tandem mass spectrometry

tandem mass spectrometry

Talanta 77 (2008) 152–159 Contents lists available at ScienceDirect Talanta journal homepage: www.elsevier.com/locate/talanta Identification of poly...

590KB Sizes 0 Downloads 28 Views

Talanta 77 (2008) 152–159

Contents lists available at ScienceDirect

Talanta journal homepage: www.elsevier.com/locate/talanta

Identification of polyoxypregnane glycosides from the stems of Marsdenia tenacissima by high-performance liquid chromatography/tandem mass spectrometry Juanjuan Chen a , Xiaoyu Li b , Cuirong Sun a , Yuanjiang Pan a,∗ , Urs Peter Schlunegger a b

Department of Chemistry, Zhejiang University, Zheda Road 38#, Hangzhou 310027, China Institute of Materia Medica, Zhejiang Academy of Medical Sciences, Hangzhou 310028, China

a r t i c l e

i n f o

Article history: Received 5 March 2008 Received in revised form 28 May 2008 Accepted 31 May 2008 Available online 8 June 2008 Keywords: Marsdenia tenacissima Polyoxypregnane glycosides HPLC-MSn Energy-resolved breakdown curves FTICR-MS

a b s t r a c t A facile method based on high-performance liquid chromatography coupled with electrospray ionization tandem mass spectrometry (HPLC/(+)ESI-MSn ) has been established for the analysis of polyoxypregnane glycosides in the stems of Marsdenia tenacissima. The data reveals the ability of MSn in the structural elucidation of polyoxypregnane glycosides including the nature of the polyoxypregnane core, the kinds of the substituents and the types of sugar residues. Offline Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) is also performed to assign accurate elemental compositions. In this study, eighteen polyoxypregnane glycosides have been investigated. Among these components, five compounds are unambiguously identified as Marsdenoside K, Tencissoside A, B, C and D; two compounds are established as novel compounds based on mass spectral data; and the other eleven compound’s structures are tentatively proposed. Furthermore, breakdown curves are constructed to distinguish five pairs of isomers among these eighteen compounds. As far as our knowledge, this is the first report on identification of polyoxypregnane glycosides in the stems of M. tenacissima by HPLC/ESI-MSn directly, which could save time and material consuming efforts in traditional phytochemistry analyses. Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved.

1. Introduction Marsdenia tenacissima (Roxb.)Wight et Arn. (Asclepiadaceae) is a perennial climber that grows from tropical to subtropical Asia. It is well known as “Tong-guang-teng” for the treatment of asthma, cancer, trachitis, tonsillitis, pharyngitis, cystitis, and pneumonia by its stems in Chinese folk medicine. Polyoxypregnane glycosides are the major bioactive constituents and rich in the stems of M. tenacissima [1–8]. To further investigate the pharmaceutical activities of these polyoxypregnane glycosides, identification of their chemical structures is necessary. However, traditional analytical protocols are tedious, laborious, time-consuming and insensitive, in which the minor constituents in plant extracts are easily ignored. Therefore, there is an increasing demand for methods for rapidly identifying and characterizing known or new structures. The method of HPLC/ESI-MSn has the advantage of high sensitivity, relatively short analysis time, considerable structural information, and low levels of sample consumption. It has the potential ability to rapidly screening of the minor constituents in

plant extracts that are difficult to obtain by conventional phytochemical means [9–12]. Glycosides in crude plant extracts have been determined by HPLC/ESI-MSn in the recent reports, demonstrating the advantages of mass spectrometry in phytochemistry [13–15]. However, to our knowledge, no previous studies of polyoxypregnane glycosides by HPLC/ESI-MSn have been published in the stems of M. tenacissima. This is the first report concerning the characterization of polyoxypregnane glycosides in M. tenacissima by HPLC/(+) ESI-MSn . Based on the fragmentations behaviors of five known polyoxypregnane glycosides, eleven compounds were tentative proposed and two compounds were established as novel compounds. Breakdown curves were utilized to differentiate five pairs of isomers among these eighteen compounds. Results showed that two pairs of positional isomers and two pairs of stereochemical isomers in this study could be distinguished obviously. 2. Experimental 2.1. Reagents and chemicals

∗ Corresponding author. Tel.: +86 571 87951285; fax: +86 571 87951629. E-mail address: [email protected] (Y. Pan).

Polyoxypregnane glycosides used for identification purposes by HPLC and MS were isolated previously from the plant M. tenacis-

0039-9140/$ – see front matter. Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2008.05.054

J. Chen et al. / Talanta 77 (2008) 152–159

153

Fig. 1. HPLC/UV (220 nm) separation and total ion chromatography (TIC) of polyoxypregnane glycosides from the extracts of the stems of Marsdenia tenacissima.

sima in the laboratory of Institute of Materia Medica, Zhejiang Academy of Medical Sciences. Methanol (chromatographic grade) used for analytical HPLC and preparative HPLC was purchased from Merck (Darmstadt, Germany). Deionized water (18 M) was obtained from a Milli-Q water purification system (Millipore, Bedford, MA, USA). The stems of M. tenacissima were obtained from Yunnan province, China.

10 ␮m) was used for preparation; the column was maintained at room temperature. The eluent for preparative HPLC was mixture of appropriate percentage of methanol and water: 72% methanol for 11, 14 and 18. The flow rate was 8.0 mL min−1 , and detection wavelength was 220 nm.

2.2. Sample preparation

HPLC/ESI-MSn analyses were performed using the Agilent HPLC system described above combined with a Bruker Esquire 3000plus ion trap mass spectrometer (Bruker–Franzen Analytik GmbH, Bremen, Germany) equipped with electrospray ionization (ESI). Instrument control and data acquisition were performed using Esquire 5.0 software. The ion source temperature was 250 ◦ C, and needle voltage was always set at 4.0 kV. Nitrogen was used as the drying and nebulizer gases at a flow rate of 10 L min−1 and a backpressure of 30 psi. Helium was introduced into the trap with an estimated pressure 6 × 10−6 mbar to improve trapping efficiency to act as the collision gas for the MSn data; the mass spectrometer was optimized in the collision energy range of 0.6–1.0 V to maximize the ion current in the spectra. The MSn spectral data of eighteen compounds are supported as the supplementary information. The offline FTICR-MS experiments were performed using an Apex III Fourier transform ion cyclotron resonance mass spectrometer with 7.0T actively shielded superconducting magnet (Bruker Daltonics, Billerica, MA, USA) combined with an Apollo electrospray ionization source operated in the positive ion mode. The solutions were infused at a rate of 3.0 ␮L min−1 using a Cole-Parmer syringe pump. Accurate mass measurements were performed using NaI as an external calibration compound. Each spectrum was an average of eight transients, each composed of 512 K points, acquired using a workstation operating XMASS version 6.1.1.

The dried and powdered stems of M. tenacissima (5 kg) were extracted three times with 95% ethanol under reflux for 2 h each time. The solvent in the extract was evaporated resulting in the ethanol extract residue which was extracted with chloroform under reflux, and a yellow residue was obtained on evaporation of the chloroform. The residue was subjected to column chromatography (silica gel, gradient CHCl3 /MeOH (50:1, 20:1, 10:1, 5:1, v/v); thus eight main fractions were obtained. A sample of fraction 8 (CHCl3 /MeOH 5:1, v/v) was dissolved in methanol, membranefiltered (0.45 ␮m), and analyzed by HPLC/ESI-MSn . The results obtained for this chloroform extract were reported and discussed in the following sections. 2.3. Analytical and preparative high-performance liquid chromatography An Agilent 1100 analytical HPLC system with a G1312 Binpump, G1314A variable-wavelength detector (VWD), model 7725 injector fitted with a 20 ␮L sample loop, along with an Agilent ChemStation data system, was used. A reversed-phase Agilent Extend C18 column (4.6 mm × 150 mm, 3.5 ␮m) was used for separation; the column was maintained at room temperature. Chromatography was carried out in gradient mode with methanol and water; the percentage of methanol was changed linearly as follows: 0 min, 60%; 10 min, 65%; 30 min, 75%; 40 min, 80%; 42 min, 60%. The flow rate was 0.8 mL min−1 , and the detection wavelength was 220 nm. Preparative HPLC system was performed using Waters 600 Separations Module equipped with a Waters model 2996 diode array detector and Waters Empower System (Waters Co., Milford, MA, USA). A reversed-phase Shim-pack PRC ODS (20 mm × 250 mm,

2.4. Mass spectrometry

2.5. NMR spectroscopy 1D and 2D NMR spectra were measured on a Bruker Advance DMX 500 spectrometer operating at 500MHz for 1 H and 125MHz for 13 C, using pyridine (C5 D5 N) as solvent and TMS as internal standard. The 13 C NMR spectral data of compounds 11, 14 and 18 are supported as the supplementary information.

154

J. Chen et al. / Talanta 77 (2008) 152–159

Fig. 2. Structures assigned to polyoxypregnane glycosides in the stems of Marsdenia tenacissima.

3. Results and discussion

3.1. Investigation of authentic compounds 6, 7, 15, 16 and 17

The HPLC/UV (220 nm) detection and total ion chromatography (TIC) of the polyoxypregnane glycosides in the extracts of stems of M. tenacissima are shown in Fig. 1. The target polyoxypregnane glycosides recorded at retention times were designed as 1 through 18. All precursor ions were observed in the positive and negative ion mode spectra. Both of them had similar fragment pathways, but in the positive mode, the fragment information of sugar chain is more abundant than that in the negative mode. So singly charged sodium adducts of the molecules [M + Na]+ were studied in the following parts. Fig. 2 exhibits the structures of the investigated molecules and high-resolution FTICR-MS data are listed in Table 1.

7, 6, 15, 16 and 17 could unambiguously be identified as Marsdenoside K, Tencissoside A, B, C and D, respectively, based on the comparison of the retention times with those of authentic reference substances and on the fragmentation behaviors observed in the MSn experiments [4,6]. Generally, these five compounds gave similar fragmentations in MSn experiments. Thus, the fragmentation patterns of 15 were discussed in detail below for elucidation of the molecular structure. The mass spectra of 15 and proposed fragmentation pathways are shown in Fig. 3 and Scheme 1, respectively. ESI-MS analysis of 15 gave the [M + Na]+ ion at m/z 1017 which was further selected for MS/MS experiment. As demonstrated in Fig. 3, product ions at

J. Chen et al. / Talanta 77 (2008) 152–159

155

Table 1 Accurate masses and assigned elemental compositions of eighteen polyoxypregane glycosides Compounds 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Composition +

C43 H68 O18 Na C46 H72 O18 Na+ C46 H72 O18 Na+ C45 H70 O19 Na+ C46 H72 O18 Na+ C48 H74 O19 Na+ C50 H72 O19 Na+ C46 H72 O18 Na+ C51 H80 O20 Na+ C51 H78 O19 Na+ C48 H74 O19 Na+ C51 H80 O19 Na+ C53 H76 O19 Na+ C50 H72 O19 Na+ C51 H78 O19 Na+ C53 H76 O19 Na+ C51 H80 O19 Na+ C53 H78 O19 Na+

Measurement (m/z)

Theoretical (m/z)

Error (ppm)

895.4318 935.4598 935.4613 937.4388 935.4655 977.4735 999.4563 935.4632 1035.5105 1017.5009 977.4738 1019.5174 1039.4873 999.4574 1017.5063 1039.4825 1019.5222 1041.5011

895.4298 935.4611 935.4611 937.4404 935.4611 977.4717 999.4560 935.4611 1035.5135 1017.5030 977.4717 1019.5186 1039.4873 999.4560 1017.5030 1039.4873 1019.5186 1041.5030

2.2 −1.4 0.2 −1.7 4.7 1.8 0.3 2.4 −2.9 −2.1 2.1 −1.2 0 1.4 3.2 −4.6 3.5 −1.8

m/z 917 (A or B, shown in Scheme 1) and 817 (C) were generated by neutral loss of 100 Da and consecutive elimination of 100 Da from the precursor ion at m/z 1017, reasonably assigned as two (E)-2methylbut-2-enoic acids from C-11 and C-12. In the MS3 spectrum of ion at m/z 917 (A or B), elimination of 100 Da produced the ion at m/z 817 (C). Product ions at m/z 759 (D) and 655 (F) corresponded to the loss of C3 H6 O (58 Da) by cleavage within the oleandropyranose residue and terminal glucopyranose residue (162 Da) from the ion at m/z 817. Fragment ion at m/z 755 (Eb ), by loss of 162 Da from ion at m/z 917, suggested the terminal glucopyranose residue. The whole sugar chain fragment ion at m/z 489 (G) resulted from the cleavage of the glycosidic bond between aglycone and the reducing end of the sugar chain, without containing the glycosidic oxygen atom. Fragment ion at m/z 345 (H) was achieved with the consecutive loss of reducing end oleandropyranose residue (144 Da) from ion at m/z 489 (G). Thus, with the analysis of the disaccharide fragment ion at m/z 345 (H) and terminal glucopyranose residue and the mass of second monosaccharide calculated as 160 Da, it could be inferred that it is consistent with allopyranose residue. Based on the above analyses of the five authentic compounds, we could conclude that their fragmentation patterns were similar related to their structural characteristics. First, it was significant to stress that all the authentic substances only eliminated R2 OH and R3 OH on the C-11 and C-12 positions in the MS2 experiments. Second, they had similar MS3 fragmentation behaviors. The key fragment ions (m/z 489 (G) and 345 (H)) and neutral loss of 162 Da provided the mass information of sugar chain. Hence, according to the literatures and these fragmentation patterns, the structures of unknown polyoxypregnane glycosides could be proposed. 3.2. Identification of unknown compounds 1, 4, 9, 12, and 18 We also examined unknown polyoxypregnane glycosides in the extracts of stems of M. tenacissima. In the MS3 spectra of compounds 1, 4, 9, 12 and 18, similar fragmentation pathways about sugar chain were observed to those of authentic compounds. Hence, the glycosyl ions of 1, 4, 9, 12 and 18 were attributed to the Glc-Allo-Ole trisaccharide chain (S, as shown in Fig. 2). The mass spectra of 1, 4, and 18 contained significant [M + Na]+ ions at m/z 895, 937 and 1041, respectively. For compound 1, neutral loss of 60 Da (acetic acid) and consecutive elimination of 18 Da (a molecular of water) were obtained from precursor ion at m/z 895 to produce fragment ions at m/z 835 and 817. These fragmentations indicated the presence of acetyl (Ac) and hydrogen atom on C-11 and C-12 positions. However, the acetyl located at C-11 or C-12

could not be confirmed by MS. As for 4, loss of one and two 60 Da fragments produced the fragment ions at m/z 877 and 817, respectively, suggesting two acetyl substituents (Ac) on C-11 and C-12 positions. For 18, the fragment ions at m/z 939, 919 and 817 were generated by neutral loss of 102 Da, 122 Da and sequential elimination of 102 Da from the precursor ion at m/z 1041, predicated the existing of benzoyl (Bz) and 2-methylbutyryl (Bu) groups. Based on fragmentation behaviors and previous studies [1,2,6,7], 1 was proposed to be 3-O-␤-d-glucopyranosyl-(1 → 4)-6-deoxy3-O-methyl-␤-d-allopyranosyl-(1 → 4)-␤-d-oleandro-pyranosyl11␣-O-acetyltenacigenin B or 3-O-␤-d-glucopyranosyl-(1 → 4)-6deoxy-3-O-methyl-␤-d-allopyranosyl-(1 → 4)-␤-d-oleandropyranosyl-12␤-O-acetyltenacigenin B. 4 was tentatively proposed as 3-O-␤-d-glucopyranosyl-(1 → 4)-6-deoxy-3-O-methyl-␤-dallopyranosyl-(1 → 4)-␤-d-oleandropyranosyl-11␣, 12␤-di-Oacetyltenacigenin B. Moreover, compound 18 has been purified by preparative HPLC and performed on 1D, 2D NMR. Based on the NMR data and previous studies[2,6], 18 was identified to be 3-O-␤-d-glucopyranosyl-(1 → 4)-6-deoxy-3-O-methyl-␤-dallopyranosyl-(1 → 4)-␤-d-oleandropyranosyl-11␣-O-2-methylbutyryl-12␤-O-benzoyltenacigenin B, which has been reported as Tenacissoside E. The [M + Na]+ ion of 9 was concluded to be m/z 1035 (see Fig. 3). In the case of 9 neutral loss of 18 Da, 100 Da and consecutive loss of 100 Da and 18 Da from the [M + Na]+ ion in the MS/MS spectrum, attributed to elimination of two molecules of water and two (E)-2-methylbut-2-enoic acid units. So it is hereby proven that there are two (E)-2-methylbut-2-enoyl (Tig) substituents on the C-11 and C-12 positions of 9. Besides, a difference between 9 and the authentic compound 15 was observed in the MS2 spectra. Compound 9 not only eliminated R2 OH and R3 OH, also loss two molecules of water. On the other hand, the 16 Da mass difference between [M + Na]+ of 9 and 15, indicated that the elemental composition of the aglycone of 9 contained one more oxygen atom than that of 15. According to the literature [5,16], the polyoxypregnane aglycone core that the 8␤-O-14␤ epoxy ring broke to form two hydroxyl groups on the C-8 and C-14 positions has been reported. So 9 was tentatively proposed as 3-O-␤-d-glucopyranosyl-(1 → 4)-6-deoxy-3-Omethyl-␤-d-allopyranosyl-(1 → 4)-␤-d-oleandropyranosyl-11␣12␤-O-di-tigloyl-5,6-dihydrosarcogenin. In the MS/MS spectrum of [M + Na]+ at m/z 1019 for 12, fragment ions at m/z 919 and 819 were derived from neutral loss of 100 Da and sequential elimination of 100 Da, corresponding to the fact that 12 consisted of two (E)-2-methylbut-2-enoyl

156

J. Chen et al. / Talanta 77 (2008) 152–159

Fig. 3. Positive ions ESI-MSn spectra: MS spectra of 15 (I) and 9 (IV); MS2 spectra of 15 (II) and 9 (V); MS3 of spectra 15 (III) and 9 (VI).

(Tig) substituents on the C-11 and C-12 positions of aglycone. The product ion at m/z 801 was generated by loss of 218 Da ([M + Na-100 Da-100 Da-18 Da]+ ) from precursor ion at m/z 1019. The mass value of 12 was 2 Da higher than that of 15. In a pre-

vious study [17], the polyoxypregnane glycoside that 8␤-O-14␤ epoxy ring on C-8 and C-14 was broken to form hydroxyl group on C-14 from aglycone has been published. Consequently 12 was tentatively proposed as 3-O-␤-d-glucopyranosyl-(1 → 4)-6-

J. Chen et al. / Talanta 77 (2008) 152–159

157

Scheme 1. Fragmentation pathways proposed for [M + Na]+ of 6, 7, 15, 16 and 17. a : the ion E of 7; b : the ions E of 6, 15, 16 and 17.

deoxy-3-O-methyl-␤-d-allopyranosyl-(1 → 4)-␤-d-oleandropyranosyl-11␣,12␤-O-di-tigloyl-5, 6-dihydrodrevogenin P. 3.3. Differentiation of isomers 2, 3, 5, 8, 10, 11, 13 and 14 by breakdown curves The [M + Na]+ ions for unknown compounds 10, 11, 13 and 14 were observed at m/z 1017, 977, 1039 and 999, respectively. It is significant to note that they had similar molecular weights and MSn spectra to authentic compounds 15, 6, 16 and 7. Thus, 10, 11, 13 and 14 were proposed to be the isomers of 15, 6, 16 and 7, respectively. Moreover, 10 had the two substituents (Tig) on C-11 and C-12 positions as 15, which may be a consequence from the stereochemical difference of some monosaccharide residue of 15. Compounds 11 and 14 have been purified by preparative HPLC and performed on 1D, 2D NMR. Based on the NMR data, it was confirmed that 11 was 3-O-␤-d-glucopyranosyl-(1 → 4)-6-deoxy-3-O-methyl-␤d-allopyranosyl-(1 → 4)-␤-d-oleandropyranosyl-11␣-acetyl, 12␤O-tigloyltenacigenin B and 14 was identified as 3-O-␤-d-

glucopyranosyl-(1 → 4)-6-deoxy-3-O-methyl-␤-d-allopyranosyl(1 → 4)-␤-d-oleandropyranosyl-11␣-O-benzoyl-12␤-O-acetyltenacigenin B. To the best of our knowledge, compound 11 and 14 were reported here for the first time, as the positional isomers of authentic compounds 6 and 7, respectively. Evaluation of breakdown curves can provide information on fragmentation mechanisms such as distinguishing between competitive and consecutive fragmentation pathways, the stability of product ions and the identification of isomers and tautomers [18–20]. In order to distinguish these isomers, breakdown curves were generated by relative abundance of selected fragment ions versus collision energy. 11 and 14 were identified as the positional isomers of 6 and 7, respectively. Their breakdown curves were displayed in Fig. 4. The fragment ion at m/z 917 by loss of R3 OH from precursor [M + Na]+ ion at 977 was the base peak in three MS2 fragment ions of 6. While product ion at m/z 877 generated by loss of R3 OH from 11 produced the highest relative abundance. The positional isomers 7 and 14 also represented similar phenomena. Therefore, it is proposed

158

J. Chen et al. / Talanta 77 (2008) 152–159

Fig. 4. Breakdown curves of MS2 ions of positional isomers 6, 11, 7 and 14.

Fig. 5. Breakdown curves of MS2 ions of 10, 15, 16 and 13.

J. Chen et al. / Talanta 77 (2008) 152–159

that product ions by loss of R3 OH had the highest relative abundance and these two pairs of positional isomers were differentiated conveniently by breakdown curves. However, the breakdown curves (illustrated in Fig. 5) of 10 and 15 are different from the positional isomers discussed above. Both 10 and 15 had two fragment ions in MS2 experiments. The ion at m/z 817 of 10 by loss of R2 OH and R3 OH had higher relative abundance than the ion at m/z 917; while for 15, the circumstance was completely opposite. Thus, it was palpable that the breakdown curves were advantageous in distinguishing the stereochemical isomers 10 and 15. The same phenomena were observed in isomers 13 and 16. Accordingly, 13 might be proposed as the stereochemical isomers of 16, with stereochemical difference in some monosaccharide residue. Unknown compounds 2, 3, 5 and 8 not only had the same [M + Na]+ ions at m/z 935, but also had similar MSn spectra. Hence, these four compounds were proposed to be isomers. In the MS2 spectra of ions at m/z 935, key fragment ions at m/z 835([M + Na100]+ ) and 817 ([M + Na-100-18]+ ) demonstrated that there were (E)-2-methylbut-2-enoyl (Tig) and hydrogen groups on the aglycone. The MS3 spectra were similar to authentic compounds. The breakdown curves of them were also investigated with little differentiation observed. It might be explained that these four compounds only have one substituent on C-11 or C-12 position, while the other four pairs mentioned above have two substituents on C-11 and C-12. However, the exact structures of these four compounds and differentiation of isomers need further studies. 4. Conclusions The results illustrated the potential advantage of the method of HPLC/(+)ESI-MSn for characterization of polyoxypregnane glycosides including the nature of the polyoxypregnane core, the kind of substituents and the types of sugar residues. Based on the fragmentation behaviors, eighteen polyoxypregnane glycosides are analyzed. Five compounds are identified by the fragmentations and authentic substances, eleven compounds are tentatively proposed and two compounds are established as novel compounds.

159

Moreover, breakdown curves assist in distinguishing the isomers, especially when the isomers yield the same product ions. Two pairs of positional isomers and two pairs of stereochemical isomers in some monosaccharide residue ascertained from this study could be promptly differentiated. Acknowledgements We gratefully acknowledge the Zhejiang Provincial Science and Technology Council (No. 2005F13028) and the Zhejiang Provincial Natural Science Foundation (No. Z206150) for financial support. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.talanta.2008.05.054. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

J. Deng, Z. Liao, D. Chen, Phytochemistry 66 (2005) 1040. J. Deng, F. Shen, D. Chen, J. Chromatogr. A 1116 (2006) 83. Q. Li, X. Wang, L. Ding, C. Zhang, Chin. Chem. Lett. 18 (2007) 831. J. Deng, Z. Liao, D. Chen, Helvetica Chim. Acta 88 (2005) 2675. S. Qiu, S. Luo, L. Lin, G.A. Cordell, Phytochemisrry 41 (1996) 1385. S. Miyakawa, K. Yamaura, K. Hayashi, K. Kaneko, H. Mitsuhashi, Phytochemistry 12 (1986) 2861. S. Luo, L. Lin, G.A. Cordell, L. Xue, M.E. Johnson, Phytochemistry 34 (1993) 1615. J. Chen, Z. Zhang, J. Zhou, Acta Bot. Yunnan 21 (1999) 369. Q. Wu, M. Wang, J.E. Simon, J. Chromatogr. A 1016 (2003) 195. L. Ding, X. Biao Luo, F. Tang, J. Yuan, M. Guo, S. Yao, Talanta 71 (2007) 668. L. Tong, Y. Wang, J. Xiong, Y. Cui, Y. Zhou, L. Yi, Talanta 76 (2008) 80. X. Huang, F. Song, Z. Liu, S. Liu, J. Mass Spectrom. 42 (2007) 1148. Y. Tai, X. Cao, X. Li, Y. Pan, Anal. Chim. Acta 572 (2006) 230. X. Cao, Y. Tai, X. Li, Y. Ye, Y. Pan, Rapid Commun. Mass Spectrom. 20 (2006) 411. J. Liu, B. Chen, S. Yao, Talanta 71 (2007) 668. F.T. Halaweish, E. Huntimer, A.T. Khalil, Phytochem. Anal. 15 (2004) 189. N.P. Sahua, N. Panda, N.B. Mandal, S. Banerjee, K. Koike, T. Nikaido, Phytochemistry 61 (2002) 383. R.D. Hiserodt, B.M. Pope, M. Cossette, M.L. Dewis, J. Am. Soc. Mass Spectrom. 15 (2004) 1462. E.R. Filho, A.M.P. Almeida, M. Tabak, J. Mass Spectrom. 38 (2003) 540. J. Makowiecki, A. Tolonen, J. Uusitalo, J. Jalonen, Rapid Commun. Mass Spectrom. 15 (2001) 1506.