Abstract
The urinary metabolites of berberine, an isoquinoline alkaloid isolated from several Chinese herbal medicines, were investigated in rats and humans. Using macroporous adsorption resin chromatography, open octadecyl silane column chromatography and preparative high-performance liquid chromatography, we isolated seven metabolites (HM1-HM7) from human urine and five metabolites (RM1-RM5) from rat urine after oral administration. Their structures were elucidated by enzymatic deconjugation and analyses of mass spectrometry, 1H NMR, and nuclear Overhauser effect spectroscopy spectra. Besides the three known metabolites demethyleneberberine-2-O-sulfate (HM1 and RM3), jatrorrhizine-3-O-sulfate (HM5), and thalifendine (RM5), six new metabolites were identified, namely, jatrorrhizine-3-O-β-d-glucuronide (HM2), thalifendine-10-O-β-d-glucuronide (HM3), berberrubine-9-O-β-d-glucuronide (HM4 and RM2), 3,10-demethylpalmatine-10-O-sulfate (HM6 and RM4), columbamin-2-O-β-d-glucuronide (HM7), and demethyleneberberine-2,3-di-O-β-d-glucuronide (RM1). These findings suggest that berberine undergoes similar biotransformation in rats and humans. Possible metabolic pathways of berberine in rats and humans are proposed.
Berberine is an isoquinoline alkaloid isolated from several Chinese herbal medicines such as rhizoma coptidis, cortex phellodendri, and caulis mahoniae. Its chemical structure is 5,8-dihydro-9,10-dimethoxy-6H-benzo(g)-1,3-benzodioxolo(5,6-a)quinolizine. Berberine exhibits a wide variety of bioactivities such as antidiarrheic (Sack and Froehlich, 1982; Taylor and Greenough, 1989), antimicrobial (Yan et al., 2008), hypolipidemic (Kong et al., 2004; Doggrell, 2005; Brusq et al., 2006; Cicero et al., 2007), hypoglycemic (Yin et al., 2002; Pan et al., 2003; Zhou et al., 2007; Turner et al., 2008; Yin et al., 2008), antiarrhythmic (Wang et al., 1994; Lau et al., 2001), anticancer (Inoue et al., 2005; Lanvers-Kaminsky et al., 2006; Lin et al., 2006, 2007; Piyanuch et al., 2007; Serafim et al., 2008; Yu et al., 2007), anti-inflammatory (Kuo et al., 2004; Lee et al., 2007), antiviral (Hayashi et al., 2007), antidepressant (Kulkarni and Dhir, 2007), and hepatoprotective (Zhang et al., 2008) effects. Its chloride salt has been used for several decades in clinical situations to treat gastroenteritis and secretory diarrhea in China. In recent years, berberine has also shown significant effects in treatment of diabetes mellitus (Ni, 1988), hyperlipemia (Kong et al., 2004), arrhythmia, and heart failure (Zeng and Zeng, 1999; Lau et al., 2001; Zeng et al., 2003). However, pharmacokinetic studies have indicated that berberine has poor oral bioavailability (Shen et al., 1993; Yu et al., 2000; Zuo et al., 2006), and a few of the metabolites have been identified in rats (Zuo et al., 2006) and in humans (Pan et al., 2002). To obtain more information about its metabolism to improve its clinical applications, we examined the biotransformation of berberine in rats and humans. In the present article we describe the isolation and identification of urinary metabolites of berberine in these two species.
Materials and Methods
Materials. Berberine chloride (purity > 99.5%) was supplied by the Northeast General Pharmaceutical Factory (Shenyang, China). Methanol was HPLC grade and water was double-distilled in our laboratory. All other reagents were of analytical grade and were purchased from Shenyang Chemical Company (Shenyang, China). Normal-phase and reverse-phase preparatory thin-layer chromatography was performed using products from Merck (Darmstadt, Germany). Macroporous resin D101 was purchased from the Chemical Plant of Nankai University (Tianjin, China) and Diaion ion exchange resin HP20 from Mitsubishi Chemical Corporation (Tokyo, Japan). Sephadex LH-20 and ODS were obtained from Pfizer (Freiburg, Germany). Spots on chromatograms were detected with Dragendorff's reagent (a solution of potassium bismuth iodide), which is usually used to visualize alkaloids that have been developed using thin-layer chromatography. Arylsulfatase (EC 3.1.6.1) and β-glucuronidase (EC 3.2.1.31) were purchased from Sigma-Aldrich (St. Louis, MO).
Subjects and Dosing Procedure. Eighty male Wistar rats, from 8 to 10 weeks old and weighing 200 to 250 g, were used in the study of the urinary metabolites of berberine chloride in rats. Rats were kept in the breeding room for 7 days before the study. Normal food and water were available at all times but were withdrawn 12 h before intragastric administration of berberine chloride. Each rat was given an oral dose (100 mg/kg b.wt.), of berberine chloride and urine samples were collected for 48 h.
For the study of the urinary metabolites of berberine chloride in humans, 12 healthy volunteers from 22 to 26 years of age and weighing 60 to 80 kg (all male) participated in this study. Subjects were judged to be in good health on the basis of a medical history, physical examination, and laboratory profiles that were obtained within a 2-week period before the start of the study. The study followed the 1964 Declaration of Helsinki for the ethical treatment of human subjects for biomedical research. Each subject was given a 300-mg oral dose of berberine chloride, three times a day for 2 days, and urine was collected between 0 and 72 h.
Isolation of Metabolites. The urine samples (rat: 3800 ml; human: 16,000 ml) were chromatographed on macroporous adsorptive resin D101 and eluted with H2O and 95% EtOH. The 95% EtOH eluate was subjected to HP 20 column chromatography using an H2O-EtOH solvent system. The fractions eluted by 20 and 30% EtOH were then fractionated by chromatography on Sephadex LH-20 and RP-18 silica gel columns with an MeOH-H2O solvent system (0-30%). The fractions containing the metabolites underwent fractionation by Sephadex LH-20 and RP-8 and RP-18 silica-gel column chromatography with an MeOH-H2O solvent system (10-60%) and then were purified by preparative HPLC.
Purification by HPLC. Preparative HPLC was performed with an ODS column (XTerra RP18, 7 μm, 19 × 300 mm) in a Waters 600 liquid chromatograph equipped with a Waters 490 UV detector. The usual detection wavelength was 340 nm. Elution was carried out with MeOH-H2O at a flow rate of 10 ml/min. Elution with MeOH-H2O (15:85) yielded RM1 (2.3 mg). Elution with MeOH-H2O (16:84) yielded HM2 (6 mg), HM3 (4 mg), and HM7 (1 mg). Elution with MeOH-H2O (22:78) yielded HM1 (40 mg), HM4 (6.1 mg), HM5 (2.2 mg), and HM6 (1.2 mg). Elution with MeOH-H2O (55:45) yielded RM5 (10 mg).
Spectroscopic Methods. Electrospray ion trap mass spectrometry in multistage full scan mode was performed on a Bruker Esquire 2000 instrument with a mass range of 25 to 2200 (the mass was calibrated). The instrument was operated in the both positive and negative ion modes, using nitrogen as the nebulizing and dry gas. The collision-induced dissociation of the quasi-molecular ion was achieved with helium as the collision gas. The ionization was performed under the following conditions: capillary temperature, 250°C; and capillary voltage, 3.0 kV. Sample solutions were directly introduced into the ESI source at a flow rate of 3 μl/min by a syringe pump.
NMR spectra were measured on a Bruker ARX-600 spectrometer, and chemical shifts are given in parts per million with tetramethylsilane as an internal standard. HM1 to HM7 and RM5 were dissolved in DMSO-d6, and RM1 was dissolved in D2O-DMSO-d6 (1:2).
Enzymatic Hydrolysis. The metabolites (HM1-HM7 and RM1, each 0.2 mg) were dissolved in 0.1 M acetate buffer (pH 5.0). Then, arylsulfatase (100 units) or β-glucuronidase (5000 units) was added, and the solution was incubated for 4 h at 37°C. After this the reaction solution was chromatographed on an ODS mini-column eluting with distilled water to remove any impurities. The subsequent MeOH eluate was evaporated to give a residue, which was analyzed by ESI-MS.
Results
Isolation and Structure Elucidation of Berberine Metabolites. By means of macroporous adsorptive resin chromatography, open ODS column chromatography, and preparative high-performance liquid chromatography, five metabolites (RM1-RM5) from rat urine and seven metabolites (HM1-HM7) from human urine were isolated after oral administration of berberine chloride, and their structures were identified on the basis of enzymatic hydrolysis, ESI-MS, 1H NMR, and NOESY spectra.
In the ESI-MS spectra (Fig. 1), removal of masses of 80 or 176 from the quasi-molecular ions [M - Cl]+ of the metabolites showed that they were sulfate or glucuronide conjugates. The evidence provides useful information to help in the structural identification of the berberine metabolites.
Sulfate-Conjugated Metabolites. HM1 (RM3) and HM5 were identified as demethyleneberberine-2-O-sulfate and jatrorrhizine-3-O-sulfate, respectively, by enzymatic hydrolysis and further comparison of MS and NMR data (Tables 1 and 2) with those in the literature (Pan et al., 2002).
HM6 (RM4) was obtained as yellow needle crystals, giving a positive reaction to Dragendorff's reagent. ESI-MS of HM6 gave the quasi-molecular ion [M - Cl]+ at m/z 404 and [M - Cl + Na - H]+ at m/z 426. Bombardment of m/z 426 afforded a fragment ion [M - Cl + Na - H - SO3]+ at m/z 346. In combination with enzymatic hydrolysis data, HM6 was deduced to be a sulfate-conjugated metabolite.
In the 1H NMR spectrum, the skeleton proton signals (Table 2) of the parent drug remained except for those of the substituent groups. The linked positions of two methoxyl groups were established by the NOESY spectrum: the methoxyl signal at δ4.24 correlated with that of H-8 at δ9.83 and the methoxyl group at δ3.95 correlated with that of H-1 at δ7.72, indicating that two methoxyl groups were linked at C-9 and C-2. Thus, the sulfate group must be linked to C-3 or C-10. Furthermore, 9-OCH3 and H-11 shifted downfield by 0.1 ppm, and H-4 hardly shifted at all, so we concluded that the sulfate group was located at C-10. Thus, HM6 was identified as 3,10-demethylpalmatine-10-O-sulfate.
Glucuronide-Conjugated Metabolites. HM2 was obtained as a yellow powder, giving a positive reaction to Dragendorff's reagent. The ESI-MS of HM2 gave the quasi-molecular ion [M - Cl]+ at m/z 514. Bombardment of 514 afforded the fragment ion [M - Cl - C6H8O6]+ at m/z 338, suggesting that HM2 might be a glucuronide conjugate. This suggestion was further confirmed by enzymatic hydrolysis of HM2, in which the aglycone ion fragment at m/z 338 was detected by ESI-MS. Comparison of the 1H NMR data (Table 1) of the aglycone of HM2 with those of berberine indicated many similarities except for the loss of the characteristic methylene signal and the appearance of a new methoxyl signal.
The linked positions of the newly formed methoxyl group and the glucuronic acid moiety were established by the NOESY spectrum, in which H-1 (δ7.73) exhibited correlations with H-13 (δ9.04) and 2-OMe (δ3.93) and H-4 (δ7.19) exhibited correlations with H-1′ (δ5.14) and H-5 (δ3.17), indicating that the glucuronic acid moiety was located at C-3 and the newly formed methoxyl group was located at C-2. The β-configuration of the glucuronic acid moiety was identified on the basis of the coupling constants of the anomeric proton (δ5.14, J = 7.0 Hz). Therefore, the structure of HM2 was identified as jatrorrhizine-3-O-β-d-glucuronide.
HM3 was obtained as a yellow powder, giving a positive reaction to Dragendorff's reagent. The ESI-MS of HM3 gave the quasi-molecular ion [M - Cl]+ at m/z 498. Bombardment of 498 afforded the fragment ion [M - Cl- C6H8O6]+ at m/z 322, suggesting that HM3 might be a glucuronide conjugate. This suggestion was further confirmed by enzymatic hydrolysis of HM3, in which the aglycone ion fragment at m/z 322 was detected by ESI-MS. In comparison with the 1H NMR spectrum of berberine, the other proton signals (Table 1) were observed in HM2 except for the loss of one methoxy group. The linked positions of the remaining methoxyl group and the glucuronic acid moiety were established from the NOESY spectrum. In this, H-11 (δ8.14) exhibited correlations with H-12 (δ7.89) and H-1′ (δ5.15), indicating that the glucuronic acid moiety was located at C-10 and the remaining methoxyl group was consequently located at C-9. The β-configuration of the glucuronic acid moiety was confirmed on the basis of the coupling constants of the anomeric proton (δ5.15, J = 7.5 Hz). Thus, the structure of HM3 was identified as thalifendine-10-O-β-d-glucuronide.
HM4 (RM2) was obtained as a yellow amorphous powder, giving a positive reaction to Dragendorff's reagent. The ESI-MS of HM4 gave the quasi-molecular ion [M - Cl]+ at m/z 498. Bombardment of 498 afforded the fragment ion [M - Cl - C6H8O6]+ at m/z 322, suggesting that HM4 was a glucuronide conjugate. This suggestion was further confirmed by enzymatic hydrolysis of HM4. Comparison of the 1H NMR data (Table 1) of HM4 with those of HM3 indicated that HM4 was an isomer of HM3. The linked positions of the methoxyl group were established from the NOESY spectrum, in which H-8 (δ10.15) exhibited correlations with H-6 (δ5.10, δ4.85) and H-11 (δ8.19) exhibited correlations with H-12 (δ8.04) and the methoxyl group (δ4.04). Therefore, the methoxyl group was located at C-10 and consequently the glucuronic acid moiety must be located at C-9. The β-configuration of the glucuronic acid moiety was confirmed on the basis of the coupling constants of the anomeric proton (δ4.82, J = 7.0 Hz). Thus, the structure of HM4 was identified as berberrubine-9-O-β-d-glucuronide.
HM7 was obtained as a yellow powder, giving a positive reaction to Dragendorff's reagent. The ESI-MS of HM7 gave the quasi-molecular ion [M - Cl]+ at m/z 514. Bombardment of 514 afforded the fragment ion [M - Cl - C6H8O6]+ at m/z 338, suggesting that HM7 might be an isomer of HM2. This suggestion was further confirmed by enzymatic hydrolysis of HM7, in which the aglycone ion fragment at m/z 338 was detected by ESI-MS. A comparison of the 1H NMR spectrum of HM7 with that of HM2 showed that there were many similarities except for the downfield chemical shifts of H-1 by δ0.36 (Tables 1 and 2). The linked positions of the methoxyl group (δ3.88) and the glucuronic acid moiety were established from the NOESY spectrum, in which H-1 (δ8.09) correlated with both H-13 (δ9.15) and H-1′ (δ5.06) and H-4 (δ7.12) correlated with the methoxyl group (δ3.88), indicating that the glucuronic acid moiety was located at C-2 and the methoxyl group (δ3.88) was consequently located at C-3. The β-configuration of the glucuronic acid moiety was confirmed on the basis of the coupling constants of the anomeric proton (δ5.06, J = 7.5 Hz). Therefore, the structure of HM7 was identified as columbamine-2-O-β-d-glucuronide.
RM1 was obtained as a yellow powder, giving a positive reaction to Dragendorff's reagent. The ESI-MS of RM1 gave a quasi-molecular ion [M - Cl]+ at m/z 676. Bombardment of 676 afforded the fragment ion [M - Cl - C6H8O6]+ at m/z 500, and consequent bombardment of 500 yielded [M - Cl - 2 × C6H8O6]+ at m/z 324, suggesting that RM1 might be a diglucuronide conjugate. This suggestion was further confirmed by enzymatic hydrolysis of RM1, in which the aglycone ion fragment at m/z 324 was detected by ESI-MS. In comparison with the 1H NMR spectrum of berberine, most of the skeleton proton signals (Table 2) remained in RM1 except for loss of the characteristic methylene signal. The linked positions of the two glucuronic acid moieties and the two methoxyl groups were established by NOESY. The two anomeric protons at δ5.12 (d, J = 6.0 Hz) and δ5.04 (d, J = 7.0 Hz) correlated with H-1 (δ7.82) and H-4 (δ7.17), respectively, indicating that the two glucuronic acid moieties were located at C-2 and C-3, respectively. The two methoxyl groups at δ3.97(s) and δ4.01(s) correlated with H-11 (δ8.03, d, J = 9.0 Hz) and H-8 (δ9.63), respectively, suggesting that they should be linked to C-10 and C-9, respectively. The β-configurations of the two glucuronic acids moieties were confirmed from the coupling constants of the two anomeric protons (δ5.12, J = 6.0 Hz; δ5.04, J = 7.0 Hz). Therefore, the structure of RM1 was finally determined as demethyleneberberine-2,3-di-O-β-d-glucuronide. Unconjugated Metabolite. RM5 was obtained as a red powder, giving a positive reaction to Dragendroff's reagent. The ESI-MS of RM5 gave the quasi-molecular ion [M - Cl]+ at m/z 322, which is 14 mass units less than that of berberine. Comparison of the 1H NMR data (Table 2) with those of berberine showed there were many similarities except for the disappearance of one methyl group. The remaining methoxyl group was assigned to 9-OMe on the basis of the NOESY spectrum, in which the methoxyl signal (δ4.05) showed a correlation with H-8 (δ9.71). After comparison of the 1H NMR data with those in the literature (Niu et al., 2006), RM5 was identified as thalifendine.
Discussion
Structural elucidation of metabolites is one of the most challenging tasks in drug metabolism studies. In recent years, comparisons of ESI-MSn data and HPLC retention times with synthetic standards usually have been used to identify the structures of metabolites. However, when the standards are difficult to synthesize, some metabolite structures deduced only from LC/MSn data may not be correct, especially if isomeric metabolites are present. In our study, three groups of isomers (HM1 and HM6, HM2 and HM7, and HM3 and HM4) were obtained, and they had identical LC/MSn data. Therefore, their exact structures could not be identified from only LC/MSn data (Zhu et al., 2007). In these cases, preparation of metabolites and further identification on the basis of NMR data are needed. Of course, direct isolation of the metabolites from urine, bile, or feces of humans or animals can be difficult, but it is the most reliable method for the identification of metabolites. We have determined the definitive structures of nine metabolites of berberine by enzymatic hydrolysis and examination of MS and NMR spectra. These results are important for a better understanding of its in vivo metabolic fate and disposition in rats and humans.
In this study, nine urinary metabolites of berberine in rats and humans were isolated and identified. On the basis of the metabolites, it is suggested that berberine is metabolized in both rats and humans mainly with the involvement of 2,3-OCH2O and 9,10-OCH3 by cleavage of the dioxymethylene five-membered ring or demethylation to form the corresponding phase I metabolites, some of which then underwent conjugation with glucuronic acid or sulfuric acid to form phase II metabolites. Most of the final metabolites were sulfate or glucuronide conjugates. These metabolites are very polar and are easily excreted.
To examine the possible site of metabolism, we analyzed the gastrointestinal contents of rats 4 h after oral administration of berberine and could not detect any metabolites, indicating that berberine is relatively stable in the gastrointestinal tract. However, we also investigated the urinary metabolites of berberine after intraperitoneal injection and obtained the same metabolites as those after oral administration, implying that the metabolites of berberine are formed after absorption rather than in gastrointestinal tract. On the basis of the structures of the metabolites isolated from the urine, possible metabolic pathways in rats and humans are proposed (Fig. 2).
Although berberine exhibits a variety of pharmacological effects in clinical situations, its poor oral bioavailability makes it hard for us to understand its clinical application for treatment of diabetes mellitus (Ni, 1988), hyperlipemia (Kong et al., 2004), arrhythmia, and heart failure (Zeng and Zeng, 1999; Lau et al., 2001; Zeng et al., 2003). This study demonstrated that the absorbed berberine is easily metabolized in vivo and its extensive biotransformation may be one of the main reasons for its low oral bioavailability.
Several reports have shown that a high concentration is necessary for berberine to display its in vitro bioactivities (Kong et al., 2004; Yin et al., 2008; Piyanuch et al., 2007). However, the blood-berberine concentration in animals or humans was very low when it was given orally in experimental or clinical doses (Shen et al., 1993; Yu et al., 2000), suggesting that the berberine metabolites might be responsible for the pharmacological effects. The bioactivities of the berberine metabolites including the corresponding phase I metabolites are now under investigation.
Acknowledgments
We thank Dr. David Jack (United Kingdom) for the language check and the editorial assistance.
Footnotes
-
This study was supported by the Educational Foundation of Liaoning Province (20060877) and the Science and Technology Foundation of Shenyang (1063306-1-00).
-
doi:10.1124/dmd.108.021659.
-
ABBREVIATIONS: HPLC, high-performance liquid chromatography; ODS, octadecyl silane; EtOH, ethanol; MeOH, methanol; ESI, electrospray ionization; DMSO-d6, deuterated dimethyl sulfoxide; MS, mass spectrometry; NOESY, nuclear Overhauser effect spectroscopy; LC/MSn, liquid chromatography-mass spectrometry at stage n.
- Received April 6, 2008.
- Accepted August 11, 2008.
- The American Society for Pharmacology and Experimental Therapeutics