In Vitro Synergistic Antioxidant Activity and Identification of Antioxidant Components from Astragalus membranaceus and Paeonia lactiflora

Many traditionally used herbs demonstrate significantly better pharmacological effects when used in combination than when used alone. However, the mechanism underlying this synergism is still poorly understood. This study aimed to investigate the synergistic antioxidant activity of Astragalus membranaceus (AME) and Paeonia Lactiflora (PL), and identify the potential antioxidant components by 1,1-diphenyl-2-picrylhydrazine (DPPH) radical spiking test followed by a high performance liquid chromatography separation combined with diode array detection and tandem mass spectrometry analysis (DPPH-HPLC-DAD-MS/MS). Eight AME-PL combined extracts (E1–E8) were prepared based on bioactivity-guided fractionation. Among them, E1 exhibited the strongest synergistic effect in scavenging DPPH radicals and reducing ferric ions (P<0.05). Moreover, E1 presented strong cytoprotection against H2O2-induced oxidative damage in MRC-5 cells by suppressing the decrease of the superoxide dismutase (SOD), glutathione peroxidase (GSH-Px) and catalase (CAT) activities. A strong correlation between the increment of total phenolic/flavonoid and synergistic antioxidant activity, especially between the increment of total flavonoid and the increase in ferric reducing power was observed. Finally, seven antioxidant substances were identified in E1 as oxypaeoniflora, catechin, calycosin-7-O-β-D-glucopyranoside, fomononetin-7-O-β-D-glucopyranoside, 9,10-dimethoxy-pterocarpan-3-O-β-D-glucopyranoside, quercetin and 2′-dihydroxy-3′,4′-dimethyl-isoflavan-7-O-β-D-glucopyranoside.


Introduction
Many medicinal herbs are believed to share a common origin with food in Chinese tradition, thus have frequently been used as functional foods or dietary supplements in the East for centuries. It is evidenced that they generally demonstrate significantly higher health-promoting effects when used in the form of multi-herb formulas than when used alone. However, interactive actions among components in these multi-herbs and the involved mechanism remain poorly understood.
Phenolic substances and flavonoids are increasingly recognized as the major bioactive components contributing to the antioxidant potency of many herbs. For example, Fattahi et al. [1] found that the significant antioxidant activity of Dracocephalum kotschyi was correlated with the flavonoid content. Misbah et al. [2] reported that the antioxidant activities of the fruits of F. deltoidea might be asserted by the phenolic content. The combination of Vernonia amygdalina and Azadirachta indica showed a positive synergism in antioxidant action, due to a boost in the flavonoid content of the extracts [3]. The mechanisms of antioxidant activity of phenolics and flavonoids can be characterized not only by directly scavenging or quenching free radicals, but also by inducing various intracellular antioxidant enzymes [4]. Nagata et al. [5] revealed that cytoprotective effect of quercetin and catechin against H 2 O 2 cytotoxicity in rat hepatocytes BL-9 was related to the activation of GPx. Leung et al. [6] provided evidence that luteolin-induced human lung carcinoma CH27 cell apoptosis was accompanied by activation of SOD and CAT. Furthermore, some antioxidant effects may be a reset of a combination of radical scavenging and the interaction with enzyme functions [7]. For instance, ethyl acetate-extracted fraction of Ficus glomerata, rich in phenolic compounds, possessed high potency to scavenge reactive oxygen species/free radicals and restore the levels of GSH, SOD and CAT [8]. However, these studies mainly focused on the cytoprotective or antioxidant effects of the phenolic compound, flavonoid or extracts from single herb, and limited information is available regarding interactive actions among them.
Atractylodes macrocephala (AME) and Paeonia lactiflora (PL) are very popular medicinal herbs in some Asian countries, which are commonly used in combination as dietary supplements. Our preliminary results indicated that AME and PL were able to result in a significant synergy in scavenging the DPPH radical, hydroxyl radical and superoxide radical anions [9]. In this paper, to elucidate the potential mechanism by which AME and PL synergistically exerted antioxidant effects, we performed a DPPH scavenging activity-guided fractionation, and investigated the protective effect of the obtained antioxidant components against H 2 O 2 -induced oxidative damage using a MRC-5 cells model.

Determination of the Antioxidant Activities
Ferric Reducing Antioxidant Power (FRAP) Assay. The FRAP assay was performed according to the method of Benzie and Strain [10] with some minor modifications. One hundred milliliter of sample was mixed with 3.9 mL FRAP reagent consisting of ferric chloride (20 mM) and TPTZ (10 mM). After 10 min, the absorbance was recorded at 593 nm using a UNICO UV-2000 spectrophotometer (Shanghai Instruments Co. Ltd., Shanghai, China). The reducing power was calculated using the absorbance difference between sample and blank and a further parallel ferrous sulfate standard solution. The results were expressed as micromoles of FeSO 4 per gram dry weight (mmol Fe 2+ /g DW).
DPPH Free Radical Scavenging Assay. The DPPH radical scavenging capacity of the samples was evaluated using the modified method reported earlier [11]. Briefly, 0.1 mL sample was added to 3.9 mL of 0.1 mM DPPH ethanol solution, and allowed to stand for 30 min in the dark at 37uC. Then, the absorbance of the reaction solutions was recorded at 517 nm immediately. Percent inhibition of the DPPH radical by the sample was calculated according to the following equation: where A sample is the absorbance of DPPH solution after reacting with a given concentration of sample, and A blank is the absorbance of DPPH solution with an ethanol blank instead of a sample. The percentage of DPPH reduced was plotted against the concentra- tion of each sample, and an SC 50 value (the concentration required to scavenge 50% of the DPPH) was calculated.

Determination of Total Phenolic and Flavonoid Contents
The determination of total phenolic content was carried out by the Folin-Ciocalteau method as reported previously [12]. Briefly, 0.1 mL of sample was mixed with 1 mL of the Folin-Ciocalteau reagent and incubated at room temperature for 5 min. Then, 1 mL of 0.1 g/mL Na 2 CO 3 solution was added to the mixture. After 90 min incubation, the absorbance of the reaction mixture was recorded at 765 nm. Total phenolic content was calculated by comparison with a standard curve generated by analyzing gallic acid. The results were expressed as gallic acid equivalents (GAE) per gram of sample.
Total flavonoid content of the samples was determined according to colorimetric method as described previously [13]. Briefly, 0.1 mL of sample was mixed with 0.3 mL of 0.05 g/mL NaNO 2 solution in a test tube and incubated for 5 min. Then, 0.3 mL of 0.1 g/mL AlCl 3 solution was added and incubated for another 6 min. The reaction was terminated by adding 2 mL of 1 M NaOH solution. The absorbance of the reaction mixture was recorded immediately at 510 nm. Total flavonoid content was calculated by comparison with a standard curve generated from rutin, and the results were expressed as the mg rutin equivalents (RE) per gram of sample.

Cell Viability Assay
The effect of different fractions on cell viability was evaluated by MTT assay as described previously [14,15], which was based on the reduction of a tetrazolium salt by mitochondrial dehydrogenases in viable cells. MRC-5 cells were seeded in 96-well plates at a density of 2610 5 cells/mL. After 24 h of preconditioning, the cells were subjected to five different treatments, i.e., H 2 O 2 (0.5 mM), The cells were incubated for additional 24 h. Subsequently, 10 mL of 5 mg/mL MTT dye solution was added to each well and the plate was incubated for 4 h at 37uC. Then, 100 mL of DMSO was added to each well and kept for 15 min to dissolve tetrazolium dye. Relative cell viability was calculated by determining the absorbance at 570 nm and untreated control cells were assigned a relative viability of 100%.

Activity of Cellular Antioxidant Enzymes and Malonaldehyde (MDA) Content
Logarithmically growing MRC-5 cells were split into six culture flasks at a density of 2610 5 cells/mL and incubated for 24 h. The medium was replaced with fresh medium containing the following samples, A: control (without H 2 O 2 or antioxidant fractions), B: H 2 O 2 (0.5 mM), C: H 2 O 2 +V E (5 mg/mL), D: H 2 O 2 +CF-AME (7 mg/mL), E: H 2 O 2 +A1 (3 mg/mL), F: H 2 O 2 +E1 (5 mg/mL), and the cells were incubated for additional 24 h. After removing the medium, the cells were suspended in 10 mM PBS (pH 7.4) and lysed suing a JY92-IIY ultrasonic cell disruption system (Beidi Experiment Instrument Inc., Nanjing, China). The lysates were centrifuged at 4,000 rpm and 4uC for 10 min. The supernatants were used for measurement of cellular MDA content and antioxidant enzymes (SOD, CAT and GSH-Px) using the commercially available assay kits (Jiancheng Biochemical Inc., Nanjing, China).

On-line DPPH-HPLC-MS/MS Assay
Identification of the potential antioxidant components in AME-PL extract was performed using a DPPH-HPLC method coupled with an on-line HPLC-DAD-MS/MS technique [16]. Briefly, 400 mL of each sample was mixed with 600 mL of DPPH ethanol solution (0.25 mg/mL) and then reacted for 30 min in the dark at room temperature. The mixture was filtered through a 0.45 mm filter membrane and subjected to HPLC separation. The control sample was prepared by adding 600 mL ethanol instead of DPPH to each sample.
The separation of the antioxidant components was conducted by an Agilent LC-1200 high performance liquid chromatograph (Agilent Inc., Palo Alto, CA) equipped with a XB-C 18 analytical column (4.6 mm6250 mm, 5 mm) and a photodiode array detector (DAD). HPLC separation was performed using a linear gradient of A (acetonitrile) and B (water) at 30uC at a flow rate of 0.8 mL/min. The solvent gradient was used as follows: 0-10% A   Table 1. The contents of the phenolics/flavonoids and the rates of increase in phenolic/flavonoid contents in eight combined extracts from AME and PL.

Determination of Combination Index
To investigate the possible interaction between different extracts, an isobolographic analysis based on the median-effect principle was performed [17]. The median-effect principle was used to calculate individual and combined AME and PL effects. Dose-effect curves for each fraction from AME or PL and their combinations with series diluted concentrations were plotted by using the median-effect equation: where D is the dose. D m is the dose required for 50% inhibition effect, which is equivalent to median-effect dose (EC 50 ). f a is the fraction affected by dose D, and m is a coefficient of the sigmoidicity of the dose-effect curve.
The medium-effect plot is based on the logarithmic form of Chow's median-effect equation [18]: where f u is the fraction unaffected, f u~1 {f a . Combination index (CI) based on the classic isobologram equation is used for data analysis of two-way combination [18]: where (D) 1 and (D) 2 are the doses of the fractions from AME and PL, respectively, in the combination system; (D x ) 1 and (D x ) 2 are the doses of the fractions from AME and PL alone, respectively. CI.1, CI = 1, and CI,1 represent antagonism, additivity or synergism, respectively. In order to analyze the interactive effect qualitatively and quantitatively, the two herbs were administered at a fixed mass ratio of 1:1.

Statistical Analysis
All experiments were carried out in triplicate and data were expressed as mean 6 standard deviations. A one-way analysis of variance (ANOVA) was performed to calculate statistical differences, and multiple comparisons of means were done by the Duncan test using statistical software SPSS 13.0 (SPSS Inc., Chicago, IL). A probability value of ,0.05 was considered significant.

DPPH-scavenging Capacity of the Combined Fractions from AME and PL
The DPPH radical-scavenging activity of different combined fractions from AME and PL are shown in Fig.1. Sixteen combined fractions exhibited a wide range of differences in scavenging DPPH free radicals. Among them, the ethyl acetate fraction of Paeonia lactiflora (EA-PL) presented the highest potency in scavenging DPPH radical when used in combination with four solvent-extracted fractions from AME (P,0.05), suggesting that EA-PL was rich in DPPH radical-scavenging activity (Fig.1A). Our results were in agreement with the previous report [19]. Herein, CIs were used to determine the possible interactive actions between the extracts or fractions. In order to calculate the CIs, dose-effect curves for the single extract or fraction were analyzed (data not shown). As shown in Fig.1B, the CIs for EA-PL+CF-AME and EA-PL+NB-AME were ,1.0, indicating these combined extracts had a synergistic effect in scavenging DPPH radical. By contrast, the CIs for several combined fractions (i.e. CF-PL+ EA-AME, CF-PL+NB-AME, and OH-PL+EA-AME) were .1.0, suggesting these combinations had antagonistic effects in scavenging DPPH radicals. As is well known, the phenolic and flavonoid compounds are the most commonly studied substances that greatly contributed to antioxidant activity of plant foods. Therefore, we measured the total phenolic and flavonoid contents in EA-PL+CF-AME combination, which exhibited the strongest activity in scavenging DPPH radicals. Results showed that the total phenolic and flavonoid contents were 603.339622.894 mg GAE/g and 121.78561.264 mg RE/g, respectively, which were significantly higher than those of other combinations (P,0.05, data not shown). These results suggested that the EA-PL fraction had the highest potency in scavenging DPPH radicals to warrant further fractionation. Thus, EA-PL was further chromatographed on a silica gel column (5.5660 cm) using a stepwise elution of methanol/chloroform (methanol/chloroform = 1:19, 1.5:18.5, 2:18, 2.5:17.5, 3:17, 4:16, 5:15, 20:0, v/v) to afford 8 fractions (A 1 -A 8 ). Then, each fraction was combined with CF-AME to yield eight AME-PL combined extracts, i.e., E 1 (A 1 +CF-AME), E 2 (A 2 + CF-AME), E 3 (A 3 +CF-AME), E 4 (A 4 +CF-AME), E 5 (A 5 +CF-AME), E 6 (A 6 +CF-AME), E 7 (A 7 +CF-AME) and E 8 (A 8 +CF-AME). These samples were examined for in vitro antioxidant activity using DPPH free radical scavenging assay and FRAP test.

In Vitro Antioxidant Activity of the Eight AME-PL Combined Extracts
The antioxidant activity might be attributed to different mechanisms, such as free radical scavenging, reducing potency, prevention of chain initiation, decomposition of peroxides and binding of transition metal ion catalysis [20]. Meanwhile, considering the complexity of the composition of herbal extracts, combined assays are needed for determination of their antioxidant activity [21]. Herein, both DPPH scavenging test and FRAP were applied for the evaluation of antioxidant activity of the AME-PL combined fractions.
DPPH Free Radical Scavenging Activity. As shown in Fig.2A, the highest DPPH scavenging efficiency was observed in the groups treated with E 1 , E 4 , E 5 and E 6 , and there was no significant difference between them. E 3 exhibited the highest SC 50 value of 0.031 g/mL, suggesting that it had the weakest ability in scavenging DPPH free radicals among the tested samples (P, 0.05). The CI values for E 1 -E 8 were calculated using CalcuSyn software and the results were illustrated in Fig.2B. It was found that E 1 and E 8 had the lowest CI values (CI = 0.789 and CI = 0.786, respectively), indicating they had the strongest synergistic effect (P,0.05). It is noteworthy that the combinations demonstrating the highest antioxidant effect did not show the lowest CI (e.g. EA-PL+CF-AME and EA-PL+EA-AME in Fig.1, and E6 in Fig.2). These findings demonstrated that the combination of two antioxidants may have strong antioxidant activity, but not necessarily engender synergistic efficacy, it may even generate antagonistic interaction (as indicated by their CIs.1.0). Previous literatures have showed that the synergism might arise from a complex interaction among single ingredients with different pharmacological functions, such that one ingredient enhances the therapeutic effect of another active ingredient [22,23] or via coalistic combinations, such that all ingredients involved are inactive individually but become active in combinations [24]. It seems that the similar interactions among ingredients did not occur in the combinations tested in this study, and thus these combinations did not exhibit the antioxidant synergism.
Total Antioxidant Power Measured by FRAP Assay. The antioxidant activity of the eight combined extracts (E 1 -E 8 ) was estimated by the FRAP assay, in which the antioxidants present in the sample reduce the Fe(III)/tripyridyltriazine (TPTZ) complex to the blue ferrous form, with an increase in absorbance at 593 nm. In order to obtain FRAP values, a calibration curve generated by analyzing standard FeSO 4 solutions was derived from the absorbance versus concentration plot (R 2 = 0.9979). RIFrapV was also calculated as the rate of increase in FRAP value in the combined extracts as compared to theoretical sum of those of the respective herbs extract. As shown in Fig.3a-A, the highest FRAP value was observed for E 1 (3.39260.015 mmol Fe 2+ /g DW), suggesting that it had the strongest ferric reducing ability. The FRAP values for other samples decreased in the order: E 4 . E 5 .E 6 .E 2 .E 3 , E 7 , E 8 . As to RIFrapV, E 1 exhibited a highest RIFrapV value of 229.68% (Fig.3a-B), showing that there was a synergistic enhancement in ferric reducing ability in E 1 . Additionally, the antioxidant activity of eight combined extracts, based on FRAP assay, was different from that based on the DPPH scavenging test. It might be explained by considering that these determinations were performed under different experimental conditions which were based on different reaction mechanisms. DPPH assay is believed to be one of the methods utilizing both hydrogen atom transfer and single electron transfer mechanism. The FRAP assay takes advantage of electron-transfer reaction and takes place at a faster rate than that for the DPPH test, whose degree of discoloration is attributed to the hydrogen donating ability of the test compounds [25].

Total Phenolic and Flavonoids Contents
The contents of total phenolic and flavonoid has been proved to be positively correlated with antioxidant capacity [26]. Therefore, the variation in the contents of phenolic and flavonoid compounds in these combined extracts was investigated. RIPC and RIFC, defined as the rate of increase in phenolic and flavonoid contents in these eight combined extracts relative to the theoretical sum of those from the respective herbs extract, were calculated. Table 1 revealed that the total phenolic and flavonoid contents in all combined extracts exhibited various increases compared to their respective theoretical sum, as indicated by the RIPC and RIFC, respectively. The correlation between RIFC/RIPC and CI values as well as between RIFC/RIPC and RIFrapV value was further analyzed. As shown in Fig.3b, both RIPC and RIFC correlated significantly with the CI values, while a higher negative correlation was found between CI value and RIFC (R CI-RIFC = 20.833, P = 0.010) than that between CI value and RIPC (R CI-RIPC = 2 0.810, P = 0.015). Meanwhile, these was a positive correlation between the rate of increase in FRAP value and RIFC (R RIFrapV-RIFC = 0.929, P = 0.001), significantly higher than that between RIFrapV value and RIPC (R RIFrap-RIPC = 0.667, P = 0.05). These findings indicate that the variations in total flavonoids content might make a more significant contribution to synergistic antioxidant effect of the combined extracts. Our results were in agreement with the previous report that the antioxidant activity of eight traditional Chinese herbs pairs was determined or highly influenced by changes of their total flavonoid content [9].

Protection Against H 2 O 2 -induced Oxidative Damage in MRC-5 Cells
Cell Viability Assay. Since E 1 exhibited a significant antioxidant synergism in both DPPH radical-scavenging and FRAP assays, we further examined its effect on H 2 O 2 -induced oxidative damage in MRC-5 cells. As shown in Table 2, treatment with 0.5 mM H 2 O 2 led to a remarkable decrease in cell viability compared with control group (P,0.05). H 2 O 2 plays a pivotal role among a great variety of ROS, and exogenous H 2 O 2 can enter the cells and induce cytotoxicity due to its high membrane permeability damage in cells [27]. All the tested samples were able to significantly alleviate H 2 O 2 -induced oxidative damage in MRC-5 cells. Moreover, the groups treated with A 1 and E 1 had higher cell viability than that treated with V E (P,0.05), indicating that they could provide stronger protection against H 2 O 2 -induced oxidative damage in MRC-5 cells. The cell viability for the tested samples and V E was decreased in the order: A 1 .E 1 .V E .CF-AME. It is noteworthy that E 1 -treated cells have a significantly higher viability compared to the theoretical sum of those from A 1 and CF-AME (P,0.05), suggesting a synergistic cytoprotection in the combination of A 1 and CF-AME. peroxidation and destruction of cell membranes, producing secondary products such as the MDA, and thus damaged the integrity of membrane and/or membrane-associated function in subcellular organelles such as mitochondria, microsomes and lysosomes [28]. The damages induced by ROS are also associated with the inactivation of endogenous antioxidant enzymes, such as SOD, CAT and GSH-Px, which are important for evaluating the effects of free radical scavenging activity and guarding against superoxide toxicity [29]. In order to gain some insights into the synergistic mechanism of AME and PL, we further investigated the effect of E 1 , A 1 and CF-AME on the antioxidant enzymes activities (SOD, CAT and GSH-Px) and MDA content in H 2 O 2 -induced MRC-5 cells.
As shown in Table 2, in the normal MRC-5 cells, the activities for Cu/Zn-SOD, Mn-SOD, GSH-Px and CAT were 16.1460.07 U/mL, 6.7061.41 U/mL, 17.1461.28 U/mL and 5.2060.08 U/mL, respectively. Upon the treatment with 0.5 mM H 2 O 2 , their activity were decreased to 7.4760.22 U/mL, 1.0060.43 U/mL, 3.1660.64 U/mL and 0.2560.03 U/mL (P,0.05), respectively. Supplementation of E 1 , A 1 , CF-AME and V E significantly inhibited the loss of these antioxidant enzymes activities in H 2 O 2 -stimulated cells (P,0.05). A 1 exhibited the extremely high efficiency in protecting the activities of Cu/Zn-SOD, GSH-Px and CAT enzymes. The highest Cu/Zn-SOD activity was observed in the cells treated with E 1 . Moreover, E 1 showed comparable potency to A 1 in elevating the activities of GSH-Px and CAT. The accumulation of MDA is a sensitive indicator of the peroxidation of cellular lipids in cultured cells. The measurement of MDA in H 2 O 2 -induced MRC-5 cells provides an alternative to evaluate the protection by the herbs extracts. Stimulation of MRC-5 cells with 0.5 mM H 2 O 2 led to a dramatic increase in MDA content from the basal level of 1.6760.11 mM to 8.8960.22 mM. All the tested samples significantly attenuated the H 2 O 2 -elicited increase in MDA level (P,0.05), of which the strongest effect was observed in the group treated with A 1 , followed by E 1 and V E . It is noteworthy that E 1 could protect the activities of almost all antioxidant enzymes and decrease MDA content better than the theoretical sum of A 1 and CF-AME, although no significant difference was observed in the activities of Mn-SOD and GSH-Px between E 1 and theoretical sum of the respective constituent.

Identification of Potential Antioxidant Components in E 1
The on-line DPPH-HPLC method has been widely used for screening of the potential radical-scavenging components from various food extracts [30]. After reaction with DPPH, the peak areas (PAs) of the radical scavenging compounds would obviously disappear or decrease in UV chromatograph, while for those without antioxidant activity, there was almost no change in their PAs. Herein, this approach was employed to screen the potential antioxidant compounds from E 1 . HPLC profiles of E 1 with or without DPPH treatment are illustrated in Fig. 4. It was found that eight peaks (1, 3, 4, 6, 7, 9, 10 and 20) were decreased in the PA after spiking with the DPPH free radicals, suggesting these components could be regarded as the potential antioxidant candidates of E 1 . Among the identified peaks, peaks 1, 3 and 4 were present in PL, while the others (6, 7, 9, 10 and 20) were from AME. Further characterization of these potential antioxidant components was carried out by HPLC-MS/MS, owing to it can provide affluent multistage fragment information for compounds with collision-induced dissociation [31]. As shown in Table 3, seven antioxidant components were unambiguously identified as Oxypaeoniflora (1), Catechin (3), Calycosin-7-O-b-D-glucopyranoside (6), Fomononetin-7-O-b-D-glucopyranoside (7), 9,10-dimethoxy-pterocarpan-3-O-b-D-glucopyranoside (9), 29-dihydroxy-39,49-dimethyl-isoflavan-7-O-b-D-glucopyranoside (10), Quercetin (20), by comparing their UV maximum absorption wavelength, MS and MS/MS data with those of the standard references, as well as the previous reports [32,33]. It is not surprising that most of these components belonged to the flavonoids family, which has been confirmed to be excellent antioxidants.
In the present study, we found that the total phenolic and flavonoid contents in combined AME and PL extracts varied differently from the theoretical sums of those from the respective herbs. A significant correlation was also observed between the increments of total phenolic/flavonoid and the elevation of antioxidant capacities in these combinations. On-line DPPH-HPLC-MS/MS analysis led to the identification of seven potential antioxidant compounds. These results suggested that the changes in the content of total phenolic and flavonoid, especially the flavonoid, might contributed greatly to the antioxidant synergism in combined AME and PL extracts.