Isolation of N-Ethyl-2-pyrrolidinone-Substituted Flavanols from White Tea Using Centrifugal Countercurrent Chromatography Off-Line ESI-MS Profiling and Semi-Preparative Liquid Chromatography

N-Ethyl-2-pyrrolidinone-substituted flavanols (EPSF) are marker compounds for long-term stored white teas. However, due to their low contents and diasteromeric configuration, EPSF compounds are challenging to isolate. In this study, two representative epimeric EPSF compounds, 5′′′R- and 5′′′S-epigallocatechin gallate-8-C N-ethyl-2-pyrrolidinone (R-EGCG-cThea and S-EGCG-cThea), were isolated from white tea using centrifugal partition chromatography (CPC). Two different biphasic solvent systems composed of 1. N-hexane-ethyl acetate-methanol-water (1:5:1:5, v/v/v/v) and 2. N-hexane-ethyl acetate-acetonitrile-water (0.7:3.0:1.3:5.0, v/v/v/v) were used for independent pre-fractionation experiments; 500 mg in each separation of white tea ethyl acetate partition were fractionated. The suitability of the two solvent systems was pre-evaluated by electrospray mass-spectrometry (ESI-MS/MS) analysis for metabolite distribution and compared to the results of the CPC experimental data using specific metabolite partition ratio KD values, selectivity factors α, and resolution factors RS. After size-exclusion and semi-preparative reversed-phase liquid chromatography, 6.4 mg of R-EGCG-cThea and 2.9 mg of S-EGCG-cThea were recovered with purities over 95%. Further bioactivity evaluation showed that R- and S-EGCG-cThea possessed in vitro inhibition effects on α-glucosidase with IC50 of 70.3 and 161.7 μM, respectively.


Introduction
Tea (Camellia sinensis, Theaceae) is one of the most consumed beverages in the world due to its health benefits and pleasant flavor [1,2]. In comparison to the more popular green and black teas, white tea (Chinese: bái chá) is a rare product, although it has become increasingly popular in recent years [3][4][5][6]. White tea is a slightly fermented tea, undergoing a prolonged withering and drying processes. White tea is classified into Silver Needle (also called Baihaoyinzhen), White Peony (also called Baimudan), Gong Mei, and Shou Mei according to the differences in the raw tea leaf materials used. In Fujian province of China, the main region producing white tea, long-term stored white tea is generally considered to have better health benefits and commands a much higher commercial price [5]. chin (EGCG), gallocatechin-gallate (GCG) and epicatechin-gallate (ECG) were effectively fractionated in the CPC experiments.

LC-ESI-MS/MS Analysis of Polyphenols in Stored White Tea
In the isolation study, the two target compounds, R-EGCG-cThea and S-EGCG-cThea (cf. Figure 1) were evaluated for optimization of suitable CPC two-phase solvent systems (cf. Section 2.2) for semi-preparative pre-fractionation. For removal of the existing predominant flavanols EGCG, GCG and ECG, the analytical investigation of the principal contents and concentrations was required. In the first step, a liquid chromatography-mass spectrometry (LC-ESI-MS) method was established to quantify the amounts of EGCG, GCG, ECG, R-EGCG-cThea and S-EGCG-cThea. As displayed in Figure 2, the ESI-mass-analysis of the two stereoisomers R-, and S-EGCG-cThea ([M + H] + , m/z 570) generated identical MS/MS fragment ion profiles, only showing a few intensity differences. The retention time analysis, using a C 18 -HPLC column, revealed that R-EGCG-cThea (t R : 28.8 min) had a slightly longer retention time than the respective epimer, S-EGCG-cThea (t R : 28 EGCG-cThea and S-EGCG-cThea. Further characteristic and principal tea polyphenols such as epigallocatechin (EGCG), gallocatechin-gallate (GCG) and epicatechin-gallate (ECG) were effectively fractionated in the CPC experiments.

LC-ESI-MS/MS Analysis of Polyphenols in Stored White Tea
In the isolation study, the two target compounds, R-EGCG-cThea and S-EGCG-cThea (cf. Figure 1) were evaluated for optimization of suitable CPC two-phase solvent systems (cf. Section 2.2) for semi-preparative pre-fractionation. For removal of the existing predominant flavanols EGCG, GCG and ECG, the analytical investigation of the principal contents and concentrations was required. In the first step, a liquid chromatography-mass spectrometry (LC-ESI-MS) method was established to quantify the amounts of EGCG, GCG, ECG, R-EGCG-cThea and S-EGCG-cThea. As displayed in Figure 2, the ESI-massanalysis of the two stereoisomers R-, and S-EGCG-cThea ([M + H] + , m/z 570) generated identical MS/MS fragment ion profiles, only showing a few intensity differences. The retention time analysis, using a C18-HPLC column, revealed that R-EGCG-cThea (tR: 28.8 min) had a slightly longer retention time than the respective epimer, S-EGCG-cThea (tR:

Evaluation and Prediction of Compound Specific Partition Ratio (K) Values by LC-ESI-MS Analysis of White Tea Extract
Because of the close HPLC retention times of EGCG, GCG, ECG, R-EGCG-cThea and S-EGCG-cThea (cf. Figure 2), we speculated that it is difficult to directly separate low-content R-EGCG-cThea and S-EGCG-cThea from high-content EGCG, GCG, and ECG using preparative C 18 -HPLC. A two-step separation method was thus used; the CPC was firstly applied to isolate and concentrate low-content EPSF compounds from white tea, then semi-preparative LC was used for epimeric separation of R-EGCG-cThea and S-EGCG-cThea.
As the CPC separation was in the so-called descending-mode using the more organic phase as stationary phase, the following equation (peak area A upper phase/ peak area A lower phase) was used for the prediction of metabolite specific K D -values (cf. Supplementary Material, Figure S1a,b). The polyphenol crude partition dissolved in the two-phase solvent systems HEMWat (experiment: CPC-1) and HEAWat (CPC-2) resulted in rather low K Dratios for the target compound mixture R-/S-EGCG-cThea, with values of 0.41 and 0.64, respectively (cf. Table 1, Supplementary Material, Figure S1a,b). The predicted K D -values for the classical tea polyphenols such as EGCG and ECG ranged at much higher K D -values (0.8 to 4.8) (cf. Table 1) and suggested a successful clean-up process for the CPC separation stage. The two selected systems also met the empirical rule that suitable K D -values of target compounds in liquid-liquid countercurrent chromatography should be in the range of 0.5 ≤ K D ≤ 1.0 [26]. As the elution-extrusion methodology of Berthod et al. [27] was applied, larger metabolite K D -values were also suitable for recovery of compounds in the extrusion-stage of the CPC operation.

Evaluation and Prediction of Compound Specific Partition Ratio (K) Values by LC-ESI-MS Analysis of White Tea Extract
Because of the close HPLC retention times of EGCG, GCG, ECG, R-EGCG-cThea and S-EGCG-cThea (cf. Figure 2), we speculated that it is difficult to directly separate lowcontent R-EGCG-cThea and S-EGCG-cThea from high-content EGCG, GCG, and ECG us-   Figure S1 Exp. Peak Range Fractions Retention Vol.

Solv. System 2 K D -Prediction (LC-MS) Cf. Details in Supplementary
Material, Figure S1 Exp. Peak Range Fractions Retention Vol.
[mL] Peak Width [mL] * The isobars 459-ECGC and 459-CGC were not base-line separated by CPC-1 and CPC-2, therefore the determined K D -ranges were estimated from the off-line ESI-MS profiling experiment.

Extraction and Liquid/Liquid Partitioning
The white tea leaves were crushed and macerated with a solvent mixture of methanol and water (70:30, v/v). The methanolic extract was diluted with water at a ratio of 1:2 (v/v), then liquid-liquid partitioned between N-hexane, dichloromethane and ethyl acetate, respectively. Thin-layer chromatography and LC-ESI-MS analysis of the N-hexane, dichloromethane, ethyl acetate, and water phase showed that R-EGCG-cThea and S-EGCG-cThea were fortified and mainly extracted to the ethyl acetate phase.

Semi-Preparative CPC Fractionation and HPLC-Purification of R-and S-EGCG-cThea
Two semi-preparative CPC separations were performed, each using 500 mg of white tea ethyl acetate partition but with two different solvent systems (HEMWat and HEAWat) (cf. Section 2.2), both being equivalent in polarity as evaluated by K D -values determined from the LC-ESI-MS analysis (cf. Section 2.2). The target CPC-fractions containing Rand S-EGCG-cThea were screened and identified by off-line ESI-MS injection profiling (cf. Section 3.5). As shown in Figure 3a Purification steps by size-exclusion chromatography on PVA 500 gel and semi-preparative C 18 -HPLC column using a previously established isocratic method [8,9] yielded pure epimers (6.4 mg R-EGCG-cThea and 2.9 mg S-EGCG-cThea) with purity > 95%.

Comparison of LC-ESI-MS Based Prediction versus CPC Experimental KD-VALUES
In this CPC study for R-/S-EGCG-cThea recovery, metabolite specific KD-values in the biphasic solvent system predicted by LC-ESI-MS were compared to the real experimental CPC based KD-values (cf. Table 1, Supplementary Material, Figure S1a,b). This process of KD-comparison required the conversion of the CPC experimental times into a partition ratio (KD)-based chromatography scale, as published by Tran et al. [24] and Grecco et al. [25]. This approach is a unique way to compare the separation performance of equipment with different technical designs, such as CPC or CCC. The complete process of KDcalculation is illustrated in the section Supplementary Material/Equations by using the experimental CPC chromatography parameters, such as retention volumes VR, CPC rotor

Comparison of LC-ESI-MS Based Prediction Versus CPC Experimental K D -VALUES
In this CPC study for R-/S-EGCG-cThea recovery, metabolite specific K D -values in the biphasic solvent system predicted by LC-ESI-MS were compared to the real experimental CPC based K D -values (cf. Table 1, Supplementary Material, Figure S1a,b). This process of K D -comparison required the conversion of the CPC experimental times into a partition ratio (K D )-based chromatography scale, as published by Tran et al. [24] and Grecco et al. [25]. This approach is a unique way to compare the separation performance of equipment with different technical designs, such as CPC or CCC. The complete process of K D -calculation is illustrated in the section Supplementary Material/Equations by using the experimental CPC chromatography parameters, such as retention volumes V R , CPC rotor column volume V C , stationary phase volume V S , and the stationary phase retention values S F with the required formulas in consecutive steps. The determination of K D in the extrusion-mode requires a second formula using the so-called switch-volume V CW to calculate the results until infinity (cf. Supplementary Material/Equation (S7b)) [28]. The off-line injection profiles are using selected single ion traces, therefore a high precision for the detection of metabolites with their respective clearly detected retention volumes V R is guaranteed (cf. Table 1) which is the basis for the K D -calculation in the CPC experiments.
The comparison revealed (cf. Table 1) that the LC-ESI-MS predicted K D -values for most of the metabolites matched well to the CPC derived K D -mean x values (using the peak centers of CPC elution), and rather small off-sets for the calculation of ∆K D values occurred, although a large discrepancy was detected for the metabolite 443-ECG recovered from the extrusion-process in the CPC runs. During extrusion both phases are recovered, thus inconsistencies in the process are more likely to occur. It was very obvious that both CPC experiments displayed wide retention volume windows (V R ), or ∆K D -ranges for metabolite elution (Table 1, Figure 3a,b). The addition of low concentrated acids to the solvent systems (such as the ion-pair forming reagent trifluoro acetic acid, TFA) might improve the CPC peak shapes, but could also lead to undesired artefact formation for recovered substances after vacuum-evaporation or lyophilization processing. Overall, both experiments with the HEMWat (CPC-1) and HEAWat (CPC-2) solvent system family displayed very similar K Dvalues in the prediction (cf. Table 1). The polarity of biphasic solvent system for all-liquid chromatography could be evaluated and compared using a method suggested by Abbott and Kleiman (1991) where Reichardt's dye indicates by the wavelength of absorbance of the polarity attributes of the phase layers [28]. The visual results for the CPC-1 and CPC-2 solvent systems are displayed in Supplementary Material Figure S2, and indicate similar attributes of polarity.

Comparing Chromatographic Performance of CPC-1 and CPC-2
For the comparison of the chromatographic performance of CPC-1 and CPC-2 on white tea metabolites, the specific CPC mean x K D -values (Table 1) were used to calculate separation factors α and resolution factors R S . The procedure of calculations is displayed in Supplementary Material/Equations, and results for α− and R S -values are given in Table 2.
Both CPC separations resulted in rather small α-values for the metabolite pair 570 R-/S-EGCG-cThea-459-EGCG (cf. Table 2, Figure 3a,b) (CPC-1: α 1.34 and CPC-2: α 1.11) with partial co-elution. These values indicated the insufficient separation of the target molecule mixture from 459-EGCG. A value of α > 1.5 would indicate a base-line separation. Nevertheless, the results from the off-line MS profile in both CPC runs proved that the largest amounts of principal polyphenols were clearly separated.
Unfortunately, the LC-ESI-MS analysis conducted on some of the respective target fractions of 570 R/S-EGCG-cThea gave no indication of pre-fractionation of Rand S-EPSFepimers by the CPC-steps using either solvent system.
The isobar pair 459-EGCG-459-GCG is partly fractionated, as the peak widths and co-elution areas could be solely estimated by interpolation of the Gaussian peak-shapes. Further LC-ESI-MS analysis with respective retention times could clarify the exact elution windows or peak widths of the diastereomers in CPC.
The compound 443-ECG was fully separated from all existent polyphenolic compounds, as seen by largest α-values and recovery from CPC in the extrusion-mode.
A certain contradiction is seen in the results for CPC-2 as a better fractionation of the pair 459-GCG-443-ECG was achieved than for CPC-1 (cf. Figure 3a,b), although the stationary phase retention S F displayed solely 76.0%, compared to 84.5% ( Table 2). The reason for this could be that the extrusion mode was started at a larger retention volume in CPC-2, and therefore better fractionation on the CPC rotor column might have occurred due to an extended time of elution.
The resolution factor values R S for CPC-1 and CPC-2 were calculated (Table 2) for the mentioned polyphenolic pairs, but were not further used for the discussion of chromatographic performance. The extremely large peak widths for elution/extrusion in CPC-1 and CPC-2 resulted in unacceptably low R S -values, as the peak broadness is a parameter in the equation (cf. Supplementary Material/Equation (S9)) [29]. The separation factor α solely uses the mean K D -value, and neglects the real peak width, as only the centers are used for calculations [29]. Nevertheless, it could be stated that α-values provide a much better assessment of the performance of broadened peak widths such as in these CPC applications. Overall, the visual inspection of separation by selected single ion traces (Figure 3a,b) clearly revealed that CPC-1 and CPC-2 provided a very efficient fractionation in the case of white tea polyphenols.

Inhibition Effect of R-EGCG-cThea and S-EGCG-cThea on α-Glucosidase
White tea was reported to have potential anti-hyperglycemic effects. For example, Xu et al. found that white teas have potent inhibitory in vitro effects on α-glucosidase and α-amylase activity, which are key enzymes related to type II diabetes [30]. The administration of white tea ethanolic extract for 14 days showed a significant effect on reducing fasting glucose level in streptozotocin-nicotinamide induced diabetic rats [31]. In addition, EGCG was reported to have potent inhibitory activity on α-glucosidase, with low IC 50 concentration [32]. In this study, the inhibition effects of the isolated compounds of R-EGCG-cThea and S-EGCG-cThea, as well as their precursor compounds (EGCG and theanine) on α-glucosidase, were evaluated ( Figure 4). The IC 50 of EGCG was calculated to be 128.8 µM (Figure 4). Theanine showed no inhibitory effect on α-glucosidase. The IC 50 of the epimer compounds Rand S-EGCG-cThea were found to be 70.3 and 161.7 µM, respectively ( Figure 4). These results indicated differences that the inhibitory effect of EGCG on α-glucosidase was retained after N-ethyl-2-pyrrolidinone substitution at the C-8 position. However, the R-configuration R-EGCG-cThea displayed a higher α-glucosidase inhibitory activity than the S-form of N-ethyl-2-pyrrolidinone-substituted EGCG (S-EGCG-cThea). These results indicated differences in bioactivity between EPSF stereoisomers which are similar to the results obtained in our previous study, where R-EGCG-cThea showed slightly stronger anti-inflammatory property (i.e., suppressed the expression of NF-κB-p65) compared to S-EGCG-cThea in lipopolysaccharide-stimulated macrophages [17].

Inhibition Effect of R-EGCG-cThea and S-EGCG-cThea on α-Glucosidase
White tea was reported to have potential anti-hyperglycemic effects. For example, Xu et al. found that white teas have potent inhibitory in vitro effects on α-glucosidase and αamylase activity, which are key enzymes related to type II diabetes [30]. The administration of white tea ethanolic extract for 14 days showed a significant effect on reducing fasting glucose level in streptozotocin-nicotinamide induced diabetic rats [31]. In addition, EGCG was reported to have potent inhibitory activity on α-glucosidase, with low IC50 concentration [32]. In this study, the inhibition effects of the isolated compounds of R-EGCG-cThea and S-EGCG-cThea, as well as their precursor compounds (EGCG and theanine) on α-glucosidase, were evaluated ( Figure 4). The IC50 of EGCG was calculated to be 128.8 μM (Figure 4). Theanine showed no inhibitory effect on α-glucosidase. The IC50 of the epimer compounds R-and S-EGCG-cThea were found to be 70.3 and 161.7 μM, respectively ( Figure 4). These results indicated differences that the inhibitory effect of EGCG on α-glucosidase was retained after N-ethyl-2-pyrrolidinone substitution at the C-8 position. However, the R-configuration R-EGCG-cThea displayed a higher α-glucosidase inhibitory activity than the S-form of N-ethyl-2-pyrrolidinone-substituted EGCG (S-EGCG-cThea). These results indicated differences in bioactivity between EPSF stereoisomers which are similar to the results obtained in our previous study, where R-EGCG-cThea showed slightly stronger anti-inflammatory property (i.e., suppressed the expression of NF-κB-p65) compared to S-EGCG-cThea in lipopolysaccharide-stimulated macrophages [17].

Preparation of Sample Extract
125 g of a three-year stored white tea (stored in Fuding city, Fujian province, China, naturally preserved in a storehouse maintained at room temperature and 25−50% relative humidity) was ground into powder using a laboratory mill (M20, IKA, Germany) and extracted with 1875 mL methanol/water solution (70:30, v/v) at 70 • C for 1 h. After filtration with a paper filter at room temperature, the solution was distilled over vacuum to remove parts of methanol. Then, the extract solution was diluted with Nanopure ® water to the ratio 1:2 (v/v) for a liquid-liquid solvent partitioning procedure. Therefore, 200 mL N-hexane eight times and 200 mL dichloromethane eight times were used for defatting and removing semi-lipophilic metabolites. The resulting methanolic water phase was further extracted with 350 mL ethyl acetate ten times. After thin-layer chromatography analysis of the N-hexane, dichloromethane, ethyl acetate and water phase, the ethyl acetate phase partition was lyophilized, yielding 30.87 g of powder.

Evaluation of Partition Ratio (K D ) Values Using LC-ESI-MS
The partition ratio (K D ) values of the target compounds (R-EGCG-cThea and S-EGCG-cThea) were evaluated by LC-ESI-ion trap-MS/MS analysis (HCT Ultra ETD II, Bruker Daltonics, Bremen, Germany). Two suitable two-phase solvent systems composed of HEMWat (N-hexane-ethyl acetate-methanol-water) and HEAWat (N-hexane-ethyl acetateacetonitrile-water) were tested in this study. Vials (20 mL) containing 1:1 mixture of the upper and lower phase (each 5 mL) of the above mentioned two-phase solvent systems were prepared. The vials were shaken vigorously with a vortex to achieve equilibration between upper and lower phases. After 10 min separation time of the phase layers, 1 mL of each phase was transferred separately into new vials and approximately 4 mg of the ethyl acetate soluble partition of white tea was added and mixed vigorously with a vortex. Aliquot volumes of the re-separated phase layers were injected to LC-ESI-MS for peak area analysis and prediction of target compound K D -values (cf. results Table 1). The K D value of each target compound was calculated as the mass intensity in areas in the stationary phase (upper phase) divided by the area value of the mobile phase (lower phase) (cf. Supplementary Material, Figure S1a,b, Equation (S1)).

CPC Isolation of the Mixture R-and S-EGCG-cThea from White Tea
The fast centrifugal partition chromatography apparatus (FCPC-Kromaton Technologies, Annony, France) was operated with a stainless semi-preparative rotor with mounted stainless steel separation disks (column volume V C : 200 mL). After gentle equilibration of the required solvents in a separatory funnel, the two phase layers were separated and briefly degassed before use in an ultrasonic bath. The prepared solvents were pumped with a preparative LC pump (solvent delivery system, K-501, Knauer Wissenschaftliche Geräte GmbH, Berlin, Germany). After complete filling of the stationary phase to the rotor column, the CPC system was started at 1000 rpm, while the mobile phase (lower phase) was pumped in descending-mode to reach the hydrodynamic equilibrium. In both CPC experiments, approximately 500 mg of CPC powder from liquid-liquid partitioning (cf. Section 3.2) was dissolved in 4.0 mL of each phase of the solvent system and filtered over a Chromafil Xtra GF-100/25 fiberglass membrane disc filter (1 µm pore size, 25 mm i.d., Macherey & Nagel, Düren, Germany), and then injected over a high-pressure injection system (Knauer Wissenschaftliche Geräte GmbH, Berlin, Germany) with a peek loop of 10 mL to the CPC system. The flow rates for CPC-1 and CPC-2 were 2.0 and 3.0 mL/min, respectively. After finishing the CPC separations in the so-called descending elution-mode (top-down direction in the rotor system) with the upper organic stationary phase and aqueous mobile phase, the biphasic solvent system which still contained metabolites in the column system was pushed out by pumping of the stationary phase at a lower spinning velocity (approx. 500 rpm) to the fraction collector. The recovery system was a SuperFrac (Pharmacia, Uppsala, Sweden), with type B racks collecting fractions during the elutionand extrusion process at every second minute. Aliquots of the in-sequence collected fractions of CPC-1-and CPC-2 were filled to HPLC vials (cf. Section 3.5) to perform the off-line metabolite elution profile by sequential ESI-MS/MS injections (cf. Figure 3a,b).

Off-Line ESI-MS/MS Injection Analysis for CPC Molecular Weight Metabolite Profiling
For a molecular weight guided detection of white tea metabolites from the semipreparative CPC experiments CPC-1 and CPC-2, aliquot volumes of recovered fractions were off-line injected to an ESI-MS/MS-ion-trap mass-spectrometer (cf. system 3.3) in the sequence of the chromatographic elution and extrusion process. By selecting the full base-peak chromatogram trace (BPC) and selected single ion target traces, the semi-preparative results were projected as molecular weight-based chromatograms, which guided the accurate fractionation process. The acquired MS-data, including MS/MS fragmentation of seven precursor ions, were recorded in single data files. For sample preparation, aliquot volumes from CPC-fractions were filled to the HPLC vials. The off-line injections to the ESI-MS/MS were conducted by an independent and programmable autosampler system (AS-2000A, Merck-Hitachi, Tokyo, Japan). The injected fraction volumes (3 µL) were delivered to the ESI-MS device by an HPLC-pump (binary pump, G1312 A, 1100 Series, Agilent, Waldbronn, Germany) at a flow rate of 0.5 mL/ min, and the make-up solvent system used was composed of ACN/H 2 O (7:3, v/v). The time interval between the re-occurring injections was set to 2 min. Every observed injection peak corresponded to one injected vial from the CPC experiments. The selected ESI-MS traces (Figure 3a,b) displayed the mass spectrometry profile information of the compounds located in the respective CPC fraction(s).
The highest ion intensities in the generated base-peak chromatograms (BPC) at the scanning range of m/z 100-2000 were observed at 4.0 × 10 7 , presenting a good signal-to-noise ratio. The N-ethyl-2-pyrrolidinone-substituted EGCG compounds yielded a sensitive ion response in the positive ESI-ionization mode.

α-Glucosidase Inhibition Assay of R-EGCG-cThea and S-EGCG-cThea
The determination of α-glucosidase inhibitory activity of R-EGCG-cThea and S-EGCG-cThea compared to EGCG and theanine was carried out according to the method described by Xu et al. [30], with slight modifications. Compound solution (50 µL) was mixed with 0.1 M phosphate buffer (100 µL, pH 6.9 with 1 U/mL α-glucosidase). Then, the mixture was incubated in 96 well plates at 25 • C for 10 min. After incubation, 50 µL of 0.1 M phosphate buffer solution (pH 6.9 with 5 mM pNPG) was added into each well at timed intervals. Subsequently, the reaction mixture was incubated at 25 • C for 5 min. Before and after incubation, absorbance values were measured at λ 405 nm by a microplate reader (Infinite ® M200 Pro, TECAN, Männedorf, Switzerland). Buffer solution instead of compound solution was utilized as the control sample. The α-glucosidase inhibitory activity was presented as inhibition percentage, and was calculated as in the following equation: inhibition ratio (%) = (1 − As/Ac) × 100, where As and Ac are the absorbance values of the compound solution and of the control, respectively.

Conclusions
In the present study, centrifugal partition chromatography (CPC) and semi-preparative HPLC chromatography were used to isolate the low-concentrated epimeric EPSF compounds R-EGCG-cThea and S-EGCG-cThea from white tea extract. All-liquid chromatography provided complementary separation characteristics to solid phase chromatography, and the loss of target compounds by chemisorption was avoided [26]. Two semi-polar two-phase solvent systems from the HEMWat and HEAWat family were optimized to reach suitable partition ratio K D values and separation factors α between the EPSF target compounds and the residual principal tea polyphenolic compounds. The metabolite profiling by in-sequence off-line injection of CPC fractions to the ESI-MS detector traced with absolute precision the very minor concentrated R-/S-EGCG-cThea metabolites in the semi-preparative fractions. The CPC process with two different semi-polar solvent systems enabled recovery at low CPC retention volumes and clearly fractionated the other tea polyphenols, which displayed much larger K D -values. This two-dimensional chromatography approach (2D: CPC x C 18 HPLC) delivered the pure diastereomers. As the retention times for C 18 -HPLC of the target metabolites appeared to be very similar (cf. Supplementary Material, Figures S1a and S2b), the 2D-approach appears to be the most effective method for large-scale isolation of these minor concentrated metabolites from white tea. A scale-up of the isolation protocol is feasible, as industrial sized CPC devices are available and the evaluated solvent systems could be easily transferred. An exemplary bioactivity evaluation showed that the in vitro inhibition IC 50 of Rand S-EGCG-cThea on α-glucosidase were 70.3 and 161.7 µM, respectively.