DOI:
10.1039/C6RA10883K
(Paper)
RSC Adv., 2016,
6, 65055-65066
A strategy for establishment of practical identification methods for Chinese patent medicine from systematic multi-component characterization to selective ion monitoring of chemical markers: Shuxiong tablet as a case study†
Received
27th April 2016
, Accepted 20th June 2016
First published on 22nd June 2016
Abstract
Authentication of Chinese patent medicine (CPM) is more challenging than that of a single herbal medicine due to the increasing complexity in the chemical matrix. A strategy from systematic multi-component characterization to selective ion monitoring of chemical markers (SMC-SIM) is presented and applied to authenticate Shuxiong tablet (SXT) as a case study. First, an ultra-performance liquid chromatography/quadrupole time-of-flight-Fast DDA (UPLC/QTOF-Fast DDA) approach was developed to comprehensively profile and characterize multi-components in SXT. Second, chemical markers were established using a home-made SXT sample. The SIM method for detection of these markers was initially established on an advanced QTOF mass spectrometer, and then transferred onto an easily accessible single quadrupole QDa detector. Third, the UPLC/QDa-SIM method was used to authenticate commercial SXT samples. Consequently, 250 compounds were characterized, and 73 of them were identified by comparison with reference standards. Two UPLC-SIM methods enabling the monitoring of eleven chemical markers (including five saponins for Notoginseng Radix et Rhizoma, two quinochalcone C-glycosides and a flavonoid O-glycoside for Carthami Flos, one phenolic acid and two phthalides for Chuanxiong Rhizoma), were developed on QTOF and QDa, respectively. Twelve batches of commercial SXT samples were identified by the UPLC/QDa-SIM method. Despite the lower sensitivity, the QDa-SIM method by simple sample preparation achieved similar authentication results, and was more easily generalized in practice than the QTOF-SIM method. In conclusion, targeted monitoring of multiple chemical markers was proven as feasible in comprehensive identification of CPM, and the SMC-SIM strategy enabled establishment of practical quality control methods for CPM that contains a complex chemical matrix.
Introduction
Formulae are the common clinical practice for traditional Chinese medicine (TCM).1 Quality control of the preparations of TCM formulae (Chinese patent medicine, CPM) is more challenging than that of single herbal medicine owing to the increasing complexity in the chemical matrix. In China Pharmacopoeia (2015 edition, ChP), thin layer chromatography (TLC) acts as a major vehicle for authentication of CPM. In general, these TLC-based methods can separate and monitor a few markers and employ complex sample pretreatment methods. Our lab has proposed a concept for the holistic quality control of TCM, that is, to deeply probe the chemical composition and biological effect, and accordingly to establish simple and practical quality standards. Following this idea, dozens of TCM standard monographs have been established and recorded in ChP, Herbal Medicines Compendium of U.S. Pharmacopoeia, and European Pharmacopoeia. However, hitherto its application in CPM is rare.
The ongoing development in instrumentation and analytical techniques provides a more practical solution to quality control of TCM. Currently, liquid chromatography/mass spectrometry (LC/MS) has been extensively applied in qualitative and quantitative analyses of TCM components due to its good sensitivity and selectivity.2–4 On one hand, the commercial availability of ultra-high performance liquid chromatography (UPLC) in 2004, and the green aspects of HPLC that uses sub-2 μm packing particles, has enabled faster separation and less consumption of organic solvent, with increasing sensitivity and resolution.5 On the other hand, the application of high-resolution mass spectrometry (HRMS), particularly quadrupole time-of-flight mass spectrometry (QTOF MS) and linear ion trap-Orbitrap mass spectrometry (LTQ-Orbitrap MS), enables high accuracy measurement of both the precursor and product ions, thus rendering the structural characterization more accurate and more convincing.6–11 QTOF allows for faster data acquisition, whilst LTQ-Orbitrap offers a higher resolution, richer structure information, more fragmentation modes, and more diverse scan methods.12–16 Additionally, to identify the minor components in TCM extracts and CPM, enhanced tandem mass spectrometric scan methods, mostly facilitated on a triple quadruple mass spectrometer or triple quadrupole-linear ion trap hybrid mass spectrometer, and hybridization of two-dimensional liquid chromatography and HRMS (2D LC/HRMS), have been reported.17–20 Interestingly, due to the application of HRMS and in silico tools based on database retrieval and scoring algorithms, new solutions facilitating automatic component characterization may be available soon.21,22
Shuxiong tablet (SXT), prepared from Notoginseng Radix et Rhizoma (NRR: the root and rhizome of Panax notoginseng), Carthami Flos (CF: the floret of Carthamus tinctorius L.), and Chuanxiong Rhizoma (CR: the rhizome of Ligusticum chuanxiong Hort.), is a popular CPM prescribed to treat coronary heart disease and stenocardia. Previous phytochemical work has revealed that NRR contains dammarane-type triterpenoid saponins with therapeutic effects against cardiovascular diseases.23,24 CF features quinochalcone C-glycosides (QCGs) and flavonoid O-glycosides (FOGs),25,26 whereas CR involves phthalide derivatives, organic acids, and alkaloids.27 Despite the fact that the quality standard of SXT compiled in ChP (2015 edition) requires the identification of all these three drugs by TLC, on one hand, the sample preparation is complex and three different TLC methods are employed, and on the other hand, only three saponin markers (noto-R1, Rg1, and Rb1) are used for quantitative evaluation of SXT. No literature is currently available for systematic characterization of multi-components in SXT. Given the differential polarity among different classes of chemicals in SXT, reverse-phase UPLC coupled with HRMS may be a highly selective, sensitive, and reliable tool for deeply studying the therapeutic basis of SXT.
We present a strategy from systematic multi-component characterization to selective ion monitoring of chemical markers (SMC-SIM) to establish practical authentication methods for CPM. The aim of this study is to validate the feasibility of the SMC-SIM strategy using SXT as a case study (Fig. 1). Targeted monitoring of multiple markers by SIM has been proven as potent in identification of a single herbal medicine from different CPMs in our lab.7 First, multi-components in SXT were globally profiled by an UPLC/QTOF-Fast DDA method. Structure characterization was performed by integrated analysis of the high-accuracy MS, Fast DDA-MS2 data and by searching in-house libraries. Second, the chemical markers representing the presence of NRR, CF, and CR, in SXT were established using a home-made SXT sample. A SIM approach enabling the monitoring of all chemical markers was developed on a QTOF mass spectrometer, and then transferred onto a single quadrupole MS detector QDa. Third, twelve batches of commercial SXT samples from different vendors were identified by means of the UPLC/QDa-SIM method. The sensitivity between QTOF and QDa in the analysis of SXT was compared as well. In particular, a large number of reference standards were used to make the multi-component characterization result more reliable.
 |
| Fig. 1 A general workflow for a strategy from systematic multi-component characterization to selective ion monitoring of chemical markers for authentication of CPM: Shuxiong tablet as a case study. | |
Experimental
Materials and reagents
A total of 73 reference standards were used for characterization of multi-components in SXT. Detailed information on the names and source of these reference standards is offered in the ESI.† Fig. 2 exhibits their chemical structures. Leucine-enkephalin was purchased from Sigma-Aldrich (St. Louis, MO, USA). HPLC-grade acetonitrile (Merck, Darmstadt, Germany), formic acid (ROE Scientific INC., USA), and ultra-pure water (18.2 MΩ cm at 25 °C) were in-house prepared by a Millipore Alpha-Q water purification system (Millipore, Bedford, USA), and were used in the mobile phase. Twelve batches of commercial SXT samples (Table 1) were purchased from different manufacturers or drug stores. Crude drug materials, comprising CF from Anhui Bozhou Shaosheng Pharmaceutical Co., Ltd. (Anhui, China), CR from Tongrentang Co., Ltd. (Beijing, China), and NRR kindly provided by Guangxi Wuzhou Pharmaceutical (Group) Co., Ltd. (Guangxi, China), were used to prepare a home-made SXT standard preparation for the establishment of chemical markers. CF, CR, and NRR were authenticated and qualified according to the quality standards recorded in ChP.
 |
| Fig. 2 Chemical structures of 73 reference standards. Compounds 170, 179, 227 and 248 possess the 20(R)-configuration, while the other saponins belong to the PPT or PPD type having the 20(S)-configuration. | |
Table 1 Information of twelve batches of commercial SXT samples
No. |
Vendor |
Batch no. |
S1 |
Zhengzhou Handu |
130408 |
S2 |
Zhengzhou Handu |
150316 |
S3 |
Jilin Jianjin |
20120601 |
S4 |
Jilin Yatai Mingxing |
20130101 |
S5 |
Beijing Jiandu |
20131201 |
S6 |
Hubei Lishizhen |
201310001 |
S7 |
Hubei Lishizhen |
201407001 |
S8 |
Hubei Lishizhen |
201501001 |
S9 |
Hubei Lishizhen |
201502001 |
S10 |
Hubei Lishizhen |
201504001 |
S11 |
Hubei Lishizhen |
201504002 |
S12 |
Hubei Lishizhen |
201412001 |
Sample preparation
Each SXT sample was pulverized into fine powder in a mortar with the coating removed. An aliquot of 0.25 g of each accurately weighed powder was placed in a 50 mL centrifuge tube and ultrasonicated (37 kHz, 1130 W) with 25 mL 70% aqueous methanol (v/v) for 30 min. After being centrifuged at 4000 rpm for 10 min, the supernatant was separated and further centrifuged at 14
000 rpm for 10 min. The obtained supernatant was used as the test solution and stored at 4 °C prior to analysis.
A standard preparation of SXT was in-house prepared according to the ingredient and procedure information recorded in ChP. One tenth of the ingredient amount was used, that is, 20 g CR, 10 g CF and 10 g NRR. Briefly, CR was decocted with 200 mL water for 2 h, and the extract was filtered. The filtrate was reserved in another container for later use. CF and the residue of CR were decocted in 200 mL water twice (2 × 1 h). These decoctions were combined and filtered. The filtrate was placed for 24 h. The supernatant was separated, filtered, and concentrated into a thick liquid. The powder of NRR was thoroughly mixed with the liquid and then dried under vacuum. The test solution of SXT standard preparation was prepared following the same sample preparation process as that for commercial SXT samples.
UPLC/QTOF-Fast DDA for multi-component characterization
An optimal UPLC/QTOF-Fast DDA method was developed to comprehensively profile and characterize the chemicals in SXT. Chromatographic separation was performed on a Waters ACQUITY I-Class UPLC® system (Waters Corporation, Milford, MA, USA) equipped with a binary solvent manager, a sample manager, a column manager, and a PDA detector. A Phenomenex Kinetex XB-C18 column (2.1 × 100 mm, 1.7 μm) equipped with an on-line filter was used and eluted by a binary mobile phase composed of acetonitrile (B) and 0.1% formic acid (v/v) following a gradient elution program: 0–1 min: 5% B; 1–2 min: 5–10% B; 2–9 min: 10–33% B; 9–15 min: 33–46% B; 15–16 min: 46–70% B; 16–21 min: 70–90% B; and 21–22 min: 90% B. The column temperature was set at 35 °C. The flow rate was 0.3 mL min−1. Each time 2 μL of the test solution was injected.
High-resolution centroided MS data were acquired on a Waters Xevo® G2-S QTOF mass spectrometer (Waters, Manchester, UK) connected to the UPLC system via a Zpray™ ESI source. Data-dependent Fast DDA in both ionization modes (ES− and ES+) was employed for structural elucidation. For the ESI source, capillary voltages of 3.5 kV and 2.5 kV were set in ES+ and ES−, respectively. A cone voltage of 60 V, cone gas flow of 30 L h−1, source temperature of 120 °C, and desolvation gas flow of 800 L h−1 at 450 °C were utilized in both ionization modes. The mass range of m/z 150–1500 was set for full-scan acquisition at a low collision energy (CE) of 6 V, and m/z 50–1500 for MS2 with a scan time of 0.1 s. The three most abundant parent ions were automatically selected to trigger MS/MS fragmentation. A CE ramp was used for MS/MS to obtain more structure information. A CE ramp of 20–60 V and 50–90 V was set for low mass and high mass, respectively, in ES−, while 20–50 V for both low mass and high mass was set in ES+. A solution of leucine-enkephalin (1 μg mL−1) was constantly infused at 5 μL min−1 as the lock mass for data calibration by reference to the ions m/z 554.2615 in ES− and 556.2771 in ES+. Data acquisition and processing were performed using MassLynx V4.1 software (Waters, Manchester, UK).
UPLC/QTOF-SIM
A SIM method by monitoring the established chemical markers was initially established on the QTOF mass spectrometer. The same UPLC chromatographic conditions, except for a meticulously modified gradient elution program, as that used for multi-component characterization was used: 0–2 min: 10% B (acetonitrile); 2–4 min: 10–14% B; 4–5.3 min: 14–18% B; 5.3–8.0 min: 18% B; 8–8.4 min: 18–21% B; 8.4–14 min: 21% B; 14–14.4 min: 21–40% B; 14.4–16.6 min: 40–43% B; 16.6–16.8 min: 43–60% B; 16.8–23 min: 60% B; 23–23.3 min: 60–90% B; and 23.3–26 min: 90% B. SIM on QTOF was performed by the MS/MS function with the CE set at 6 V. Detailed information with regard to the SIM settings is given in Table 2.
Table 2 Information of SIM settings on UPLC/QTOF and UPLC/QDa
Markers |
ESI mode |
Theoretical precursors (m/z) |
QTOF |
QDa |
Time window (min) |
Selected ions by Q (m/z) |
Scan range (Da) |
Time window (min) |
Selected ions by Q (m/z) |
Deprotonated precursor ion. Formic acid adduct precursor ion. Protonated precursor ion. FA: ferulic acid. KR: kaempferol 3-O-rutinoside. ZL: Z-ligustilide. LA: levistilide A. |
HSYAa |
ES− |
611.1612 |
3.4–3.9 |
611.16 |
100–616 |
2.8–3.4 |
611.16 |
FAa,d |
ES− |
193.0501 |
6.1–6.5 |
193.05 |
100–198 |
5.8–6.3 |
193.05 |
KRa,e |
ES− |
593.1506 |
7.9–8.4 |
593.15 |
100–598 |
7.3–7.8 |
593.15 |
AnHSYBa |
ES− |
1043.2669 |
8.4–8.8 |
1043.27 |
100–1048 |
7.8–8.3 |
1043.27 |
Noto-R1b |
ES− |
977.5321 |
11.7–12.2 |
977.53 |
100–983 |
11.1–11.6 |
977.53 |
Rg1b |
ES− |
845.4899 |
13.6–14.4 |
845.49 |
100–850 |
12.9–13.4 |
845.49 |
Reb |
ES− |
991.5478 |
14.0–14.8 |
991.55 |
100–996 |
13.1–13.6 |
991.55 |
Rb1b |
ES− |
1153.6006 |
15.4–15.9 |
1153.60 |
100–1158 |
15.3–15.8 |
1153.60 |
Rdb |
ES− |
991.5478 |
16.0–16.5 |
991.55 |
100–996 |
15.8–16.3 |
991.55 |
ZLc,f |
ES+ |
191.1072 |
18.5–18.8 |
191.11 |
100–196 |
18.3–18.6 |
191.11 |
LAc,g |
ES+ |
381.2066 |
22.6–23 |
381.21 |
100–386 |
22.6–22.8 |
381.21 |
UPLC/QDa-SIM for authentication of commercial SXT samples
The SIM method established on QTOF was transferred onto another set of LC-MS instruments involving the Waters ACQUITY I-Class UPLC® system and a single quadruple MS detector QDa. The same chromatographic column and mobile phase as those used in UPLC/QTOF-MS were applied. SIM on QDa was achieved in SIR mode at a low cone voltage of 15 V with a sampling frequency of 8 Hz. Detailed information concerning the UPLC/QDa-SIM settings is given in Table 2. Notably, due to the difference in system volume of the UPLC system, the time windows were slightly modified. The UPLC/QDa-SIM method enables the simultaneous monitoring of all chemical markers within a single run because of the rapid switching between ES− and ES+.
Results and discussion
Optimization of UPLC/QTOF-Fast DDA for multi-component analysis of SXT
To obtain good separation and high sensitivity, the UPLC/QTOF-Fast DDA method for systematic multi-component analysis of SXT was optimized in the first step. The chromatographic conditions, including the stationary phase, additive in the mobile phase, column temperature, flow rate, and QTOF parameters such as the capillary voltage and CE were compared. Among six columns (HSS T3, Synergi Fusion, BEH C18, BEH Shield RP-18, Zorbax SB C18, and Kinetex XB C18) from different vendors, the Kinetex XB C18 column gave a high resolution and good peak shape (Fig. S1†). Acetonitrile–H2O containing 0.1% formic acid was selected as the mobile phase due to the best resolution and ion response (Fig. S2†). The influence resulting from alternation in the column temperature (from 25 °C to 45 °C) was negligible, and 35 °C was used (Fig. S3†). The flow rate was set at 0.3 mL min−1 to maintain an appropriate system pressure and analytical efficiency (Fig. S4†). The gradient elution program was also optimized by considering the chromatographic performance in both ES− and ES+.
Key QTOF parameters, involving the capillary voltage affecting the ion response and CE determining the MS/MS fragmentation, were optimized. The components used for ion response assessment were ten representative compounds representing four structure classes in SXT, including noto-R1, Rg1, Re, Rb1, and Rd (saponin), hydroxysafflor yellow A (HSYA) and anhydrosafflor yellow B (AnHSYB, QCG), and kaempferol-3-O-rutinoside (FOG) detected in ES−, together with levistilide A and Z-ligustilide (phthalide) in ES+. The increase of capillary voltage from 1.5 kV to 3.5 kV resulted in different effects to saponins, QCGs, and FOG in ES−, but a constant enhancement on phthalides in ES+ (Fig. S5†). As a compromise, the capillary voltages of 2.5 kV and 3.5 kV were finally set for ES− and ES+, respectively. Ramping the CE can facilitate mass-dependent dynamic energies and thus allows for good MS/MS fragmentation for different masses of structures,7 which is especially suitable for complex matrix analysis like CPM. We firstly probed the most suitable CE range for different structure classes by comparing the MS/MS spectra obtained under constant CEs from 10 V to 110 V (Fig. S6†). As a result, preferable CE ranges of 30–50 V for kaempferol-3-O-rutinoside (m/z 593), 30 V for HSYA (m/z 611), 30–70 V for AnHSYB (m/z 1043), 50–70 V for Re (m/z 991), Rg1 (m/z 845), Rb1 (m/z 1153), and Rd (m/z 991), and 30–70 V for noto-R1 (m/z 977) were acquired. Subsequently, several different CE ramps were tested, and the low mass CE ramp of 20–50 V and high mass CE ramp at 50–90 V were set in ES−. Given that the CE range of 30–50 V was suitable for both Z-ligustilide (monomer, m/z 191) and levistilide A (dimer, m/z 381) (Fig. S7†), the CE ramp of 20 V was set for both low mass and high mass in ES+, to characterize phthalide derivatives. The MS2 spectra obtained under the optimized CE ramp of these ten compounds are given in Fig. S8.†
Ultimately, an optimized UPLC/QTOF-Fast DDA method was developed and further employed to profile and characterize the multi-components in SXT. Fig. 3 shows the base peak chromatograms of SXT obtained in ES− and ES+, respectively.
 |
| Fig. 3 Base peak chromatograms (BPCs) of a home-made SXT sample in ES− (A) and ES+ (B) on the UPLC/QTOF instrument. Those identified peaks by comparison with reference standards are annotated. | |
Collision-induced dissociation of 73 reference standards
In addition to the provision of retention evidence for peak identification, the collision-induced dissociation (CID) features of the reference standards were investigated to deduce the common fragmentation pathways and diagnostic product ions (DPIs) useful for the comprehensive characterization of different subclasses of chemicals in SXT. A large number of reference standards representing five structure subclasses, involving 39 saponins, 5 QCGs, 17 FOGs, and 7 phenolic acids detected in ES−, together with 4 phthalide derivatives in ES+, are studied here. The nomenclature of product ions is consistent with our previous reports for FOGs and saponins.28,29
Saponins. Dammarane-type tetracyclic triterpenoid saponins are the major bioactive components in P. notoginseng, which are composed of 1–6 sugar residue(s) and a sapogenin of protopanaxadiol (PPD), protopanaxatriol (PPT), or their variants on the C17 side chain.23,24,29 The known sugar residues involve glucose (Glc), glucuronic acid (GluA), rhamnose (Rha), xylose (Xyl), and pyran-/furan-arabinose (f-Ara/p-Ara). These saponin reference standards contain six sapogenins: PPT (C30H52O4), PPD (C30H52O3), 20-dehydrated PPT (C30H50O3), OA (oleanolic acid, C30H48O3), 5-ene-PPD (C30H50O3), and 7-OH-5-ene-PPD (C30H50O4). The DPIs associated with sapogenin characterization were observed at m/z 475.38 for PPT, m/z 459.38 for PPD, m/z 457.37 for both 20-dehydrated PPT and 5-ene-PPD, m/z 455.35 for OA, and m/z 473.37 for 7-OH-5-ene-PPD. Additional DPIs of PPD and PPT were observed at m/z 375.29 and 391.29, respectively, due to the neutral loss of C6H12 (84.09 Da) on the C17 side chain. Additionally, the sugar residues were characterised by neutral elimination, including 162.05 Da for Glc, 176.03 Da for GluA, 146.06 Da for Rha, and 132.04 Da for both Xyl and Ara (p-/f-). Interestingly, the low mass region (m/z < 400) gave direct DPIs for characterization of attached sugar residues or oligosaccharide chains, consistent with the results obtained by high energy collision-induced dissociation (HCD) on an LTQ-Orbitrap mass spectrometer.205,6-Didehydroginsenoside-Rb1 (169) is a characteristic ginsenoside having a sapogenin of the 5-ene-PPD type.30 It gave a rich formic acid adduct ion ([M + HCOO]−) at m/z 1151.59 in ES− (Fig. 4). CID of m/z 1151.59 produced a balanced MS2 spectrum showing product ions at m/z 1105.58 ([M − H]−), m/z 943.53 ([M − Glc − H]−), m/z 781.47 ([M − 2Glc − H]−), m/z 619.42 ([M − 3Glc − H]−), and m/z 457.37 for the sapogenin, and m/z 221.07 for the fragmented disaccharide chain ([GlcGlc + H2O − 120 Da − H]−). 20(S)-Sanchirhinoside-A5 (122) is a typical PPT-type saponin, with the formic acid adduct ion ([M + HCOO]−) at m/z 977.53. Analogous sequential elimination of sugar residues to that of 5,6-didehydro-Rb1 was observed by CID of the precursor ion. The secondary product ion of PPT at m/z 391.29 was also observed (Fig. 4).
 |
| Fig. 4 The MS and MS2 spectra and proposed fragmentation pathways for six reference standards. 5,6-Didehydroginsenoside Rb1 (169) and 20(S)-sanchirhinoside A5 (122), 6-hydorxyapigenin-6-O-glucoside-7-O-glucuronide (63), isosafflomin C (119), 3,5-di-O-caffeoylquinic acid (100), and senkyunolide I (114) belong to the saponin, flavonoid O-glycoside, quinochalcone C-glycoside, phenolic acid, and phthalide types, respectively. | |
FOGs. For FOGs isolated from C. tinctorius (CF), the aglycone moiety contains 6-hydroxyapigenin, 6-hydroxykaempferol, 6-hydroxyquercetin, kaempferol, quercetin, (2S)-4′,5,6,7-tetrahydroxyflavanone, and 3,7,3′,4′-tetrahydroxy-5-methoxyflavonol, and the sugar moieties involve Glc, Rha, and GluA.31–33 CID of FOG reference standards in ES− exhibited similar features to those of saponins. Deprotonated flavonoid aglycone ion species showed a diagnostic significance in differentiation of FOG subclasses and glycosylation sites. First, flavonol 3-O-glycosides can be easily differentiated from 7-O-glycosidic flavonols and flavones by the ratio of Y0− to [Y0 − H]−˙ generated due to heterolytic and homolytic cleavages.28 For example, the aglycone ions of isomeric quercetin-7-O-glucoside (83) and quercetin-3-O-glucoside (84) were remarkably different in Y0−/[Y0 − H]−˙ (m/z 301.03 and 300.03): approximately 2.5 and 0.5 for 83 and 84, respectively (Fig. S9†). Second, additional diagnostic information for flavonol 3-O-glycosides was a higher than 50% value of [Y0 − CH2O]− to Y0− (or [Y0 − H]−˙), which was less than 30% for flavonol 6,7-di-O-glucoside (63), flavonol 7-O-glucoside (83), flavone 7-O-glucoside (85), and flavanone 6-O-glucoside (88). Third, [Y0 − 2H]− was diagnostic for flavonol 3,7-di-O- or 3,6-di-O-glycosides (59), and could be used to differentiate from flavonol 3-mono-O-diglycosides (75) (Fig. S10†).34 6-Hydroxyapigenin-6-O-β-D-glucoside-7-O-β-D-glucuronide (63) is a typical flavone di-O-glycoside, showing a dominant deprotonated precursor ion at m/z 623.13. Its CID generated the base peak aglycone ion at m/z 285.04 together with weak Y07− at m/z 447.09 (Fig. 4). No abundant [Y0 − H]−˙ and [Y0 − 2H]− product ions were obtained, which was remarkably different from the CID of flavonol 3-mono-O-glycosides or 3,6-di-O-/3,7-di-O-glycosides.
QCGs. QCGs are the major bioactive components associated with the blood-invigorating effect of CF.25,26 The negative mode CID of five QCGs (monomers: HSYA, isosafflomin C, safflomin C, and saffloquinoside A; dimer: AnHSYB) was investigated. CID of QCGs in ES− featured the DPI of m/z 119.05 (C8H7O−) and cross-cleavage of the Glc characteristic for C-glycosides.7,19 CID of HSYA (39) was representative of QCG monomers (Fig. 4). Different from saponins and FOGs, neutral elimination of a whole Glc residue by 162.05 Da rarely occurred to QCGs, but the combined neutral loss of Glc + H2O + CO by 208.06 Da and crossing cleavage of Glc (0,2X0) by 120.04 Da were very common. CID-MS/MS of the precursor ion m/z 611.16 generated product ions at m/z 491.12 (0,2X0 of Glc), 403.10 [–(Glc + H2O + CO)], 325.07 (0,2X060,2X04–CO–H2O), 283.06, 205.01, and 119.05.
Phenolic acids. The negative mode CID of three caffeoylquinic acid derivatives (neochlorogenic acid, chlorogenic acid, and 3,5-di-O-caffeoylquinic acid) was studied. The neutral elimination of caffeoyl (C9H6O3, 162.03 Da) and generation of deprotonated quinic acid (m/z 191.06) together with its secondary product ions by elimination of H2O and H2O + CO2 were the common CID features.35 In addition, these components generally show characteristic UV absorption around 325 nm.36 For 3,5-di-O-caffeoylquinic acid (100), a deprotonated precursor ion was observed at m/z 515.12 (Fig. 4). Its CID could undergo successive loss of two caffeoyls with the remaining residues at m/z 353.09 and 191.06. Moreover, the secondary product ions of quinic acid at m/z 179.04 and 135.05 were ascribed to the neutral loss of H2O and H2O + CO2, respectively.
Phthalide derivatives. The phthalide derivative reference standards, including Z-ligustilide, senkyunolide H, senkyunolide I, and levistolide A, gave an abundant ion response in ES+, with weak sodium adduct precursor ions as well as the in-source fragmentation product ion due to the cleavage of H2O.37 In the case of senkyunolide I (114), the sodium adduct ion at m/z 247.10 and base peak ion at m/z 207.11 ([M + H − H2O]+) were observed in the full-scan spectrum (Fig. 4). Further CID of the ion m/z 207.10 generated diverse product ions due to the sequential elimination of 2H2O (m/z 189.09), 2H2O + CO (m/z 161.10), 2H2O + 2CO (m/z 133.07), 2H2O + CO + C3H6 (m/z 119.09), 2H2O + 3CO (m/z 105.07), and 2H2O + CO + C3H6 + H2O (m/z 91.05).
Comprehensive characterization of multi-components from SXT
Comparative analysis of the CID features of 73 reference standards representing five structure subclasses (saponin, FOG, QCG, phenolic acid, and phthalide) in SXT led to a MS data interpretation guideline (Fig. 5). Convenient primary characterization of multi-components in SXT could be achieved. In-house libraries of NRR, CF, and CR were established for reliable structural characterization of unknown components. Characterization of saponins, FOGs, QCGs, and phenolic acids from SXT was based on the ES− information, whist that for phthalide derivatives by the ES+ information.
 |
| Fig. 5 MS data interpretation guideline for multi-component characterization of SXT based on the negative and positive mode Fast DDA-MS2 data. | |
Characterization of saponins. In general the adduct ion forms of precursors can be used for primary subclass classification: (1) predominant formic acid adduct ions ([M + HCOO]−) concomitant with weak deprotonated precursor ions ([M − H]−) are diagnostic for neutral saponins, and (2) only rich [M − H]− precursors are common for malonyl-ginsenosides and OA-type saponins. An in-house library of NRR (95 known saponins are recorded) was retrieved for identity confirmation. A total of 106 saponins, involving 33 PPD, 40 PPT, 16 C17 side chain variations, 9 OA, and 8 miscellaneous ones, were identified or tentatively characterized. Of them, 39 were identified by comparison with reference standards.Compound 162 (tR 10.27 min) was characterized as a PPT-type saponin, with a rich formic acid adduct ion at m/z 845.49 (Fig. S11†). Its molecular formula was thus determined as C42H72O14. CID of the precursor ion at m/z 845.49 generated a high intensity of product ions at m/z 799.49 ([M − H]−), m/z 637.43 ([M − H − Glc]−), the sapogenin ion at m/z 475.38 ([PPT − H]−), and its secondary fragment at m/z 391.28 due to the neutral elimination of C6H12, together with the oligosaccharide chain fragments at m/z 323.10 ([GlcGlc − H]−) and 221.07 ([GlcGlc + H2O − H − 120 Da]−). We could infer the presence of a disaccharide chain of GlcGlc attached to PPT. Given its different retention to that of Rf (160, tR 10.09 min), the disaccharide chain should be at 20-OH. By searching the NRR library, compound 162 matched noto-U [PPT-20-Glc(6,1)Glc]. Compound 180 (tR 10.27 min) was identified as a malonyl-ginsenoside since easy neutral loss of CO2 (−43.99 Da), C3H2O3 (−86.00 Da), and C3H2O3 + H2O (−104.01 Da) was observed by CID of the deprotonated precursor ion at m/z 1193.59 (Fig. S11†).29,38 In addition, the other MS/MS product ions were the same as those obtained by CID of Rb1. Therefore, compound 180 was characterized as PPD-(Glc–Glc)–Glc–Glc–malonyl, one isomer of malonyl-ginsenoside-Rb1 (176, tR 11.14 min). Interestingly, since no hit was obtained by searching the in-house library, it should be a potential new malonyl-ginsenoside that has not been isolated from P. notoginseng. The saponins with varied C17 side chains were diverse in SXT, however, their characterization was rather challenging due to the unexpected structural diversity. Characterization of compound 156 (tR 9.61 min) is illustrated as an example given in Fig. S11.†
Characterization of FOGs. The FOGs in SXT were characterized by the sequential elimination of sugar residues and generation of flavonoid aglycone ion species (Y0−, [Y0 − H]−˙, and [Y0 − 2H]−˙) together with their secondary product ions by typical neutral loss or RDA fragmentation. The composition of the aglycone ion species is diagnostic for identification of the glycosylation sites. Compound 27 (tR 3.68 min) was a highly polar FOG with the precursor ion at m/z 801.17 (Fig. S12†), and the MS/MS product ions at m/z 625.14, 463.09, 301.04, and 271.03 were ascribed to [M − H − GluA]−, [M − H − GluA − Glc]−, Y0−, and [Y0 − CH2O]−, respectively. The known aglycone matching Y0− at m/z 301.04 included 6-hydroxykaempferol and quercetin. The product ions of the aglycone moiety, including abundant [Y0 − 2H]− (m/z 299.02) and Y0−/[Y0 − H − CH2O]−, were quite analogous to those of 3,6-di-O-glycosidic 6-hydroxykaempferol (Fig. S13†). Ultimately, compound 27 was tentatively identified as 6-hydroxykaempferol-3,6-di-O-glucoside-7-O-glucuronide or its isomer.
Characterization of QCGs. Characterization of QCGs from SXT was mainly based on the DPI at m/z 119.05 and typical neutral eliminations. Compound 61 (tR 5.22 min) was used as an example. The deprotonated precursor ion at m/z 611.16 indicated the molecular formula C27H32O16. Some key product ions at m/z 521.13, 287.06, and 119.05, detected in the MS/MS spectrum (Fig. S14†) were assigned as 0,3X, the aglycone moiety, and p-hydroxystyrene residue (C8H7O−). The aglycone ion was the same as those of AnHSYB (91, tR 6.67 min), isosafflomin C (119, tR 7.76 min), and safflomin C (125, tR 7.95 min). By searching the library of CF, 61 was characterized as hydroxysafflor yellow B or C or isomers. Characterization of QCGs from C. tinctorius has been discussed in detail in our recent report.19 As a result, 32 QCGs were identified or tentatively identified, including 27 monomers and 5 dimers.
Characterization of phenolic acids. A group of caffeoylquinic acids (or the glycosides) were characterized from SXT according to their typical neutral loss of caffeoyl (−162.03 Da) and generation of rich product ion corresponding to deprotonated quinic acid (m/z 191.06). Compounds 14 (tR 3.11 min, 5-caffeoylquinic acid), 33 (tR 3.92 min, 3-caffeoylquinic acid), and 100 (tR 6.92 min, 3,5-di-O-caffeoylquinic acid) were definitely identified by comparison with reference standards. The unknown compounds 9 (tR 2.73 min) and 37 (tR 4.06 min), showing the same deprotonated molecule at m/z 353.09 (C16H18O9) and very similar CID features to 14 and 33, were characterized as two mono-caffeoyl quinic acid isomers. On further comparison of the elution order and the MS/MS spectra with literature,35,39,40 compounds 9 and 37 were tentatively identified as 1-O-caffeoylquinic acid and 4-O-caffeoylquinic acid, respectively (Fig. S15†). Meanwhile, four unknown compounds 56 (tR 4.92 min), 92 (tR 6.70 min), 98 (tR 6.83 min), and 112 (tR 7.31 min), giving a deprotonated precursor ion at m/z 515.12 (C25H24O12), were di-caffeoyl quinic acids by comparison with the reference compound 100. However, we could not exactly identify the substituting sites only by MS information. Compounds 7 (tR 2.57 min), 12 (tR 2.98 min), 18 (tR 3.26 min), 22 (tR 3.52 min), and 29 (tR 3.78 min), showing the same [M − H]− at m/z 515.14 (C22H28O14), were not the isomers of di-caffeoyl quinic acids, but glycosides, thanks to the high-resolution mass measurement facilitated by QTOF. CID of these five compounds all produced the product ion at m/z 191.06 and underwent neutral elimination of a caffeoyl group. For instance, CID of 22 produced a high intensity of product ions m/z 323.08, 191.06, and 161.02, assigned as [Glc + cafferoyl − H]−, deprotonated quinic acid molecule, and [caffeic acid − H2O − H]−, respectively (Fig. S14†). And thus it was tentatively characterized as a caffeoylquinic acid glycoside.
Characterization of phthalide derivatives. Hydroxylated phthalides, in contrast to those hydroxyl-free ones, showed weaker retention and protonated precursors ([M + H]+), but higher intensity of [M + H − H2O]+ and weaker [M + Na]+ in ES+. Characterization of compound 89 (tR 6.56 min) was illustrated as an example. The ions present in the MS spectrum, including dominant [M + H − H2O]+, weak [M + H]+ and [M + Na]+, were observed at m/z 209.12, 227.13, and 249.11, respectively, based on which we could determine its molecular formula as C12H18O4 (Fig. S14†). Abundant MS/MS product ions at m/z 163.11, 153.06, and 107.05, were obtained by CID of the ion of m/z 209.12, due to the neutral loss of H2O + CO, 2CO, and C5H10O2 (by combined fragmentations), respectively. Since the fragmentation pathways of 89 were similar to those of senkyunolide H (124, tR 7.90 min) but with additional 2 Da in m/z values, we could infer a H2-added senkyunolide H. A further search of the in-house library of CR helped us to characterize 89 as senkyunolide J or N.As a result, a total of 250 compounds were identified or tentatively characterized, involving 106 saponins, 32 QCGs, 39 FOGs (including three free flavonoid aglycones), 26 phthalide derivatives, 28 phenolic acids, and 19 miscellaneous ones. Detailed information of these characterized components is given in Table S1.† A 2D plot of the characterized compounds by tR versus m/z is shown in Fig. S16,† in which clear grouping among different structure classes is observed. On one hand, FOGs, QCGs, and phenolic acids are eluted earlier (with tR < 10 min) than saponins, due to their larger polarity. On the other hand, saponins, phthalides, and QCG monomers are clearly grouped by the dimension of m/z or tR. These distribution features are beneficial to the establishment of targeted analytical methods.
Selection of chemical markers for authentication of SXT
Taking into account the pharmacological property, chemical representativeness and the content, 3–5 markers from each drug are selected to construct the chemical markers of SXT. Those known markers in ChP (2015 edition) for quality control of NRR, CF, CR, and SXT are given the highest priority. The selected markers are then confirmed by their presence in a home-made SXT sample and three compositional drugs.
For NRR, five main saponins, noto-R1, Rg1, Re, Rb1, and Rd, account for about 7.5% (w/w) of the drug material.41 Among them, noto-R1, Rg1, and Rb1 are the quantitative makers of NRR, while noto-R1, Rg1, Re, and Rb1 are used for identification of NRR in ChP. In addition, Rd was also considered as a marker for quantitative evaluation of NRR and as a neuroprotective agent to attenuate ischemic stroke damage.42 Therefore, these five saponins were used for identification of NRR from SXT. In the case of CF, HSYA and kaempferol (obtained by acid hydrolysis) are the quantitative markers for CF. An extract of C. tinctorius mainly containing kaempferol 3-O-rutinoside and AnHSYB showed potential in treatment of Parkinson’s disease.43 Our recent research by fingerprinting analysis of 20 batches of C. tinctorius samples could testify their presence as the common components (Fig. S17†).7 We therefore chose HSYA, AnHSYB, and kaempferol 3-O-rutinoside as the markers for identification of CF from SXT. Levistilide A and ferulic acid are the ChP markers for CR. Z-Ligustilide is a characteristic, rich phthalide component in CR and has been reported to have an anti-inflammatory effect.44,45 Thereby, levistilide A, Z-ligustilide, and ferulic acid were used as the markers for CR. These eleven chemical markers could represent five classes of bioactive components (saponin, FOG, QCG, phenolic acid, and phthalide), and were used for the identification of NRR, CF, and CR from SXT. Clearly, these chemical markers could be detected in the home-made SXT sample and three corresponding drugs (Fig. S18†).
Establishment of the UPLC/QTOF-SIM method
SIM, by selective monitoring and intuitive exhibition of the chemical marker, can give a higher sensitivity than full-scan methods,7 and thus is particularly suitable for authentication of CPM. To enhance the analytical efficiency and get better resolution of these markers, the gradient elution program was carefully modified. The mass range for each marker was set from 100 Da to the target m/z with an expansion by 5 Da, to ensure the simultaneous acquisition of the product ion of leucine-enkephalin (for MS/MS data calibration) and the precursor ion of each analyte. Consequently, a UPLC/QTOF-SIM method facilitating the detection of eleven chemical markers was established. Two chromatographic runs separately in ES− (noto-R1, Rg1, Re, Rb1, Rd, ferulic acid, HSYA, AnHSYB, and kaempferol 3-O-rutinoside) and ES+ (levistilide A and Z-ligustilide) are necessary to monitor these markers. By the UPLC/QTOF-SIM approach, all eleven markers could be detected in the home-made SXT sample (Fig. 6), with good resolution and intuitive exhibition. On the other hand, this SXT sample was analyzed under three different conditions by reference to the fingerprinting analysis reports of a Notoginseng total saponins preparation (Xueshuantong), CF, and CR,7,37,46 and no interference from the major isomers was observed (Fig. S19†). It indicated that the developed SIM method was specific for the detection of eleven markers suitable for qualitative identification of SXT samples.
 |
| Fig. 6 The combined SIM spectra of a home-made SXT sample (A) and two batches of commercial SXT samples (B-S9 and C-S5) obtained with UPLC/QTOF and UPLC/QDa, showing eleven chemical markers for authentication of SXT. HSYA: hydroxysafflor yellow A; FA: ferulic acid; KR: kaempferol-3-O-rutinoside; AnHSYB: anhydrosafflor yellow B; ZL: Z-ligustilide; LA: levistolide A. | |
Method transfer from QTOF to QDa
Taking into account the practicability, the SIM approach should be transferred onto an easily accessible instrument such as a UPLC/QDa system. QDa is a modular single quadrupole MS detector. SIM on QDa was accomplished in SIR mode by setting the precursor ions selected by Q and proper time windows under a low cone voltage. In contrast to QTOF, the advantages of QDa can be embodied in three aspects: (1) fast switching between ES− and ES+, facilitating the simultaneous monitoring of all chemical markers by a single chromatographic run; (2) easy parameter setting, with only the cone voltage and sampling frequency being pre-defined; and (3) the much lower cost and less space occupation, rendering QDa suitable in routine analysis by more users. The disadvantages of UPLC/QDa compared to UPLC/QTOF otherwise a involve lower resolution and sensitivity, and the sensitivity of the SIM methods on QTOF and QDa was assessed by comparing the signal to noise ratio (S/N) of six markers in ES− at different concentrations (Fig. S20†). Generally QDa gave a lower sensitivity than QTOF, and the ratio of S/N varied between 4.93% (0.008 mg mL−1 of noto-R1) and 81.0% (0.09 mg mL−1 of ferulic acid). The limit of detection (defined as S/N = 3) of QDa was estimated at 46 pg for noto-R1, 66 pg for HSYA, 1800 pg for ferulic acid, 56 pg for kaempferol 3-O-rutinoside, 334 pg for AnHSYB, and 24 pg for Rb1, respectively. It is the first report that compares the performance of UPLC/QDa and UPLC/QTOF in analysis of herbal components. In addition, QDa exhibited a better performance than the UV detector and comparable performance to a tandem mass spectrometer in qualitative and quantitative analyses of herbal components and biosamples.47–49
Identification of commercial SXT samples by UPLC/QDa-SIM
Authentication of twelve batches of commercial SXT samples collected from different vendors was performed using the established UPLC/QDa-SIM method (Fig. 6 and Fig. S21†). These SXT samples were of differentiated qualities. Samples S6 and S12 contained a high intensity of eleven markers. Only AnHSYB was of a low content in S1 and S2, while only FA was rare in S3, S5, S7, S8, and S11. However, for S4 and S9, three polar markers (HSYA, ferulic acid, and AnHSYB) were extremely rare. It may be caused by different preparation process or use of poor quality drug materials. The identification results obtained on QDa were similar to those on QTOF, indicating the practicability of UPLC/QDa-SIM for routine authentication work of SXT.
Conclusions
Complying with a concept of in-depth investigation-guided establishment of simple quality standards, the first practice in CPM was achieved by authentication of SXT following a strategy from systematic multi-component characterization to selective ion monitoring of chemical markers. UPLC/QTOF-Fast DDA was proven as very powerful in global profiling and characterization of multi-components in SXT, and as many as 250 compounds were identified or tentatively characterized, including 73 identified by comparison with reference standards. The highly selective and sensitive SIM method, by monitoring eleven chemical markers for identification of NRR, CF, and CR from SXT, was firstly established on the advanced QTOF instrument and then successfully transferred onto easily accessible QDa detection. In spite of the lower sensitivity, the UPLC/QDa-SIM approach enabled similar identification results to QTOF-SIM, but was more easily generalized in practice by more users. It indicates the feasibility of the SMC-SIM strategy in establishment of practical identification approaches for CPM. We believe that UPLC-SIM can be a solution to the comprehensive identification of CPM, and is also promising in its quantitative evaluation.
Conflict of interest
The authors declare no conflict of interest.
Acknowledgements
We gratefully acknowledge the financial support from National Natural Science Foundation of China (81503240 and 81473344) and the National Science and Technology Major Project for Major Drug Development (2014ZX09304-307-001-007).
References
- H. Chu, A. H. Zhang, Y. Han and X. J. Wang, World J. Tradit. Chin. Med., 2015, 1, 26–32 CrossRef.
- H. F. Xu, J. Guo, S. L. Chen, X. Liu, Y. Zhou, X. P. Zhang and X. D. Xu, J. Pharm. Biomed. Anal., 2013, 72, 267–291 CrossRef PubMed.
- D. Steinman and M. Ganzera, J. Pharm. Biomed. Anal., 2011, 55, 744–757 CrossRef PubMed.
- W. Xu, F. R. Song, D. J. Guo, J. Mi, Q. J. Qin, Q. Yu and S. Y. Liu, Curr. Anal. Chem., 2012, 8, 43–66 CrossRef.
- J. Cielecka-Piontek, P. Zalewski, A. Jelinska and P. Garbacki, Chromatographia, 2013, 76, 1429–1437 CAS.
- H. P. Wang, C. Chen, Y. Liu, X. W. Yang and H. B. Xiao, RSC Adv., 2015, 5, 80583–80590 RSC.
- W. Si, W. Z. Yang, D. A. Guo, J. Wu, J. X. Zhang, S. Qiu, C. L. Yao, Y. J. Cui and W. Y. Wu, J. Pharm. Biomed. Anal., 2016, 117, 510–521 CrossRef CAS PubMed.
- H. Zhang, D. Zheng, H. H. Li, H. Wang, H. S. Tan and H. X. Xu, Anal. Chim. Acta, 2016, 912, 85–96 CrossRef CAS PubMed.
- H. Q. Pan, W. Z. Yang, Y. B. Zhang, M. Yang, R. H. Feng, W. Y. Wu and D. A. Guo, Anal. Bioanal. Chem., 2015, 407, 6057–6070 CrossRef CAS PubMed.
- J. Y. Zhang, C. Li, Y. Y. Che, J. R. Wu, Z. J. Wang, W. Cai, Y. Li, Z. G. Ma and P. F. Tu, RSC Adv., 2015, 5, 80816–80828 RSC.
- J. X. Zhang, S. H. Guan, J. H. Sun, T. Liu, P. Chen, R. H. Feng, X. Chen, W. Y. Wu and D. A. Guo, Anal. Bioanal. Chem., 2015, 407, 581–595 CrossRef CAS PubMed.
- M. Hoicapek, R. Jirasko and M. Lisa, J. Chromatogr. A, 2012, 1259, 3–15 CrossRef PubMed.
- C. J. S. Lai, T. Tan, S. L. Zeng, L. W. Qi, X. G. Liu, X. Dong, P. Li and E. H. Liu, J. Pharm. Biomed. Anal., 2015, 109, 184–191 CrossRef CAS PubMed.
- M. Yang, Z. Zhou and D. A. Guo, Anal. Chim. Acta, 2015, 894, 44–53 CrossRef CAS PubMed.
- W. Z. Yang, W. Y. Wu, M. Yang and D. A. Guo, Chin. J. Nat. Med., 2015, 13, 867–872 Search PubMed.
- J. Y. Zhang, Z. J. Wang, Q. Zhang, F. Wang, Q. Ma, Z. Z. Lin, J. Q. Lu and Y. J. Qiao, Talanta, 2014, 124, 111–122 CrossRef CAS PubMed.
- X. Qiao, X. H. Lin, S. Ji, Z. X. Zhang, T. Bo, D. A. Guo and M. Ye, Anal. Chem., 2016, 88, 703–710 CrossRef PubMed.
- Y. L. Song, Q. Q. Song, J. Li, N. Zhang, Y. F. Zhao, X. Liu, Y. Jiang and P. F. Tu, J. Chromatogr. A, 2016, 1429, 238–247 CrossRef CAS PubMed.
- W. Z. Yang, W. Si, J. X. Zhang, M. Yang, H. Q. Pan, J. Wu, S. Qiu, C. L. Yao, J. J. Hou, W. Y. Wu and D. A. Guo, RSC Adv., 2016, 6, 495–506 RSC.
- S. Qiu, W. Z. Yang, X. J. Shi, C. L. Yao, M. Yang, X. Liu, B. H. Jiang, W. Y. Wu and D. A. Guo, Anal. Chim. Acta, 2015, 893, 65–76 CrossRef CAS PubMed.
- L. Ridder, J. J. J. van der Hooft, S. Verhoeven, R. C. H. de Vos, J. Vervoort and R. J. Bino, Anal. Chem., 2014, 86, 4767–4774 CrossRef CAS PubMed.
- J. X. Zhang, W. Z. Yang, S. R. Li, S. Yao, P. Qi, Z. Yang, Z. J. Feng, J. J. Hou, L. Y. Cai, M. Yang, W. Y. Wu and D. A. Guo, Anal. Bioanal. Chem., 2016, 408, 3881–3890 CrossRef CAS PubMed.
- J. J. Liu, Y. T. Wang, L. Qin, Y. Y. Yu and C. M. Wang, Expert Opin. Invest. Drugs, 2014, 23, 523–539 CrossRef CAS PubMed.
- W. Z. Yang, Y. Hu, W. Y. Wu, M. Ye and D. A. Guo, Phytochemistry, 2014, 106, 7–24 CrossRef CAS PubMed.
- X. D. Zhou, L. Y. Tang, Y. L. Xu, G. H. Zhou and Z. J. Wang, J. Ethnopharmacol., 2014, 151, 27–43 CrossRef CAS PubMed.
- S. J. Yue, Y. P. Tang, S. J. Li and J. A. Duan, Molecules, 2013, 18, 15220–15254 CrossRef CAS PubMed.
- W. X. Li, Y. P. Tang, Y. Y. Chen and J. A. Duan, Molecules, 2012, 17, 10614–10651 CrossRef CAS PubMed.
- W. Z. Yang, X. Qiao, T. Bo, Q. Wang, D. A. Guo and M. Ye, Rapid Commun. Mass Spectrom., 2014, 28, 385–395 CrossRef CAS PubMed.
- W. Z. Yang, M. Ye, X. Qiao, C. F. Liu, W. J. Miao, T. Bo, H. Y. Tao and D. A. Guo, Anal. Chim. Acta, 2012, 739, 56–66 CrossRef CAS PubMed.
- J. B. Wan, Q. W. Zhang, S. J. Hong, J. Guan, W. C. Ye, S. P. Li, M. S. Lee and Y. T. Wang, Molecules, 2010, 15, 8169–8176 CrossRef CAS PubMed.
- L. Fan, H. Y. Zhao, R. Pu, J. Han, B. R. Wang and D. A. Guo, Chin. Pharm. J., 2011, 46, 333–337 CAS.
- V. A. Kurkin and A. V. Kharisova, Chem. Nat. Compd., 2014, 50, 446–448 CrossRef CAS.
- S. Y. Lim, A. Y. Hwang, H. R. Yoon and Y. S. Paik, J. Appl. Biol. Chem., 2007, 50, 304–307 CAS.
- K. Ablajan, Z. Abliz, X. Y. Shang, J. M. He, R. P. Zhang and J. G. Shi, J. Mass Spectrom., 2006, 41, 352–360 CrossRef CAS PubMed.
- J. Y. Zhang, Z. J. Wang, Y. Li, Y. Liu, W. Cai, C. Li, J. Q. Lu and Y. J. Qiao, Talanta, 2016, 147, 16–27 CrossRef CAS PubMed.
- L. Zhang, Z. C. Tu, H. Wang, Z. F. Fu, Q. H. Wen and D. Fan, Food Chem., 2015, 186, 123–132 CrossRef CAS PubMed.
- X. L. Zhang, L. F. Liu, L. Y. Zhu, Y. J. Bai, Q. Mao, S. L. Li, S. L. Chen and H. X. Xu, J. Pharm. Biomed. Anal., 2014, 88, 391–400 CrossRef CAS PubMed.
- W. Z. Yang, T. Bo, S. Ji, X. Qiao, D. A. Guo and M. Ye, Food Chem., 2013, 139, 762–769 CrossRef CAS PubMed.
- K. Schutz, D. Kammerer, R. Carle and A. Schieber, J. Agric. Food Chem., 2004, 52, 4090–4096 CrossRef PubMed.
- M. N. Clifford, K. L. Johnston, S. Knight and N. Kuhnert, J. Agric. Food Chem., 2003, 51, 2900–2911 CrossRef CAS PubMed.
- S. P. Li, C. F. Qiao, Y. W. Chen, J. Zhao, X. M. Cui, Q. W. Zhang, X. M. Liu and D. J. Hu, J. Chromatogr. A, 2013, 1213, 302–307 CrossRef PubMed.
- S. F. Nabavi, A. Sureda, S. Habtemariam and S. M. Nabavi, J. Ginseng Res., 2015, 39, 299–303 CrossRef CAS PubMed.
- R. T. Ren, C. Y. Shi, J. Cao, Y. Sun, X. Zhao, Y. F. Guo, C. Wang, H. Lei, H. J. Jiang, N. Ablat, J. M. Xu, W. Li, Y. C. Ma, X. R. Qi, M. Ye, X. P. Pu and H. B. Han, Sci. Rep., 2016, 6, 22135 CrossRef CAS PubMed.
- Y. F. He, Q. Li and K. S. Bi, J. Sep. Sci., 2015, 38, 1090–1099 CrossRef CAS PubMed.
- L. Liu, Z. Q. Ning, S. Song, K. Zhang, T. Deng, X. P. Lu and Y. Y. Cheng, Planta Med., 2005, 71, 808–813 CrossRef CAS PubMed.
- W. Z. Yang, J. X. Zhang, C. L. Yao, S. Qiu, M. Chen, H. Q. Pan, X. J. Shi, W. Y. Wu and D. A. Guo, J. Pharm. Biomed. Anal., 2016, 128, 322–332 CrossRef CAS PubMed.
- L. Veryser, L. Taevernier, N. Roche, P. Blondeel and B. De Spiegeleer, J. Pharm. Biomed. Anal., 2015, 115, 594–602 CrossRef CAS PubMed.
- J. Galba, A. Michalicova, V. Parrak, M. Novak and A. Kovac, J. Pharm. Biomed. Anal., 2016, 117, 85–90 CrossRef CAS PubMed.
- D. Spaggiari, F. Mehl, V. Desfontaine, A. G. Perrenoud, S. Fekete, S. Rudaz and D. Guillarme, J. Chromatogr. A, 2014, 1371, 244–256 CrossRef CAS PubMed.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra10883k |
‡ These authors contributed equally to this work. |
|
This journal is © The Royal Society of Chemistry 2016 |
Click here to see how this site uses Cookies. View our privacy policy here.