A 42-Markers Pharmacokinetic Study Reveals Interactions of Berberine and Glycyrrhizic Acid in the Anti-diabetic Chinese Medicine Formula Gegen-Qinlian Decoction

Herbal medicines are commonly used as compound formulas in clinical practice to achieve optimal therapeutic effects. However, the combination mechanisms usually lack solid evidence. In this study, we report synergistic interactions through altering pharmacokinetics in Gegen-Qinlian Decoction (GQD), an anti-diabetic Chinese medicine formula. A multi-component pharmacokinetic study of GQD and the single herbs was conducted by simultaneously monitoring 42 major bioactive compounds (markers) in rats plasma using LC/MS/MS within 30 min. GQD could remarkably improve the plasma concentrations of berberine (BER) and other alkaloids in Huang-Lian by at least 30%, and glycyrrhizic acid (GLY) from Gan-Cao played a major role. A Caco-2 cell monolayer test indicated that GLY improved the permeability of BER by inhibiting P-glycoprotein. Although GLY alone did not show observable effects, the co-administration of GLY (ig, 50 or 80 mg/kg) could improve the anti-diabetic effects of berberine (ig, 50 mg/kg) in db/db mice in a dose-dependent manner. The blood glucose decreased by 46.9%, whereas the insulin level increased by 40.8% compared to the control group. This is one of the most systematic studies on the pharmacokinetics of Chinese medicine formulas, and the results demonstrate the significance of pharmacokinetic study in elucidating the combination mechanisms of compound formulas.


Experimental
Herbal materials and extracts.
The four component herbs, Puerariae Lobatae Radix (Ge-Gen, GG), Scutellariae Radix (Huang-Qin, HQ), Coptidis Rhizoma (Huang-Lian, HL), and Glycyrrhizae Radix et Rhizoma Praeparata cum Melle (Gan-Cao, GC) were purchased from Tianheng Pharmacy. GQD was prepared by decocting the four component herbs (Ge-Gen 25.04 g, Huang-Qin 9.43 g, Huang-Lian 9.43 g, Gan-Cao 6.31 g) in 400 mL of water for three times (1.5 h, 1.5 h, 0.5 h) to obtain the extracts, followed by a pre-extraction of Ge-Gen using 200 mL of water for 0.5 h. The decoctions were combined, filtered, and concentrated in vacuum at 50 °C. GQD-GC was prepared using the same procedure, except that Gan-Cao was removed from the formula. Final concentration of the GQD-GC extract was equivalent to 1.0 g of the crude drugs per mL. For single herb extracts, the herbs were decocted in 8-fold volume of water for three times (1.5 h, 1.5 h, 0.5 h) and then concentrated. Final concentrations of the extracts were 1.0 g/mL for Ge-Gen, and 0.5 g/mL for Huang-Qin, Huang-Lian and Gan-Cao.

Animals
For pharmacokinetic study, Male Sprague-Dawley rats (220-250 g) were obtained from the Laboratory Animal Center of Peking University Health Science Center. The rats were bred in a cage (465 × 300 × 200 mm) in a breeding room at 25 °C, 60 ± 5% humidity, and a 12-h dark-light cycle for 3 days. The rats had free access to tap water and soy-free custom diet (Ke'ao Xieli Co.). All procedures were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Research Council, 2011).
For anti-diabetic study, Eight-week-old male db/db mice (30-35 g) were purchased from Card Vince Laboratory animal Co., Ltd. The mice were bred in a cage (318 × containing glucose (25 mM) and HEPES (10 mM). The final concentration of berberine was 5 µM. The final concentration of DMSO was below 1% (v/v).
Verapamil (5 µM) and rifampicin (5, 10, and 20 µM) were used as P-gp inhibitor and inducer of Caco-2 cells, respectively. For the inhibitor experiments, the Caco-2 cells were cultivated in DMEM until the 21 st day. After removal of the medium, the cells were incubated for 30 minutes with HBSS containing 5 µM of verapamil. For the inducer experiments, the cells were cultivated in DMEM until the 18 th day, and then cultivated in DMEM containing rifampicin (5, 10, and 20 µM for 24 h). The medium was removed on the 21 st day, and the cells were then incubated for 30 min with HBSS.
Berberine and glycyrrhizic acid were added to the system following HBSS incubation.
For all bidirectional transport assays, samples (300 µL) were collected after 30, 60, 90, 120, 150 and 180 min of incubation, and the culture was immediately replenished with an equal volume of HBSS. The samples were freeze-dried and then dissolved in 300 µL of methanol. The solutions were filtered through 0.22-µm membranes before LC/MS/MS analysis.

Inhibitor and inducer experiments
Verapamil and rifampicin were used as P-gp inhibitor and inducer of Caco-2 cells, respectively. For the inhibition experiments, Caco-2 cells were cultivated in DMEM until the 21 st day. After removal of the medium, the cells were incubated for 30 minutes with HBSS containing verapamil (5 µM). For induction experiments, the cells were cultivated in DMEM until the 18 th day, and then cultivated in DMEM containing rifampicin (5, 10, and 20 µM) for 24 h. The medium was removed on the 21 st day, and the cells were then incubated for 30 min with HBSS. Berberine and glycyrrhizic acid were added to the system following HBSS incubation.
For all bidirectional transport assays, samples (300 µL) were collected after 30, 60, 90, 120, 150 and 180 min of incubation, and the culture was immediately replenished with an equal volume of HBSS. The samples were freeze-dried and then dissolved in 300 µL of methanol. The solutions were filtered through 0.22-µm membranes before LC/MS/MS analysis.

Experimental details for the PK study 1. Preparation of internal standard solution, calibration standard solutions, and quality control solutions
Butein 4-O-glucoside (0.1 µg/mL) was dissolved in 67% methanol to prepare the internal standard solution. Reference standard compounds were dissolved in the internal standard solution to prepare individual stock solutions (1.0 mg/mL). These stock solutions were mixed to form a standard solution, containing 23800 ng/mL of each compound. The standard solution was then serially diluted to obtain calibration standard solutions (11900, 5950, 2975, 1190, 595, 297.5, 119, 59.5, 29.75, 11.9, 5.95, 2.975, and 1.19 ng/mL for each compound). Quality control stock solutions were prepared at three concentration levels as high QC (HQC), middle QC (MQC), and low QC (LQC), based on linear ranges of the analytes.

Sample preparation for pharmacokinetic study
The calibration standard and quality control solutions for pharmacokinetic study were prepared with the same procedures for plasma samples as described below. For these samples, 150 µL of stock solution was evaporated to dryness under a gentle stream of nitrogen, and then 150 µL of blank rat plasma, 150 µL of internal standard solution, and 300 µL of acetonitrile were added. The mixture was vortexed (2200 rpm) for 2 min, and ultrasonicated in a water bath for 5 min. Then, the mixture was vortexed for another 2 min, and centrifuged (9000 rpm, 4 °C) for 10 min. The supernatant was separated and evaporated to dryness at 37 °C using a speedvac concentrator. The residue was stored at -80 °C and redissolved in 150 µL of 67% methanol before analysis. All samples were filtered through 0.22-µm membranes. A 5-µL aliquot was injected for LC/MS/MS analysis.

Sample preparation for anti-diabetic activity study
On day 28, db/db mice were fasted for 12 h, and blood was collected and centrifuged to obtain serum for insulin and biochemical tests. After that, 150 µL of plasma was prepared by adding 150 µL of internal standard solution, 150 µL of methanol, and 300 µL of acetonitrile. The processing method was consistent with sample preparation for pharmacokinetic study.

Optimization of chromatographic separation
Different solvents (67% methanol, and 100% methanol) were compared to effectively extract both hydrophilic and hydrophobic compounds of GQD. 67% methanol showed better extraction recovery for most compounds. A number of columns were tested, and Waters XTerra C 18 column was selected to obtain the best peak shapes and chromatographic separation of the 42 markers. Different organic HPLC mobile phases (acetonitrile, and acetonitrile containing 2% (v/v) methanol) were compared. Good peak shapes were obtained when acetonitrile containing 2% (v/v) methanol was used.

Optimization of MS/MS conditions
(+)-ESI mode was used to obtain high ionization efficiency of alkaloids. To optimize MS parameters, 42 pure compounds were individually injected into (+)-ESI source by continuous infusion. MS fragment ions, tube lens offset voltage, and collision energy (CE) were optimized for all the 42 analytes. The optimized parameters are given in Table S1. To ensure the precision of sample extraction and MS detection, butein 4-O-glucoside was employed as internal standard.

Method validation
The method was validated in terms of specificity, linearity, accuracy, precision, matrix effect, extraction efficiency, and stability in accordance with the USA Food and Drug Administration (FDA) bioanalytical method validation guidance (U.S. Food and Drug Administration, 2013).

Specificity
The specificity of the method was investigated by analyzing blank plasma samples from six rats. The chromatogram of each blank plasma sample was tested for interference using the proposed extraction procedure and compared with the spiked rat plasma. As a result, no significant endogenous interference was observed in the blank rat plasma (Fig. S1).

Linearity
The linearity of each calibration curve was determined by plotting the analyte / internal standard peak area ratio (y) against the concentrations of analytes (x). All the calibration curves showed good linearity with correlation coefficients (r 2 ) of >0.9900 (Table S2). The lower limit of quantification (LLOQ) was determined at signal-to-noise ratios of 10. The LLOQs varied from 1.19-29.75 ng/mL for the analytes. The regression equations, correlation coefficient, dynamic ranges and LLOQs were listed in Table S2.

Intra-and inter-day precision and accuracy
The intra-and inter-day precisions and accuracy were investigated by determining QC samples at three different concentrations (six replicates for each concentration level) in the same day for five times and on three consecutive days. The precision of the method at each QC concentration was expressed as the relative standard deviation (RSD) and the accuracy was described as relative error (RE). The RSD values of intra-and inter-day precisions ranged from 0.50% to 15.53% and 0.57% to 21.98%, respectively. The RSD and RE values of accuracy for LQC, MQC, and HQC ranged from 0.37% to 15.12% and -18.02% to 18.12%, respectively (Table S3).

Recovery and matrix effect
The extraction recoveries and the matrix effects of analytes were determined as we had previously reported (Qiao et al., 2012). The method recoveries were determined by comparing the nominal concentration of plasma QC samples to the measured concentration of QC samples (analytes were diluted in methanol). The extraction and method recoveries at three different QC concentrations were between 82.60%-118.68% and 81.56%-118.67%, suggesting that the method was accurate (Table S4). The matrix effects were not significant for most analytes, with ion suppression ranged from -11.34% to 7.33% (Table S4).

Stability
The freeze and thaw stability, short-term and long-term stabilities were determined at high and low concentrations after 24 h storage in the sample tray, after one, and twenty-day storage at −80 °C, respectively. All the analytes showed variations between -17.26% and 17.44% (Table S5). The results indicated that these analytes in rats plasma were all stable for twenty-day storage at −80 °C, 24 h in the auto-sampler (4 °C) and three freeze-thaw cycles.

Qualitative and quantitative analysis of Gegen-Qinlian Decoction
An