Screening and Identification of Glycosides in Biological Samples

(2) The EGNLS simultaneously measures the screened compounds' OCE, which not only are essential parameters for further LC/MS/MS analysis but also carr...
0 downloads 0 Views 215KB Size
Download PDF
Anal. Chem. 2004, 76, 2239-2247

Screening and Identification of Glycosides in Biological Samples Using Energy-Gradient Neutral Loss Scan and Liquid Chromatography Tandem Mass Spectrometry Jun Qu, Quonglin Liang, Guoan Luo,* and Yiming Wang

Department of Chemistry, School of Life Science and Technology, Tsinghua University, Beijing, 100084, P. R. China

A rapid, selective, and reliable strategy has been developed for the screening and identification of glycosides in biological samples: a crude extract was directly infused to a triple-quadrupole MS/MS, and major glycosides were screened out with high confidence by an energygradient neutral loss scan (EGNLS) for the loss of sugar(s); then these glycosides were further identified with LC/MS/MS. The proposed EGNLS method was established and optimized with 16 representative glycosides (including ginsenosides and the glycosides of flavones, anthraquinones, and terpenoids). The EGNLS method has two major advantages over the conventional fixed-energy neutral loss scan: (1) The latter is liable to ‘“omit” some target compounds due to the usual mismatch between the preset collision energy and interested compounds’ optimal collision energy (OCE), while EGNLS solves this problem by scanning over an energy range. (2) The EGNLS simultaneously measures the screened compounds’ OCE, which not only are essential parameters for further LC/MS/MS analysis but also carry some structural information, as proved by this study. This strategy has been successfully demonstrated with the analysis of glycosides in Scutellaria viscidula Bge and transformed Panax hairy roots (the glycoside constitutions of both had not been studied before): without laborious separation processes; comprehensive glycoside information on those two plants was obtained by a rapid and simple procedure. This strategy is valuable for the study of glycosides in complex samples. Glycosides are one of the most widespread classes of biological compounds and of great importance in such fields as phytochemistry, pharmacology, biochemistry, and nutriology. These compounds are metabolized from flavonoids, anthraquinones, terpenes, etc., and most of them are biologically active.1-3 Numerous studies have been conducted in the past to analyze certain glycosides in biological samples. The methodologies include * Corresponding author. E-mail: [email protected]. Fax: 8610-62781688. Tel: 86-10-62781688. (1) Pratt, D. E.; Hudson, B. J. F. In Food Antioxidants; Hudson, B. J. F., Ed.; Elsevier Applied Science: New York, 1990; pp 173-179. (2) Frankel, E. N.; Kanner, J.; Genman, J. B.; Parks, E.; Kinsella, J. E. Lancet 1993, 341, 454-461. (3) Kim, H. S. Pharmacol. Biochem. Behav. 1996, 53, 185-190. 10.1021/ac030413t CCC: $27.50 Published on Web 03/13/2004

© 2004 American Chemical Society

colorimetric methods,4 thin-layer chromatography,5 high-performance liquid chromatography (HPLC),6-9 capillary electrophoresis,10,11 gas chromatography-mass spectrometry (GC/MS),12,13 MALDI-TOF MS,14 LC/MS/MS,15-20 and LC/NMR/MS.21 However, most of those methods could not be employed to screen glycosides (both known and unknown) in biological samples. Since the constitutions of glycosides in biological samples are often complicated, a method that is capable of screening out glycosides in complex matrixes is valuable because of the clues it can provide. We reported in a recent paper22 a rapid search and identification of glucuronides and their aglycons in a plant extract: the screening of glucuronides by a MS/MS neutral loss scan for the loss of a glucosiduronic acid, followed by the identification of the glucuronides and their aglycons by LC/MS/MS. In the present study, we proposed a universal strategy for the rapid and reliable screening and identification of glycosides in biological matrixes: the interested glycosides were screened out and the optimal (4) Tatsuma, T.; Komori, K.; Yeoh, H. H.; Oyama, N. Anal. Chim. Acta 2000, 408, 233-240. (5) Tanaka, H.; Putalun, W.; Tsuzaki, C.; Shoyama, Y. FEBS Lett. 1997, 404, 279-282. (6) Rehwald, A.; Meier, B.; Sticher, O. J. Chromatogr., A 1994, 677, 25-33. (7) Deliorman, D.; Calis, I.; Ergun, F.; Tamer, U. J. Liq. Chromatogr. Relat. Technol. 1999, 22, 3101-3114. (8) Merken, H. M.; Beecher, G. R. J. Agric. Food Chem. 2000, 48, 577-599. (9) Derksen, G. C. H.; van Beek, T. A.; de Groot, A. E.; Capelle, A. J. Chromatogr., A 1998, 816, 277-281. (10) Tomasbarberan, F. A. Phytochem. Anal. 1995, 6, 177-192. (11) Frazier, R. A.; Ames, J. M.; Nursten, H. E. Electrophoresis 1999, 20, 31563180. (12) Mazur, W.; Fotsis, T.; Wahala, K.; Ojala, S.; Salakka, A.; Adlercreutz, H. Anal. Biochem. 1996, 233, 169-180. (13) Chassagne, D.; Crouzet, J.; Bayonove, C. L.; Baumes, R. L. J. Agric. Food Chem. 1998, 46, 4352-4357. (14) Wang, J.; Sporns, P. J. Agric. Food Chem. 2000, 48, 1657-1662. (15) Vanbreemen, R. B.; Huang, C. R.; Lu, Z. Z.; Rimando, A.; Fong, H. H. S.; Fitzloff, J. F. Anal. Chem. 1995, 67, 3985-3989. (16) Wang, X. M.; Sakuma, T.; Asafu-Adjaye, E.; Shiu, G. K. Anal. Chem. 1999, 71, 1579-1584. (17) Nielsen, S. E.; Freese, R.; Cornett, C.; Dragsted, L. O. Anal. Chem. 2000, 72, 1503-1509. (18) Kite, G. C.; Howes, M. J. R.; Leon, C. J.; Simmonds, M. S. J. Rapid Commun. Mass Spectrom. 2003, 17, 238-244. (19) Wang, X. M.; Plomley, J. B.; Newman, R. A.; Cisneros, A. Anal. Chem. 2000, 72, 3547-3552. (20) Mellon, F. A.; Bennett, R. N.; Holst, B.; Williamson, G. Anal. Biochem. 2002, 306, 83-91. (21) Lommen, A.; Godejohann, M.; Venema, D. P.; Hollman, P. C. H.; Spraul, M. Anal. Chem. 2000, 72, 1793-1797. (22) Qu, J.; Wang, Y.; Luo, G.; Wu, Z. P. J. Chromatogr., A 2001, 928, 155-162.

Analytical Chemistry, Vol. 76, No. 8, April 15, 2004 2239

collision energies (OCE) for those compounds were obtained by an energy-gradient neutral loss scan (EGNLS) with MS/MS; then the glycosides were further identified with database searching and further confirmed by LC/MS/MS. Neutral loss scan using MS/MS is a useful approach for bioanalysis, and there have been some reports on the applications of this technique.23-25 So far a neutral loss scan must be employed under a preset collision energy. However, since the OCE for a loss of the same moiety are often diverse among different types of compounds, the compounds whose OCE are quite different from the preset collision energy will not be efficiently fragmented into the expected products and thus may give poor signal or no signal at all. As a consequence, using any fixed collision energy for neutral loss scan might result in the “omission” of some compounds, low sensitivity for some others, or both. This is one of the major reasons why presently the neutral loss scan is not popularly used, despite the convenience it can provide.26 In this report, we developed an EGNLS method, which not only avoids the omission of interested compounds caused by collision energy mismatch but also can measure the OCE of those compounds, providing additional structural information on the screened compounds. The method was developed and optimized using 16 representative O-glycosides. To examine the practicability of the approach, we assayed two plants: Scutellaria viscidula Bge, and a biologically produced Panax ginseng hairy roots (The glycoside constituents of both plants had not been sufficiently studied before this work.). Using the rapid and simple procedure established in this study, the glycosides of interest in those crude extracts were readily screened out and identified, and it was found that the absolute and relative contents of some glycosides in the two plants were significantly different from those in their congeneric species. EXPERIMENTAL SECTION Materials and Reagents. Rhapontin, proscillaridin, quercetin3-O-glucoside, salicin, aesculin, and rutin were purchased from Sigma (St. Louis, MO). Luteolin- 7-O-glucoside was from Beijing Institute of Medicinal Plant Research (Beijing, China), and the emodin-8-O-glucoside, phloretin-2-O-glucoside and gossypin were from Kunming Institute of Phytochemistry (Yunan, China). All other authenticated glycoside standards were from the Chinese State Drug Administration (SDA, Beijing, China). HPLC grade acetonitrile and methanol were from Tedia Inc. (Fairfield, OH). The dry roots of Scutellaria baicalensis Georgi (cultivated in Henan Province, China), Scutellaria amoena C. H. Wrignt (cultivated in Kuichow Province, China), S. viscidula Bge (wild type, picked up in northeastern China), P. ginsen C. A. Meyer (4 year old, cultivated in northern China), Panax quinquefolius L. (4 year old, cultivated in the United States), and the transformed Panax ginseng hairy roots (cultivated by a company in Shanghai, China) were provided and authenticated by the Pharmaceutical and Biological Products Control Laboratory of SDA. (23) Yates, J. R.; Eng, J. K.; Mccormack, A. L.; Schieltz, D. Anal. Chem. 1995, 67, 1426-1436. (24) Jackson, P. J.; Brownsill, R. D.; Taylor, A. R.; Walther, B. J. Mass Spectrom. 1995, 30, 446-451. (25) Schlosser, A.; Pipkorn, R.; Bossemeyer, D.; Lehmann, W. D. Anal. Chem. 2001, 73, 170-176. (26) Arnott, D. In Protome Research: Mass Spectrometry; James, P., Ed.; Springer: New York; 2001; pp 23-34.

2240

Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

Sample Preparation. All dry plant samples were separately ground to fine powder by a pulverizer. As for the Scutellaria plants, a 0.2-g amount of powder was placed in a 25-mL capped conical flask and 5 mL of methanol was added. The flask was closed, and the extraction was carried out for 10 min in an ultrasonic washer. The temperature ranged from 35 to 45 °C after extraction. Then the extract was centrifuged for another 10 min at 8000 rpm, and the supernatant was transferred to a beaker. The above extraction procedure was repeated for 4 times. All the supernates were combined. The extraction procedures for the Panax plants were the same as those for the Scutellaria plants, except that the extracts were diluted 4-fold with 10 mM ammonia acetate in 20: 80 water/methanol before analysis. Instrumentation. An Applied Biosystems (Toronto, Canada) API 3000 triple-quadrupole tandem mass spectrometer, equipped with a Turbo Ionspray interface and an Agilent 1100 binary HPLC system, was used for EGNLS and LC/MS/MS analysis. Energy-Gradient Neutral Loss Scan. Standard mixtures or plant extracts were directly infused to the MS/MS at a rate of 0.3 mL/h. To minimize dead volume, samples were introduced to the spray needle via a 150 × 0.075 mm i.d. fused-silica capillary. The MS/MS was operated under negative ion mode, and neutral loss scan for the loss of proper sugar(s) was performed over the m/z range of 200-1200 with a step size of 0.3 amu and a dwell time of 2 ms. The mass spectrometer was programmed to perform neutral loss scans continuously over a collision energy range from 0 to 100 eV and the total scan time was ∼23 min/sample. This was realized by increasing the Quad 2 rod offset (RO2) voltage linearly from 11 to 111 V with an increment of 0.5 V for each scan, while the Quad 3 rod offset (RO3) and Stubbies 3 (ST3) were set dependent to RO2 with the intercepts respectively of 2 and 20 V. The Inter Quad 1 lens (IQ1) was set at 11 V and the voltage difference of (RO2-IQ1) was defined as the collision energy. A compromised orifice potential at 80 V was used, and the ion spray voltage and ring focus voltage were set at -4000 and -160 V, respectively. The pressure of target gas (argon) for collisionally activated dissociation (CAD) was 4.8 mTorr. The obtained intensity (Y axis) versus time (X axis) spectrum was transformed to intensity (Y axis) versus collision energy (X axis) spectrum by calculating the collision energy values on the basis of the duration of each scan and the step size for the increment of the collision energy. The screened glycosides were displayed on the neutral loss spectra extracted at appropriate collision energies. All of the extracting ion currents (XIC) of screened glycosides were investigated, and the OCE of each ion was determined as the collision energy corresponding to the peak summit on that ion’s intensity versus energy XIC. Database searching for the possible ID of the screened glycosides were performed using a commercial natural product database CD with reference to their molecular weights and plant sources. Further Identification by LC/MS/MS. Separations of all plant extracts were carried out on a Dikama Diamonsil C18 (particle size 5 µm, 150 × 2.1 mm) column maintained at 40 °C. The mobile phase for the assay of the Scutellaria plants consisted of two solvents: water and acetonitrile. The proportion of acetonitrile was 15% in the first 5 min and then linearly increased to 40% in 25 min, and the 40% acetonitrile was maintained until the end of analysis. The gradient conditions for the analysis of ginsenosides

in Panax plant extracts were as follows: solvent A was 12 mM ammonium acetate in water (buffered to pH 7.5 by aqueous ammonia). and solvent B was 12 mM ammonium acetate in 20:80 water/acetonitrile. The elution was 7 min isocratic 10% solvent B, followed by linear gradients from 10 to 33% B (8-35 min), 33 to 39% B (35-48 min), and 39 to 85% B (48-65 min). The flow rate was 0.2 mL/min, and the injection volume was 20 µL. For MS/MS, multiple reactions monitoring (MRM) was used for the detection of the glycosides found by EGNLS. The transitions and their corresponding collision energies for MRM of screened glycosides were obtained from the previous EGNLS step. The dwell time for each transition was 200 ms, and the pause time for the change of parameters was 6 ms. RESULTS AND DISCUSSION Establishment and Optimization of the EGNLS Method. (a) Investigation of Sugar Cleavage Efficiencies of Glycosides. The CAD sugar cleavage efficiencies were investigated with 16 glycosides (shown in Figure 1, representing some of the most common classes of glycosides such as glycosides of flavonoids, flavonoid derivatives, anthraquinones, terpenes, iridoids, coumarins, and cardiac glycosides) under both positive and negative ion modes. To study the influence of the types of aglycons on glycoside sugar cleavage efficiencies, we only focused on the cleavages of two common sugar linkages: the linkages with glucose (loss of 162) and rhamnose (loss of 146). In some other experiments, we found that glycosides with other sugars (such as galactose, arabinose, etc.) exhibited cleavage efficiencies similar to those with glucose and rhamnose (data not shown). All glycosides studied here exhibited neutral losses of sugar(s) in both ionization modes. This opens the possibility that glycosides can be detected with neutral loss scan for the loss of sugar(s). Negative ion mode was found to be more suitable for neutral loss scan of glycosides than positive mode: first, the fragments with a negatively charged ether oxygen on the aglycon, which result from the loss of sugar(s), are quite stable in negative ion mode; this ensured the sensitivity of neutral loss scan. Second, sodium or potassium adduct ions ([M + Na]+ or [M + K]+) of glycosides were often observed in positive ion mode, and in some cases, those ions were much more intensive than the [M + H]+; the adduct ions would complicate the neutral loss spectra; however, the adduct ions were relatively rare in negative mode. (b) Reproducibility and Specificity of OCE Values. In our experiments, we found the OCE values for the loss of sugar(s) from glycosides were reproducible. We investigated the reproducibility of sugar cleavage OCE for the 16 glycosides, at different concentrations and in both neat solvent solutions (in methanol) and spiked human serum samples (representing a complex matrix). Representative results are presented in Table 1. The OCE values for all the glycosides studied here were found to be quite reproducible and independent of analyte concentrations and sample matrixes. The authors believe that it is due to the constancy of the CAD reaction’s activation energy (Ea) and relevant thermodynamic parameters. Besides the reproducibility of OCE values, we also found that the OCE values were specific to the aromaticities of glycosides. When a group of glycosides has similar aromaticities, their OCE values are dependent on the m/z of parent ions, as suggested by a previous study;27 however when two groups of glycosides with different aromaticities are

compared, the aromaticities appear to play a more important role in determining glycosides’ OCE (Table 1). For example, for the same CAD transition [M - H]- f [M - H - Glc]-, the OCE of a smaller aliphatic glycoside is higher than that of a bigger aromatic glycoside (Table 1, icariin (V) vs proscillaridin (XII)). For the glycosides investigated in this study, the OCE values for single sugar loss from the aromatic glycosides ranged from 17 to 28 eV, while those for the aliphatic glycosides ranged from 46 to 68 eV. The OCE values for more than one sugar loss from glycosides were found higher than those for the single sugar loss (e.g., rutin in Table 1; other data not shown). The demonstration of both the reproducibility and the specificity of OCE in this study enables the use of OCE as an index to help to identify compounds and, thus, the establishment of the EGNLS method. (c) Energy-Gradient Neutral Loss Scan Method. For a conventional neutral loss scan analysis of glycosides, it is problematic that the OCE for the sugar loss are diverse among different types of glycosides: a collision energy value that is optimal for one type of glycosides may not be optimal for another. As a consequence, any preset collision energy applied to the neutral loss scan for sugar loss may result in the failure to detect some glycosides. This is especially troublesome when the composition of a sample is uncertain, because it is difficult to accurately predict the types of glycosides in the sample. In this report, we proposed an EGNLS method, which not only solves the above problem but also provides additional structural information on the screened compounds. Instead of using a fixed collision energy, we applied a linear collision energy gradient when performing neutral loss scan with MS/MS. Similar to a LC/MS/ MS chromatogram, an EGNLS spectra can be used not only to obtain the neutral loss spectra at any collision energy value but also to extract the intensity versus energy extracting ion currents of each screened glycoside. Using the neutral loss spectra at different collision energies, glycosides in a complex sample can be screened out thoroughly regardless of their OCE; with the intensity versus energy XIC of the screened glycosides, the OCE of each screened glycosides can be obtained. The obtained OCE values have two important usages: first, they are essential parameters for further LC/MS/MS analysis of the screened glycosides; second, some structural information on the aglycon (i.e., its aromaticity) can be obtained from the glycoside’s OCE value. The EGNLS spectra for an assay of the standard glucosides mixture are shown in Figure 2. Glycosides with different types of aglycons can be resolved by this EGNLS method (Figure 2A-D). These results also suggest that the problem of omitting some glycosides encountered by fixed-energy neutral loss scan could be avoided with EGNLS. For example, the neutral loss spectra extracted at 20 eV showed all the aromatic glycosides but most aliphatic glycosides were not shown (Figure 2D), whereas at 65 eV only the aliphatic glycosides were found (Figure 2E). This implies that while fixed-energy neutral loss scan using either of the two collision energies would result in omission of some glycosides and severe sensitivity loss for some others, all the glycosides with a glucose loss could be screened out by a EGNLS scan. (27) Keski-Hynnila, H.; Luukkanen, L.; Taskinen, J.; Kostiainen, R. J. Am. Soc. Mass Spectrom. 1991, 10, 537-545.

Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

2241

Figure 1. Structures of the 16 representative glycosides (molecular weight): I, aesculin (340); II, salicin (286); III, emodin-8-O-glucoside (432); IV, gossypin (480); V, icariin (676); VI, luteolin-7-O-glucoside (448); VII, gardenoside (404); VIII, quercetin-3-O-glucoside (464); IX, phloretin2-O-glucoside (436); X, rhapontin (420); XI, rutin (610.5); XII, proscillaridin (530.5); XIII, ginsenoside Rb1 (1108); XIV, Re (946); XV, Rg3 (784); XVI, saikosaponin A (780). *Glc, glucose; Rha, rhamnose; Gal, galactose. 2242

Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

Table 1. Investigation of the Reproducibility and Specificity of Optimal Collision Energy (OCE) Values with Some Glycosidesa spiked to a human serumb (n ) 6)

in standard solutions (n ) 6)

glycoside aromatic salicin gossypin icariin rutin aliphatic gardenoside proscillaridin ginsenosides Re saikosaponin A

CAD transition (m/z)

OCE (eV)

intraday precision (CV %)

interday precision (CV %)

OCEc (eV)

intraday precision (CV %)

interday precision (CV %)

285 f 123 479 f 317 675 f 513 609 f 301

18 24 27 35

17 12 9 14

11 16 15 10

17 23 29 34

14 10 11 7

16 17 14 12

403 f 241 529 f 367 945 f 783 779 f 617

47 53 65 63

9 9 4 2

10 4 9 8

46 55 68 61

8 10 5 3

5 7 6 7

a All OCE values were obtained under negative ion mode. b The OCE of a glycoside is the average (n ) 6) from duplicate determinations, respectively, at glycoside concentrations of 0.5, 5.0, and 80 µg/mL in methanol. c A human serum was spike with the glycoside level. respectively. at 0.5, 5.0. and 80 µg/mL and then extracted by methanol. The OCE of a glycoside is the average (n ) 6) from duplicate determinations of the spiked samples, respectively, at the three concentrations.

Figure 2. Selected spectra by an EGNLS (loss of 162 amu) of glucoside standards mixture. (A) total ion current (TIC); (B) extracting ion current (XIC) of emodin- 8-O-glucoside (an aromatic glycoside, m/z 431); (C) XIC of ginsenoside Rg3 (an aliphatic glycoside, m/z 783); (D) the neutral loss spectra extracted at 20 eV; (E) the neutral loss spectra extracted at 65 eV. I-X and XIII-XVI denoted in Figure 1.

Problems Associated with EGNLS. (a) Carryovers among Samples. As a result of direct infusion of samples after the simple extraction procedure, contamination of the tubing and interface may occur. This may cause carryovers among samples. To avoid this, two approaches had been taken: first, we used a small-i.d.

fused-silica capillary to introduce samples toward the spraying needle to minimize the dead volume of tubing; second, we flushed the system at a flow rate of 0.2 mL/min between two successive assays, according to the following sequence, methanol (5 min), 50 mM aqueous ammonia (5min), and methanol (10min). Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

2243

Figure 3. Selected EGNLS (loss of 176 amu) spectra for the analysis of glucuronides in Scutellaria plants. (A) TIC of plant 1; (B) XIC of 445 of plant 1; (C) XIC of 475 of plant 1; the neutral loss spectra extracted at 23 eV, respectively, of (D) plant 1, (E) plant 2, and (F) plant 3.

(b) Low-Abundant Glycosides. Low-abundant glycosides refer to those glycosides with either low concentrations in the extract or low responses by this EGNLS method (for example, some glycosides without a free hydroxyl group on aglycons). In the case where such glycosides were of interest, we first located the approximate OCE for those compounds on the TIC/XIC of the EGNLS spectra and then accumulated the neutral loss spectra at that fixed collision energy for 20-50 scans (2-6 min). Due to the low noise character of this method, the signal-to-noise ratios of the accumulated spectra were usually satisfying. Analysis of Glucuronides in S. viscidula Bge. The roots of Scutellaria plants have been used as an important crude herbal medicine for the treatment of hepatitis, diarrhea, and inflammations, and their major active compounds are glucuronides of flavonoid derivatives.28 The plant for this study was S. viscidula Bge (plant 1), whose glycoside constituents had not been sufficiently investigated. For comparison, we also studied the roots of other two common Scutellaria species: S. baicalensis Georgi (the most extensively used Scutellaria species, plant 2) and S. amoena C. H. Wrignt (plant 3). The EGNLS spectra of plant 1 for the loss of 176 (loss of a glucosiduronic acid moiety) are shown in Figure 3. Four ions, respectively, at m/z 429, 445, 459, and 475 were observed in the OCE range of 22-25 eV. These ions were also detected in plants 2 and 3. The relative abundances among the four ions were similar between the two common Scutellaria species (plants 2 and 3) but quite different from those of plant 1: for plant 1, the relative abundances of the ions at m/z 429 and 475 were significantly higher than those of the other two plants (Figure 3D-F). Therefore, this method shows the potential for rapidly distinguishing one plant from its conspecific ones using the relative abundances of glycosides. On the basis of the database searching 2244 Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

results and the known information on the conspecific plants, we presumed these four ions might be from some possible glucuronides such as baicalin, wogoninoside, etc. This needed further LC/MS/MS confirmation. Some minor ions were also found in the OCE range from 55 to 70 eV (Figure 3A); however, since those aliphatic glucuronides were not major active compounds, we did not investigate them in this study. The LC/MS/MS confirmation was performed using the screened glucuronides’ OCE and parent/product ion pairs previously obtained by the EGNLS procedure. Representative LC/MS/ MS chromatograms are shown in Figure 4. For plant 1, two structural isomers for the ion at m/z 445, two for m/z 459, and three for m/z 475 were observed, while only one compound was found for the ion at m/z 429 (Figure 4A-D). By comparing the precursor/product ions pairs and retention times against those of the standards provided latter by SDA, the compounds were identified as following: m/z 429, chrysin-3-O-glucuronide (1); m/z 445, baicalin (2) and 5, 8 dihyroxyflavone-7-O-glucuronide (3); m/z 459, wogoninoside (4) and oroxylin A-7-O-glucuronide (5); m/z 475, 5,2′-dihydroxy-6-methoxyflavone-7-O-glucuronide (6); 5,2′-dihydroxy-6′-methoxyflavone-7-O-glucuronide (7); and 5,2′dihydroxy-8-methoxyflavone-7-O-glucuronide (8). Those glucuronides were also found in both plant 2 and plant 3. Much higher absolute abundances of 1 and 8 were observed in the extract of plant 1 than in the other two plants. This agreed well with the EGNLS spectra. Besides the absolute abundances, the relative abundances of some isomeric glucuronides were also found to be quite different among the three plants. For plant 1, 7 is predominant among its three isomeric glucuronides (Figure 4D); however, the relative abundances among those three isomers do (28) Huang, H. C.; Wang, H. R.; Hsieh, L. M. Eur. J. Pharmacol. 1994, 251, 91-93.

Figure 4. Further LC/MS/MS identification of the glucuronides found by EGNLS in Scutellaria plants: (A)-(D) XIC of m/z 429 f 253, XIC of m/z 44 f 269, XIC of m/z 459 f 283, and XIC m/z 475 f 299 of plant 1; (E) XIC of m/z 475 f 299 of plant 2; (F) XIC of m/z 475 f 299 of plant 3.

not differ greatly in plants 2 and 3 (Figure 4E and F). This implies that the differences in relative abundances of those isomeric glucuronides might be significant enough to distinguish plant 1 against plants 2 and 3. The findings from this experiment are valuable for the study of the constituents and biological activities of those plants. Analysis of Ginsenosides in the Extract of P. ginseng Transformed Hairy Roots. Ginsenosides are the major effective compounds of ginseng products, which are among the most popularly used herbal medicines worldwide for their stimulating and tonic properties.15-16 In recent years, a new technique, the hairy roots culture after infection of some Panax species, has been extensively studied in order to enable the rapid and inexpensive production of ginsenosides.29 This technique shows potential for mass production of ginsenosides without time-consuming agricultural cultivation; however, the productivity of ginsenosides is greatly affected by the technical efficiency of the cultural process.30 In this report, we applied the EGNLS method for fast acquisition of the ginsenoside profiles in hairy roots and other Panax products. The transformed hairy roots (plant 4) used in this study were cultured in hormone-free media for 7 weeks after the inoculation of P. ginseng C.A. with Agrobacterium rhizogenes. For comparison,

two commonly used ginsengs, P. quinquefolius L. (American ginseng, plant 5) and P. ginseng C.A. Meyer (Korean ginseng, plant 6) were also investigated with this method. Since all major ginsenosides have terminal glucoses, those compounds can be screened out by applying EGNLS for the loss of a glucose moiety (loss of 162 amu). Representative EGNLS spectra for plants 4-6 are shown in Figure 5. Some ions with the OCE ranged from 58 to 67 eV were observed, and most of these ions could be attributed to the ions of ginsenosides (Figure 5, the labeled ginsenosides were later confirmed by LC/MS/MS). Several minor ions of aromatic glycosides with the OCE from 25 to 33 eV, were also found (Figure 5A), since these glycosides are not ginsenosides (which are aliphatic), we did not investigate them in this study. It was found the relative abundances of ginsenoside ions varied significantly among the three plants (shown in Figure 5D-F, the extracted neutral loss spectra at 63 eV): the spectrum of plant 4 is characterized by high relative abundances of the ions at m/z 945 (Re+Rd) and 799 (Rg1+Rf); plant 5 is characterized (29) Washida, D.; Shimomura, K.; Nakajima, Y.; Takido, M.; Kitanaka, S. Phytochemistry 1998, 49, 2331-2335. (30) Mallol, A.; Cusido, R. M.; Palazon, J.; Bonfill, M.; Morales, C.; Pinol, M. T. Phytochemistry 2001, 57, 365-371.

Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

2245

Figure 5. Selected EGNLS (loss of 162 amu) spectra for the analysis of ginsenosides in Panax plants: (A) TIC of plant 4; (B) XIC of m/z 945 of plant 4; (C) XIC of m/z 799 of plant 4; the neutral loss spectra extracted at 66 eV, respectively, of (D) plant 4, (E) plant 5, and (F) plant 6.

by high relative abundance of m/z 1107 (Rb1) and low relative abundance of m/z 799 (Rg1); plant 6 is characterized by high relative abundances of m/z 1107 (Rb1), 945 (Re+Rd+Glc-Rc), and 799 (Rg1+Rf). This implies that the three plants can be rapidly distinguished from one another using the ginsenoside profiles obtained by EGNLS. The results also suggest that the EGNLS method could provide a rapid estimation of ginsenosides’ relative contents in Panax plants with the relative abundances among ginsenoside ions. The absolute abundances of most ginsenoside ions in plant 4 are remarkable lower than those in both plants 5 and 6 (Figure 5), suggesting that the ginsenoside content in plant 4 may be much lower than in the other two plants. This needs confirmation with quantitative data. Fourteen major ginsenosides, respectively, Rb1, Rb2, Rb3, Rc, glc-Rc, Rd, Re, Rf, Rg1, Rg2, Rg3, Ro, Ra1, and Ra2, were further confirmed in those plants by LC/MS/MS. It was found that the relative abundances among some isomeric ginsenosides varied significantly in the three plants: for example, Rf was detected with remarkable amount in both plants 4 and 6, but not detected in plant 5 (Figure 6A-C); and the relative abundances among the three isomers of Rc, Rb2, and Rb3 are quite different from one plant to another (Figure 6D and E). The results obtained by EGNLS and LC/MS/MS for plants 5 and 6 agreed well with previously published results.16,31 Quantitative Data of Four Major Ginsenosides in the Panax Plants. We employed the method previously established by Wang et al. for the quantitation of four major ginsenosides in the three Panax plants.16 The quantitative data for the three plants are shown in Table 2. The concentrations of all four ginsenosides in the transformed hairy roots are much lower than those in the other two plants and also significantly lower than those in the successfully cultivated transformed hairy roots reported elsewhere.30 This suggests that the transformed hairy roots studied 2246 Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

here were not successful in the production of ginsenosides, and the culture conditions needed further optimization to improve productivity. CONCLUSIONS In this report, upon the demonstration of the reproducibility and specificity of OCE, an EGNLS approach for rapid and selective screening of glycosides has been established. This approach employs a collision energy gradient while performing neutral loss scans with a MS/MS and has been successfully demonstrated with the rapid screening of glycosides of interest in two crude plant extracts. Compared with the conventional fixed-energy neutral loss scan, the EGNLS has several advantages: first, the former is liable to such problems as low sensitivity or omission of some interesting compounds; this is due to the usual mismatch between the preset collision energy and some target compounds’ OCE; however, the EGNLS method overcomes this problem by scanning continuously over a collision energy range; second, EGNLS simultaneously measures the OCE of the screened compounds and therefore provides extra structural information on the screened compounds and helps to locate the compounds of interest, as demonstrated with the two applications in this report. In addition, the measured OCE of screened compounds are also essential parameters for further LC/MS/MS analysis of those compounds. It has been shown that this approach offers a rapid and selective screening of glycosides in crude biological extracts: the glycosides in crude extracts were screened out with EGNLS over the collision energy range from 0 to 100 eV, and among them the interested glycosides are sorted out based on their OCE; then by using some parameters obtained previously by the EGNLS (i.e., (31) Chan, T. W. D.; But, P. P. H.; Cheng, S. W.; Kwok, I. M. Y.; Lau, F. W.; Xu, H. X. Anal. Chem. 2000, 72, 1281-1287.

Figure 6. Further LC/MS/MS analysis of isometric ginsenosides in Panax plants: XIC of m/z 799 f 637, respectively, of (A) plant 4, (B) plant 5, and (C) plant 6; XIC of m/z 1077 f 915, respectively, of (D) plant 4, (E) plant 5, and (F) plant 6. Table 2. Contents of Four Major Ginsenosides in Three Panax Plants ginsenoside contents (mg/g of dry roots, n ) 4)

transformed P. ginseng roots P. ginseng C.A. Meyer P. quinquefolius L.

Rb1 (CV %)

Re (CV %)

Rc (CV %)

Rg1 (CV %)

0.43 (17)

1.1 (10)

0.12 (20)

0.35 (14)

2.1 (17) 2.8 (10)

1.9 (19) 0.93 (12)

4.6 (14) 29 (11)

1.8 (9) 13 (7)

the OCE values and the m/z of the parent and product ions of the screened glycosides), a LC/MS/MS method can be easily established for the further analysis of the screened glycosides. The two applications in this report have demonstrated that this strategy could rapidly provide unambiguous information on the glycosides in crude plant extracts, without using conventional timeconsuming and laborious separation procedures (such as phytochemical separation methods). In that case where glycosides with more complex sugar links are of interest, the neutral loss masses for EGNLS can be easily

modified according to the possible losses of various types and numbers of sugars. This approach might not be applicable to C-glycosides, whose sugar losses were minimal (data not shown). This strategy also has the potential for screening and identification of other types of compounds in complex matrixes, provided a constant and intensive neutral loss was found for the type of target compounds. ACKNOWLEDGMENT This work was financially supported by Chinese National Key Foundation of Science (Grant 1999054404) and an internal funding from SDA. NOTE AFTER ASAP POSTING This paper was posted on the Web on 3/13/04. Subsequently, the list and order of contributing authors was changed. The paper was reposted on 3/24/04. Received for review December 16, 2003. February 6, 2004. AC030413T

Analytical Chemistry, Vol. 76, No. 8, April 15, 2004

2247