Comparison of Various Solvent Extracts and Major Bioactive Components from Portulaca oleracea for Antioxidant, Anti-Tyrosinase, and Anti-α-Glucosidase Activities

Portulaca oleracea is a well-known species for traditional medicine and food homology in Taiwan. In traditional medicine, P. oleracea is also used to treat gastrointestinal disorders, liver inflammation, fever, severe inflammation, and headaches. We investigated antioxidant, anti-tyrosinase, and anti-α-glucosidase activities of various solvent extracts and major bioactive components from P. oleracea. Ethanol and acetone extracts showed potent DPPH, ABTS, and hydroxyl radical scavenging activities. Chloroform and n-hexane extracts displayed significant superoxide radical scavenging activity. Furthermore, ethyl acetate and acetone extracts of P. oleracea showed potent anti-tyrosinase and anti-α-glucosidase activities. Examined and compared to the various solvent extracts for their chemical compositions using HPLC analysis, we isolated seven major compounds and analyzed their antioxidant, anti-tyrosinase, and anti-α-glucosidase activities. Seven active compounds of P. oleracea, especially quercetin, rosmarinic acid, and kaempferol, exhibited obvious antioxidant, anti-tyrosinase, and anti-α-glucosidase activities. The molecular docking model and the hydrophilic interactive mode of tyrosinase and α-glucosidase revealed that active compounds might have a higher antagonistic effect than commonly inhibitors. Our result shows that the active solvent extracts and their components of P. oleracea have the potential as natural antioxidants, tyrosinase and α-glucosidase inhibitors. Our results suggest that the active solvent extracts of P. oleracea and their components have potential as natural antioxidants, tyrosinase and α-glucosidase inhibitors.


Reverse-Phase HPLC
Reversed-phase separations were performed using a LiChrospher ® 100 RP-18 Endcapped (5 µm; column of dimensions 4.6 × 250 mm) purchased from Merck KGaA, Darmstadt, Germany. HPLC-PDA chromatographic fingerprints were obtained with an Agilent 1260 Infinity II HPLC instrument equipped with a 1260 Infinity II quaternary pump, a 1260 Infinity II degasser, a 1260 Infinity II vialsampler, a 1260 Infinity II column thermostat, a 1260 Infinity II diode array detector HS, and a PC with the Agilent ChemStation software. All of them were purchased from Agilent Technologies (Waldbronn, Germany). Gradient separation using 0.2% acetic acid in water (v/v) (solvent A) and acetonitrile (solvent B) as mobile phase was as follows: 0-15 min, linear gradient from 95 to 88% A; 15-35 min, 88% A with isocratic elution; 35-55 min, 75% A with isocratic elution; 55-65 min, 60% A with isocratic elution; 65-90 min, 40% A with isocratic elution; 90-95 min, linear gradient from 40 to 0% A; 95-100 min, back to initial condition at 95% A; and 100-105 min, at 95% A. The flow rate was 1.0 mL/min, and the injection volume was 500 µL. Peaks were detected at 280 nm. Different compounds were identified by retention time. To guarantee peak purity, DAD acquisition from 200-650 nm was conducted to register UV-spectra. For the quantitative analysis of seven compounds in the extracts, aliquots of samples were dispersed in 10 mL of a methanol solution by sonication for 5 min. Following this, the samples were centrifuged for 15 min at 3500 rpm, and the supernatant extracts were filtered through 0.45 µm PTFE syringe filters (Zhejiang Sorfa Medical Plastic Co., Ningbo, China). Quantification of seven components from Portulaca oleracea in each solvent extract was described as above.

Determination of Total Phenolic Content
The total phenolic content (TPC) of various solvent extracts was measured as described with Folin-Ciocalteu's method with a slight modification [18].

Determination of Total Flavonoid Content
The total flavonoid content (TFC) of different solvent extracts was determined with slight modification of the aluminum chloride colorimetric method described by Chang et al. [19]. In brief, the extract sample was dissolved with ethanol to 10 mg/mL. The calibration curve was prepared by diluting quercetin in methanol (0-100 mg/mL). The dissolved extract or quercetin (2.0 mL) was mixed with 0.1 mL of 10% (w/v) aluminum chloride solution and 0.1 mL of 0.1 mM potassium acetate solution. The mixture was kept at room temperature (25 • C) for 30 min. Then the maximum absorbance of the mixture was measured at 415 nm using a UV-VIS spectrophotometer. The TFC was expressed as milligram quercetin equivalent per gram of dry extract (mg QCE/g dry extract).

DPPH Radical Scavenging Activity
The DPPH radical scavenging assay was determined by the reference method [18].

ABTS Cation Radical Scavenging Activity
The ABTS cation radical scavenging assay was carried out using the reference method [18].

Superoxide Anion Radical Scavenging Activity
Superoxide anion radical (O 2 •− ) scavenging activity was described earlier with slight changes [18]. Briefly, the superoxide radical was generated in 16 mM Tris-HCl buffer (pH 8.0) containing 50 µL of NBT (300 µM), 50 µL of PMS (120 µM), and 50 µL of different concentrations of test sample. The reaction was initiated by adding 50 µL of NADH (468 µM) solution to the mixture. After incubating at room temperature for 5 min, the activity of samples was determined by calculating the decrease in absorbance measured at 560 nm by the following equation, Superoxide anion radical scavenging activity (%) = (A 0 − A 1 )/A 0 × 100, where A 1 is the absorbance of the test sample and A 0 is the absorbance of the control.
IC 50 values for the tested activities were determined by linear regression of the percentage of remaining superoxide anion radicals against sample concentration.

Hydroxyl Radical Scavenging Activity
The samples were evaluated for hydroxyl radical scavenging activity based on the previously described method [18].

Anti-Tyrosinase Activity Assay
The substrate was premixed with the sample at various concentrations in potassium phosphate buffer (50 mM, pH 6.5). After incubating at room temperature for 10 min, 20 µL of mushroom tyrosinase (1000 units/mL) was added, and the mixtures were incubated for another 30 min. Absorbance was measured at 490 nm and the percentage of anti-tyrosinase activity of sample was calculated from, Inhibition (%) = (A 0 − A 1 )/A 0 × 100, where A 1 is the absorbance of the test sample and A 0 is the absorbance of the control. Arbutin was used as positive control.

Anti-α-Glucosidase Activity Assay
α-Glucosidase inhibitory activity was established in accordance with the described method [18] with slight modifications. In brief, 100 µL of different concentrations of the sample, 380 µL of p-nitro-phenyl-α-D-glucopyranoside (p-NPG) (0.53 mM), and 20 µL of α-Glucosidase solution (1 units/mL) were mixed in 0.1 M potassium phosphate buffer (pH 6.8). After incubating at 37 • C for 20 min, 500 µL of sodium carbonate (0.1 M) was added in order to quench the reaction. The anti-α-glucosidase activity of the sample was determined by calculating the decrease in absorbance measured at 405 nm by the following equation, Inhibition (%) = (A 0 − A 1 )/A 0 × 100, where A 1 is the absorbance of the test sample and A 0 is the absorbance of the control. Acarbose was used as positive control.

Molecular Modeling Docking Study
All calculations were performed by Discovery Studio 2019 (San Diego, CA, USA) software. Primarily, the structure of the docked compound is energy minimized until the default derivative convergence criterion of 0.01 kcal/mol is met. The crystal structure (PDB: 3A4A or 2Y9X) is retrieved from the Protein Databank and hydrogen atoms are added to prepare the docked receptor. This protein structure is subsequently used in the CDocker program to dock the compound into the active site. Ten different docking poses are calculated and ranked by using the PLP score scoring function. The top-ranked docking solution is visually analyzed to determine the binding mode of the docked compound.

Statistical Analysis
All data are expressed as mean ± SD. Statistical analysis was carried out using the Mann-Whitney U test. A probability of 0.05 or less was considered statistically significant. All the experiments were performed at least 3 times.
The comparative evaluation of the total phenolic content (TPC) and total flavonoid content (TFC) of various solvent extracts (n-hexane, chloroform, dichloromethane, ethyl acetate, acetone, ethanol, and methanol) from the whole plant of P. oleracea is first conducted in our study. This can provide a guide for the selection of appropriate solvents in TPC and TFC extraction applications.
Comparative evaluation of antioxidant assays (DPPH, ABTS, superoxide, and hydroxyl radical scavenging) of various solvent extracts (n-hexane, chloroform, dichloromethane, ethyl acetate, acetone, ethanol, and methanol) from the whole plant of P. oleracea is first proposed in our study. This can provide an indication for the selection of appropriate solvents in antioxidant extraction applications.
The comparative assessment of anti-tyrosinase and anti-α-glucosidase assays of various solvent extracts (n-hexane, chloroform, dichloromethane, ethyl acetate, acetone, ethanol, and methanol) from the whole plant of P. oleracea is first conducted in our study. This can provide an indication for the selection of appropriate solvents in natural tyrosinase and α-glucosidase inhibitors extraction applications.

Quantitation of Active Components in Different Solvent Extracts
The HPLC methods using reverse-phase column for the quantification of seven components isolated from P. oleracea were verified regarding linearity, LOD, and LOQ. The linearity was validated by the data from six different concentrations (1.0, 5.0, 10.0, 25.0, 50.0, and 100.0 µg/mL) of the standard solutions. The linear regression parameters of calibration curves, correlation coefficient, LOD, and LOQ were shown in Table S1. Six concentrations of each standard were analyzed in triplicate to generate respective calibration curve. The linearity (R 2 > 0.9993) between Y (the peak area of the analytes with external standard) and X (concentration of the standards) was achieved in the tested stage.
Figures S1-S8 displayed the quantification of active components in different solvent extracts from P. oleracea by reverse-phase HPLC analyses. The contents of seven compounds in each solvent extract were shown in Table 4. Total quantities of seven compounds in each extract ranged from a maximum of 50.90 ± 1.55 mg/kg (EtOH extract) to a minimum of 10.85 ± 0.66 mg/kg (EtOAc extract) in succeeding order of ethanol > chloroform > methanol > n-hexane > dichloromethane > acetone > ethyl acetate. The EtOH extract exhibited more of the seven active components than the other extracts. Among the seven active compounds in the organic solvent extract, kaempferol was the most abundant, followed by quercetin, trans-ferulic acid, chlorogenic acid, p-coumaric acid, caffeic acid, and rosmarinic acid ( Figure 1).

Antioxidant Activities of Isolated Components
Compounds isolated from P. oleracea were measured for their antioxidant activities, including DPPH, ABTS, superoxide, and hydroxyl radical scavenging assays. The results are shown in Table 5. Except for p-coumaric acid, most of the compounds have much higher DPPH free-radical scavenging activity than the positive control BHT (IC50 = 31.04 ± 2.12 μg/mL). Furthermore, all seven compounds showed high ABTS radical scavenging
In conclusion, all seven compounds isolated from P. oleracea exhibited good antioxidant activity against DPPH, ABTS, superoxide, and hydroxyl radical scavenging. The comparative evaluation of antioxidant assays of active compounds from the whole plant of P. oleracea is first conducted in our study.

Molecular Modeling Docking Study
According to the experimental data (Table 6), the most potent compounds, rosmarinic acid and kaempferol, were selected to determine their binding abilities to the crystal structures of tyrosinase and α-glucosidase, respectively, by molecular docking. The IC 50 value was defined as half-maximal inhibitory concentration, and was expressed as mean ± SD (n = 3); b Arbutin and acarbose were used as positive controls; * p < 0.05, ** p < 0.01, and *** p < 0.001 compared with the control.
The 3D crystal structure (PDB: 2Y9X) of tyrosinase from mushroom complexed with tropolone, an inhibitor for tyrosinase, shows that the substrate binding site of tyrosinase is primarily formed by five α-helices as well as several loops, and it mainly contains hydrophilic residues such as six histidines, His 61, 85, 94, 259, 263 and 296, which interact with two copper ions [21]. Once tropolone enters the substrate binding pocket, it is surrounded by some hydrophilic and hydrophobic residues including six histidines, Cys 83, Phe 90, Glu 256, Asn 260, Phe 264, Met 280 and Val 283. Tropolone enters the substrate binding pocket by locating its cycloheptyl core scaffold in the middle of the pocket. It mainly interacts with the pocket by its 1-keto group acting as the H-bond acceptor to interact with His 61 and His 263. Additionally, the cyclohepta-2,4,6-triene scaffold of tropolone makes an essential hydrophobic π-π interaction with His 263. These interactions result in tropolone's antagonistic activity against tyrosinase.
To understand how rosmarinic acid (Figure 2b) might make interaction with tyrosinase from mushroom to exert its antagonistic effect, the docking models of rosmarinic acid was generated by the Discovery Studio 2019 (Accelrys, San Diego, CA, USA) CDocker modeling program. The 3D crystal structure (PDB: 2Y9X) of tyrosinase from mushroom was utilized to perform the docking study. In the Figure 3, the docking model of rosmarinic acid showed that rosmarinic acid resided in the substrate binding pocket by locating its 3 ,4dihydroxyphenylacetic acid moiety to the middle of the pocket as the core scaffold of tropolone, and leaning its A ring toward to the right side of the pocket. By this orientation, rosmarinic acid not only bound to the pocket in the similar manner as tropolone but also exhibited additional interaction with the pocket. As shown in the Figure 3a, the essential interactions rosmarinic acid exerting included (1)  As the positive control, the docking model for arbutin (Figure 2a) is also generated to compare its binding mode with rosmarinic acid's. Arbutin bound to the catalytic site in the similar manner as tropolone by leaning its phenolic ring toward to the tropolone binding position. In the substrate binding pocket, arbutin made three important interactions including (1) the 4-hydroxyl group on the phenolic ring interacted with His 85 by acting as the H-bond donor; (2) the phenolic aromatic ring made the π-π interaction with His 263; (3) the 2 -hydroxymethyl group on the glycosyl ring interacted with Asn 280 by acting as the H-bond acceptor. This model also supported the importance of arbutin's glycosyl ring due to its contribution to the binding affinity, since deoxyarbutin has been shown to have less binding affinity than arbutin. hydroxyl group on the B ring interacted with the backbone of Val 283 by acting as the Hbond donor; (2) the 4′-hydroxyl group on the B ring interacted with His 263 by acting as the H-bond acceptor; (3) the phenyl B ring of rosmarinic acid made the π-π interaction with His 263; (4) the oxygen atom on the 9′-carboxylate group acted as the H-bond acceptor to interact with the side chain of Asn 260; (5) the 4-hydroxyl group on the A ring served as the H-bond donor to contact with the backbone of Ala 323.  As the positive control, the docking model for arbutin (Figure 2a) is also generated to compare its binding mode with rosmarinic acid's. Arbutin bound to the catalytic site in the similar manner as tropolone by leaning its phenolic ring toward to the tropolone binding position. In the substrate binding pocket, arbutin made three important hydroxyl group on the B ring interacted with the backbone of Val 283 by acting as the Hbond donor; (2) the 4′-hydroxyl group on the B ring interacted with His 263 by acting as the H-bond acceptor; (3) the phenyl B ring of rosmarinic acid made the π-π interaction with His 263; (4) the oxygen atom on the 9′-carboxylate group acted as the H-bond acceptor to interact with the side chain of Asn 260; (5) the 4-hydroxyl group on the A ring served as the H-bond donor to contact with the backbone of Ala 323.  As the positive control, the docking model for arbutin (Figure 2a) is also generated to compare its binding mode with rosmarinic acid's. Arbutin bound to the catalytic site in the similar manner as tropolone by leaning its phenolic ring toward to the tropolone binding position. In the substrate binding pocket, arbutin made three important The 3 ,4 -dihydroxyphenylacetic acid moiety of rosmarinic acid and the phenolic group of arbutin both can locate at the tropolone binding position of tyrosinase's active site to make the same π-π interaction with His 263 as tropolone. Furthermore, the phenolic group and the glucosyl group of arbutin make two H-bond interactions with His 85 and Asn 260. However, the 3 ,4 -dihydroxyphenylacetic acid moiety of rosmarinic acid makes three H-bond interactions with Asn 260, His 263 and Val 283, so rosmarinic acid exerts more hydrophilic interaction than arbutin in the tropolone binding site. Additionally, the caffeic acid moiety of rosmarinic acid makes one H-bond interaction with Ala 323 and can exert weak interaction with the loop of the substrate binding pocket to enhance the binding affinity by its ethenyl group with His 85 and 3-hydroxyl group with Asn 81.
Based on the results mentioned above, it is highly suggested that rosmarinic acid should have better antagonistic activity than arbutin.
The 3D crystal structure of α-glucosidase complexed with acarbose showed that it mainly contains numerous structural domains including N-terminal domain, the barrel domain in which the active site is located and C-terminal domain. The α-glucosidase substrate binding site is primarily formed by numerous β-sheets and several loops or α-helices. The crystal structure of α-glucosidase from Ruminococcus obeum complexed with voglibose (PDB: 6C9X) reveals that its active site is formed by six β-sheets, two loops and one α-helice. Once voglibose enters the substrate binding pocket, it is surrounded by some hydrophilic and hydrophobic residues including Pro 75, To further study the interaction between kaempferol (Figure 4b) and the α-glucosidase of Saccharomyces cerevisiae and to try to interpret how kaempferol might exert its antagonistic effect, the docking model of kaempferol was generated by the Discovery Studio 2019 (Accelrys, San Diego, USA) CDocker modeling program. The crystal structure of α-glucosidase from Saccharomyces cerevisiae is not available now, so the crystal structure (PDB: 3A4A) of Saccharomyces cerevisiae containing 72% sequence homology with the α-glucosidase from Saccharomyces cerevisiae is usually used to perform the docking study and also employed in this study [22]. In the crystal structure (PDB: 3A4A), the configuration of its substrate binding site is quite similar to that of the α-glucosidase from Ruminococcus obeum but it is deep and narrow. In the active site, its co-crystalized ligand, α-D-glucopyranose, locates deeply in the substrate binding pocket and makes three essential H-bond interactions including when (1)    As shown in Figure 5, the docking model of kaempferol indicated that kaempfero did not enter the active site in the flat conformation due to the narrow entrance Alternatively, the A ring of kaempferol leaned toward to the binding position of α-D glucopyranose. Importantly, kaempferol made some significant hydrophilic interaction including when (1) the 5-hydroxyl group on the A ring served as the H-bond donor t interact with Asp 69 and acted as the H-bond acceptor to make contact with Arg 442; (2 7-hydroxyl groups on the A ring behaved as the H-bond donor to contact with Asp 215 (3) the 3-hydroxyl group on the B ring served as the H-bond donor to interact with Glu 411; (4) the 4'-hydroxyl groups on the C ring acted as the H-bond donor to interact with Asp 307. Apart from the hydrophilic interaction, kaempferol also made the importan hydrophobic contact, the π-π interaction between its A ring and Phe 178. As shown in Figure 5, the docking model of kaempferol indicated that kaempferol did not enter the active site in the flat conformation due to the narrow entrance. Alternatively, the A ring of kaempferol leaned toward to the binding position of α-D-glucopyranose. Importantly, kaempferol made some significant hydrophilic interactions including when (1) the 5-hydroxyl group on the A ring served as the H-bond donor to interact with Asp 69 and acted as the H-bond acceptor to make contact with Arg 442; (2) 7-hydroxyl groups on the A ring behaved as the H-bond donor to contact with Asp 215; (3) the 3-hydroxyl group on the B ring served as the H-bond donor to interact with Glu 411; (4) the 4'-hydroxyl groups on the C ring acted as the H-bond donor to interact with Asp 307. Apart from the hydrophilic interaction, kaempferol also made the important hydrophobic contact, the π-π interaction between its A ring and Phe 178.
including when (1) the 5-hydroxyl group on the A ring served as the H-bond donor to interact with Asp 69 and acted as the H-bond acceptor to make contact with Arg 442; (2) 7-hydroxyl groups on the A ring behaved as the H-bond donor to contact with Asp 215; (3) the 3-hydroxyl group on the B ring served as the H-bond donor to interact with Glu 411; (4) the 4'-hydroxyl groups on the C ring acted as the H-bond donor to interact with Asp 307. Apart from the hydrophilic interaction, kaempferol also made the important hydrophobic contact, the π-π interaction between its A ring and Phe 178. As the positive control, the docking model for acarbose (Figure 4a) is also generated to compare its binding mode with kaempferol's. Acarbose bound to the catalytic site in the similar manner as kaempferol by leaning its A ring toward to the binding position of As the positive control, the docking model for acarbose (Figure 4a) is also generated to compare its binding mode with kaempferol's. Acarbose bound to the catalytic site in the similar manner as kaempferol by leaning its A ring toward to the binding position of α-D-glucopyranose and located its D ring to protrude out of the entrance of the active site. Unlike acarbose, kaempferol can reside in the middle of the substrate binding pocket and occupy the whole pocket. Especially when kaempferol makes four H-bond interactions with the residues on different secondary structures in the pocket to allow it to stay long enough for exerting its antagonistic effect. For acarbose, its A ring can locate at the similar position as the A ring of kaempferol but its D ring partially protrudes out of the substrate binding pocket. The binding mode of acarbose shows that acarbose highly replies on its A ring to make contact with the residues nearby the substrate binding site. However, the B and C rings of acarbose exert only one hydrophilic interaction. More importantly, the B and C rings of acarbose are surrounded by the hydrophobic residues including Tyr 158, Phe 159 and Phe 178, so their numerous hydroxyl moieties might not be able to exhibit significant hydrophobic interaction as the A ring of kaempferol. Furthermore, the D ring of acarbose protrudes out of the pocket and exerts its interactions with the residues outside the pocket. These interactions out of the substrate binding pocket might not contribute significant binding affinity to the antagonistic effect. All the results mentioned above highly indicate that kaempferol should be able to exert more antagonistic effect than acarbose. However, to generate the crystal structure of the α-glucosidase from Saccharomyces cerevisiae complexed with kaempferol or acarbose, further research is needed to further establish the nature of their respective interaction.

Discussion
Several methods have been developed and utilized to extract natural products, i.e., plants, fungi and herbs, as an alternative to modern medicines. The most traditional method applied for many years is to boil or make a decoction with water, which is a very simple and economical procedure. Today, studies based on medicinal plants have been published and many studies are ongoing as many metabolites with health benefits are found in natural products. More recently, in more advanced studies, organic solvents have been used to obtain natural product extracts including various metabolites depending on the polarity and property of the components of interest [23]. Other factors that may affect the natural product extraction process include the type of solvent to be used, the temperature set during extraction, the property of plant material, and the target metabolites [24]. Changes in solvent polarity lead to significant differences in phytochemical composition and biological activity. Therefore, we utilized solvents of various polarities to extract the whole plant of P. oleracea in an effort to evaluate these different metabolites. We found that various metabolites have different degrees of biological activity due to differences in solvent polarity.
DPPH and ABTS radical scavenging assays have been widely utilized to assess the antioxidant activities of natural components. Both assays are mainly related to the proton radical scavenging or hydrogen donating ability of the target compound [25]. Superoxide radical scavenging activity is measured by the PMS-NADH-NBT system. Superoxide anion radicals generated from dissolved oxygen by the PMS/NADH coupling reaction reduce NBT. The decrease of absorbance at 560 nm with antioxidants indicates the reduction of superoxide anion radicals in the reaction mixture [26]. The ferric reducing antioxidant According to the above data, the docking scores of rosmarinic acid, kaempferol, and quercetin were higher than those of arbutin and acarbose, indicating their better binding capability. In this study, the active ingredients, rosmarinic acid, kaempferol and quercetin, possessed not only anti-tyrosinase and anti-α-glucosidase activity, but also the better binding potential with the active sites of A. bisporus tyrosinase and S. cerevisiae αglucosidase. This indicated that these compounds may deserve further investigation as natural tyrosinase and α-glucosidase inhibitors.

Discussion
Several methods have been developed and utilized to extract natural products, i.e., plants, fungi and herbs, as an alternative to modern medicines. The most traditional method applied for many years is to boil or make a decoction with water, which is a very simple and economical procedure. Today, studies based on medicinal plants have been published and many studies are ongoing as many metabolites with health benefits are found in natural products. More recently, in more advanced studies, organic solvents have been used to obtain natural product extracts including various metabolites depending on the polarity and property of the components of interest [23]. Other factors that may affect the natural product extraction process include the type of solvent to be used, the tempera-ture set during extraction, the property of plant material, and the target metabolites [24]. Changes in solvent polarity lead to significant differences in phytochemical composition and biological activity. Therefore, we utilized solvents of various polarities to extract the whole plant of P. oleracea in an effort to evaluate these different metabolites. We found that various metabolites have different degrees of biological activity due to differences in solvent polarity.
DPPH and ABTS radical scavenging assays have been widely utilized to assess the antioxidant activities of natural components. Both assays are mainly related to the proton radical scavenging or hydrogen donating ability of the target compound [25]. Superoxide radical scavenging activity is measured by the PMS-NADH-NBT system. Superoxide anion radicals generated from dissolved oxygen by the PMS/NADH coupling reaction reduce NBT. The decrease of absorbance at 560 nm with antioxidants indicates the reduction of superoxide anion radicals in the reaction mixture [26]. The ferric reducing antioxidant power (FRAP) measures the antioxidant potential of each extract through the reduction of ferric iron (Fe 3+ ) complex to ferrous iron (Fe 2+ ) complex by antioxidants present in the samples [27]. In our study, the ethanolic extract of the whole plant of P. oleracea showed high antioxidant activities among all solvent extracts via DPPH and hydroxyl radical scavenging assays. The differences in antioxidant capacities of the extracts may be owing to the different extents of TPC or the composition of antioxidant compounds in the extracts.
The comparative evaluation of the total phenolic content (TPC), total flavonoid content (TFC), and antioxidant assays (DPPH, ABTS, superoxide, and hydroxyl radical) of various solvent extracts (n-hexane, chloroform, dichloromethane, EtOAc, acetone, EtOH, and MeOH) from the whole plant of P. oleracea is first mentioned in this study. This can provide a guide for the selection of appropriate solvents in TPC, TFC, and antioxidant extraction applications. According to the antioxidant data, rosmarinic acid and quercetin displayed potent antioxidant properties. In addition, the contents of rosmarinic acid and quercetin in the ethanolic extract were the highest among all solvent extracts. This was consistent with the result that the ethanolic extract has high antioxidant activity.
Tyrosinase is an enzyme mainly involved in the biosynthesis of melanin and catalyzes melanin biosynthesis in human skin, which causes diseases of epidermal hyperpigmentation [10,11]. Anti-tyrosinase agent could suppress the activity of tyrosinase. In addition, antioxidants can prevent or delay pigmentation by different mechanisms, e.g., by scavenging ROS and RNS, or by reducing o-quinones or other intermediates in melanin biosynthesis, thus delaying oxidative polymerization [28]. Based on our data, rosmarinic acid showed strong antioxidant and anti-tyrosinase activities and was supposed to possess some potential in treatment of diseases related to melanin.
Anti-α-glucosidase agent could suppress the activity of glucosidases in small intestines which split the glycosidic bonds in carbohydrate so that it decreases the glucose release from food. The inhibitors studied were classified into non-sugar and sugar-mimicking types on the basis of their chemical structure. The anti-α-glucosidase drugs for clinical treatment such as voglibose, acarbose, and miglitol all belong to the sugar-mimicking type. Nevertheless, the α-glucosidase inhibitors with the non-sugar type have received the attention of investigators due to the limitations of sugar-mimicking inhibitors. According to the results of an anti-α-glucosidase assay, kaempferol exhibited the most potent antiα-glucosidase activity among all isolated compounds. Thus, the interaction between α-glucosidase and kaempferol was evaluated by molecular modeling docking. As the result of molecular docking, kaempferol exhibited high affinity with α-glucosidase. Similar experimental results could also be found in past studies [29].

Conclusions
Various solvent extracts of P. oleracea were investigated with various antioxidant systems, anti-tyrosinase and anti-α-glucosidase activity assays. In our study, the ethanol extract of P. oleracea displayed the highest total phenol contents (TPC) and relatively high total flavonoid contents (TFC). The overall antioxidant capacity of the ethanolic extract of P. oleracea was superior to that of the other solvent extracts, consistent with the results of TPC and TFC assay. Furthermore, the acetone extract of P. oleracea showed the highest antiα-glucosidase activity among all solvent extracts. EtOAc and acetone extracts of P. oleracea showed more potent anti-tyrosinase activity than other solvent extracts.
Biological activity analysis showed that these seven compounds had potent antioxidant activity in general on the scavenging of DPPH, ABTS, superoxide and hydroxyl radicals, especially rosmarinic acid, kaempferol and quercetin. Kaempferol and quercetin showed strong anti-α-glucosidase activity. Rosmarinic acid, quercetin, and p-coumaric acid had a potent inhibitory effect against tyrosinase. Further molecular model docking analysis also confirmed their binding sites and binding abilities.
In conclusion, this study demonstrated that extraction solvent for P. oleracea affects total phenolic and flavonoid contents, antioxidant activities, and bioactive component levels. Ethanol is the most suitable solvent for extracting the effective components of P. oleracea among all solvents. The ethanol extract and its active components (especially rosmarinic acid, kaempferol and quercetin) of P. oleracea can be used as natural antioxidant, anti-tyrosinase, and anti-α-glucosidase agents.