Valeriana rigida Ruiz & Pav. Root Extract: A New Source of Caffeoylquinic Acids with Antioxidant and Aldose Reductase Inhibitory Activities

Valeriana rigida Ruiz & Pav. (V. rigida) has long been used as a herbal medicine in Peru; however, its phytochemicals and pharmacology need to be scientifically explored. In this study, we combined the offline 2,2-diphenyl-1-picryl-hydrazyl-hydrate (DPPH)-/ultrafiltration-high-performance liquid chromatography (HPLC) and high-speed counter-current chromatography (HSCCC)/pH-zone-refining counter-current chromatography (pH-zone-refining CCC) to screen and separate the antioxidants and aldose reductase (AR) inhibitors from the 70% MeOH extract of V. rigida, which exhibited remarkable antioxidant and AR inhibitory activities. Seven compounds were initially screened as target compounds exhibiting dual antioxidant and AR inhibitory activities using DPPH-/ultrafiltration-HPLC, which guided the subsequent pH-zone-refining CCC and HSCCC separations of these target compounds, namely 3-O-caffeoylquinic acid, 4-O-caffeoylquinic acid, 5-O-caffeoylquinic acid, 3,4-O-di-caffeoylquinic acid, 3,5-O-di-caffeoylquinic acid, 4,5-O-di-caffeoylquinic acid, and 3,4,5-O-tri-caffeoylquinic acid. These compounds are identified for the first time in V. rigida and exhibited remarkable antioxidant and AR inhibitory activities. The results demonstrate that the method established in this study can be used to efficiently screen and separate the antioxidants and AR inhibitors from natural products and, particularly, the root extract of V. rigida is a new source of caffeoylquinic acids with antioxidant and AR inhibitory activities, and it can be used as a potential functional food ingredient for diabetes.


V. rigida Material and Preparation of Plant Extract
The dried root of V. rigida was obtained from the department of La Libertad in Peru in a local market and preserved at the Center for Efficacy Assessment and Development of Functional Foods and Drugs, Hallym University. The specimen was authenticated by Paul H. Gonzales Arce from the Museo de Historia Natural, Universidad.
For preparation of the plant extract, the dried root of V. rigida (70 g) was ground to powder and extracted using 2 L of 70% MeOH aqueous solution assisted by sonication for 4 h at room temperature (around 22 • C). The extraction was carried out twice, and the extraction solutions were combined, filtered (Advantec #2), and evaporated by rotary evaporation at 37 • C to gain about 16 g of extract powder.

Antioxidant Assay 2.4.1. DPPH Radical Scavenging Assay
The experiment was carried out as previously described [39]. In brief, 180 µL of DPPH solution (in MeOH, 0.32 mM) and 20 µL of sample solution (in MeOH, extract 100-800 µg/mL, compounds 125-2000 µm) were mixed in a 96-well plate and allowed to react for 20 min at 25 • C in the darkness. Thereafter, the absorbance was measured using a microplate reader (EL800, Bio-Tek Instruments, Winooski, VT, USA) at 570 nm. Trolox, a reference antioxidant, was used to make a calibration curve derived from the DPPH radical scavenging activity (%) against the final concentrations of Trolox (6.25-100 µm), and the results (n ≥ 3) were presented as Trolox equivalent antioxidant capacity (TE, µm Trolox/µg extract or µm Trolox/µM compound). The inhibitory activity (%) of the samples against DPPH radical was calculated using Equation (1): where A control is the absorbance of DPPH solution free of samples, A sample is the absorbance of DPPH solution incubated with a sample, A blank1 is the absorbance of the test sample free of DPPH solution, and A blank2 is the absorbance of MeOH.

ABTS Radical Scavenging Assay
The experiment was conducted as previously described [39]. In brief, 3.5 mM of potassium persulfate aqueous solution was used to prepare 0.2 mM of ABTS diammonium salt. The solution was diluted 10-fold using distilled water and allowed to produce ABTS radicals (ABTS + ) by keeping the solution in the dark for 14 h at room temperature.
Thereafter, 10 µL of sample (prepared in MeOH, extract 25-400 µg/mL, compounds 31.25-1000 µm) was mixed with 290 µL of ABTS + solution in a 96-well plate and incubated in the dark for 10 min at 25 • C, which was followed by measuring the absorbance at 750 nm using the same microplate reader. Trolox, a reference antioxidant, was used to make a calibration curve derived from the ABTS + scavenging activity (%) against the final concentrations of Trolox (1.04-16.67 µm), and the results (n ≥ 3) were presented as Trolox equivalent antioxidant capacity (TE, µm Trolox/µg extract or µm Trolox/µM compound). Equation (1) was used to calculate the ABTS + scavenging activity (%), where A control is the absorbance of ABTS + solution free of samples, A sample is the absorbance of ABTS + solution incubated with a sample, A blank1 is the absorbance of the test sample free of ABTS + solution, and A blank2 is the absorbance of the diluted potassium persulfate solution.

Oxygen Radical Absorbance Capacity Assay
The oxygen radical absorbance capacity (ORAC) assay was performed as previously described [39]. Briefly, 0. 1 M PBS of pH 7.4 was used to prepare AAPH (40 mM) and fluorescein sodium salt (117 nm) shortly before use. Thereafter, 20 µL of sample (prepared in MeOH, extract 6.25-25 µg/mL, compounds 12.5-50) and 120 µL of fluorescein sodium salt (117 nm) were mixed and incubated for 15 min at 37 • C in a black 96-well plate. Next, 60 µL of AAPH (40 mM) was added to generate peroxyl radicals to initiate the reaction. The fluorescence intensity (λ ex = 485 nm, λ em = 538 nm) was monitored for 90 min (1 time per min) using a Fluoroskan Ascent FL microplate reader (Thermo, Waltham, MA, USA) maintained at 37 • C. Trolox was used to make a calibration curve derived from the net AUC (the area under the curve) values against the final concentrations of Trolox (1-10 µm), and the results (n ≥ 3) were expressed as Trolox equivalent antioxidant capacity (TE, µm Trolox/µg extract or µm Trolox/µM compound). The net AUC value was calculated by subtracting the AUC of the AAPH group (free of samples) from that of a sample group (with AAPH and a sample) using Equation (2): where f n is the fluorescence intensity measured at n min and f 0 is the fluorescence intensity measured at 0 min.

Hypochlorous Acid Scavenging Assay
The HOCl scavenging assay was adapted from a previous study using 1,8-diamino naphthalene as a fluorescence probe for hypochlorite [40]. Briefly, 90 µL of 0.1 M PBS of pH 7.4, 20 µL of sample (prepared in water or MeOH aqueous solution, extract 62.5-250 µg/mL, compounds 62.5-1000 µm), and 40 µL of HOCl (10 µm in 0.1 M PBS of pH 7.4) were mixed and allowed to react for 5 min at room temperature in a black 96-well plate. Thereafter, 50 µL of fluorescence probe (1,8-diaminonaphthalene, 240 µm in distilled water) was added and allowed to react for 2 min. Next, the fluorescence intensity (λ ex = 360 ± 20 nm, λ em = 460 ± 20 nm) was immediately measured using a microplate fluorescence reader (FLx800; Bio-Tek, Winooski, VT, USA). Trolox (final concentration 25-100 µg/mL) was used as a reference antioxidant. Notably, DMSO remarkably interferes with the result, and therefore, should not be used in HOCl assay. The inhibitory activity (%) of the sample against HOCl was calculated using Equation (3) and presented as mean ± standard deviation (n = 3) and IC 50 value (half-maximal inhibitory concentration), which was calculated using linear regression analysis: where f p , f hp , and f s are the fluorescence intensities of the probe alone (f p ), the mixture of HOCl and probe (f hp ), and the mixture of sample, HOCl, and probe (f s ), respectively.

AR Inhibition Assay
The experiment was approved by the Institutional Animal Care and Use Committees (IACUC) of Hallym University (Hallym-2016-95). The eyes of 10-week Sprague-Dawley rats (250-280 g) were removed and frozen at −70 • C before use. Thereafter, the lenses were removed from the eyes, ground in a mortar (precooled at −70 • C), and extracted using 0.1 M PBS of pH 6.2 (around 0.5 mL of buffer per two frozen rat lenses). Then, the extraction solution was centrifuged using a 5417R centrifuge (Eppendorf, Germany) for 30 min at 10,000× g and 4 • C to get the rat lens AR homogenate (in the supernatant).
The AR inhibitory activities of the extract and the separated compounds were determined as described previously [39]. were used as positive controls, including two strong natural AR inhibitors, quercetin, and quercitrin [39], and one proved AR inhibitor drug, epalrestat [41]. The DMSO used for sample preparation was less than 0.4% (v/v) in the reaction system. The AR inhibitory activity (%) of the sample was calculated using Equation (4), and the results were presented as mean ± standard deviation (n = 3) and IC 50 value, which was calculated using linear or logarithmic regression analysis depending on which one offered a better regression coefficient (r 2 ): where Slope b , Slope c , and Slope s are the slopes derived from the OD 340 nm against the reaction time (min)-dotted lines of blank group (without enzyme or sample), the control group (without sample), and the sample group (with enzyme and sample), respectively. |Slope| is the absolute value of slope.

Screening of Antioxidants from the Extract Using Offline DPPH-HPLC
The offline DPPH-HPLC strategy was used to screen the potential antioxidants in the extract before separation. Briefly, 150 µL of DPPH solution (2.5 mg/mL in MeOH) and 50 µL of the extract solution (10 mg/mL in MeOH) were mixed and incubated for 30 min at 37 • C. Thereafter, the reaction solution (injection volume 10 µL) was subjected to HPLC assay. MeOH was used to replace the DPPH solution for incubation with the sample to be used as a DPPH-free control group. The compounds with reduced HPLC peak areas from DPPH group compared with those from the DPPH-free group were assigned as potential antioxidants.

Screening of AR Inhibitors from the Extract Using Ultrafiltration-HPLC
The enzyme-ligand binding affinity-based ultrafiltration was used to screen the potential AR inhibitors from the extract prior to separation. Briefly, 250 µL of 0.1 M phosphatebuffered saline (PBS, pH 7.0), 60 µL of human recombinant AR (0.05 units/mL in 0.1 M PBS of pH 6.2), and 20 µL of quercitrin (used as an enzyme blocker, 0.5 mg/mL) were mixed in a 1.5 mL tube and preincubated for 10 min at 37 • C. Thereafter, 40 µL of the extract (1 mg/mL in water) was added into the reaction mixture, and it was further incubated for 20 min at 37 • C. Next, the reaction mixture was ultrafiltrated through a Amicon ®® Ultra 10 kDa membrane (Merck Millipore Ltd., Tullagreen, Ireland) for 20 min at 13,000 rpm (10,770× g; Micro-12, Hanil Science Industrial Co., Incheon, South Korea) at 20 • C. Moreover, the filtrate was individually collected, and the centrifugal membrane was washed by adding 200 µL of 0.1 M PBS (pH 7.0) and further centrifuged for 20 min at 13,000 rpm.
The two-time filtrates were combined and evaporated using nitrogen gas, which was then re-dissolved using 200 µL of 50% MeOH aqueous solution and subjected to HPLC assay with an injection volume of 60 µL. The AR and AR-free experiments were carried out simultaneously with and without AR, respectively, in the absence of quercitrin. Furthermore, the compounds with reduced HPLC peak areas in the AR group compared with those from the AR-free group and quercitrin-blocked AR group were assigned as AR inhibitors.

Separation of Target Compounds by pH-Zone-Refining CCC
The target compounds screened via ultrafiltration-HPLC were sensitive to the acid and base added into the CCC solvent systems, thereby indicating that these compounds are ionizable compounds and, therefore, they are suitable for separation via pH-zone-refining CCC [32].
2.8.1. Screening of pH-Zone-Refining CCC Solvent System The solvent system was screened via HPLC and evaluated according to the partition coefficient (K value) of the target compounds on the principle introduced by Ito [32]. Briefly, four solvent systems comprising EtOAc, n-BuOH, and H 2 O were first acidified by using formic acid to 208 mM or basified by using ammonia solution to 29 mM, and then the corresponding K values of the target compounds under acidic (K acid ) or basic (K base ) conditions were determined, as previously described [36]: each acidified or basified solvent system was partitioned to upper layer and lower layer. Thereafter, the extract (about 1-2 mg) was prepared in a 1.5 mL tube and dissolved by adding equal volumes (each 500 µL) of the upper layer and lower layers, which were mixed by a vortex mixer and centrifuged for about 20 s using a C1301 Mini Centrifuge (Labnet International, South Korea). Next, the upper and lower phase sample solutions were respectively pipetted (each 200 µL) into a new 1.5 mL-tube and evaporated by nitrogen gas. Each sample residue was redissolved using 200 µL of MeOH and subjected to HPLC analysis with an injection volume of 20 µL. The K value was calculated as K = A upper /A lower , where A upper and A lower are the HPLC peak areas of the target component in the upper and lower phases, respectively.

Preparation of pH-Zone-Refining CCC Solvent System and Sample Solution
The solvent system EtOAc/n-BuOH/H 2 O (2:3:5, v/v) was selected as the solvent system for pH-zone refining CCC separation, which was prepared in a separating funnel and partitioned to upper and lower layers after equilibration. The upper layer was acidified by using formic acid (208 mM) as the stationary phase and the lower layer was basified by using ammonia solution (29 mM) as the mobile phase. Both the mobile and stationary phases were degassed by sonication before use. The sample solution was prepared by dissolving the extract (about 3.54 g) in the biphasic solvents containing 16 mL of the formic acid-acidified stationary phase and 3 mL of the ammonia-free mobile phase. Thereafter, an extra 208 mM of formic acid was added to the sample solution to further improve the solubility of the target compounds in the organic phase via the protonation effect. Next, the sample solution was centrifuged (UNION 32R PLUS; Hanil Science Industrial Co., Kimpo, Korea) for 10 min at 3720× g (4000 rpm) and only the supernatant was used for sample loading.

pH-Zone-Refining CCC Separation
The separation was performed on TBE 300C HSCCC equipment (Tauto Biotech. Co., Ltd., Shanghai, China) with three polytetrafluoroethylene multilayer coils (ID: 2.6 mm; total volume: 300 mL). A Biotage Isolera FLASH purification system (Uppsala, Sweden) was equipped with HSCCC equipment as a pump, a UV monitor, and an auto fraction collector. In brief, the stationary phase was first introduced to fill the HSCCC coil column at 50 mL/min, and then, the flow rate was set at 4 mL/min and the rotation speed of the coils was adjusted to 800 rpm. Thereafter, the sample solution was loaded, and the separation was initiated by introducing the mobile phase at 4 mL/min. The eluate was monitored and automatically collected by the Isolera FLASH purification system according to the changes of UV absorbance at 254 nm. Eventually, the stationary phase retention ratio was calculated as the volume of the stationary phase collected from the HSCCC coil column relative to the total volume of the HSCCC coil column after separation.
2.9. Separation of Target Compounds 1 and 3 by Conventional HSCCC 2.9.1. Preparation of HSCCC Solvent System and Sample Solution The solvent system n-BuOH/H 2 O (1:1, v/v) was prepared in a separating funnel and modified by adding acetic acid to 8.7 mM, followed by thorough mixing, and then partitioning to upper and lower phases after settling. The partitioned upper and lower phases were used as mobile and stationary phases, respectively, and were degassed by sonication before use. Approximately, 42 mg of the mixture of compounds 1 and 3 concentrated by pH-zone-refining CCC was dissolved in biphasic solvents comprising equal volumes (each 7 mL) of the mobile and stationary phases.

HSCCC Separation
The stationary phase was first pumped to fill the HSCCC coil column; thereafter, the rotation rate of the HSCCC coil column was gradually regulated to 850 rpm. Next, the mobile phase was pumped in at 5 mL/min until a steady elution of the mobile phase from the column outlet line was observed. Then, the sample was loaded and eluted by the mobile phase at 5 mL/min and monitored at 254 nm. When the separation was complete, the stationary phase retention ratio was determined, as previously described.
2.11. Quantification of the Major Compounds 5, 6, and 8 Although the three major compounds 5, 6, and 8 have different maximum UV absorbances, to simplify the quantification process, their contents in the extract were quantified using the same HPLC condition (injection volume 10 µL) described in Section 2.3. These three purified components (5, 6, and 8) were dissolved in MeOH at 1 mg/mL and diluted appropriately using MeOH to prepare standard solutions for making calibration curves and method validation. Calibration curves were obtained by plotting the HPLC peak areas (y) versus the corresponding concentrations (x, µg/mL) by triplicate injection of at least nine different concentrations of standard solutions 3,5-diCQA (5; 12.50-400.00 µg/mL), 4,5-diCQA (6; 12.50-400.00 µg/mL), and acacetin (8; 6.25-400.00 µg/mL). The limit of detection (LOD) and limit of quantification (LQD) were measured by signal-to-noise ratios of three (S/N = 3) and ten (S/N = 10), respectively. The repeatability and reproducibility of the quantification method were examined by measuring the relative standard deviation (RSD) values of the peak areas of each compound (50.00 and 200.00 µg/mL) determined by HPLC at intraday (n = 6) and interday (n = 3). The accuracy of the quantification method was examined by determining the spike recovery of each standard solution spiked in the extract solution. The standard solutions of 100.00 and 300.00 µg/mL and the extract of 1.00 mg/mL (in MeOH) were used for sample spiking by mixing 0.2 mL of the extract solution and 0.2 mL of each individual standard solution. The extract solution and the spiked solution were detected by HPLC in triplicate to calculate the spike recovery, as described previously [42], using Equation (5), and the content (µg/mg = mg/g) of each compound in the extract was calculated as the concentration (C 1 , µg/mL) from the corresponding calibration curve of each compound/extract concentration (1.00 mg/mL).
where C 0 , C 1 , and C 2 are the concentrations of each compound tested in the standard solution, extract solution, and the spiked solution, respectively. V 0 , V 1 , and V 2 are the volumes of the standard solution used for sample spiking, the extract solution used for sample spiking, and the spiked sample solution, respectively. In this study, V 0 = V 1 = 0.2 mL; V 2 = V 0 + V 1 = 0.4 mL; C 1 (µg/mL) and C 2 (µg/mL) are calculated from the corresponding calibration curve of each compound, and C 0 is the concentration of the standard solution (100.00 and 300.00 µg/mL) used for sample spiking.

Statistical Analysis
The data obtained from DPPH and ABTS assays were analyzed using one-way analysis of variance (ANOVA) followed by a LSD's multiple comparison Test (SPSS version 25; IBM, New York, NY, USA), whereas the data obtained from ORAC assay were analyzed using one-way ANOVA followed by a tamhane T2 Test (SPSS version 25), since the group variances are not equal (F = 0.03 < 0.05 by Levene's test). A value of p < 0.05 was considered as statistically significant.

Antioxidant and AR Inhibitory Activity of the 70% MeOH Root Extract of V. rigida
The 70% MeOH root extract of V. rigida exhibited comparable antioxidant activities toward Trolox against DPPH (0.75 µm Trolox equivalents/µg extract, TE), ABTS (0.82 TE), and peroxyl (2.60 TE) (ORAC assay) radicals (Table S1). Moreover, the extract was able to scavenge HOCl radicals (IC 50 16.52 µg/mL), although it was weaker than Trolox (IC 50 8.52 µg/mL) (Table 1). Notably, the AR inhibitory activity of the extract (IC 50 0.478 µg/mL) was remarkably higher than quercetin (IC 50 4.536 µg/mL), which is a popular natural AR inhibitor with anti-diabetic potential [15]; however, the extract was less active than quercitrin (IC 50 0.046 µg/mL), which is one of the most active natural AR inhibitors [13], and epalrestat (IC 50 0.016 µg/mL), one proved AR inhibitor drug (Table 1). In general, the extract exhibited remarkable antioxidant and AR inhibitory activities, thereby suggesting that the components it contains may be used as new sources of antioxidants and AR inhibitors with health-promoting benefits including anti-diabetic properties.

Screening of Antioxidants and AR Inhibitors from the Extract Using Offline DPPH-and Ultrafiltration-HPLC
To screen the antioxidants and AR inhibitors from the extract prior to separation, offline DPPH-HPLC and ultrafiltration-HPLC were performed. As illustrated in Figure 1, after reaction with DPPH radicals, a significant reduction in the HPLC peak areas of compounds 1-7 was found in DPPH group compared with that of DPPH-free group; therefore, these seven compounds were screened as antioxidants in the 70% MeOH extract of V. rigida root via DPPH-HPLC method. The AR inhibitors in the extract were screened using the ultrafiltration-HPLC based on enzyme-ligand binding affinity. Potential AR inhibitors would bind to the active site of AR, forming macromolecular enzyme-ligand complexes during incubation with AR, and thus, they cannot pass through the ultrafiltration membrane via centrifugation, thereby leading to decreased concentration of the AR inhibitors in the filtrate of the ARcontaining group (AR group) compared with those from the AR-free group. Thereafter, the resulting concentration difference was determined by analyzing the centrifugal filtrates of AR and AR-free groups using HPLC as Section 2.7 described, by which compounds 1-7 were preliminarily screened as potential inhibitors since the HPLC peak areas of these components in the centrifugal filtrate of AR group reduced compared with those in the AR-free group ( Figure 2B). Moreover, considering that false positives could arise from nonspecific binding of compounds to AR nonfunctional sites, AR was preincubated with quercitrin, a strong AR inhibitor, to block the active site of AR, thereby reducing the possibility of other AR inhibitors from binding to the AR active site. As illustrated in Figure 2C, blocking AR with quercitrin (quercitrin-blocked AR group) resulted in an increase in the HPLC peak areas of AR inhibitors in the centrifugal filtrate compared with those from the AR group (free of quercitrin), thus verifying compounds 1-7 as AR inhibitors. Consequently, by DPPH-HPLC and ultrafiltration-HPLC methods, compounds 1-7 were screened as target components with dual antioxidant and AR inhibitory activities, which could guide further HSCCC separation.

Selection of Solvent System and Separation of Target Compounds by pH-Zone-Refining CCC and Conventional HSCCC
A successful HSCCC separation mainly depends on the selection of a suitable solvent system, which is expected to satisfy the K values of the target compounds with 0.5-2.0, and the separation factor (α) of two objective compounds higher than 1.5 (α = K a /K b ≥ 1.5, K a ≥ K b ) [43]. Several solvent systems consisting of n-hexane, EtOAc, MeOH, and H 2 O in different proportions were initially tested. As indicated in Table S2, multiple runs of conventional HSCCC using several solvent systems would be required to separate all the target compounds, since their K values could not be covered by a single HSCCC solvent system (0.5 ≤ K ≤ 2.0); however, this would be time-consuming, require more solvent, and laborious. Thereafter, we tried to modify the solvent system n-hexane/EtOAc/MeOH/H 2 O (0.2:5:1.8:5, v/v) by adding a little formic acid. Surprisingly, the K values of target compounds markedly increased, indicating that these compounds are acid compounds since their solubility toward the organic phase significantly increased after protonation by formic acid (Table S2). Accordingly, we decided to separate the target compounds using pH-zonerefining CCC, which is particularly suitable for separating ionizable analytes [43].
As for pH-zone-refining CCC, a good solvent system is expected to satisfy K 1 under acidic condition (K acid 1) and K 1 under basic condition (K base 1) for acid compounds [32]. We first modified the solvent systems EtOAc/H 2 O (1:1, v/v), EtOAc/n-BuOH/H 2 O (4:1:5, v/v), EtOAc/n-BuOH/H 2 O (3:2:5, v/v), and EtOAc/n-BuOH/H 2 O (2:3:5, v/v) with 208 mM formic acid and with 30 mM ammonia, respectively, and then, the K acid and K base values of the target compounds were determined using the modified solvent systems. As listed in Table 2, the K acid values of the target compounds offered by the 208 mM formic acid-acidified EtOAc/n-BuOH/H 2 O (2:3:5, v/v) solvent system were greater than those obtained from other solvent systems, whereas the K acid values of the target compounds offered by the 30 mM ammonia-basified EtOAc/n-BuOH/H 2 O (2:3:5, v/v) solvent system were significantly lower than 1, thereby indicating that EtOAc/n-BuOH/H 2 O (2:3:5, v/v) is a suitable pH-zone-refining CCC solvent system for separation of the target compounds. As described in Section 2.8.2, the solvent system EtOAc/n-BuOH/H 2 O (2:3:5, v/v) was prepared in a separating funnel and divided into upper and lower phases. The upper phase was acidified using formic acid at a final concentration of 208 mM to be used as the mobile phase, and the lower phase was basified using ammonia at a final concentration of 29 mM to be used as the stationary phase. Moreover, formic acid retains the target compounds (acid compounds) in the stationary phase (organic phase) via the protonation effect, whereas ammonia elutes the target compounds (acid compounds) in the mobile phase (aqueous phase) through the deprotonation effect. Furthermore, the target compounds were separated as described in Section 2.8.3. As illustrated in Figure 3, by a single run of pH-zone-refining CCC, target compounds 1 (55.7 mg), 2 (57.8 mg), 4 (59.5 mg), 5 (146.5 mg), 6 (300.2 mg), and 7 (56.0 mg) were separated from the centrifugal supernatant of the sample solution (about 3.54 g of sample was used to prepare sample solution) with purities over than 93% by HPLC assay at 254 nm; however, target compound 3 was eluted as a mixture of 3 and 1 (42.0 mg). After separation, the stationary phase retained in the CCC coil column was collected to calculate the stationary phase retention ratio (about 30%). Notably, the major nontarget compound 8 was found to precipitate in the form of crystals in the collected stationary phase during storage in the hood. The remaining solvent system was carefully discarded, and then, the crystals were washed out using MeOH and evaporated to obtain high-purity compound 8 (25.7 mg, Figure 3B).
The mixed compounds 1 and 3 were further separated using conventional HSCCC. As listed in Table 3, the 8.7 mM acetic acid-modified n-BuOH/H 2 O (1:1, v/v) provided satisfactory K values and α value for target compounds 1 and 3 (K 1 = 0.73, K 3 = 1.57, α = K 3 /K 1 = 2.15 > 1.5) and was therefore used for HSCCC separation of compounds 1 and 3. The separation was conducted as described in Section 2.9.2. The target compounds 1 (10.8 mg) and 3 (13.1 mg) were successfully separated from the mixture of compounds 1 and 3 (about 42 mg) (Figure 4). After the separation was complete, the final volume retention ratio of the stationary phase was determined to be 63%.    (7) [45], and acacetin (8) [46]. The structures are illustrated in Figure 5. Notably, all these components are reported for the first time in V. rigida. Moreover, according to the results obtained from offline DPPH-HPLC and ultrafiltration-HPLC analysis, the seven caffeoylquinic acids (target compounds 1-7) are deduced to be the main antioxidants and AR inhibitors in the 70% MeOH root extract of V. rigida.

Antioxidant and AR Inhibitory Activity of the Target Compounds 1-7
The antioxidant and AR inhibitory activities of the target compounds 1-7 were further verified using DPPH, ABTS, ORAC, and HOCl radical scavenging assays (Tables 4 and 5) and AR inhibition assay (Table 5). Table 4. Antioxidant activity of the seven target components from the 70% MeOH root extract of V. rigida using DPPH, ABTS, and ORAC assays.
The spike recovery of these three compounds was within 96.00-101.21% (Table 7). These results suggested that the quantification method established was reliable and accurate for quantifying the content of compounds 5, 6, and 8 in the extract. As summarized in Table 7, the 70% MeOH root extract of V. rigida was proved to contain high contents of 5 (66.47 mg 3,5-diCQA/g extract), 6 (112.95 mg 4,5-diCQA/g extract), and 8 (23.26 mg acacetin/g extract).
where C 0 , C 1 , and C 2 are the concentrations of each compound tested in the standard solution, extract solution (the extract was prepared as 1.00 mg/mL), and the spiked solution, respectively. C 1 and C 2 are calculated from the corresponding calibration curve of each compound; V 0 , V 1 , and V 2 are the volumes of the standard solution used for sample spiking, the extract solution used for sample spiking, and the spiked sample solution, respectively. The content (µg/mg = mg/g) of each component in the extract was measured as C 1 (µg/mL)/extract concentration (1.00 mg/mL).
In addition, the present study suggests that the combination of offline DPPH-, ultra filtration-HPLC, and HSCCC/pH-zone-refining CCC can be used for efficient screening and separation of antioxidants and AR inhibitors from natural products. Compounds 1-7 were screened as target compounds with dual antioxidant and AR inhibitory activities using DPPH-HPLC and ultrafiltration-HPLC (Figures 1 and 2), and their activities were further confirmed by antioxidant and AR inhibition assays (Tables 4 and 5). In contrast, the nontarget compound 8 exhibited very weak antioxidant [67] and AR inhibitory activities in this study (Table 5), thus proving that as antioxidants and AR inhibitor screening tools, offline DPPH-HPLC and ultrafiltration-HPLC can improve the hit. Chemical reaction-based DPPH offline has been widely used to couple with HPLC [27,28] or even GC [68] for screening antioxidants, which also functioned well in this study, whereas AR-ligand binding affinity based ultrafiltration-HPLC has been less used [69][70][71]; however, no competitive experiments were carried out for those ultrafiltration-HPLC studies, which may lead to nonspecific binding of compounds with enzymes or centrifugal membrane, causing false positives [30,72]. The inactivation of AR by heating results in loss of its binding ability to AR inhibitors, which has been recently used to verify the screening result by comparing it with that obtained from the active AR group [73]. Nevertheless, the enzyme precipitates in the solution after heat deactivation in our test, thereby affecting the screening result by blocking the centrifugal membrane, as described previously [72]. Our result proves that the preincubation of AR with quercitrin, one of the most active natural AR inhibitors, can block the active site of AR, reducing the possibility of other AR inhibitors from binding to AR active site ( Figure 2). This suggests that introducing a quercitrin-blocked AR in the control group for verification can be used to reduce the false positives caused by nonspecific binding in AR ultrafiltration-HPLC assay, as those have been carried out in other enzymes [30,72].

Conclusions
In conclusion, the present study is the first to report that the 70% MeOH root extract of V. rigida possesses considerable in vitro antioxidant and AR inhibitory activities, and moreover, it demonstrated that the antioxidant and AR inhibitory activities of the extract were mainly attributed to the presence of seven caffeoylquinic acids 3-CQA (1), 4-CQA (3), 5-CQA (2), 3,4-di-CQA (4), 3,5-diCQA (5), 4,5-diCQA (6), and 3,4,5-triCQA (7) using offline DPPH-/ultrafiltration-HPLC analysis and DPPH, ABTS, ORAC, HOCl, and AR inhibition assays. The results indicated that the caffeoylquinic acids-rich root extract of V. rigida can act as a potential functional food or medicine ingredient for diabetes; therefore, further animal studies may be required to further assess its activity. Moreover, the results also suggested that the offline DPPH-/ultrafiltration-HPLC and HSCCC/pH-zone-refining CCC can be combined to efficiently screen and separate antioxidants and AR inhibitors from natural products.