible chromatin condensation and fragmentation using vital Hoechst staining. While MG administration triggered autophagy in the first phase of treatment, data reported in Figure 7A showed that when prolonging the exposure up to 48 h, remarkable nuclear modifications associated with cell death occurred in the presence of the phytocompound.

When exploring possible changes in DNA damage markers, we provided evidence that the phytochemical produced a consistent increase in γH2AX, as well as p53 (Figure 7B). Meanwhile, the pre-incubation with NAC counteracted MG effects on both chromatin condensation and γH2AX and p53 activation.

To elucidate the type of cell death induced by MG, we performed flow cytometry analyses using Annexin V/PI double staining (Figure 8A). At 48 h, about 19% of early (Q3 quadrant) and late (Q2 quadrant) apoptotic cells were found using 30 μg/mL of the compound. Such a value amounted to 60% apoptotic cells when the higher dose of MG was employed. We then checked the status of specific apoptotic markers, such as caspase-3 and its target PARP1. Specifically, we found that MG promoted a decrease in the pro-enzymatic form of caspase-3 and the appearance of the fragmented and activated forms at 19 and 17 kDa, respectively. Such an effect was accompanied with the fragmentation of PARP1, which is a caspase-3 target (Figure 8B).

#### *3.5. MG Treatment Induced Early Upregulation of p53 Related to the Molecular Switch between Autophagy and Apoptotic Cell Death*

In light of these results, we wondered about the possible role of p53 and γH2AX in the mechanism analyzed. Time course studies (4–24 h) provided evidence that both doses of MG promoted an early upregulation of these factors, which were already visible after 4 h of incubation with the compound (Figure 9A). As it is known, p53 protein is activated in key responses to genotoxic stresses and DNA damage. In these scenarios, p53 translocates to the nucleus, boosting a tumor suppressor program via the cell cycle arrest or apoptosis via the direct transcriptional activation of specific pro-apoptotic targets, such as Apaf-1, Puma, Bax and Noxa [52,53]. By performing subcellular fractionation experiments, we found p53 accumulation in the nuclear fraction already at 16 h of incubation with MG (Figure 9B).

**Figure 6.** MG exposure provoked an autophagic process in the early phases of treatment. (**A**) Monodansylcadaverine (MDC) staining that enabled the visualization of autophagic vacuoles as dot-like structures was performed in MG-treated cells. HCT116 cells were incubated with MG for the indicated time periods and autophagic vacuoles were highlighted via fluorescence microscopy using a Leica microscope equipped with a DAPI filter. Representative fluorescence microscopy images were taken at a magnification of 400×, as reported in Section 2. (**B**) Immunoblots of autophagic markers performed in MG-treated HCT116 cells. Proteins were detected using different antibodies directed against the LC3-I and LC3-II forms, p62, Beclin 1, Atg1/Ulk1 and Atg7. γ-tubulin was used as the loading control. All graphs show the mean ± SD of three independent experiments. (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with the untreated cells.

**Figure 7.** MG-induced DNA damage. (**A**) Morphological analysis of HCT116 cells after vital staining with Hoechst 33342. Following Hoechst staining, cells were treated for 48 h with different doses of MG in the presence or absence of NAC. Chromatin fragmentation and condensation were observed under a fluorescence microscope. The images (original magnification at 200×) were acquired with a DAPI filter using an inverted fluorescence microscope and processed with Leica Q Fluoro Software. (**B**) Analysis of DNA damage markers: γH2AX and p53. After treatment with MG in the presence or absence of NAC for 48 h, cells were lysed and proteins were analyzed using Western blotting. The γ-tubulin blot was reported as a loading control. The blots and histograms of densitometric analyses reported are representative of three independent experiments. (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with untreated cells. (#) *p* < 0.05 compared with the MG-treated condition.

To further dissect the underlying role of p53 since its increase was already observed in the early phases of treatment (4–16 h) when oxidative injury and the autophagy process occurred, we explored a possible interplay between p53 and MG-induced autophagy. Compelling evidence showed that p53 can positively or negatively impact autophagy in a context-dependent manner or via its subcellular localization [54–57].

By using acridine orange (AO) staining, which corroborated the accumulation of acidic vesicular organelles (AVO, orange fluorescence) in the cytoplasm of MG-treated conditions, we observed that the addition of antioxidant NAC counteracted the autophagy, while Bafylomicin A1, which is an inhibitor of autophagosome-lysosome fusion, markedly inhibited the autophagic process (Figure 10A). On the other hand, pre-incubating the cells with pifithrin-α (PFT-α), which is a specific inhibitor of p53 transcriptional activity, the AVO accumulation induced by MG further increased (Figure 10A). Such an effect was also accompanied by an upregulation of the LC3-II form (Figure 10B), thus suggesting that p53 can negatively affect autophagy in response to MG.

**Figure 8.** MG-induced apoptosis in colon cancer cells. (**A**) Annexin V/PI staining to evaluate apoptosis. HCT116 cells were treated with MG and compared with untreated cells. The rate of apoptosis was assessed via flow cytometry using the Annexin V/PI double staining assay. The data represent one of three independent experiments. (**B**) MG treatment evoked an increase in apoptotic markers. Caspase-3 activation and PARP1 fragmentation were analyzed using Western blotting. The relative quantification was assessed after densitometric analysis of bands and normalization to γ-tubulin used as a loading control. Histograms of densitometric analyses report the average values of three independent experiments. (\*) *p* < 0.05, (\*\*\*) *p* < 0.001 and (\*\*\*\*) *p*< 0.0001 compared with the untreated sample. ns, not significant.

Accordingly, when we pre-incubated HCT116 cells for 2 h in the presence of PFT-α, followed by a co-treatment with MG for 48 h, PTF-α played a protective role against the cytotoxicity showed by the phytocompound alone. As shown in Figure 11A, the residual viability that amounted to only 22% with 90 μg/mL MG rose to about 60% when PFT-α was added.

Such prevention of MG cytotoxic effects was also found when PFT-α was co-administered and morphological changes were evaluated using light microscopy (Figure 11B).

To further ascertain whether PFT-α can inhibit MG-induced apoptosis, we explored its effect on both p53 and caspase-3. The results showed that compared with the control group, PFT-α counteracted p53 upregulation and, in the same experimental conditions, also suppressed caspase-3 activation, inhibiting the production of its active fragments (19–17 kDa) induced by the MG exposure (Figure 11C).

310

**Figure 9.** Time course analysis of the MG action on DNA damage markers and p53 nuclear accumulation. (**A**) MG treatment provoked an early upregulation of DNA damage markers p53 and γH2Ax in HCT116 cells. After treatment with MG for various periods, p53 and γH2Ax were detected using Western blotting analyses. Data were normalized to γ-tubulin, which was used as the loading control. The blots and histograms of densitometric analyses reported are representative of three independent experiments. (\*) *p* < 0.05, (\*\*) *p* < 0.01 and (\*\*\*) *p* < 0.001 compared with the untreated sample. (**B**) Subcellular fractionation for cytosolic and nuclear protein extract displayed nuclear enrichment in the p53 content after the MG incubation. The relative quantification was assessed after densitometric analysis of the bands and normalization to the correspondent loading control. Lamin B or GADPH was used to assess the possible changes in the loaded protein amount for the cytosolic and nuclear fractions, respectively. (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with the untreated sample.

**Figure 10.** Effects of p53 inhibition on MG-induced autophagy. (**A**) The inhibition of p53 transcriptional activity by PFT-α enhanced the MG-induced autophagy in HCT116 cells. Cells were pre-incubated with NAC (5 mM), BafA1 (100 nM) or PFT-α (20 μM) in the presence or absence of MG for 24 h; then, the production of AVOs showing bright red fluorescence was evaluated via AO staining using Leica Q Fluoro software. (**B**) PFT-α/MG co-treatment enhanced the levels of the autophagic protein LC3. For the analysis of the LC3 forms, cells were treated with MG in the presence or absence of PFT-α (20 μM), followed by a Western blot analysis. The relative quantification was assessed after a densitometric analysis of bands and normalization to γ-tubulin. Data reported in the histograms were the average of three independent experiments. (\*\*) *p* < 0.01 compared with the untreated sample. (#) *p* < 0.05 compared with the MG-treated sample.

**Figure 11.** PFT-α, which is a p53 inhibitor, negatively affected the MG-induced apoptosis in HCT116 cells. (**A**) PFT-α inhibited the MG-induced cytotoxic effect. The pre-incubation of cells with PFT-α was performed as reported in the Results section; then, different MG concentrations were added and incubation was protracted for 48 h. Cell viability was analyzed using an MTT assay. Values reported in the line chart represent the mean of three independent experiments ± SD; (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with the untreated conditions. (#) *p* < 0.05 and (##) *p* < 0.01 compared with the MG-treated sample. (**B**) Micrographs showing the PFT-α effect on morphological changes induced by the MG treatment. Pictures were taken using a Leica inverted microscope as reported in the Materials and Methods section. (**C**) Effect of PFT-α on the p53 and caspase-3 levels. Protein lysates were prepared as reported in the Materials and Methods section and resolved using SDS-PAGE. Blots were detected using specific antibodies directed against the proteins of interest and their level was normalized to γ-tubulin, which was used as the loading control. The blots and histograms of densitometric analyses reported are representative of three independent experiments. (\*) *p* < 0.05 and (\*\*) *p* < 0.01 compared with the untreated sample; (#) *p* < 0.05 and (##) *p* < 0.01 compared with the MG-treated sample.

#### **4. Discussion**

Moderate levels of ROS are crucial regulators of signaling pathways relevant for life in normal cells, while their increased generation can seriously impair cellular redox balance, contributing to many pathologic conditions, such as cancer. Many different studies highlighted that tumor cells, compared with their normal counterpart, exhibit an increased oxidative burst that favors cancer transformation, metabolic remodeling and increased generation of ROS [58]. Indeed, an escalated production of ROS positively impacts cancer through pro-tumorigenic signaling that enhances cell survival, DNA damage, genetic instability, hypoxia adaptation and resistance to the most common chemotherapeutics [59]. In this intricate scenario, malignant cells thrive in and counterbalance the ROS overload by boosting a plethora of enzyme-based scavenger systems to detoxify themselves with an oxidative burst and maintain their pro-tumorigenic profile.

However, despite these aspects, the role of ROS in cancer and anti-cancer strategies is still widely debated. A growing body of evidence supports the view that ROS conceal an oncojanus nature, either activating pro-tumorigenic or anti-tumorigenic signaling. Such pathways can be differently orchestrated in cancer treatment to preferentially kill tumor cells [58,60]. The fine-tuning of some anti-neoplastic therapies that promote an escalated ROS generation that overwhelms the scavenging tumor ability seems to be the other side of the coin that could be exploited as an Achilles' heel and drive tumors to different death pathways, such as apoptosis, autophagy or necroptosis [61].

Based on these rationales, some chemotherapeutics as natural or synthetic compounds (i.e., platinum-based compounds, anthracyclines, taxanes and sesquiterpenoids) have been extensively used with this purpose to treat tumors [62,63].

In particular, the investigations developed in this study focused on methyl gallate (MG), which is a well-known phytochemical that harbors strong anti-neoplastic properties in many different tumor systems, but no data are available on colorectal cancer. Here, we provide evidence for the first time that MG inhibits the growth of colon cancer cells, sparing differentiated Caco-2 cells, which is a model of polarized enterocytes. In particular, the cytotoxic action of MG could be ascribed to a deadly liaison occurring between ROS and p53 that, unavoidably, dictates the cell fate toward apoptosis. In the first phase of treatment, MG stimulated a stress-associated program characterized by the precocious increase in ROS content, along with an upsurge in both ER stress markers (PERK, phospho-PERK, Grp78 and CHOP) and the intracellular calcium level. As a consequence of this oxidative burst, probably to serve a stress defense response, MG-treated cells also upregulated the level of the ROS scavenger enzymes MnSOD and catalase.

In this complex scenario, we do not know the origin of intracellular calcium increase at this time. As it is well known, changes in intracellular calcium content that increase beyond the normal threshold can be ascribed to its release by the ER or dysfunctional mitochondria. Since some of our preliminary studies also suggested an involvement of mitochondria in MG-induced mechanism, we cannot exclude this possibility and aim to better clarify such an aspect in our future directions.

Overall, the early ROS generation drove cells in the first phases of treatment (16–24 h) along an autophagy process, as testified by the appearance of autophagic vacuoles to MDC and AO staining and significant changes in autophagic markers, such as LC3, p62, Beclin 1 and Atg7. However, when analyzing the timeline of MG exposure up to 48 h, such an autophagic flux was interrupted, leading cells to an apoptotic demise characterized by DNA fragmentation and caspase activation. These findings were in accordance with Huang's data [30], demonstrating that the extensive oxidative injury induced by MG treatment is a causative event in the apoptosis triggered by MG in HCC cells. Indeed, in our experimental condition, the addition of NAC, which is an antioxidant sulfidryl compound, prevented the toxic effect of MG in colon cancer cells, inhibited the autophagic process and counteracted the DNA fragmentation occurring during the apoptotic cell death. On the other hand, the observation that oxidative stress could play a role in the analyzed events was shown

by data that reported a complex interconnection between ROS production, ER stress and autophagy [64,65].

Interestingly, our results also indicated that a pivotal role in the mode of action of MG was played by the tumor suppressor protein p53. p53 represents a key player that is capable of monitoring a plethora of cellular pathways related to the control of cell cycle progression, genome stability and apoptosis [66]. For all these multifunction properties, it has been named the "guardian of the genome". Accumulating evidence demonstrated that under normal conditions, p53 is maintained at very low levels by its negative regulator MDM2 that targets p53 degradation to 26S proteasome [67]. Differently, when cells are under stress conditions (nutrient deprivation, hypoxia, DNA damage), the p53 level increases and the protein translocates to the nucleus, where, as a transcription factor, it regulates the expression of a subset of genes functioning in cell cycle progression, cell metabolism, autophagy, tumor microenvironment and apoptosis. ROS and p53 were shown to establish a versatile partnership [68]. Indeed, ROS generated as by-products of cellular metabolism can act either upstream of p53, promoting its expression, or downstream, triggering apoptotic cell death pathways. In our experimental conditions, MG stimulated p53 upregulation, and such an effect seemed to be strictly intertwined with ROS generation. Its increased level, which was already visible in the first hours of exposure to MG (4–8 h), was maintained at a high level up to 48 h, when DNA damage occurred and cells collapsed via apoptosis. On the other hand, the addition of NAC counteracted the increase in both p53 and γH2Ax DNA damage markers, demonstrating that their upregulation can be ascribed to the impairment of redox balance induced by the phytocompound. Clearly, we believe that p53 was crucial in the context of the mechanism studied when monitoring both autophagy and apoptotic cell demise. Such an observation was supported by the experiments performed using PFT-α, which is a specific inhibitor of p53 transcriptional activity. Indeed, when PFT-α was co-administered with MG, we observed an enhancement of autophagy flux, as well as in LC3-II form, showing that p53 could inhibit autophagy in our condition. The relationship between p53 and autophagy has been widely discussed and still appears to be controversial since p53 can act as a rheostat system that adjusts the authophagy rate (through a positive or negative regulation) in a context-dependent fashion [56].

On the other hand, p53 also represents an active player in MG-induced apoptosis in colon cancer cells. Indeed, when exposing cells to a PFT-α/MG combo treatment, we observed that the p53 level dropped and the cytotoxic action of MG, as well as caspase-3 activation, were prevented by PFT-α.

A schematic representation describing the intricate mechanism of MG anticancer activity on colon cancer cells is reported in Figure 12.

**Figure 12.** Timeline of the antitumor signaling pathway activated by MG in colon cancer cells. MG exposure triggered autophagy and apoptotic cell demise. The represented processes are orchestrated by an intertwined liaison between oxidative injury and p53. The early generation of ROS, accompanied with intracellular calcium increase and ER stress, stimulated p53 to switch the autophagy toward apoptotic cell death.

#### **5. Conclusions**

As a whole, the findings reported in this paper fit well with the current scientific literature on the anti-tumor potential of methyl gallate. In particular, our results shed light on the potential action of MG in preferentially targeting colon cancer cells with respect to enterocyte-like cell models. The biochemical characterization of MG signaling in these cells reveals a new clue to its underlying mechanism, thus highlighting an intertwined relationship between oxidative stress and p53 as a causative event in apoptotic demise. The data obtained here represent a reason to use MG in future investigations and explore whether this phytochemical can be used alone or in combinatorial studies as a possible adjuvant in traditional chemotherapy.

**Author Contributions:** Conceptualization, A.D. and M.L.; methodology, A.N., D.D.L. and A.A.; investigation, D.C. and S.E.; data curation, A.D.B. and S.E.; software, D.C., A.N. and D.D.L.; writing original draft preparation, A.D. and M.L.; writing—review and editing, M.L., L.T., M.G., M.A. and A.D.; supervision, all authors; funding acquisition, G.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was partially sustained by Finalized Research Funding (FFR 2022), FFR-D15 D'Anneo and FFR-D03-Lauricella, Università degli Studi di Palermo, Palermo, Italy.

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data reported in this paper are available on request from the corresponding author.

**Acknowledgments:** We thank Giovanni Perconti for providing the PERK, phospho-PERK and eiF2α antibodies and Roberto Chiarelli for the LC3 and p62 antibodies.

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

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Evaluation of Proanthocyanidins from Kiwi Leaves (***Actinidia chinensis***) against Caco-2 Cells Oxidative Stress through Nrf2-ARE Signaling Pathway**

**Ji-Min Lv 1, Mostafa Gouda 1,2,\*, Xing-Qian Ye 1, Zhi-Peng Shao <sup>3</sup> and Jian-Chu Chen 1,\***


**Abstract:** Proanthocyanidins (PAs) are considered to be effective natural byproduct and bioactive antioxidants. However, few studies have focused on their mode of action pathways. In this study, reactive oxygen species (ROS), oxidative stress indices, real-time PCR, Western blotting, confocal microscopy, and molecular docking were used to investigate the protective effect of purified kiwi leaves PAs (PKLPs) on Caco-2 cells' oxidative stress mechanisms. The results confirmed that pretreatment with PKLPs significantly reduced H2O2-induced oxidative damage, accompanied by declining ROS levels and malondialdehyde (MDA) accumulation in the Caco-2 cells. The PKLPs upregulated the expression of antioxidative enzymes (GSH-px, CAT, T-SOD) and the relative mRNA (Nrf, HO-1, SOD-1, CAT) of the nuclear factor erythroid 2-related factor (Nrf2) signaling pathway. The protein-expressing level of the Nrf2 and its relative protein (NQO-1, HO-1, SOD-1) were significantly increased (*p* < 0.05) in the PKLPs pre-treatment group compared to the model group. In conclusion, the novelty of this study is that it explains how PKLPs' efficacy on the Nrf2-ARE signaling pathway, in protecting vital cells from oxidative stress, could be used for cleaner production.

**Keywords:** antioxidant response element (ARE); Nrf2 signaling pathway; bioactive byproducts; proanthocyanidins; oxidative stress mechanisms

### **1. Introduction**

With the increase in demand for natural safe alternatives to synthetic chemicals, scientists are searching for efficient techniques to evaluate the functionality of phytochemicals [1,2]. The correlation between phytochemicals and cellular antioxidant enzyme activity, based on their chemical and molecular gene expressions, can provide accurate information about their functionalities [3–5]. For instance, proanthocyanidins (PAs), which are abundant in kiwi (*Actinidia chinensis*) fruits and leaves, are a byproduct of kiwi fruit production, and are secondary plant metabolites that belong to the class of flavan-3-ols, with several biological activities and a wide range of health-related benefits [6,7].

For instance, as a natural safe extract, kiwi PAs have vital antioxidant, antidiabetic, and antimicrobial properties, and outstanding anticancer effects [8,9]. Our previous work found that purified kiwi leaves PAs (PKLPs) were relatively high, with a yield of 6.23% (dry weight), making kiwi leaves byproducts a commercially viable source of PAs. Moreover, the PAs isolated from kiwi leaves are mainly composed of (epi)afzelechin, (epi)catechin and (epi)gallocatechin [6]—in which regard, this structure is unique compared to other PAs' plants sources (like baobab seeds, grape seeds, and hazelnut skin).

**Citation:** Lv, J.-M.; Gouda, M.; Ye, X.-Q.; Shao, Z.-P.; Chen, J.-C. Evaluation of Proanthocyanidins from Kiwi Leaves (*Actinidia chinensis*) against Caco-2 Cells Oxidative Stress through Nrf2-ARE Signaling Pathway. *Antioxidants* **2022**, *11*, 1367. https://doi.org/10.3390/ antiox11071367

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 20 June 2022 Accepted: 12 July 2022 Published: 14 July 2022

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

**Copyright:** © 2022 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/).

The bioactive potential of PAs—such as their antioxidant, antimicrobial activity—make them applicable in functional food additives for the therapeutic intervention of human disorders [2]. Thus, this novel health-related natural molecule could be used in different medicinal foods applications: for instance, the effects of acid-hydrolyzed PKLPs on the viability of Caco-2 cancer cells, which indicate that the degree of polymerization (DP) of PAs has a significant effect on cancer cells [7]. However, the key antioxidative bioactive PAs fraction mode of action still requires further study.

Therefore, it is of great significance to explore the main bioactive ingredients of PKLPs as edible and potent viable antioxidant food ingredients. Based on the mean degree of polymerization (mDP) of PAs complexes, previous study has investigated the relationship between the antioxidant activity and PAs' structure compositions. The DP significantly affected the cellular absorption of PAs fractions. Ou and Gu [10] indicated that DP > 4 of PAs were not absorbable as a result of their large molecular dimensions. Additionally, a high DP influences the bioactivity of PAs. For instance, Li, et al. [11], reported a remarkable antiproliferative activity induction of PAs on Caco-2 cells with increasing PAs DP. Additionally, Li, Chen, Li, Liu, Liu, and Liu [11] reported that PAs' cellular antioxidant activities on Caco-2 cells decreased with the increase of their molecular weight. Under normal conditions, there is a balance between antioxidative defense and the generation of reactive oxygen species (ROS). Oxidative stress occurs when ROS production exceeds the extent of cellular antioxidative defense [12]. The increased ROS are able to disturb barrier integrity, damage cell membranes, and enhance incidences of endotoxemia and inflammation [13]. Thus, the over-production of ROS has a deleterious effect on human health, which needs to be controlled. As PAs have a significant impact on regulating central transcription factors, they have the ability to reduce ROS' generation of Caco-2 cells [14]. Koudoufio, et al. [15], reported that PAs have a protective impact on differentiated intestinal Caco-2/15 cells' oxidative stress (OxS) and inflammation. They reported that PAs significantly reduced malondialdehyde, as a lipid peroxidation biomarker, and raised the relative antioxidant enzymes via increasing the ratio of the Nrf2/Keap1. The uniqueness of using Caco-2 cells in studying vital antioxidant activity has been reported by Kellett, et al. [16], who noted that, due to differences in the active membrane transport among the cell types, Caco-2-based cellular antioxidant activity measurements appeared to be a more suitable method for phytochemicals bioactivities studies compared to other cell lines, like hepatocarcinoma (HepG2) cells.

The importance of studying the nuclear factor erythroid 2-related factor 2 (Nrf2) as a cellular antioxidant marker comes from its high ability to bind with Kelch-like ECHassociated protein (Keap1, the cysteine-based mammalian intracellular sensor for electrophiles and oxidants) in the cytoplasm, under normal circumstances [17]. Upon oxidative stress, the Nrf2 migrates to the nucleus, and combines with the antioxidant response element (ARE), which can upregulate the transcription of cell defense-related genes, including drug metabolizers, detoxifying enzymes, and antioxidant proteins [18]. Several studies have reported that the Nrf2 signaling pathway plays a crucial role in the activation of cytoprotective genes in response to xenobiotics, and protecting cells against oxidative stress. However, the underlying mechanisms of the antioxidative action were unclear. In particular, the relationship between the chemical structure of PAs and their antioxidant mechanisms has not been investigated before.

Arroyave-Ospina, et al. [19], noted that intracellular antioxidant activity is an effective assay for measuring ROS and oxidative stress indices. Moreover, real-time PCR, in combination with Western blot and florescent confocal microscopy, could reflect the molecular level of the genotype and phenotype relationship, through antioxidant enzymes gene expression [20]. Additionally, molecular docking is an effective tool for investigating the best intermolecular framework between bioactive phytochemicals, cellular proteins, and biological macromolecules, in order to clarify the potential mechanisms of their interactions [21,22].

Thus, the aim of this study was to characterize kiwi leaves PAs' viable antioxidant activity against Caco-2 cells, and to emphasize the potential pathways that cause that mode of action through their binding efficiency to the ARE from their impacts on NRF2–Keap1 complexes. This could enhance the functional application of these kinds of functional bioactive components that are produced as byproducts of leaves from kiwi fruit production. Thus, it could enhance the applicability of PAs in green and sustainable industrial applications.

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

#### *2.1. Chemicals and Materials*

RNase Free dH2O, ethanol (analytical grade), and acetone (analytical grade) were purchased from Macklin Biochemical Technology Co., Ltd., (Shanghai, China). Hydrogen peroxide, Nile Red, Dichlorodihydrofluorescein diacetate (DCFH-DA), and Dihydroethidium (DHE) were purchased from Sigma–Aldrich (St. Louis, MO, USA). AB-8 Macroporous resin was obtained from Solarbio Science & Technology Company (Beijing, China), and Sephadex LH-20 was purchased from GE Healthcare Bio-Sciences (Uppsala, Sweden). Dulbecco's Modified Eagle Medium (DMEM), fetal bovine serum (FBS), penicillin/streptomycin (P/S), and phosphate-buffered solution (PBS), were supplied by Gibco company (Grand Island, NE, USA). SYBR Green PCR Master Mix and TRIzolTM were obtained from Thermo Fisher Scientific, Inc., (Cleveland, OH, USA). The commercial PVDF membrane (0.45 μ, immobilon) was purchased from Solarbio Science & Technology Co., Ltd., Beijing, China.

Kiwi (*Actinidia chinensis*) leaves were collected from Zhuji farm, Shaoxing, Zhejiang Province, China, during October. Then, they were freeze-dried, and grounded into powder for extracting PAs.

#### *2.2. Extraction and Purificantion of PAs from Kiwifruit Leaves*

Fresh kiwifruit leaves PAs were extracted by using the optimized conditions of ultrasound-assisted extraction, following the method described by Lv, Gouda, Zhu, Ye, and Chen [6], with some modifications. Firstly, freeze-dried kiwi leaves powders (10 g) were sonicated by JY92-IIDN (Ningbo Scientz Biotechnology Co., Ningbo, China) under the following optimum conditions: 30 mL/g dry weight solvent to solid ration; 40% ultrasoundamplitude; and 70 ◦C sonication temperature for 15 min. Secondly, the crude PAs were extracted by 0.4 L aqueous acetone solvent (80%, *v*/*v*). Then, the acetone was removed through an evaporation process, under vacuum and 40 ◦C temperature, by using rotary evaporator (Dragon RE100-pro, Beijing, China). Afterwards, hexane was used to remove the non-polar components from the obtained aqueous phase. Thirdly, sugars, proteins, and pigments were removed from the extracted PAs by using an AB-8 Macroporous resin column (Solarbio Science & Technology, Beijing, China). Afterwards, to further purify the obtained PAs, according to the previous study (with some modifications) of Chai, et al. [23], the above-obtained PAs (1 g, freeze-dried) were loaded onto a Sephadex LH-20 column (GE Healthcare Bio-Sciences, Uppsala, Sweden); then, methanol (50%, *v*/*v*) was used to wash the column and remove the impurities. Subsequently, 90% methanol was used to elute the fraction A (FA) anthocyanins. After that, 50% acetone was used to collect as fraction B (FB). Both fractions (FA and FB) were freeze-dried as PKLPs.

#### *2.3. Reversed-Phase HPLC-QTOF-MS/MS Analysis of PKLPs*

The separation of the phenolic compounds was performed following our previous study [7]. The column used for analysis was the Luna HILIC column (Phenomenex, Torrance, CA, USA; 250 × 4.6 mm; 5 μm), with a flow rate of 0.35 mL/min at 30 ◦C. Injections of 10 μL of each purified extract were injected into a Waters 2489 HPLC with a UV-Vis detector (Waters Corp., Milford, MA, USA), using a mobile phase comprising a linear gradient of 99.5% acetonitrile and 0.5% acetic acid (solvent A),96.9% acetonitrile, 3% water, and 0.1% acetic acid (solvent B). The detection wavelength was set at 280 nm to monitor all phenolic compounds. For comparison, the elution conditions for solvent B were: 0–10 min, 15%; 10–20 min, 15–20%; 20–70 min, 20–60%; and 70–80 min, 60–100%. The

separated compounds were fractionated and defined by mass spectra, using a Triple-TOF 5600+ ion trap mass spectrometer (AB scientific, Framingham, MA, USA). Three replicates of each sample were collected for data analysis.

#### *2.4. Cell Culture and Treatment*

The Caco-2 (human colonic carcinoma) cell line was provided by the Cell Resource Center, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. The cells were cultured at 37 ◦C under 5% CO2, in a DMEM medium (Gibco) with 1% P/S (Gibco) and 20% FBS (Gibco). The cells were sub-cultured 3–4 times a week, and replaced in a fresh medium to keep the cells in a good growth state.

#### *2.5. Injured Cell Model Induced by H2O2*

The antioxidant activity of the H2O2-induced cell death of the Caco-2 cells was determined according to a previous study [24]. Briefly, the Caco-2 cells were seeded in cell culture 96-well plates, with a density of 1 × <sup>10</sup><sup>5</sup> cells/mL, and cultivated for 24 h at <sup>37</sup> ◦C. Afterwards, the medium was removed, and the wells were washed 3 times with PBS, before a new, fresh medium (containing 200 μM/L H2O2) was added. The group treated by the same cell medium without H2O2 was taken as the control group. The H2O2 treatment lasted for 4 h, and the results were expressed as cell viability [7].

#### *2.6. Intracellular Antioxidant Activity Assay*

The effect of PKLPs' antioxidant activity on the Caco-2 cells was determined according to previous literature, with some modification [25]. Briefly, the Caco-2 cells were seeded in cell culture 96-well plates, with a density of 1 × <sup>10</sup><sup>5</sup> cells/mL, and cultivated for 24 h at 37 ◦C. The experimental groups were treated with various concentrations of samples (50 μg/mL of FA; 50 μg/mL of FB; 75 μg/mL of catechin; each for 100 μL) 24 h before H2O2 treatment. The control group was normally cultivated without H2O2 or PAs, and the model group was only treated with H2O2 to cause the oxidant damage. The result was expressed as the cell viability determined by the cell counting kit (CCK-8 assay), as described in the literature [7].

#### *2.7. Determination of Reactive Oxygen Species (ROS)*

The cellular ROS was determined according to previously described methods, with slight modifications [26]. Briefly, Caco-2 cells were seeded into 12-well plates, at a density of 1 × 105 cells/mL, for 24 h cultivation at 37 ◦C. The experimental groups were treated with various concentrations of samples (50 μg/mL of FA; 50 μg/mL of FB; 75 μg/mL of catechin; each for 100 μL) for 24 h cultivation, followed by treatment with (200 μM/L) H2O2 for another 4 h. The control group was normally cultivated at the same time, and the model group was only treated with (200 μM/L) H2O2 for 4 h. After incubation with 10 μM dichlorofluorescein diacetate (DCF-DA) at ambient temperature in the dark for 30 min, the cells were instantly washed 3 times by PBS. And then, the fluorescence value was measured by the fluorescence microscope (Nikon, Tokyo, Japan) at an excitation wavelength of 485 nm, and an emission wavelength of 525 nm. The fluorescence intensity was calculated by image analysis software ImageProPlus 6.0 (Media Cybernetics, Inc., Rockville, MD, USA), and expressed as mean DCF fluorescence intensity.

#### *2.8. Determination of Oxidative Stress Indices*

Malondialdehyde (MDA, #S0131M), catalase (CAT, # S0051), total superoxide dismutase (T-SOD, #S0101S), and glutathione peroxidase (GSH-Px, #S0056), were measured by the commercial kits purchased from Beyotime Biotechnology Company (Shanghai, China), according to the manufacturer's instructions.

#### *2.9. RNA Extraction, Reverse Transcription, and Quantitative Real-Time PCR*

Total RNA was extracted from the cells using TRIzolTM (Thermo Fisher Scientific, Inc.), and diluted to 1 μg/μL. qPCR was performed according to a previously described method. Firstly, cDNA was synthesized from 1.0 μg of total RNA using the PrimeScript RT reagent Kit (TaKaRa, Japan), in a final volume of 20 μL with 10 μL master mix (4 μL RNase Free dH2O, 4 μL 5× PrimeScript Buffer, 1 μL RT Mix, and 1 μL PrimeScript RT Enzyme Mix). Then, 1 μL of cDNA template, 1 μL of upstream and downstream primers, 10 μL SYBR Green PCR MasterMix, and 7 μL of RNase Free dH2O were mixed, before being carried out in a Stepone Plus qPCR instrument (Thermo Fisher Scientific, Inc.). The PCR conditions were set as: 95 ◦C maintained for 5 min, followed by 40 cycles of 95 ◦C for 15 s and 60 ◦C for 40 s. The primer sequences are listed in Table 1.

**Table 1.** Primer sets for quantitative real-time PCR.


Relative mRNA expression was normalized to the control group. The 2 <sup>−</sup>ΔΔCt formula was used to quantify, using *β-actin* as a reference gene [27]. All the results were obtained from at least three independent experiments.

#### *2.10. Immunofluorescence*

Samples were blocked with bovine serum (5%, *w*/*v*) for 30 min at ambient temperature, and repaired with 10.2 mM sodium citrate buffer. Then, the following primary antibodies against the Nrf2 (1:200), and the HO-1 (1:200) were used for overnight incubation at 4 ◦C. Afterwards, the samples were washed with PBS before incubation with anti-rabbit secondary antibodies DAPI staining at ambient temperature for 1 h. The treated Caco-2 cells were visualized using a Zeiss LSM 780 confocal microscope (Carl Zeiss SAS, Jena, Germany), and the DP2-TWAN image-acquisition system (Olympus Corp., Tokyo, Japan).

#### *2.11. Molecular Docking*

The initial crystal structures of Keap1 and the Nrf2 (PDBID: 2FLU) were obtained from the RCSB PDB database (http://www.rcbs.org, accessed on 15 February 2022), as reported in previous literatures [28,29]. The 3D structure of the PAs was built by Chem3D Ultra 12.0, and energetically minimized with an MM2 force field. The interaction between Keap1, the Nrf2, and the PAs was investigated by using docking analysis. Schrodinger® docking suits were selected for the molecular docking studies, and a Glide® receptor grid generator was used to create the grid sites with default parameters. Both the protein and the ligand structures were refined using an OPLS3e forcefield, to get correct formal charges and protonation states. A receptor grid was then generated with the prepared structure, with correct formal charges and protonation states. Finally, the ligand was docked into the corresponding protein structure. After docking, the results were ranked according to a scoring function, combining GlideScore with Prime energies, and the complex features of the protein-ligand were visualized with Pymol (Delano Scientific LLC, San Carlos, CA, USA).

#### *2.12. Western Blot*

Western blot was conducted, following the method of Su, et al. [30]. In brief, the cytoplasmic protein was separated by NE-PER kit (n. 78833, Thermo Fisher Scientific, Rockford, IL, USA). The separated protein contained 1% phenylmethanesulfony fluoride (PMSF, ST506, Beyotime Biotech, China) for preventing the degradation of protein, and the concentrations were measured using the bicinchoninic acid method (BCA, P0012, Beyotime Biotech, China). The separated protein was produced by electrophoresis on SDS-polyacrylamide gels, then transferred to 0.45 μm polyvinylidene fluoride (PVDF) membranes (immobilon, Solarbio Science & Technology Co., Ltd., Beijing, China). After blocking with 5% non-fat dry milk, in PBS containing 0.1% Tween-20, the membrane was incubated with the primary antibody at 4 ◦C for 14 h. Then, the membrane was incubated with horseradish peroxidase-conjugated secondary antibodies (ab97205 and ab97215, Abcam, Cambridge, MA, USA) for 1–2 h after washing 3 times. Next, a chemiluminescent HRP substrate was used to visualize the immunoreactive protein bands (Millipore, WBKLS0100), after washing 3 times. Primary antibodies were used as follows: the nuclear factor erythroid 2-related factor 2 (Nrf2) for humans (ab62352), heme oxygenase-1 (HO-1) for humans (ab13243), and Kelch-like ECH-associated protein 1 (Keap1) for humans (ab139729) were procured from Abcam (Cambridge, MA, USA); superoxide dismutase 1 (SOD-1) for humans (sc-17767), quinone oxidoreductase 1 (NQO1) for humans (sc-271116), and B-cell lymphoma 2 (Bcl-2) for humans (sc-7382), were procured from Santa Cruz Biotechnology Inc. (California, MA, USA).

#### *2.13. Statistical Analysis*

Experiments were conducted in triplicate, and the results were presented as mean ± standard deviation. Data were further analyzed via one-way analysis of variance (ANOVA), using SPSS 19.0 (Chicago, IL, USA). Duncan and Least Significant Difference (LSD) analyses at *p* < 0.05 level were used to significantly differentiate the studied treatments. IC50 was calculated based on the regression equation.

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

#### *3.1. Comparation of the Chemical Composition of PKLPs (FA and FB) by Reversed-Phase HPLC-QTOF-MS/MS*

According to our previous study [7], FA contains small molecule flavonoids, including quercetin, isoquercetin, and polyphenols, such as procyanidins (Figure 1a). In the present study, we focused on analyzing the composition of PAs fractions without the acid hydrolysis post-purification process. In addition, we marked the retention time (Rt) of PAs with different degrees of polymerization. We also marked the main ionized fragments of PAs in QTOF-MS2: molecular ions at *m*/*z* 447, with major fragment ions at *m*/*z* 301 ([M − H − 146]−) and *m*/*z* 109 ([M − H − 338]−), which were identified as quercetin based on their fragmentation patterns and retention time (Table 2). According to the results of QTOF-MS2, the FA included one molecular ion peak [M − H]<sup>−</sup> at *<sup>m</sup>*/*<sup>z</sup>* 463, with ionic fragmentation MS<sup>2</sup> at *m*/*z* 301 and 463, which was consider as isoquercetin. Fraction A contained catechin, epicatechin, and procyanidins, with its molecular ion peak ([M − H]−) at *m*/*z* 289 and 577, respectively, while tandem mass spectrometry yielded typical fragment ions at *m*/*z* 163, 137, 245, and 287, 289, 425, 451, respectively. The ionic fragmentations at *m*/*z* 163 and 451 ([M − H − 126]−) resulted from the loss of a neutral molecule, A-ring (1,3,5-trihydroxybenzene, 126 Da), through heterocyclic ring fission (HRF). The typical fragment ions at *m*/*z* 137 and 425 derived from retro-diels-alder (RDA) reaction through the loss of a neutral fragment containing the B-ring (152 Da) [31].

**Figure 1.** Chemical structure of eluted fractions: (**a**) FA; (**b**) FB. Enhancement of cell viability and protection of Caco-2 cells from H2O2-induced oxidant damage by pre-treatment with PKLPs: (**c**) cell viability; (**d**) intracellular generation of ROS; (**e**) malondialdehyde (MDA) level of Caco2 cells. The reported values are represented as mean ± SD (*n* = 3). Columns marked with different lowercase letters indicate the significant differences among treatments by using Duncan analysis.


Fraction B contained monomer procyanidins and polymer PAs, such as catechin, epicatechin, procyanidins, prodelphindins, and propelargonidins (Figure 1b). For instance, FB showed the fragment ion ([M − H − 288]−) at *m*/*z* 289 and 577, that derived from cleavage of the trimer procyanidins; it also contained one molecular ion peak [M − H]<sup>−</sup> at *m*/*z* 865, with MS<sup>2</sup> yielding typical fragment ions at *m*/*z* 577, 289, 739, and 713. The base ion [M − H − 126] at *m*/*z* 739 came from HRF. The ion of *m*/*z* 713 was formed through an RDA fission, which characterizes hydroxvyinylbenzenediol elimination ([M − H − 152]−). Compared with FA, the chemical composition of FB was purer. The concentration of procyanidins as extension units in FB was much higher than that in FA, as well as the quantity of monomers units in FB. On the other hand, there was a significant difference in the DP of the two fractions. Specifically, the mean degree of polymerization (mDP) values

of the eluted fractions of FA and FB were 3.2 and 5.9, respectively, which implied that FA had a smaller molecular weight than FB.

#### *3.2. PKLPs (FA and FB) Suppressed H2O2-Induced Oxidative Stress in Caco-2 Cells*

As hydrogen peroxide (H2O2) was a product of the cellular oxygen metabolism that was a feature of the various metabolic and signaling cascades [12,32], the control of the physiological H2O2 intracellular concentration, as an antioxidant indicator, has a significant relationship to the cells' functionality and viability [33]. Therefore, the potential antioxidant activity of PKLPs on the cellular H2O2-induced oxidation model was established. In our previous study, we found that using 200 microM of H2O2 showed the most suitable impacts on Caco-2 cells, compared to 10, 25, 50, and 100 μM, which each caused too low an injury on the cells [6]. Therefore, 200 μM H2O2 was selected for the injury cell model.

In this study, a significant inhibition (*p* = 0.01; 47.25 ± 5.72%) was observed after using 200 μM of H2O2 on treating Caco-2 cells for 4 h. On the other hand, pre-treatment with PKLPs observably increased the Caco-2 cell viability after using the same concentration of ROS. Meanwhile, FA cell viability (73.16 ± 7.27%) was higher than FB, with cell viability of 69.10 ± 7.31% (Figure 1c). This phenomenon was confirmed by the fluorescence microscope, which clearly defined the increase in the number of viable cells by FA pre-treatments compared to FB (Figure 2a).

**Figure 2.** Effects of PKLPs on: (**a**) ROS level express mean fluorescence intensity; (**b**) GSH-px; (**c**) CAT; (**d**) T-SOD activities in H2O2-induced Caco-2 cells. The cells were pre-treated with catechin (75 mg/mL) and PKLPs (50 mg/mL) before being stimulated with H2O2 (200 μM/L) for 4 h. The reported values are represented as mean ± SD (*n* = 3). Columns marked with different lowercase letters indicate the significant differences among treatments by using Duncan analysis (*p* < 0.05).

The distinction between FA and FB came from the chemical structure of PKLPs with a higher DP of FB, which decreased its functionality [15]. Rauf, et al. [34], reported that extracted PAs revealed effective antioxidant activity as a result of their functional hydroxyl groups positions, compared to other phenolic compounds. Additionally, Koudoufio, Feldman, Ahmarani, Delvin, Spahis, Desjardins, and Levy [15] mentioned that pretreatment with PAs could significantly prevent the Caco-2/15 cell from oxidative damage to its biological macromolecules due to its exposure to the strong oxygen-radicals (Fe/Asc).

For the cytoprotective effect, the two fractions of PKLPs were evaluated for their intracellular free radicals scavenging ability on the ROS and MDA cellular levels (Figure 1d,e): the ROS levels reflected the antioxidant enzyme activity, and the MDA level reflected

the cellular lipid oxidation degree that was damaged by the excessive exposure to free radicals [35,36].

This study shows that ROS levels significantly (*p* = 0.03) declined from 10.8 ± 0.8 × 103 florescence/mg protein in the model group to 6.6 ± 0.6 × <sup>10</sup><sup>3</sup> in the FA group (Figure 1d). Meanwhile, the Caco-2 treated with FB showed a significant decrease (*p* = 0.04) in the release of ROS compared to the normal cells (Figure 1d). Furthermore, the MDA of FB decreased (*p* < 0.05) to 18.58 ± 3.34 nM/mg protein compared to the model with 41.59 ± 4.23 nM/mg protein. Zhou, et al. [37], reported that PAs contain an abundance of hydroxyl groups, and can release H+ to form cross-linkages with the radicals, which can significantly attenuate oxidative damage, by reducing the contents of MDA and ROS. Furthermore, Koudoufio, Feldman, Ahmarani, Delvin, Spahis, Desjardins, and Levy [15] reported that PAs can decrease lipid peroxidation through their ability to regulate the expression of TNFα, COX2, and NF-κB. On the other hand, Li, Liu, Li, McClements, Fu, and Liu [8] reported that reduction of the influence of PAs on Caco-2 cells viability is dependent on the specific mDP fractions that stimulate the apoptotic pathways of caspase-9, caspase-3, and caspase-8, which are generally increased by ROS generation. They also noted that the bioavailability of PAs can be increased through the hormesis effect, if it is under a lower mDP. Therefore, the study of PAs' specific fraction vital influence should be emphasized.

#### *3.3. Protective Effect of PKLPs (FA and FB) in H2O2-Induced Caco-2 Cells*

Glutathione peroxidase (GSH-Px) activity, catalase (CAT) activity, and total superoxide dismutase (T-SOD) activity are known to be a crucial enzyme-driven antioxidant defense system in organisms, which can scavenge free radicals to maintain the redox balance in cells [38]. As catechin has significant antioxidant activity, as reported previously, we selected it for the positive treatment group. In this study, there was a significant increase in the antioxidant enzyme biomarkers of the PKLPs compared to the oxidative stress model, in which regard, pre-treatment of H2O2-induced cells with PKLPs (FA or FB) significantly promoted GSH-Px activity when compared to the model group (Figure 2b). Meanwhile, FA exhibited a more noteworthy effect on enhancing antioxidant enzyme activity than FB, which significantly increased to 12.36 ± 2.70 U/mg protein in the FA group, and increased to 10.91 ± 0.93 U/mg protein in the FB group, respectively, when compared to the model group (4.39 ± 1.41 U/mg protein) (*p* < 0.05). Fujimaki, et al. [39], noted that PAs contain (+)-catechin as an upper unit that mainly increases its bioactivity in its unique structural combination with (−)-epicatechin-(4β→8)-(−)-epicatechin 3-*O*-gallate subunits. Moreover, Caco-2 treated with FB increased the release of CAT significantly (*p <* 0.05), to 23.21 ± 4.05 U/mg protein compared to the model group (Figure 2c).

As shown in Figure 2c,d, CAT and T-SOD activities in the PKLPs treatment group exhibited similar antioxidant activity to GSH-Px activity. From the results, we found that PKLPs pre-treatment could effectively improve the antioxidant status of cells exposed to free radicals damage by increasing the GSH-Px, CAT, and T-SOD activities, which indicated that PKLPs could activate an antioxidant signaling pathway [34]. Our findings are consistent with the previous study of Su, Li, Hu, Xie, Ke, Zheng, and Chen [30]: PAs can attenuate oxidative stress by scavenging excessive ROS, and protect the organism by increasing the activity of antioxidant enzymes.

#### *3.4. Effect ofPKLPs (FA and FB) on the Nrf2 and Its Downstream Target Genes Transcription*

The employment of micro(m)RNA expression data as molecular markers could explain the antioxidant and anticancer activities potential pathway through the transcription factors [40]. For instance, the nuclear factor erythroid 2-related factor (Nrf2) is a transcription factor that plays a significant role in response to xenobiotics and oxidative stress, by binding to the antioxidant response element (ARE) [41].

In this study, the mRNA levels of the Nrf2 and Keap1, and the downstream target genes, were further investigated in Caco-2 cells for their potential response to various treatments. Quantitative real-time PCR (qRT-PCR) analysis of relative mRNA expression is shown in Figure 3. There were significant increases in the antioxidant mRNA biomarkers of the PKLPs pre-treated groups compared to the model group, in which regard, the Nrf2 levels were significantly increased to 1.86 ± 0.22 mRNA expression in the FA group compared to the catechins with 1.18 ± 0.11 mRNA expression (Figure 3a). Meanwhile, the Caco-2 treated with FA increased the release of HO-1 significantly (*p =* 0.03), to 1.29 ± 0.12 mRNA expression, compared to FB, with 0.97 ± 0.11 mRNA expression (Figure 3b), in which regard, the changes in FA and FB that impacted on the mRNA antioxidant expression could be ascribed to the higher molecular weight of FB, which may have blocked its accessibility into the Caco-2 cell, and therefore decreased its impacts [10] (Figure 3c). Ge, et al. [42], reported that PAs' impact on Caco-2 cell monolayers and cholesterol could explain their different dimers' impacts on the signaling pathways through the absorption affinity. The same trend was observed for the mRNA level of NQO1, SOD-1, CAT, and Bcl-2 (Figure 3c–e,g), while the mRNA expression of Keap1 and Bax showed different responses (Figure 3f,h). For instance, the Caco-2 pre-treated with FA significantly decreased the mRNA expression of Bax (*p* = 0.02), with 1.43 ± 0.13, compared to the model with 2.37 ± 0.18 (Figure 3h). Contrastingly, the PKLPs pre-treatment groups remarkably reversed the expression level of these genes compared to the model group, which indicates that PKLPs pre-treatment could significantly inhibit oxidative stress, by alleviating the alteration in Bcl-2 and Bax gene transcription (Figure 3g,h). Siddiqui, et al. [43], mentioned that Bax, as a pro-apoptotic protein, could promote the release of apoptotic molecules into the cytoplasm by competing with CAT, in which its expression was affected by the antioxidant molecules against ROS.

**Figure 3.** Effects of PKLPs on Nrf2-mediated antioxidants signaling transcripts were analyzed by qRT-PCR. The gene expression of the Nrf2 and its relative or downstream molecules: (**a**) Nrf2 mRNA; (**b**) HO-1 mRNA; (**c**) NQO1 mRNA; (**d**) SOD-1 mRNA; (**e**) CAT mRNA; (**f**) Keap1 mRNA; (**g**) Bcl-2 mRNA; (**h**) Bax mRNA. Data are shown as mean ± standard deviation (*n* = 3). Columns marked with different lowercase letters indicate the significant differences among treatments by using Duncan analysis (*p* < 0.05).

Our results are consistent with previous results, which confirmed that polyphenols extracted from plants or herb could increase antioxidative enzymes based on the upregulation of their antioxidative mRNA genes expression, such as NQO-1, HO-1, Nrf2, and SOD [44]. Hilary, et al. [45], reported that PAs' mode of action came from their hydroxyl groups, which could direct supporting antioxidant reactions or even mediate the occurrence of oxidant events. This phenomenon could be explained as, under normal cellular condition, the Nrf2 binds to Keap1 to form suitable complexes in cytoplasm; on the other hand, ROS cause the separation of the Nrf2 from Keap1 and, as a result, the Nrf2 transfers into the nucleus; meanwhile, Keap1 degrades in the cytoplasm [46] (Figure 4a). The reason for the

dissociation of the Nrf2–Keap1 complex in the cytoplasm is the sulfhydryl modification and Nrf2 phosphorylation, which results in an uncoupled Nrf2 and Keap1 [47].

**Figure 4.** Comparison between the computational geometries of the binding of the Nrf2 and PKLPs with a pocket of Keap1 protein: (**a**) Proposed schematic diagram of the Nrf2 signaling pathway mechanism of the PKLPs in the H2O2-induced Caco-2 cells; (**b**) Magnification of the interaction site of the Nrf2–Keap1 complex motif, and its binding geometry; (**c**) Magnification of the interaction between the Keap1 pocket and the PKLPs. Hydrogen bonds are represented by yellow dashed lines, and the related amino acid residues and peptides are highlighted in the picture.

#### *3.5. Effect of PKLPs (FA and FB) on the Nrf2 and Its Downstream Protein Expression*

To further confirm the results of the Nrf2 and HO-1 expressions, the signaling pathway was further investigated through the molecular mechanism to underline the protective effect of PKLPs on H2O2-induced oxidative stress. The related antioxidant protein levels of the Nrf2, Keap1, SOD-1, HO-1, and NQO-1 were studied by Western blotting assay, and their changes were evaluated by confocal microscopy technology (Figures 5 and 6).

**Figure 5.** Effects of PKLPs on the protein expression of the Nrf2 signaling pathway: (**a**) Western blotting estimation of the Nrf2 signaling pathway protein expression in Caco-2 cells after pre-treatment with catechin (75 μg/mL) and PKLPs (50 μg/mL) for 24 h before stimulation with H2O2 (200 μM/L) for 2 h; (**b**–**g**) Quantification of the above-mentioned Western blots, using β-actin as a loading control. The results are presented as mean ± SD (*n* = 3). Columns marked with different lowercase letters indicate the significant differences among treatments by using Duncan analysis (*p* < 0.05).

As shown in Figure 5, compared to the normal control group, the antioxidant proteins were increased (*p* < 0.05) by pre-treatment of the PKLPs, when compared to the model that used the oxidative damage hydrogen peroxide as a negative control, in which regard, the Nrf2 levels were significantly increased to a 1.40 ± 0.10 Nrf2/*β*-actin relative level in the FA group, compared to the model with a 0.773 ± 0.05 Nrf2/*β*-actin relative level (Figure 5b). Additionally, the FA increased the release of HO-1 significantly (*p* < 0.05) to a 1.24 ± 0.10 HO-1/*β*-actin relative level compared to the catechins with a 0.91 ± 0.11 HO-1/*β*-actin relative level (Figure 5c). As shown in Figure 6, the confocal images showed that the distribution of the Nrf2 was increased inside the nucleus for the FA group compared to the model and catechin treated groups. It demonstrated that Nrf2 expression plays a critical role in oxidative stress response. This action is through the dissociative Nrf2 transfer from cytoplasm to nucleus, and binds to the antioxidant response element in the nucleus to exert function by activation of the gene transcription of the antioxidant enzyme [48].

Meanwhile, the obtained results revealed that the protein levels of the Nrf2 and its downstream protein were significantly increased in the FA pre-treated group, compared to those in other pre-treatment groups, such as the FB pre-treated group and the catechin pretreated group. Based on our study, FA contained procyanidins, quercetin, and isoquercetin, while FB only contained polymer PAs. The relative antioxidative mRNA transcription and protein expression results implied that oligomer PAs and quercetin could work synergistically to regulate multiple and interactive molecular targets related to H2O2-induced oxidative damage and the involvement of Nrf2-associated pathways, which was consistent with previous reporting [30]. As for FB, the fraction had more chemical adducts that were causing bigger mDP, which might inhibit the bioavailability of PAs, and further inhibit their vital impact on enzyme activity and other bioactivities. Therefore, in this study, FA had overall activities higher than FB.

The mechanism of defense oxidative damage is clarified in Figure 4a. The insertion of FA inside the cells inhibited the ROS negative impacts, which significantly increased the excretion of Nrf2 and HO-1 antioxidant enzymes expressions. Similarly, the corresponding expression of the antioxidant enzymes (SOD-1, HO-1, NQO-1, and Bcl-2) were upregulated (*p* < 0.05) in Caco-2 cells against the H2O2 exposed group. The results of the molecular mechanisms, by qRT-PCR and the Western blot, indicated that both FA and FB of the PKLPs dramatically rescued previous effects caused by oxidative damage. H2O2 exposure inhibited the expression of the Nrf2 and its related or downstream genes such as SOD-1, HO-1, NQO-1, and Bcl-2 in organisms. Zhou, Chang, Gao, and Wang [3] reported that extracted plants' PAs protect the organisms against apoptosis caused by H2O2, through their significant impacts on the Nrf2-ARE pathway. Furthermore, the current results agree with the previous investigations, that H2O2 treatment suppressed the translocation of the Nrf2 signaling pathway [49].

**Figure 6.** Confocal microscopy of single Caco-2 cells protein expression of (**a**) the Nrf2 and (**b**) HO-1 signaling pathways after pre-treatment with catechin (75 μg/mL) and PKLPs (50 μg/mL).

Molecular modelling of the interaction between Keap1, Nrf2, and PAs is shown in Figure 4b,c by using a molecular docking analysis with the binding mode of the Nrf2 to Keap1. The phenylalanine 83 (PHE-83) of the peptide was stabilized by a hydrogen bond with the asparaginate 382 (ASN-382) of the Keap1. Arginine 380 (ARG-380) and tyrosine 334 (TYR-334) from the Keap1 protein were simultaneously held together with glutamic acid 82 (GLU-82) on the Nrf2 protein by hydrogen bond. Moreover, serine 602 (SER-602) from the surface of the Keap1, was bonded to threonine 80 (THR-80) from the Nrf2 protein.

From the molecular docking simulation, it can be seen that PAs were simultaneously held together with the active cites amino acids, like aspartic acid 382 (ASN-382), arginine 380 (ARG-380), serine 602 (SER-602), and glutamine 530 (GLN-530), on Keap1 by hydrogen bonding [50]. It can be seen that PAs compete with the Nrf2 to promote the separation of the Nrf2 from the Nrf2–Keap1 complexes. The Nrf2 is then transported into the nucleus, where it binds to the antioxidant response element, and produces different antioxidant enzymes by activating relative antioxidant genes. Ma, et al. [51], noted that PAs bind with large molecular-weight proteins like β-casein through the interaction of their hydrophobic residues with the amino acid hydrophobic residues, including Ala-192, Pro-196, and Pro-201. Therefore, the increase in the affinity of carbonylic groups of the peptide backbone forming hydrogen bonds with other amino acids from Keap1 could be the best explanation of how the PAs stimulate the separation of the Nrf2 from Keap1 [47]. Under normal physiological conditions, Keap1 acts as a binder in the cytoplasm, which binds to the Nrf2 through Cul3 ubiquitin ligase, to regulate the ubiquitination of the Nrf2. As a result, the free Nrf2 is maintained at a low level in the cytoplasm.

#### **4. Conclusions**

Two PAs fractions (FA and FB) were considered as byproducts that were extracted from the leaves waste of kiwi fruit production. These compounds showed significant functional and vital antioxidant activity for green and sustainable production. The antioxidative mechanism of both fractions (FA and FB) was studied by using ROS measurement, oxidative stress indices, quantitative real-time PCR, Western blot analysis, confocal microscopy, and molecular docking in Caco-2 cell lines. The present study indicates that PLKPs have a potential key impact on promoting cells signaling pathways to construct the linkage between the Nrf2 and ARE in Caco-2 cells. In comparison with the models, the FA group had a more positive and significant effect compared to FB and catechin. Therefore, this study proposes the use of PAs for promoting the antioxidant capacity of life cells through enhancing antioxidant-related enzyme genes expressions; in which regard, their potential application as a medicinal bioactive ingredient in food cleaner production may enhance their lifestyle-related protective impact against oxidant-related diseases like cancers.

**Author Contributions:** Conceptualization, J.-M.L., M.G. and J.-C.C.; methodology, J.-M.L., M.G., X.-Q.Y. and J.-C.C.; software, J.-M.L. and X.-Q.Y.; validation, J.-M.L., M.G. and J.-C.C.; formal analysis, J.-M.L., M.G., X.-Q.Y., Z.-P.S. and J.-C.C.; investigation, J.-M.L., M.G., X.-Q.Y., Z.-P.S. and J.-C.C.; resources, X.-Q.Y., Z.-P.S. and J.-C.C.; data curation, J.-M.L. and M.G.; writing—original draft preparation, J.-M.L., M.G. and J.-C.C.; writing—review and editing, J.-M.L., M.G., X.-Q.Y., Z.-P.S. and J.-C.C.; visualization, X.-Q.Y., Z.-P.S. and J.-C.C.; supervision, J.-C.C.; project administration, X.-Q.Y. and Z.-P.S.; funding acquisition, X.-Q.Y. and J.-C.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research and the APC were funded by the National Key Research and Development Program of China (2017YFD0400704).

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

**Informed Consent Statement:** Not applicable.

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

**Acknowledgments:** The authors acknowledge the support of Zhejiang University and the National Key Research and Development Program of China throughout the study.

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

#### **Abbreviations**

Proanthocyanidins: PAs, purified kiwifruit leaves PAs: PKLPs, fetal bovine serum: FBS, Dulbecco's Modified Eagle Medium: DMEM, penicillin/streptomycin: P/S, phosphate-buffered solution: PBS, cell counting kit-8: CCK-8, fraction A: FA, fraction B: FB, weight basis: WD, hydrogen peroxide: H2O2, reactive oxygen species: ROS, malondialdehyde: MDA, glutathione peroxidase: GSH-px, nuclear factor erythroid 2-related factor: Nrf2, Quantitative real-time PCR: qRT-PCR, dichlorofluorescein diacetate: DCF-DA, Kelch-like ECH-associated protein 1: Keap1, heme oxygenase-1: HO-1, antioxidant response element: ARE, quinone oxidoreductase 1: NQO-1, superoxide dismutase 1: SOD-1, catalase: CAT, B-cell lymphoma 2: Bcl-2, and Bcl-2-associated X protein: Bax, HRF: heterocyclic ring fission, RDA: Retro-diels-alder, QM: quinone methide.

#### **References**


### *Article* **Investigation of Antioxidant Synergisms and Antagonisms among Phenolic Acids in the Model Matrices Using FRAP and ORAC Methods**

**Danijela Skroza 1,\*, Vida Šimat 2, Lucija Vrdoljak 1, Nina Joli´c 1, Anica Skelin 1, Martina Cagalj ˇ 2, Roberta Frleta <sup>3</sup> and Ivana Generali´c Mekini´c <sup>1</sup>**


**Abstract:** The total antioxidant potential of a sample cannot be predicted from the antioxidant activity of its compounds; thus, scientists usually explain the overall activity through their combined effects (synergistic, antagonistic, or additive). Phenolic compounds are one of the most powerful and widely investigated antioxidants, but there is a lack of information about their molecular interactions. This study aimed to investigate the individual and combined antioxidant activity of equimolar mixtures (binary, ternary, quaternary, and quinary) of 10 phenolic acids (protocatechuic, gentisic, gallic, vanillic, syringic, *p*-coumaric, caffeic, ferulic, sinapic, and rosmarinic acid) at different concentrations using ferric reducing antioxidant power (FRAP) and oxygen radical absorbance capacity (ORAC) assays. Gallic acid showed the highest antioxidant activity, determined using the FRAP assay (494–5033 μM Fe2+) and rosmarinic acid with the ORAC assay (50–92 μM Trolox Equivalents (TE)), while the lowest antioxidant potential was observed for *p*-coumaric acid (FRAP 24–113 μM Fe2+ and ORAC 20–33 μM TE). The synergistic effect (by FRAP) in the equimolar mixtures of hydroxybenzoic acids was confirmed for a large number of tested mixtures, especially at low concentrations. All mixtures containing gentisic acid showed a synergistic effect (28–89% difference). Using the ORAC method, only two mixtures of hydroxybenzoic acids showed an antagonistic effect, namely a mixture of gentisic + syringic acids (−24% difference) and gallic + vanillic acids (−30% difference), while all other mixtures showed a synergistic effect in a range of 26–236% difference. Among mixtures of hydroxycinnamic acids, the highest synergistic effect was observed for the mixtures of *p*-coumaric + ferulic acids and caffeic + sinapic acids with differences of 311% and 211%, respectively. The overall antioxidant activity of phenolic acids could be explained by the number or position of hydroxyl and/or methoxy functional groups as well as the compound concentration, but the influence of other parameters such as dissociation, intramolecular hydrogen bonds, and electron donating or withdrawing effect should not be neglected.

**Keywords:** phenolic acids; phenolic mixtures; interaction effect; antioxidant activity; FRAP; ORAC

### **1. Introduction**

The research effort concerning the antioxidant behavior of phenolic compounds has significantly increased in recent decades, but the knowledge about their interaction in model mixtures is still scarce. Among the diverse and complex groups of phenolics that include simple phenols, flavonoids, stilbenes, tannins, and others, phenolic acids are the most distributed in nature. They have been found in various plants, fruits, vegetables, beverages, and agro-food by-products where they contribute to organoleptic attributes such as color, flavor, and odor but their true merits are numerous positive biological activities

**Citation:** Skroza, D.; Šimat, V.; Vrdoljak, L.; Joli´c, N.; Skelin, A.; Cagalj, M.; Frleta, R.; Generali´ ˇ c Mekini´c, I. Investigation of Antioxidant Synergisms and Antagonisms among Phenolic Acids in the Model Matrices Using FRAP and ORAC Methods. *Antioxidants* **2022**, *11*, 1784. https://doi.org/ 10.3390/antiox11091784

Academic Editors: Antonella D'Anneo and Marianna Lauricella

Received: 8 August 2022 Accepted: 7 September 2022 Published: 9 September 2022

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

**Copyright:** © 2022 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/).

such as antibacterial, anti-inflammatory, antiallergenic, anticancer, cytotoxic, antitumor, cardioprotective, and antioxidant, which is among the most investigated [1–10].

Phenolic acids are represented by two main classes: hydroxybenzoic and hydroxycinnamic acids, containing seven (C1–C6) and nine (C3–C6) carbon atoms, respectively. Each phenolic acid is composed of an aromatic ring with hydroxyl (–OH) and carboxyl (–COOH) groups, and the main difference in the structure of these groups is the presence of one additional double bond between the –COOH group and the aromatic ring [3,11–13]. The phenolic acids also differ in type, number, and position of the attached functional groups on the aromatic ring (–OH, methoxy (–OCH3)), and the research on their distribution is commonly used to find a relationship between structural features and compound activity, known as quantitative structure–activity relationship (QSAR) [14]. However, the knowledge of the mechanisms by which these molecules and their parts act in different reactions is limited. Scientific research indicated several factors with a possible impact on the mechanisms behind the compound's activity. Among them, the number and position of hydroxyl groups and their methylation, the distance between phenyl and carboxylic groups, and the concentration of the compound are suggested [15–21].

Like other phenolics, phenolic acids demonstrate different mechanisms of antioxidant action such as reduction of agents by hydrogen donation, quenching of singlet oxygen, or acting as chelators and trappers of free radicals, so usually, methods used to analyze their antioxidant activity are based on different mechanisms [3,14,22,23]. These methods may be generally classified as electron transfer (ET) and hydrogen atom transfer (HAT)-based assays [24]. The most accepted and widely used assays for the determination of antioxidant activity are Folin–Ciocalteu, FRAP (ferric reducing antioxidant power), ABTS/TEAC (2,2- -azinobis (3-ethylbenzothiazoline 6-sulfonate radical scavenging activity/Trolox equivalent antioxidant capacity), DPPH (2,2-diphenyl-1-picrylhydrazyl radical scavenging activity), and ORAC (oxygen radical absorbance capacity) [23–26]. Among these, Folin–Ciocalteu, FRAP, ABTS/TEAC, CUPRAC (cupric reducing antioxidant capacity), and DPPH methods are ET-based assays that provide information about reducing the capacity of an antioxidant, while the ORAC method is based on the HAT reaction mechanism. The FRAP method is often used to measure the reducing power of different samples and is considered one of the fastest, simplest, and less expensive methods, with reproducible results in a wide range of concentrations. On the other hand, the ORAC method uses a biologically relevant radical source (peroxyl radical), thus, the obtained activity could be used for interpreting activity in various biological systems [23,24,27]. For these reasons, the results obtained using different methods must be interpreted carefully, as due to differences in their mechanisms, the correlations between the obtained result often fail [23,26,28].

Although the antioxidant activity of phenolic acids is well studied using both in vitro and in vivo methods, the mechanisms of their action remain unclear and/or undefined [17–19]. An important factor that should be considered is their mutual interactions which can be synergistic, antagonistic, or additive (no interaction). Several studies aimed to investigate these interactions among phenolic acids using different antioxidant assays and confirmed both the occurrence of synergistic as well as antagonistic interactions [13,14,26,27,29–34]. The authors emphasized the influence of chemical structure and used concentrations on the overall activity of the tested mixtures. The efficiency of these interactions is also widely used to explain the activities of phenolic-rich extracts, where the dominant components cannot be identified as carriers of the total antioxidant activity [10,31,35].

In this regard, this study aimed to evaluate the antioxidant potential (reducing and free scavenging activity) of individual phenolic acids (protocatechuic, gentisic, gallic, vanillic, syringic, *p*-coumaric, caffeic, ferulic, sinapic, and rosmarinic) and their interactions in binary, ternary, quaternary, and quinary equimolar mixtures at different concentrations using FRAP and ORAC assays.

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

#### *2.1. Preparation of Standard Solutions and Model Mixture*

All used reagents and solvents were analytical or higher grade and purchased from Sigma (Sigma-Aldrich GmbH, Steinheim, Germany), Alkaloid AD (Skopje, North Macedonia), Merck (Darmstadt, Germany), Fluka (Buch, Switzerland), and Kemika (Zagreb, Croatia). The solutions of hydroxybenzoic acids (protocatechuic, gentisic, gallic, vanillic, and syringic) and hydroxycinnamic acids (*p*-coumaric, caffeic, ferulic, sinapic, and rosmarinic) (Sigma-Aldrich GmbH, Steinheim, Germany) shown in Table 1 were dissolved in an ethanol/water mixture (80:20, by volume) to the final concentration of 1000 μM. The experiment was divided in two parts. Firstly, all phenolic acids were individually tested for antioxidant activity at the concentrations of 2.5 and 5 μM in the ORAC assay, and 100, 500, and 1000 μM in the FRAP assay. Thereafter, the phenolic acids were mixed in binary, ternary, quaternary, and quinary equimolar combinations to reach the concentrations of 5 μM for ORAC and 100, 500, and 1000 μM for the FRAP assay.

**Table 1.** List and structural features of the investigated phenolic acids.


#### *2.2. Evaluation of the Antioxidant Activity*

#### 2.2.1. FRAP (Ferric Reducing Antioxidant Power) Assay

The reducing power of the samples detected with the FRAP method was measured according to the procedure described by Skroza et al. [36] and measurements were performed on a Tecan MicroPlate Reader, model Sunrise (Tecan Group Ltd., Männedorf, Switzerland). Analyses were completed in triplicates and results are expressed as μM of Fe2+.

#### 2.2.2. ORAC (Oxygen Radical Absorbance Capacity) Assay

Fluorimetric measurements in the ORAC assay were recorded on a microplate reader (Synergy HTX Multi-Mode Reader, BioTek Instruments, Inc., Winooski, VT, USA) following the procedure of Cagalj et al. [ ˇ 37]. The reaction was observed for 80 min and the results of three replicates are expressed in μM of Trolox Equivalents (μM TE).

#### *2.3. Interaction and Statistical Analysis*

The obtained results were analyzed using GraphPad Prism Version 4.03 for Windows (GraphPad Software, San Diego, CA, USA).

The interactions between phenolic acids were described as the difference in antioxidant activity between experimental and theoretical (calculated) values using the equation (Equation (1)) [36,38]:

$$\text{Difference (\%)} = \text{((Combination } ab \times 100) / (\text{Individual } a + \text{Individual } b)) - 100 \tag{1}$$

where combination *ab* is an experimentally obtained result for the binary mixture, while each *a*/*b* value was calculated individually for each compound. The theoretical values for each compound were calculated by dividing the experimental values by the number of compounds in the mixtures. Likewise, for ternary, quaternary, and quinary mixtures, the difference was calculated by subtracting the average of the individual three, four, or five compounds from the combination (Equations (2)–(4)):

$$Difference\left(\%\right) = \left( \left( \text{Combination } abc \times 100 \right) / \left( a+b+c \right) \right) - 100 \tag{2}$$

$$\text{Difference (\%)} = \text{((Combination abcd \times 100)/(a+b+c+d))} - 100\tag{3}$$

$$\text{Difference (\%)} = \{ (\text{Combination abcde} \times 100) / (a+b+c+d+e) \} - 100 \tag{4}$$

The obtained interactions (% difference) were used to determine potential synergistic (positive values, difference (%) > 0) or antagonistic (negative values, difference (%) < 0) effects. The additive effect was considered for the difference (%) ∼= 0 ± 5% when it can be considered that there was no interaction.

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

#### *3.1. Antioxidant Activity of Individual Phenolic Acids*

Previous studies confirmed that the number and arrangement of –OH and –OCH3 groups (their mutual position) affect the antioxidant activity of the phenolic acids [14,27,31,39], but also other parameters, such as ionization, dissociation, and the rate constants of radical scavenging, resonance, solvent solvation effects, intramolecular hydrogen bonds, bond dissociation enthalpy, etc., should also be taken in consideration as reported by Sroka [16], Hanscha at al. [15], Foti et al. [17] Lucarini and Pedulli [18], Hang et al. [21], and Biela et al. [20].

The antioxidant activities of individual phenolic acids at different concentrations are provided in Table 2. Among the hydroxybenzoic phenolic acids, gallic acid showed the highest FRAP value, while the lowest activity was detected for vanillic acid. This was observed for all tested concentrations and linear dependence was confirmed, as expected. The highest reducing power of gallic acid could be related to its chemical structure and three –OH groups located at positions 3, 4, and 5. The gentisic acid, with two –OH groups in *para*-position to each other (at positions 2 and 5), and protocatechuic acid with two –OH groups at positions 3 and 4 (catechol structure) showed similar FRAP values at low concentrations, while at higher concentrations gentisic acid was superior. Considering the molecular structure, the activity of gentisic acid, an active metabolite of salicylic acid degradation, was investigated by Mardani-Ghahfarokhi and Farhoosh [40], and the authors reported its higher activity in comparison to α-resorcylic acid with the same type and number of substituents, but in *meta*-position. Sroka and Cisowski [16] reported that acids with two –OH groups exhibited higher antioxidant activity than those possessing only one. Cuvelier et al. [41] concluded that the introduction of –OH in *para*- or *ortho*position enhances compound activity which increases also with the –OCH3 substitution in *ortho*-positions to –OH group. The –COOH group, as one of the most common electronwithdrawing groups (containing an atom with a positive charge directly attached to a benzene ring) with a relatively strong effect, can affect other substituents and it has been known that this effect is the strongest on –OH groups in the *para*-position, and weaker on those in the *meta*-position [20,42]. Biela et al. [20] reported that phenolic acids in aqueous solutions are mostly dissociated (fully deprotonated) to carboxylate anions so the key role in the activity is carried by their phenolic –OH groups. It has been reported that the carboxylic

substituent is a weak donor in the *meta*-position and has no effect in the *para*-position in comparison to the –COOH. Furthermore, the presence of other substituents, such as carboxylate anions, –OH and –OCH3 can also form intramolecular hydrogen bonds that can affect the activity. The authors concluded also that –OH and –OCH3 show an electron-donating effect in the *para*-position and an electron-withdrawing effect in the *meta*-position. Syringic acid, with two –OCH3 groups at positions 3 and 5 (*meta*-position), and one –OH at the *ortho*position, showed an increase in activity at higher concentrations reaching 3207 μM Fe2+ at 1000 μM, while vanillic acid, with an –OH group in the *para*-position, and –OCH3 group in the *meta*-position had the lowest FRAP value. Spiegel et al. [27] investigated 22 phenolic acids using an FRAP assay and reported that the main structural feature of good antioxidants are two or more –OH groups in *ortho*- and *para*-positions, but the importance of the inductive effect of the carboxylic group should not be neglected. Protocatechuic acid has two –OH groups, the same as gentisic acid, but it showed a lower reducing effect so it can be concluded that again, besides the number and arrangement of attached functional groups (in this case, *ortho*- or *para*-position), the intramolecular hydrogen bonds between two –OH groups also have a notable impact. Additionally, intramolecular hydrogen bonds between the –OH and –COOH groups may affect antioxidant activity. The obtained results showed that hydroxylation at positions 2 and 5 and one intramolecular hydrogen bond between them contributes to the reducing ability of the compound [17,18,27]. Rice-Evans et al. [43] reported that the insertion of an additional –OH group at position 2 of hydroxybenzoic acids decreases the overall antioxidant capacity, while Sroka and Cisowski [16] showed that the antioxidant activity of phenolic acids correlates with the number of –OH groups in *ortho*- and *para*-positions, but also reported the importance of the position of –COOH and acetyl group near the –OH groups. Foti et al. [17], except for position and number of substituents, also indicate the importance of resonance stabilization and intramolecular hydrogen bonds between them, while Biskup [11] reported that the functional group binding site and the type of substitute affect the activity. Spiegel et al. [27] observed that the position of the second –OH group affected the reducing capacity and that two or more –OH groups located either in the vicinal position or in the opposite position to each other resulted in higher antioxidant activity. They also explained differences in antioxidant activity of phenolic acids using resonance stabilization of radicals by intermolecular hydrogen bonds between functional groups and a polar solvent. The influence of the hydrogen bonds is also discussed by Foti et al. [17] where the authors reported that only compounds that are non-hydrogen-bonded (free) possess activity (electron transfer mechanism) and that the rate of reaction depends on the strength of the hydrogen bond as well as on the used solvent (methanol or ethanol).


**Table 2.** Antioxidant activity of individual phenolic acids at different concentrations using the FRAP and ORAC methods.

The FRAP values of the hydroxycinnamic acids also showed a linear correlation with concentration (Table 2). Similarly, FRAP values increased with the introduction of –OH and –OCH3 groups, which is in correlation with previous reports [14,27]. Among hydroxycinnamic acids, caffeic and rosmarinic acids had the highest FRAP values, while *p*-coumaric acid, with one –OH group, showed the lowest FRAP at all tested concentrations. Exceptionally good reducing power was also observed for sinapic and ferulic acids. In addition to one –OH group, these two acids also have –OCH3 groups, sinapic acid two and ferulic only one. The rosmarinic acid, an ester of caffeic acid and 3,4-dihydroxyphenyllactic acid (diphenolic compound), also stands out, with FRAP values 2-fold higher than those obtained by caffeic acid alone at 1000 μM. This can be also connected with its structure (two phenolic rings with two –OH groups in the *ortho*-position and an unsaturated double bond and –COOH between them). Cao et al. [44] investigated the antioxidant activity of this compound given its molecular structure and reported a stronger electron-donating capability of the B ring. According to the authors, its activity is a result of the H-abstraction reactions on both rings. Therefore, the activity of rosmarinic acid can be related to two phenolic groups in both rings. The good reducing power of caffeic acid can be related to the catechol structure and distance between the –COOH group and functional groups [14,16,17,27]. It has also been reported that hydroxybenzoic acids have lower antioxidant activity in comparison to hydroxycinnamic acids when they have the same substituents at positions 2–6 [14,20,27,41,43,45]. These observations are also confirmed in the present study where syringic acid showed higher reducing activity than sinapic acid, and ferulic acid was superior to vanillic acid. In the case of the ORAC method, the values for ferulic and vanillic acids at 2.5 μM were similar, but at higher concentrations better activity was detected for vanillic acid.

The results of antioxidant activity tested using the ORAC method at two concentrations (2.5 and 5 μM) are shown in Table 2. Among hydroxybenzoic acids, protocatechuic acid showed the highest activity, while gallic acid had the lowest activity. At higher concentrations, 21–55% higher activity was detected for all acids, with exception of the gallic acid which had 19% higher activity. Protocatechuic acid, with two –OH groups at positions 2 and 3, had higher activity than other acids with the same number of –OH groups at other positions (gentisic acid) or even –OCH3 groups (one in vanillic acid and two in syringic acid structure). Sroka and Cisowski [16] in their study also reported higher activity of protocatechuic acid against DPPH free radicals in comparison to the gentisic acid. Hang et al. [21] reported the influence of –OH and/or –OCH3 groups at position 3 and/or 5 on the hydrogen transfer mechanism (characterized by bond-dissociation energy) using the hydroperoxyl radical scavenging assay. They conclude that the intramolecular hydrogen bond between the –OH group at the *ortho*-position and the –COOH group could be the main reason for the highest reduction of bond-dissociation energy, which indicated weaker antioxidant activity (radical scavenging activity followed the order: syringic acid > gentisic acid > gallic acid > vanillic acid > protocatechuic acid). For higher antioxidant activity measured using ORAC, the location of –OH groups in vicinal or in the opposite position side of the ring seems to be a more important factor than only the number of –OH groups.

The ORAC test for hydroxycinnamic acids (Table 2) showed the lowest results for *p*-coumaric acid, while rosmarinic acid had the highest antioxidant effect at tested concentrations. This could be again related to its chemical structure and four –OH groups. *p*-coumaric acid was previously described as a weak antioxidant [29], what is also confirmed in this research where the obtained results showed that caffeic, ferulic, and sinapic acids had 2-fold higher activity than this compound. While *p*-coumaric acid has only one –OH group, other acids have additional –OH and/or –OCH3 groups. At the concentration of 5 μM, all acids showed approximately 45% higher values. Ferulic (one –OH and one –OCH3) and sinapic (one –OH and two –OCH3) acids showed almost the same activity, indicating that the additional –OCH3 group does not have a significant impact on the activity. Interestingly, caffeic acid with a catechol group at the same position as protocatechuic acid showed similar antioxidant activity as this hydroxybenzoic acid, in accordance with the

results of Sroka and Cisowski [16] who also confirmed a similar free radical scavenging activity. The authors pointed out that although these compounds have a similar model of substitution of the –OH and the incorporation of –CH2 between –COOH and the phenyl group does not increase the antiradical activity of 3,- and 4- substituted acids. This is also in accordance with the conclusions that the catechol group enhances the radical scavenging activity of the compound [26,27,29]. However, ferulic acid was not effective in scavenging peroxyl radicals such as vanillic acid with the same structural features (–OH and –OCH3 group), not in accordance with previous reports [26,27,29]. Additionally, greater distances between –COOH groups from the methoxylated ring do not enhance the antioxidant effect, as previous studies by Mathew et al. [19] and Spiegel et al. [27] suggested. Lucarini and Pedulli [18] reported the importance of the reaction medium. In their study on free radical scavenging activity of peroxyl radicals in autooxidation reactions, they reported the connection of bond dissociation enthalpies and rate constants with the antioxidant compound structure. The authors would like to point out once more the lack of systematic research on the antioxidant activity of phenolic acids using the ORAC method and investigations of their structure–activity relations.

#### *3.2. Antioxidant Activity of Equimolar Mixtures of Phenolic Acids*

The results of the antioxidant activity of the equimolar mixtures of two or more phenolic acids tested at different concentrations are shown in Tables 3–6. Based on the obtained data, the potential interaction (synergistic/additive/antagonistic) was determined and expressed as a percentage of the difference (%) between the experimental and theoretical (calculated) FRAP and ORAC values.

In the binary mixtures of hydroxybenzoic acids, at 100 μM, all mixtures containing gentisic acid showed a synergistic effect (28–89% difference). The mixture of protocatechuic and syringic acid showed an additive effect, while all others showed an antagonistic effect (up to −58% difference). At a concentration of 500 μM the synergic effect was observed only for the mixture of gentisic + syringic acids, while at 1000 μM the synergistic effect was not confirmed. Among ternary mixtures, it is interesting to highlight the mixture protocatechuic + gentisic + syringic acids, which showed the highest reducing power and the greatest synergistic effect (174% difference) at the lowest tested concentration. In comparison to the mixture of protocatechuic + gentisic acids and gentisic + syringic acids that also showed a high synergistic effect, and the mixture of these three phenolic acids showed a higher overall reducing capacity. When protocatechuic, gallic, and vanillic acids were combined they retained the antagonistic effect observed for mixtures of protocatechuic + gallic acid acids and protocatechuic + vanillic acids at all tested concentrations. Again, the mixtures containing either protocatechuic, gallic, and/or syringic acid along with gentisic acid show a synergistic effect leading to a conclusion that gentisic acid is a key component for the synergistic effect of these mixtures. However, the synergistic effect for protocatechuic + gentisic + vanillic acids and protocatechuic + gentisic + syringic acids observed at the concentration of 100 μM was confirmed also at 500 μM but not at a higher concentration where it was only detected for the protocatechuic + gentisic + gallic acids mixture. These results indicated that along with the compound ratio in the mixture, the compound concentration is also an important factor. The quaternary and quinary equimolar combinations showed a synergic effect only at the concentration of 100 μM, with the exception of the mixture of gallic + vanillic + syringic + protocatechuic acids which showed an antagonistic effect at all tested concentrations.



#### *Antioxidants* **2022** , *11*, 1784




**Table 5.** Comparison of theoretical and experimental ORAC values and the interaction of equimolar phenolic mixtures (% difference) of hydroxybenzoic acids at a concentration of 5 μM.

P—protocatechuic acid; Ge—gentisic acid; G—gallic acid; V—vanillic acid; Sy—syringic acid; *p*C—*p*-coumaric acid; C—caffeic acid; F—ferulic acid; Si—sinapic acid; R—rosmarinic acid; a difference (%) > 0 indicates a potential synergistic effect; a difference (%) < 0 shows an antagonistic and a difference (%) ∼= 0 or ± 5% shows an additive effect (no interaction).

The results of the reducing capacity for the mixtures of hydroxycinnamic acids are shown in Table 4. In contrast to the results for interactions obtained for hydroxybenzoic acids, a large number of tested mixtures of cinnamic acids showed lower antioxidant activity compared to the expected theoretical values, which indicated an antagonistic effect between these compounds. Among binary combinations, ferulic + sinapic acids showed the highest reducing power at 100 μM (501 μM Fe2+), while at other concentrations the mixture of *p*-coumaric + ferulic acids exhibited the lowest activity. At the concentration of 100 μM, the greatest positive difference between theoretical and expected FRAP values, indicating the higher synergistic effect, was observed for the following mixtures: *p*-coumaric + sinapic acids (72% difference) and ferulic + sinapic acids (102% difference), while other mixtures showed an antagonistic effect. With the increase in concentration, the number of mixtures showing a synergistic effect rose to five at 500 μM and nine at 1000 μM. At the first higher concentration, only the combination of ferulic + rosmarinic acids had an antagonistic effect while at 1000 μM they showed only an additive interaction. The addition of *p*-coumaric acid to the mixture of ferulic and rosmarinic acid, at both concentrations of 500 and 1000 μM, resulted in a synergistic effect of the mixture (127 and 48% difference, respectively) which was strange due to the low FRAP value of this compound and only one –OH group in its structure. In ternary mixtures, only *p*-coumaric + caffeic + rosmarinic acids showed a synergistic effect at all tested concentrations, but the decrease in the difference with the

increase in the concentration was recorded (127% at 100 μM > 48% at 500 μM > 11% at 1000 μM). The mixture of caffeic + sinapic + rosmarinic acids, with the highest number of –OH (seven) and –OCH3 (two) groups showed an antagonistic effect at 100 μM and a synergistic effect at 500 and 1000 μM. In mixtures of *p*-coumaric + caffeic + sinapic acids and caffeic + sinapic + rosmarinic acids, an antagonistic effect passed to synergistic at 1000 μM. Among quaternary and quinary mixtures, at 1000 μM only mixtures of caffeic + ferulic + sinapic + rosmarinic acids, mixtures of caffeic + sinapic + rosmarinic + *p*-coumaric acids and *p*-coumaric + caffeic + ferulic + sinapic + rosmarinic acids showed a weak synergistic effect.


**Table 6.** Comparison of theoretical and experimental ORAC values and the interaction of equimolar phenolic mixtures (% difference) of hydroxycinnamic acids at a concentration of 5 μM.

P—protocatechuic acid; Ge—gentisic acid; G—gallic acid; V—vanillic acid; Sy—syringic acid; *p*C—*p*-coumaric acid; C—caffeic acid; F—ferulic acid; Si—sinapic acid; R—rosmarinic acid; a difference (%) > 0 indicates a potential synergistic effect; a difference (%) < 0 shows an antagonistic and a difference (%) ∼= 0 or ± 5% shows an additive effect (no interaction).

Hajimehdipoor et al. [32] confirmed the synergistic effect of the caffeic + rosmarinic acids mixture (38% difference) in different binary combinations. The authors mixed binary combinations of hydroxybenzoic and hydroxycinnamic acids (gallic, rosmarinic, caffeic, and chlorogenic), alone and with flavonoids (quercetin and rutine) and reported that binary mixtures show stronger synergistic effects than their ternary combinations. Olszowy-Tomczyk [23] also reviewed the available information in the literature about interactions among compounds in the binary mixtures of phenolic acids with other phenolics (flavonoids, catechins, stilbenes, etc.). Differences between the experimental and theoretical values for antioxidant activity among phenolic acids were observed also in binary mixtures

of gallic + protocatechuic acids, gallic + vanillic acids [31], rosmarinic + caffeic acids [29], and gallic + caffeic acids [23].

The results of the theoretical and experimental ORAC values of phenolic mixtures, as well as their interactions, are presented in Tables 5 and 6. Only two mixtures of hydroxybenzoic acids showed an antagonistic effect, namely a mixture of gentisic + syringic acids (−24% difference) and gallic + vanillic acids (−30% difference), while all other mixtures had a synergistic effect. The antioxidant activities and synergistic effects of binary mixtures were higher in comparison to mixtures of three or more acids. The highest synergistic effect was observed for the mixture of gentisic + syringic acids with a total of two –OCH3 and three –OH groups while the mixture of gallic + syringic acids with the highest number of substituents, four –OH and two –OCH3 groups, showed an antagonistic effect. In the ORAC method, the presence of gentisic acid and protocatechuic acid resulted in a synergistic effect of the mixtures that contain these substances which could indicate that the presence of two –OH groups on the benzene ring (in *ortho*- or *para*-positions) is most likely responsible for this effect. In ternary, quaternary, and quinary mixtures, all combinations showed a synergistic effect except the mixture of protocatechuic + gallic + syringic acids which showed an additive effect. When gentisic acid was added to this mixture the effect was again synergistic.

The ORAC results obtained for the mixtures of hydroxycinnamic acids were lower and only a few mixtures showed the synergistic effect. The best antioxidant potential was confirmed for the mixture of *p*-coumaric + ferulic acids (160 μM TE) with a difference of 311% and a mixture of caffeic + sinapic acids (162 μM TE) with a difference of 211%. A positive interaction was observed also for the mixture of ferulic + rosmarinic acids (127%) and caffeic + rosmarinic acids (115%). Peyrat-Maillard et al. [29] also confirmed synergistic interaction between caffeic and rosmarinic acid at concentrations up to 5 μM by ORAC assay but in their study the concentration showed no effect on the interaction. In ternary, quaternary, and quinary mixtures of hydroxycinnamic acids only mixtures of *p*-coumaric + caffeic + ferulic acids, *p*-coumaric + caffeic + rosmarinic acids, and rosmarinic + *p*-coumaric + ferulic + sinapic acids showed a synergistic effect, while others showed a slight antagonistic or additive effect. In contrast to hydroxybenzoic acids, the number of the −OCH3 group in the structure of hydroxycinnamic acids (e.g., in ternary, quaternary, and quinary mixtures with ferulic or sinapic acids) cannot be related to their higher antioxidant activity.

Palafox Carlos et al. [31] suggested that gallic, protocatechuic, and vanillic acids interact in a synergic way. Using a DPPH assay, the authors also confirmed the synergistic effect of the gallic and protocatechuic acid mixture relating this effect to the chemical structure of the compounds and the presence of the hydroxyl group. On the other hand, in their study, the mixture of protocatechuic and vanillic acid showed an antagonistic effect. Some authors suggested that interactions are concentration-related, rather than structurerelated [39] or that presence or absence of the catechol group in the chemical structure of the compounds from the mixtures contributes to their synergic effect [13,46]. They investigated the interaction effect between caffeic, ferulic, and rosmarinic acid at different concentrations (50, 100, 200, and 250 μM) using the Briggs–Rauscher assay and reported the synergistic effect of the mixtures at concentrations ranging from 50 to 200 μM, and strong antagonism at 250 μM. The authors concluded that the antioxidant activity depends on compound structure (number and distribution of substituents) and concentration, which was opposite to some of the results obtained in the present study.

#### **4. Conclusions**

The results indicated that differences in antioxidant activity of the tested phenolic acids depend on their structure, as expected, regarding not only the type, number, and arrangement of substituents but also the compound concentration. The additional number of the –OCH3 groups in the same positions in the phenolic ring in the hydroxybenzoic acids resulted in higher activity in comparison to the hydroxycinnamic acids with the same structural features. Among individual hydroxybenzoic acids, gallic acid showed the highest reducing activity, while the lowest activity was recorded with the ORAC assay. Among hydroxycinnamic acids the *p*-coumaric acid showed the lowest activity, using both methods at all tested concentrations. In the mixtures, synergistic effects were detected in several combinations, but special attention should be devoted to hydroxybenzoic acid mixtures containing gentisic acid, especially at lower concentrations where in all cases the positive differences were calculated. Similarly, in the ORAC method the presence of gentisic acid resulted in a synergistic effect of the mixtures, while low activity of the gallic acid obviously influenced the overall mixture activity as lower antioxidant or antagonistic effects are detected. Furthermore, it is obvious that other parameters such as the applied antioxidant method and solvent medium, the position of functional groups in relation to the –COOH group and other groups attached to the ring, ionization and bond dissociation enthalpies, intramolecular hydrogen bonding, etc., that were discussed but not investigated in this study, should be taken in consideration in further studies since they might have an impact on the overall antioxidant activity of the compounds and their mixtures.

**Author Contributions:** Conceptualization: D.S., V.Š. and I.G.M.; methodology and formal analysis: L.V., N.J., A.S., M.C. and R.F.; data curation: D.S. and I.G.M.; supervision: D.S. and I.G.M.; writing— ˇ original draft: D.S., V.Š. and I.G.M.; writing—review and editing: D.S., V.Š., I.G.M., M.C. and R.F. All ˇ authors have read and agreed to the published version of the manuscript.

**Funding:** This research is supported by Croatian Science Foundation (grant number IP-2014-09-6897) and the PRIMA program under project BioProMedFood (Project ID 1467). The PRIMA program is supported by the European Union.

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

**Informed Consent Statement:** Not applicable.

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

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

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland www.mdpi.com

*Antioxidants* Editorial Office E-mail: antioxidants@mdpi.com www.mdpi.com/journal/antioxidants

Disclaimer/Publisher's Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Academic Open Access Publishing

www.mdpi.com ISBN 978-3-0365-8605-2