Identification of metabolites of Radix Paeoniae Alba extract in rat bile, plasma and urine by ultra-performance liquid chromatography-quadrupole time-of-flight mass spectrometry.

Ultra-performance liquid chromatography coupled with quadrupole time-of-flight tandem mass spectrometry (UPLC-Q-TOF/MS) was developed to identify the absorbed parent components and metabolites in rat bile, plasma and urine after oral administration of Radix Paeoniae Alba extract (RPAE). A total of 65 compounds were detected in rat bile, plasma and urine samples, including 11 parent compounds and 54 metabolites. The results indicated that glucuronidation, hydroxylation and methylation were the major metabolic pathways of the components of RPAE. Furthermore, the results of this work demonstrated that UPLC-Q-TOF/MS combined with MetaboLynx™ software and mass defect filtering (MDF) could provide unique high throughput capabilities for drug metabolism study, with excellent MS mass accuracy and enhanced MSE data acquisition. With the MSE technique, both precursor and fragment mass spectra can be simultaneously acquired by alternating between high and low collision energy during a single chromatographic run.

Introduction

Radix Paeoniae Alba, the roots of plant Paeonia lactiflora Pall (Family Ranunculaceae), referred as Baishao in China, is a commonly used medicinal herb in traditional Chinese medicine (TCM). Modern phytochemical and pharmacological studies have shown that the major bioactive components of Radix Paeoniae are monoterpene glycosides, galloyl glucoses and phenolic compounds [1], [2], [3]. It displays a range of pharmacological activities such as anti-inflammatory [4], anti-oxidant [5], antiallergic [6], anti-thrombosis [7], immunoregulating [8], cognition-enhancing [9], antihyperglycemic effects [10] and kidney protection effect in diabetic rats [11].

The components of RPAE have been identified thoroughly in vitro [12], [13], [14], [15]. Li et al. [16] have identified 40 components in RPAE with UPLC–Q-TOF/MS. As the analytical method in Li's [16] research was well developed to identify the components in RPAE. In our study, the same analytical method with minor changes including the high energy in MSE mode and MS/MS mode, was used to analyze the in-vitro and in-vivo samples. The pharmacokinetic parameters of RPAE have been investigated [17], but the studies on metabolism of RPAE are limited. There have been some reports on the metabolism studies of the components of RPAE, such as paeoniflorin, one main component of RPAE. Paeoniflorin possesses wide pharmacological effects in the nervous system. It has been used in the treatment of epilepsy, cerebral ischemia and neurodegenerative disorders such as Alzheimer's disease. And it also exhibits anticoagulant, neuromuscular blocking, cognition-enhancing and immunoregulating and antihyperglycemic effects. In Su et al.'s [18] study, metabolite hydroxyl-paeoniflorin was identified in rat plasma after oral administration of Shaofu Zhuyu decoction active extract. Previous studies had reported low bioavailability of orally administered paeoniflorin in rats (approximately 3–4%) [19], [20], metabolism of paeoniflorin is an important mechanism that is responsible for poor bioavailability of paeoniflorin. Administered paeoniflorin was extensively metabolized into paeoniflorgenin, paeonimetabolins I and II [21], [22], [23]. Most of the components in RPAE possess the same parent structure and just differ in substituents such as thionyl, hydroxy and methyl. Besides, the content of paeoniflorin is larger compared to other monoterpene glycosides in RPAE and the reference of paeoniflorin could be purchased. Therefore, paeoniflorin is chosen as the representative to study the metabolism behaviors of monoterpene glycosides in RPAE. In our study, pure paeoniflorin was administered to rats to characterize the metabolic pathways of monoterpene glucosides. The results elucidate the analysis of the various metabolites of other monoterpene glucosides from RPAE.

In order to get more metabolism information of RPAE from the complex biological matrics, it is necessary to develop effective and reliable analytical methods for detection. Recently, UPLC–ESI-Q-TOF/MS system with automated MSE data analysis software (Metabolynx™) has been proved to be a powerful and reliable analytical approach for in vivo metabolite identification studies [24], [25], [26]. With this efficient tool, we studied thoroughly the absorbed components and metabolites in rat bile, plasma and urine after oral administration of RPAE. These absorbed components and metabolites might be the bioactive substances of RPAE in vivo and should be screened on pharmacological models to pave a way for research on the effect related to the mechanism of action.

Materials and methods

Chemicals and regents

The standards of albiflorin, paeoniflorin and catechin (purity >98%, see their structures in Fig. 1) were purchased from the National Institute for Drug Control of China (Beijing, China). Acetonitrile (Merk, Germany), methanol (Merk, Germany) and formic acid (Fisher Scientific, Farilawn, NJ, USA) were of HPLC grade, and other chemicals and regents were of analytical grade. De-ionized water was prepared by a Milli-Q purification system (Millipore, Bedford, MA, USA). Waters Oasis HLB Solid-phase extraction cartridge (30 mg/mL) was purchased from Waters Corp.

Fig. 1

Chemical structures of 14 parent compounds of RPAE.

Preparation of RPAE

The dried powder of Radix Paeoniae Alba (Bozhou, Anhui, China) extract was supplied by Tianjin Tasly Pharmaceutical Co. Ltd (Tianjin, China). The preparation process was as follows: the pulverized samples of Radix Paeoniae Alba (50.0 g) was immersed in 400 mL water and extracted in a reflux water bath for 2 times, 2 h for each time. Proper amount of 95% ethanol was added into the extracts to adjust the concentration of ethanol to 70%. The solution was kept at 4 °C for 12 h then filtered. Cake was dried and went through 80 mesh sieve to yield RPAE. The RPAE was dissolved in methanol and filtered through a 0.22 μm membrane before analysis.

Animal experiments

Animal studies were conducted according to the protocols approved by the Animal Ethics Committee of Tasly Group. Animal experiments adhered to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Twenty four male Wistar rats (weighing 200±20 g) were purchased from Beijing Vital River Laboratories (Beijing, China) and kept under controlled environmental conditions (temperature 22±2 °C; humidity 50±10%) with free access to the standard laboratory food and water. The animals were divided into four groups with 6 rats in each group. One group was used for collection of bile after oral administration of paeoniflorin at the dose of 400 mg/kg. The other three groups were used for collection of bile, urine and plasma samples after oral administration of RPAE at the dose of 3 g/kg RPAE (namely, 18.75 g/kg for crude drug) respectively. The dose of RPAE was picked according to the publications on the pharmacokinetic studies of RPAE.

The rats were fasted for 12 h with free access to water prior to the experiments. Blank bile, plasma and urine samples were collected before administration with drugs. The rats were anesthetized with 20% urethane (m/m) at the dose of 0.7 mL/100 g, then fixed on wooden plates. An abdominal incision was made and bile duct was cannulated with PE tubing for collection of bile samples continually for 12 h after administration. The blood samples (0.5 mL each time) were collected from ophthalmic veins of the rats by sterile capillary tube at 0.5, 1, 2, 4 and 6 h after administration. The blood samples were centrifuged at 3500 rpm for 10 min immediately to get plasma. Urine samples were collected continually for 12 h after administration. Meanwhile the rats were kept free access to water. All the samples were stored at −20 °C.

Preparation of urine, plasma and bile samples

A solid-phase extraction (SPE) method was used for plasma and urine pretreatment. Before applying to sample, a SPE cartridge was treated with 1 mL of water, 1 mL of methanol and 1 mL of water successively. An aliquot of 0.5 mL plasma and urine sample was vortexed, loaded, and allowed to flow through the SPE cartridge with pressure. The SPE cartridge was then washed with 1 mL of water and 2 mL of methanol successively. The methanol eluate was collected and evaporated under a stream of N2 at room temperature. The dried sample was dissolved in 200 μL methanol for analysis. The bile samples were mixed with acetonitrile (1:1, v/v) and centrifuged at 12,000×g for 10 min. The supernatant was used for analysis.

UPLC–Q-TOF/MS analytical system

Chromatographic experiments were performed on Waters Acquity UPLC system (Waters Corp., Milford, MA, USA) equipped with a 2998 photodiode array detector (PDA) together with a quaternary pump, an auto-sample injector, an on-line degasser and an automatic thermostatic column oven. The in-vitro and in-vivo samples were separated on a Waters Acquity UPLC HSS T3 column (2.1 mm×100 mm; 1.8 μm) with the column temperature maintained at 35 °C. A linear gradient elution of solvent acetonitrile (A) and 0.1% formic acid aqueous solution (B, v/v) was applied with the following program: 2–5% A (0–2 min), 5–12% A (2–4 min), 12–20% A (4–8 min), 20–30% A (8–9 min), 30–50% A (9–10 min), and 50–100% A (10–12 min) at 0.3 mL/min. UV spectra were recorded from 190 to 400 nm. The temperature of autosampler tray was maintained at 4 °C through the analysis process. The injection volume was 2 μL for in-vitro samples and 10 μL for in-vivo samples.

The MS instrument consisted of a Waters Synapt G2 Q-TOF/MS (Waters Corp., Milford, MA, USA) equipped with an electrospray ionization (ESI) source. Both positive and negative ion modes were tested, and it was found that the sensitivities for most components in RPAE were higher in the negative ion mode. Ultrahigh purity argon (Ar) was used as the collision gas and high purity nitrogen (N2) as the nebulizing gas. The nebulization gas was set to 650 L/h desolvation temperature at 300 °C, while the cone gas and the source temperature were set to 50 L/h and 90 °C, respectively. The capillary voltage and sample cone voltage were set to 2700 V and 35 V, respectively. The Q-TOF Premier acquisition rate was set to 0.3 s, with a 0.05 s inter-scan delay. Argon was employed as the collision gas at a pressure of 7.1×10−3 Pa. The energies for collision-induced dissociation (CID) were set at 5 eV in low energy and 30–50 eV ramp voltage in high energy in MSE mode respectively for fragmentation information. For some peaks, the MS/MS mode at 30–50 eV ramp voltage was further employed to get the fragmentation information. All MS data were acquired using the LockSpray to ensure mass accuracy and reproducibility. The [M−H]− ion of leucine-enkephalin at m/z 554.2615 was used as the lock mass in negative electrospray ionization mode. The concentration of leucine-enkephalin was 2 ng/μL and the infusion flow rate was 5 μL/min. Sodium formate solution at the concentration of 0.5 mM was used to create the calibration file. The data were acquired in centroid mode with the mass ranged from 100 to 1000 Da. The Elemental Composition function of the software Metabolynx™ was used to predict the formulas of chemicals. Screening and identification of the metabolites of PRAE was assisted by the software Metabolynx™ (Waters Corp., Milford, MA, USA).

Software to calculate n-octanol/water partition coefficient (log P)

Software chembiodraw ultra 11.0 (Cambridgesoft®, Cambridge, MA) based on theoretical calculations was used to predict n-octanol/water partition coefficient (log P). The prediction of log P is based on 222 atomic contributions calculated from 1868 molecules by least squares analysis. This method allows a calculation of log P with a standard deviation of 0.43 log P units and can handle molecules containing hydrogen, oxygen, nitrogen, sulfur, halogens and phosphorus atoms. If this method is applied to molecules with internal hydrogen bonds, the standard deviation is 0.83 log P units.

Results and discussion

Identification of components from RPAE in vitro

RPAE was dissolved in methanol and filtered through a 0.22 μm membrane filter unit, and 2 μL was injected for analysis. Compounds albiflorin, paeoniflorin and catechin in RSE were unambiguously identified by the comparison of retention time (tR), UV spectrum, and fragments information as shown in Table 1 with their standards. The rest peaks in the chromatograms of RPAE were identified by comparing their accurate masses, retention time, UV spectra and fragmentation information with those reported in literature [12], [13], [14], [15], [16]. Chemical structures of 14 parent compounds of RPAE are shown in Fig. 1.

Table 1

The compounds identified from bile, plasma and urine after oral administration of the RPAE by UPLC–ESI-Q-TOF/MS.

No.	RT (min)	m/z [M–H]− Found	PPM	Metabolite name	Formula [M–H]−	Fragment ions	Parent drug	Identification	Bile	Plasma	Urine
M1	1.02	802.2350	1.1	+GSH+O	C33H45N3O18S	784,766,511,306,272,254,210,143,128	PAE		+	−	−
M2	1.03	453.0736	7.2	Hy–Glc+Glu	C16H22O13S	179,277,	OPS		+	−	−
M3	1.04	262.9858	−1.5	+CH2+SO3	C8H8O8S	183,168,150,124,203,96	GaA	3-Hydroxy-5-methoxy-4-(sulfooxy)benzoic acid	+	−	−
M4	1.05	671.1819	−0.7	+O+Glu	C29H36O18	495,465,447,333,137	PAE		+	+	−
M5	1.06	575.1063	−1.4	+O+SO3	C23H28O15S		PAE		+	+	−
M6	3.09	262.9839	−8.7	+CH2+SO3	C8H8O8S	183,168,150,124,203,96	GaA	3-Hydroxy-4-methoxy-5-(sulfooxy)benzoic acid	+	−	−
P1	3.15	359.1335	−1.9	–	C16H23O9	179,197	–	1-O-β-d-Glucopyranosyl-paeonisuffrone	−	+	+
M7	3.51	453.0682	−4.7	Hy–Glc+Glu	C16H22O13S	179,277,	OPS		+	−	−
M8	3.64	359.0607	−2.1	+CH2+Glu	C14H16O11	183,168,125,227,314,175,193	GaA		+	−	−
M9	3.74	345.0434	−7	+Glu	C13H14O11	169,125,	GaA		+	−	−
M10	3.91	373.0779	2.1	+2×CH2+Glu	C15H18O11	197,182,175,167,178,354,305	GaA		+	−	−
M11	3.97	671.1837	2	+O+Glu	C29H36O18	495,465,375,345,281,195,177,152,137	IsP		+	−	−
M12	4.05	802.2358	2.1	+GSH+O	C33H45N3O18S	784,766,511,306,272,254,210,143,128	PAE		+	−	−
M13	4.16	233.0131	4.7	+2×CH2–CO2+SO3	C8H10O6S	109,124,153	GaA	2,6-Dimethoxyphenyl hydrogen sulfate	+	−	−
M14	4.18	495.1513	2.1	+O	C23H28O12	137,165,177,195,281, 283, 327,345,	PAE		+	+	−
P2	4.2	559.1122	0	–	C23H28O14S	137,161,247,259,421	–	Oxypaeoniflorin sulfonate	+	+	+
M15	4.25	671.1831	1.1	+O+Glu	C29H36O18	495,465,447,333,137	PAE		+	+	−
M16	4.31	802.2358	2.1	+GSH+O	C33H45N3O18S	784,766,511,306,272,254,210,143,128	PAE		+	−	−
M17	4.37	701.1921	−1.2	+Glu	C30H38O19	363,453,525	ME		+	−	−
M18	4.4	262.9850	−4.5	+CH2+SO3	C8H8O8S	183,168,150,124,203,96	GaA	4-Hydroxy-3-methoxy-5-(sulfooxy)benzoic acid	+	−	+
M19	4.43	717.1884	0.8	+O+Glu	C30H38O20	225,255,312,327,461,651,669,687,493,543	ME		+	−	−
M20	4.5	373.0770	−0.3	+2×CH2+Glu	C15H18O11	197,182,175,167,153	GaA		+	−	−
M21	4.61	701.1934	0.7	+Glu	C30H38O19	239,327,363,495,525	ME		+	−	−
P3	4.68	589.1221	−1.1	–	C24H30O15S	167,259,331,375,421,535,574	–	Isomaltopaeoniflorin sulfonate	+	+	+
M22	4.74	277.0017	−0.5	+2×CH2+SO3	C9H10O8S	107,124,167,182,197	GaA	3,5-Dimethoxy-4-(sulfooxy)benzoic acid	+	−	−
M23	4.88	621.1166	6.5	+O+SO3	C24H30O17S	225,327,346,361,445,511,541	ME		+	−	−
M24	4.91	495.1498	−1	+O	C23H28O12	137,165,177,195,281, 283, 327,345,	PAE		+	+	−
M25	5	655.1887	1.9	+Glu	C29H36O17	319,481	IsP		+	−	−
M26	5.01	605.1142	−5.7	+O	C24H30O16S	183,241,525,544	MES	6-Hydroxy mudanpioside E sulfonate	+	−	−
M27	5.01	701.1924	−0.8	+Glu	C30H38O19	285,525	ME		+	−	−
M28	5.02	575.1044	−4.7	+O+SO3	C23H28O15S	137,495,465	PAE		+	+	−
M29	5.04	717.1868	−1.5	+O+Glu	C30H38O20	284,511,543,574	ME		+	+	−
P4	5.16	705.1695	−0.8	–	C29H38O18S	121,259,421,543,583	–	Mudanpioside E sulfonate	+	+	+
M30	5.28	575.1054	−2.9	+O+SO3	C23H28O15S	137,495,465	PAE		+	+	−
M31	5.3	605.1144	−5.4	+O	C24H30O16S	183,495,525	MES	2-Hydroxy mudanpioside E sulfonate	+	−	−
P5	5.32	543.117	−0.5	–	C23H28O13S	121,213,259,375,421	–	Paeoniflorin sulfonate	+	+	+
M32	5.32	479.1189	−0.2	+CH2+Glu	C22H24O12	202,217,244,259,285,303	Cat		+	+	−
M33	5.38	717.1868	−1.5	+O+Glu	C30H38O20		ME		+	+	−
M34	5.43	605.1143	−5.5	+O	C24H30O16S	183,277,403,425,439,525,	MES	5-Hydroxy mudanpioside E sulfonate	+	−	−
M35	5.54	479.1190	0	+CH2+Glu	C22H24O12	219,244,259,285,303	Cat		+	+	−
M36	5.55	383.0439	0.5	+CH2+SO3	C16H16O9S	303,285,259,244,216,202,164,137	Cat		+	+	−
M37	5.55	621.1121	−0.8	+O+SO3	C24H30O17S	210,255,285,327,346,361,445,493,407,541	ME		+	−	−
M38	5.63	449.1465	3.8	–CH2–O	C22H26O10	165,327,273,309,291,386,430,447	IsP		+	−	−
M39	5.65	373.0757	−3.8	+2×CH2+Glu	C15H18O11	197,182,193,167,175,250,255	GaA		+	−	−
M40	5.66	479.1203	2.8	+CH2	C22H24O12	219,244,259,285,303	Cat		+	+	−
M41	5.67	701.1915	−2	+Glu	C30H38O19	363,525	ME		+	−	−
P6	5.79	495.1488	−3	–	C23H28O12	137,165,289, 407,421,465	–	Oxypaeoniflorin or oxypaeoniflorin isomer	+	−	−
M42	6.13	277.0023	1.7	+2×CH2+SO3	C9H10O8S	107,124,167,182,197	GaA	3,4-Dimethoxy-5-(sulfooxy)benzoic acid	+	−	−
M43	6.34	495.1483	−4	+O	C23H28O12	137,165, 283, 327,345,	PAE		+	−	+
M44	6.37	621.1127	0.2	+O+SO3	C24H30O17S	345,427,445,523,541,	ME		+	−
M45	6.47	479.1191	0.3	+CH2+Glu	C22H24O12	219,244,259,285,303	Cat		+	−	+
M46	6.6	621.1140	2.3	+O+SO3	C24H30O17S	511,523,541	ME		+	−	−
M47	6.72	233.0103	−7.3	+2×CH2–CO2+SO3	C8H10O6S	109,124,153	GaA	2,3-Dimethoxyphenyl hydrogen sulfate	+	−	−
M48	6.86	383.0431	−1.6	+CH2+SO3	C16H16O9S	303,285,259,244,216,202,164,137	Cat		+	−	−
M49	6.96	655.1865	−1.4	+Glu	C29H36O17	319,481	IsP		+	−	−
M50	7.14	449.1446	−0.4	–CH2–O	C22H26O10	165,327,273,309,291,386,430,447	IsP		+	+	−
M51	7.29	577.1219	−1.5	+H2+O	C23H30O15S	272,382,400,497	OPS		+	−	−
P7	7.33	479.1562	1.7	–	C23H28O11	121,195,283, 327,357,375	–	Albiflorin	+	+	+
M52	7.49	383.0431	−1.6	+CH2+SO3	C16H16O9S	303,285,259,244,216,202,164,137	Cat		+	+	−
M53	7.7	655.1886	1.8	+Glu	C29H36O17	319,481	IsP		+	−	−
M54	7.79	449.1444	−0.9	–CH2–O	C22H26O10	165,327,273,309,291,386,430,447	IsP		+	+	+
P8	7.81	479.1543	−2.2	–	C23H28O11	121,165,283, 327,345,431,449	–	Paeoniflorin	+	+	+
P9	8.25	495.1502	−0.2	–	C23H28O12	137,169,325,443	–	Ortho-oxypaeoniflorin	+	−	−
P10	9.46	631.1661	−0.3	–	C30H32O15	169,313,399,	–	Galloylpaeoniflorin or galloylalbiflroin or their isomers	+	−	−
P11	10.24	647.1447	1.9	–	C30H32O14S	121,169,213,259,313,479,525,631	–	Benzoylpaeoniflorin sulfonate	+	+	+

Cat: catechin; GaA: gallic acid; IsP: isomer of paeoniflorin; ME: Mudanpioside E; MES: Mudanpioside E sulfonate; OPS: Oxypaeoniflorin sulfonate; PAE: paeoniflorin.

–CO2: Decarboxylation; +SO3: sulfate conjugation; +Glu: glucuronide conjugation; –CH2: demethylation; +CH2: methylation; –O: loss of oxygen; +O: hydroxylation; +GSH: glutathione conjugation; Hy: hydrolisis; –Glc: deglucose; +H2: reduction.

P: the parent components in RPAE; M: metabolites; PPM: parts per million.

+: detected, −: not detected.

Identification of parent components in bile, plasma and urine after oral administration of RPAE

By comparing the accurate masses, retention time and fragmentation information of peaks appearing in the chromatograms of bile, plasma and urine with those previously identified in RPAE in vitro, 11 peaks were detected as prototype components of RPAE. Their extracted ion chromatograms are shown in Fig. 2.

Fig. 2

Extracted ion chromatograms in negative mode of 11 parent compounds (P1–P11) detected in bile (B), plasma (P) and urine (U). P1–P11 are 1-O-β-d-glucopyranosyl-paeonisuffrone, oxypaeoniflorin sulfonate, isomaltopaeoniflorin sulfonate, mudanpioside E sulfonate, paeoniflorin sulfonate, oxypaeoniflorin or oxypaeoniflorin isomer, albiflorin, paeoniflorin, ortho-oxypaeoniflorin, galloylpaeoniflorin or galloylalbiflroin or their isomers, benzoylpaeoniflorin sulfonate, respectively.

Characterization of metabolites of single standard

RPAE contains multiple compounds with significant varieties in structural types, physiochemical properties, and relative amounts. When they were administered by oral route, lots of compounds may get into circulation and then be converted into a mixture of metabolites of unknown origin. Meanwhile, chemicals in herbal medicine usually present in the form of a series of analogs with the same skeleton but different functionalities and their metabolites may undergo similar fragmentation and produce some of the same ions. Understanding the metabolic pathway of certain parent drugs will greatly facilitate metabolite identification of their analogs. Therefore, in our study paeoniflorin was chosen as the representative to study the metabolic pathways of monoterpene glycosides. The fragmentation pathways of paeoniflorin (shown in Fig. 3) were investigated thoroughly which will throw light upon the identification of the metabolites. In addition, as only one single compound was fed to rats, the metabolic profiles were easy to be elucidated. The result indicated that hydroxylation, glucuronidation, glutathione (GSH) conjugation and sulfation predominated the metabolism of paeoniflorin.

Fig. 3

Proposed fragmentation pathways of paeoniflorin.

n-octanol/water partition coefficient (log P)

It was common that several metabolites were assigned as isomers in biological matrices. To elucidate the exact structures of these isomers, such as the positions of conjugation or hydroxylation, were challenging due to little differences present in their MS spectra. Log P, an important parameter in the process of designing new drugs, provides the direct information on hydrophobicity that describes the tendency of distribution of a drug from aqueous phase into biological membranes [27]. Software chembiodraw ultra 11.0 was used to predict Log P. Generally, for compounds with log P<4, the calculated log P values are almost the same as the experimental measurements. The larger log P is the less retention time it would have on the reverse phase chromatography [28], [29].

Identification of metabolites from bile, urine, plasma after oral administration of RPAE

Besides the 11 prototype compounds, 54 peaks were tentatively presumed to be metabolites of RPAE, and could be generally divided into three groups: monoterpene glycosides-related, catechin-related and gallic acid-related metabolites.

Identification of monoterpene glycoside-related metabolites

Identification of paeoniflorin-related metabolites

A total of 11 compounds detected in rat bile, plasma and urine were tentatively assigned as metabolites originating from paeoniflorin. From the results shown in Table 1, we can see that paeoniflorin underwent hydroxylation before undergoing phase II metabolic reaction. They all shared the same fragment ion at m/z 137 corresponding to the hydroxylated benzene ring.

M14, M24 and M43 shared a series of characteristic ions at m/z 165, 283, 327 and 345 with paeoniflorin. In addition, according to the accurate molecular mass, M14, M24 and M43 were 16 Da higher than paeoniflorin. Thus, M14, M24 and M43 were assigned as hydroxylation products of paeoniflorin. The shorter retention time of M14, M24 and M43 than that of paeoniflorin (tR=7.81 min) on the RP-UPLC column also supported the tentative identification as oxidation metabolites of paeoniflorin.

M5, M28 and M30 shared a series of characteristic ions of hydroxylated paeoniflorin at m/z 495, 465 and 137. In addition, the molecular weight of M5, M28 and M30 was 80 Da (SO3) higher than that of hydroxylated paeoniflorin, indicating that they were sulfate conjugates of hydroxylated paeoniflorin. M4 and M15 shared a series of characteristic ions of hydroxylated paeoniflorin at m/z 447 and 333. The molecular weight of M4 and M15 was 176 Da higher than that of hydroxylated paeoniflorin, indicating that they were glucuronide conjugates of hydroxylated paeoniflorin.

For metabolites M1, M12 and M16, a series of characteristic ions of glutathione at m/z 306, 272, 254, 210, 143 and 128 were observed. Their molecular weight was 307 Da higher than hydroxylated paeoniflorin, so they were tentatively identified as glutathione conjugates of hydroxylated paeoniflorin. The proposed metabolic pathways of paeoniflorin are shown in Fig. 4 and the extracted ion chromatograms of paeoniflorin-related metabolites are shown in Fig. 5.

Fig. 4

Proposed metabolic pathways of paeoniflorin-related metabolites.

Fig. 5

Extracted ion chromatograms of paeoniflorin-related metabolites from rat bile in negative mode.

Identification of mudanpioside E-related metabolites

Metabolites M23, M37, M44 and M46 gave [M−H]− ions at m/z 621.1166, 621.1121, 621.1127 and 621.1140 respectively, which were calculated as C30H38O19 by the Elemental Composition function of software Metabolynx™. Their different retention times suggested they were isomers. They shared a series of characteristic ions at m/z 345, 511, 523 and 541 with mudanpioside E. The [M−H]− ion at m/z 621 yielded a product ion at m/z 541 with a loss of 80 Da (SO3), suggesting that they were sulfate conjugates. In addition, the ion at m/z 541 is 16 Da higher than that of mudanpioside E, indicating that M23, M37, M44 and M46 were sulfate conjugates of hydroxylated mudanpioside E.

According to the accurate mass, the molecular weight of M17, M21, M27 and M41 was 176 Da higher than that of mudanpioside E. In addition, their fragments ions were similar to that of mudanpioside E, suggesting that they were glucuronide conjugates of mudanpioside E. While the molecular weight of M19, M29 and M33 was 16 Da higher than that of M17, suggesting that they were glucuronide conjugates of hydroxylated mudanpioside E. The proposed metabolic pathways of mudanpioside E are shown in Fig. 6 and the extracted ion chromatograms of mudanpioside E-related metabolites are shown in Fig. 7.

Fig. 6

Proposed metabolic pathways of mudanpioside E-related metabolites.

Fig. 7

Extracted ion chromatograms of mudanpioside E-related and mudanpioside E sulfate-related metabolites from rat bile in negative mode.

Identification of mudanpioside E sulfate-related metabolites

M26, M31 and M34 yielded an ion at m/z 183, indicating that the benzene ring was hydroxylated. In addition, their [M−H]− ion at m/z 605 gave a product ion at m/z 525 with a loss of 80 Da (SO3) and their molecular weight was 16 Da higher than that of mudanpioside E sulfate, so they were tentatively identified as hydroxylated mudanpioside E sulfate. There were three places in the benzene ring of mudanpioside E sulfate that could be hydroxylated. According to the fact that one with a larger log P value would have less retention time on the reverse phase chromatography, M26 (tR=5.01 min, log P=−3.648), M31 (tR=5.3 min, log P=–3.898) and M34 (tR=5.43 min, log P=–3.928) were tentatively identified as 6-hydroxy mudanpioside E sulfonate, 2-hydroxy mudanpioside E sulfonate and 5-hydroxy mudanpioside E sulfonate, respectively. The extracted ion chromatograms of mudanpioside E-related metabolites are shown in Fig. 7.

Identification of oxypaeoniflorin sulfonate-related metabolites

Metabolites M2 and M7 gave a [M−H]− ion at m/z 453.0682 and 453.0736 respectively, calculated as C16H22O13S. In addition, M2 and M7 gave an ion at m/z 277 with a loss of 176 Da, suggesting that they were glucuronide conjugates. Based on the information above, M2 and M7 were tentatively identified as glucuronide conjugates after oxypaeoniflorin sulfonate undergoing hydrolysis and deglucose reactions. M51 gave a [M−H]− ion at m/z 577.1219 (C23H30O15S), which yielded a product ion at m/z 497 with a loss of 80 Da (SO3), corresponding to the characteristic loss of oxypaeoniflorin sulfonate. Thus M51 was tentatively identified as the reduction and hydroxylation product of oxypaeoniflorin sulfonate. The extracted ion chromatograms of oxypaeoniflorin sulfonate-related metabolites are shown in Fig. 8.

Fig. 8

Extracted ion chromatograms of oxypaeoniflorin sulfate and isomer of paeoniflorin-related metabolites from rat bile in negative mode.

Identification of isomer of paeoniflorin-related metabolites

Three isomers of paeoniflorin (m/z=479.1543, C23H28O11) were identified in vitro as albiflorin, isopaeoniflorin and albiflorin R1 according to their fragmentation information. M25, M49 and M53 produced fragment ions at m/z 319 and 481, corresponding to loss of 176 Da and 162 Da. So they were tentatively identified as glucuronide conjugates of isomer of paeoniflorin. The molecular weight of M11 was 16 Da (O) higher than that of M25 and also shared the loss of 176 Da. Thus M11 was tentatively identified as glucuronide conjugate of hydroxylated isomer of paeoniflorin. M38, M50 and M54 gave [M−H]− ions at m/z 449.1465, 449.1446 and 449.1444 respectively, which were calculated as C22H26O10. They shared the same characteristic ions at m/z 165 and 327 with paeoniflorin. Moreover, compared with the molecular weight of paeoniflorin, they were 30 Da less, which was tentatively assigned to CH2 (14 Da) and O (16 Da). Thus M38, M50 and M54 were tentatively identified as demethylation and loss of oxygen products of isomer of paeoniflorin. The extracted ion chromatograms of isomer of paeoniflorin-related metabolites are shown in Fig. 8.

Identification of catechin-related metabolites

M36, M48 and M52 gave [M−H]− ions at m/z 383, which produced a product ion at m/z 303 with a loss of 80 Da (SO3), indicating that they were sulfate conjugates. The presence of ions at m/z 259 and 244 suggested that there was a methyl group (15 Da). Thus M36, M48 and M52 were tentatively identified as methylation and sulfation conjugates of catechin. M32, M35, M40 and M45 gave a [M−H]− ion at m/z 479, which produced a product ion at m/z 303 with a loss of 176 Da, indicating that they were glucuronide conjugates. The presence of product ions at m/z 259 and 244 suggested that there was a methyl group (15 Da). Thus M32, M35, M40 and M45 were tentatively identified as methylation and glucuronide conjugates of catechin. The extracted ion chromatograms of catechin-related metabolites are shown in Fig. 9.

Fig. 9

Extracted ion chromatograms of catechin-related metabolites from rat bile in negative mode.

Identification of gallic acid-related metabolites

M22 and M42 gave a [M−H]− ion at m/z 277, which gave a product ion at m/z 197 with a loss of 80 Da (SO3), suggesting that they were sulfate conjugates. The presence of product ions at m/z 197,182 and 167 indicated that there were two methyl groups. So M22 and M42 were assigned as sulfation and double methylation conjugates of gallic acid. According to the fact that one with larger log P value would have less retention time on the reverse phase chromatography, M22 (tR=4.73 min, log P=–0.189) and M42 (tR=6.13 min, log P=–0.539) were tentatively identified as 3,5-dimethoxy-4-(sulfooxy)benzoic acid and 3,4-dimethoxy-5-(sulfooxy)benzoic acid, respectively.

As for M13 and M47, the [M−H]− ion at m/z 233.0131 (C8H10O6S) was 44 Da less than that of M22, suggesting that they were the decarboxylation products of M22. Thus M13 and M47 were assigned as decarboxylation products after sulfation and double methylation of gallic acid. M13 (tR=4.16 min, log P=–0.004) and M47 (tR=6.72 min, log P=–0.354) were tentatively identified as 2,6-dimethoxyphenyl hydrogen sulfate and 2,3-dimethoxyphenyl hydrogen sulfate, respectively. M3, M6 and M18 gave a [M−H]− ion at m/z 263, which yielded product ions at m/z 183 and 168, with a loss of 80 Da (SO3) and 15 Da (CH3). Thus M3, M6 and M18 were assigned as methylation and sulfate conjugates of gallic acid.

M9 gave a [M−H]− ion at 345.0434 (C13H14O11), which led to a product ion at m/z 169 with the loss of 176 Da, suggesting that it was glucuronide conjugate of gallic acid. The molecular weight of M8 was 14 Da (CH2) higher than that of M9, indicating that M8 was methylated M9. So M8 was tentatively identified as methylation and glucuronide conjugate of gallic acid.

M10, M20 and M39 yielded [M−H]− ions at m/z 373, which could lead to fragment ion at m/z 197 with a loss of 176 Da, suggesting that they were glucuronide conjugates. Moreover, the presence of product ions at m/z 197, 182 and 167 implied that there were two methyl groups. Thus M10, M20 and M39 were tentatively identified as glucuronide and double methylation conjugates of gallic acid. The proposed metabolic pathways of gallic acid are shown in Fig. 10 and the extracted ion chromatograms of gallic acid-related metabolites are shown in Fig. 11.

Fig. 10

Proposed metabolic pathways of gallic acid-related metabolites.

Fig. 11

Extracted ion chromatograms of gallic acid-related metabolites from rat bile in negative mode.

Discussions

Isomers are the groups of compounds with the same exact masses and elemental compositions. In most cases, these compounds are similar in structure and polarity and also have similar cleavage pathways, which is difficult for analysts to deal with [30]. As for LC–MS method, there are three ways to distinguish them, including the retention time, the differences of fragment ions and UV spectra. For some isomers, they could be distinguished by analyzing their retention time and n-octanol/water partition coefficient. The larger the n-octanol/water partition coefficient is, the smaller the retention time is on the reverse phase chromatography. The software chembiodraw ultra 11.0 could be used to calculate the n-octanol/water partition coefficient. The maximum absorption wavelength in UV spectra will change according to the different positions of these moieties. So the UV spectra can help pinpoint the position of conjugations. However, in our study, the polarities of the metabolites are very close that make it impossible to separate them completely on chromatography, so the UV spectra could not be used to distinguish them. For most of the isomers, the difference of fragmentation information could hardly be found as the structures seem so alike. Therefore, judging from the factors illustrated above, some metabolites in our study cannot be given exact structures. Further studies such as 1H and 13C NMR are needed to clarify the exact structures of these isomers.

Conclusion

In our study, a total of 54 metabolites were detected and the metabolic pathways were proposed. This is the first study to establish the metabolic profile of RPAE, which is helpful in revealing the bioactive components of RPAE. Moreover, it was demonstrated that the high-speed and sensitive UPLC–Q-TOF/MS analytical system combined with the software Metabolynx™ was a useful tool to identify metabolites in herbal medicine.