Anal. Chem. 2004, 76, 4332-4336
A Method for Extraction and Quantification of Ginkgo Terpene Trilactones Chen Ding,† Erqin Chen,*,‡ Weijia Zhou,‡ and Robert C. Lindsay†
Department of Food Science, and Wisconsin Center for Space Automation and Robotics, University of WisconsinsMadison, Madison, Wisconsin 53711
A method was developed for the extraction and quantification of pharmacologically active terpene trilactones (ginkgolides, bilobalide) from the tissues of Ginkgo biloba L. and pharmaceutical ginkgo products by RPHPLC, based on the theory of terpene trilactones ionization. Four ginkgolides (GA, GB, GC, GJ) and bilobalide (BB) from both the ginkgo leaves and commercially available ginkgo extracts were quantitatively extracted by using this method. The recovery rate of the method was 97.5-100% with RSD of 1.2-2.8%. The detection limit was 0.05-0.1 µg, and the linear range was 0.1-12 µg. This detection limit represents a marked improvement over previously reported methods, suggesting the new method is a viable technique for routine analysis of ginkgo terpene trilactones in natural and commercial samples. The method reported by van Beek et al. in 1991 (van Beek, T. A.; Scheeren, H. A.; Rantio T.; Melger, W. C.; Lelyveld, G. P. J. Chromatogr. 1991, 543, 375-387.) was used as a reference method to monitor the accuracy of extraction and analysis in this study. SSI-MS technique was used to identify isolated target components. Carbohydrase treatment and solubility of terpene trilactones in various solvents were also discussed. The leaf and fruit of Ginkgo biloba L have been used for the treatment of vascular diseases in traditional Chinese medicine and the earliest medical application was cited as far as 5000 years ago.1 During the past 20 years, EGb761, a standard G. biloba L. extract, was subjected to scientific research and clinical trials in order to verify its theraputic effect and mechanism.2,3 Increasing research evidence supports that the bioactive components in Egb761 have significant theraputic effect on age-related physical and mental deteriorations and on cerebral vascular insufficiency; e.g., Alzheimer’s and cardiovascular diseases.4-6 * To whom correspondence should be addressed. Phone: (608) 262-4588. Fax: (608) 262-9458. E-mail:
[email protected]. † Department of Food Science. ‡ Wisconsin Center for Space Automation and Robotics. (1) Braquet, P. In Ginkgolides: Chemistry, Biology, Pharmacology and Clinical Perspctives; Braquet, P., Ed.; J. R. Prous Science Publishers: Barcelona, Spain, 1988; pp XV-XXXIV. (2) Sierpina, V. S.; Wollschlaeger, B.; Blumenthal, M. Am. Family Physican 2003, 68, 923-926. (3) Mix, J. A.; Crews, W. D. Hum. Psychopharmscol.-Clin. Exp. 2002, 17, 267277. (4) Pietri, S.; Maurelli, E.; Drieu, K.; Culcasi, M. J. Mol. Cell. Cardiol. 1997, 29, 733-742.
4332 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
Figure 1. Chemical structure of ginkgo terpene trilactones.
Important bioactives in G. biloba L. are ginkgolides and bilobalide, or ginkgo terpene trilactones (Figure 1), having three γ-lactones, which is one of the unique characteristics of G. biloba L. Several methods have been developed to analyze these compounds in various sources,7 and most of them employed organic solvents to extract terpene trilactones from tissues or pharmaceutical products based on the high solubility of ginkgo terpene trilactones in organic solvents. In fact, the solvent extraction resulted in complex extract mixtures due to the lack of selectivity of organic solvents and resulted in time-consuming follow up purifications, serious interference in quantification, unsatisfactory detection limits, and poor accuracy and reproducibility. Moreover, these methods were found expensive at largescale analysis. Van Beek et al.8 have reported a widely accepted method, and their work made a significant contribution to ginkgo terpene trilactone extraction and quantification by HPLC-RI. Neverthless, the method included a time-consuming purification step, which limited application of the method in large-scale analysis. An improved method9 was developed with supercritical fluid chro(5) Le Bars, P. L.; Velasco, F. M.; Ferguson, J. M.; Dessain, E. C.; Kieser, M.; Hoerr, R. Neuropsychobiology 2002, 45, 19-26. (6) Diamond, B. J.; Shiflett, S. C.; Feiwel, N.; Matheis, R. J.; Noskin, O.; Richards, J. A.; Schoenberger, N. E. Arch. Phys. Med. Rehabil. 2000, 81, 668-678. (7) van Beek, T. A. J. Chromatogr., A 2002, 967, 21-55. (8) van Beek, T. A.; Scheeren, H. A.; Rantio T.; Melger, W. C.; Lelyveld, G. P. J. Chromatogr. 1991, 543, 375-387. 10.1021/ac049809a CCC: $27.50
© 2004 American Chemical Society Published on Web 06/17/2004
Table 1. Terpene Trilactone Content (mg/g of DW) from Different Extraction Methodsa 1
2
3
4
5
6
7
8
sample
terpene trilactone
reference method
water•2 + RM
water•2 + EA
water•1 + RM
water•1 + EA
water•2 + CH2Cl2
BB
1.48 5.74 2.33 0.314 4.90 1.94 0.240 1.90 1.09 0.282 2.79 1.08 0.144 1.96 0.590
1.50 5.34 2.25 0.321 4.87 1.91 0.241 1.87 1.08 0.286 2.77 1.12 0.150 1.98 0.623
1.49 5.50 2.38 0.320 4.91 1.89 0.242 1.84 1.09 0.288 2.89 1.15 0.154 2.17 0.650
0.991 2.30 1.05 0.302 5.01 1.95 0.239 1.89 1.05 0.286 2.77 1.12 0.155 1.95 0.653
0.990 2.35 1.03 0.312 4.90 1.92 0.242 1.90 1.07 0.287 2.80 1.08 0.149 1.99 0.625
0.890 3.10 1.19 0.311 5.48 1.99 0.240 1.82 1.10 UD 0.234 UD UD 0.109 UD
leaf supplement 1 supplement 2 leaf supplement 1 supplement 2 leaf supplement 1 supplement 2 leaf supplement 1 supplement 2 leaf supplement 1 supplement 2
GA GB GC GJ
a Water•1, extracted with ddH O once and Na HPO twice; Water•2, extracted with ddH O and Na HPO twice; RM, cleaned up with reference 2 2 4 2 2 4 method; EA, partitioned by ethyl acetate at pH 4.5; CH2Cl2: partitioned by CH2Cl2 at pH 4.5. Supplement 1, tables (Boehringer Ingelheim pharmaceuticals, Inc. Ridgefield, CT); supplement 2, capsules (Sundown Vitamins, Boca, Raton, FL, and Alert Health Products, Porterville CA).
matography (SFC). This method achieved a low limit of detection (LOD ) 10 ng) in quantification and simplification of extraction. However, application of this method was restricted because a specific SFC system and an evaporative light scattering detector were needed. Applying the ionization constants of ginkgolide B reported by Zekri et al.,10 Lang and Wai11 have reported a method using aqueous-phase extraction followed by derivative analysis. Unfortunately, the extraction method had several pitfalls due to lack of a reference method and an improper experimental design, which resulted in a loss of over 80% of ginkgolide C and J and 40-60% loss of bilobalide. Van Beek12 pointed out the weak points in this method with critical comments. Although much effort has been made in the analysis of ginkgo terpene trilactones, timeconsuming extraction, low sensitivity and poor reproducibility are still the major obstacles in quantitative analysis. Demand for nutraceutical and pharmaceutical products from G. biloba L. is continually increasing worldwide due to the increased awareness of their significant theraputic effects. A viable routine analytical method, with high sensitivity, accuracy and excellent reproducibility, will be significantly important to scientific research, clinical trial and phytopharmaceutical industry. In this paper, a rapid and sensitive RP-HPLC method was reported. Using van Beek’s method8 as a reference method to monitor accuracy of extraction and analysis, and based on ionization characteristic of ginkgolide B,10 the developed method employed sequential aqueous extractions at different pH values. The new method of extraction demonstrated an excellent selectivity to ginkgo terpene trilactones and significantly reduced extraction time. This method was applied to analyze samples of G. biloba L. leaf and commercial supplement products, and the obtained results were in good agreement with those from reference method. In addition, enzymatic extraction was also employed in this study (9) Strode, J. T. B.; Taylor, L. T.; van Beek T. A. J. Chromatogr.. A 1996, 738, 115-122. (10) Zekri, O.; Boudeville, P.; Genay, P.; Perly, B.; Braquet, P.; Jouenne, P.; Burgot, J. L. Anal. Chem. 1996, 68, 2598-2604. (11) Lang, Q. Y.; Wai C. M. Anal. Chem. 1999, 71, 2929-2933. (12) van Beek, T. A. Anal. Chem. 2000, 72, 3396.
to validate the reported extraction procedure, and the sonic spray ionization source (SSI)-LC/MS technique was used for the identification of target compounds. MATERIALS AND METHODS Materials. Leaves were collected from both male and female ginkgo trees on Walnut St., Madison, WI, and the Arboretum areas of University of WisconsinsMadison during August and September 2000. G. biloba phytopharmaceuticals, capsules (Sundown Vitamins, Boca Raton, FL, and Alert Health Products, Porterville, CA), ginkoba tables (Boehringer Ingelheim Pharmaceuticals, Inc. Ridgefield, CT), and soft gels (Whitehall-Robins Healthcare, Madison, NJ, and Nature Made Nutritional Products, Mission Hills, CA) were obtained from a local grocery store. Authentic ginkgolide A, B, and bilobalide, benzyl alcohol, and HPLC grade organic solvents (methanol, ethanol, dichloromethane, ethyl acetate, ethyl ether) were purchased from Sigma-Aldrich Chemical Co. (St. Louis. MO). Authentic ginkgolide C was kindly provided by Dr. Jian Zhang (Qinghua University in P. R. China). The purity was determined by means of HPLC, UV spectroscopy, and 400-MHz NMR spectroscopy. Sodium phosphate was obtained from Fisher Scientific (Chicago, IL). The enzymes cellulase, hemicellulase, and pectinase were obtained from ICN Biomedicals Inc. (Aurora, OH), Sigma-Aldrich Chemical Co., and EMD Biosciences, Inc. (La Jolla, CA), respectively. Water used for experiments was double-distilled water (ddH2O). High-Performance Liquid Chromatography (HPLC). Chromatographic analyses were performed on a Hitachi liquid chromatograph system (D-7000 model, Hitachi Instrument Inc., San Jose, CA). Detection was carried out with a refractive index detector (RI, model L-7490, 16 × 10-6 refractive index units). An autosampler (model L-7200) was used for sample injection while the column used was a C18 analytical column (3 µm Microsorb 3 Spherical, ODS2, 100 × 4.6 mm id, Waters Co., Milford, MA) and a guard column (5 µm, ODS2, Waters Co.). The eluent was MeOH/H2O (29.5:70.5 v/v) at a flow rate of 0.5 mL/min. All separations were carried out isocratically at 37 °C. The chromatoAnalytical Chemistry, Vol. 76, No. 15, August 1, 2004
4333
graphic separations were monitored with Hitachi data collection software. Mass Spectrometry. Hitachi M-8000 LC/3DQ ion trap mass spectrometer system (Hitachi Instrument Inc.) was used to identify extracted ginkgo terpene trilactone. SSI was used for ionization with temperatures of 150 °C for cover plate and plate aperture 1 and 120 °C for aperture 2. The discharge voltages were 0.8 kV and 50, 40, and 450 V for SSI chamber, drift, focus, and detector, respectively. Nitrogen gas pressure was at 360 KPa. All LC/MS analyses were carried out in the positive standard ms mode from m/z of 200-700. For injection, the stock solution used for HPLC analysis was diluted 10 times with final concentration of 50% methanol and 0.002% acetic acid. Injection volume was 10 µL. Reference Method. The extraction and analytical protocols reported by van Beek et al.8 was used as a reference method to monitor accuracy of the extraction and analytical method developed in this study. Sample Preparation. Fresh ginkgo leaves were washed with distilled water and cut excluding petioles and air-drying at room temperature (21 °C). Dry ginkgo leaves and ginkgo phytopharmaceutical tablets were pulverized to a fine powder (300-420 µm) and stored in covered glass bottles in a glass desiccator until used. The wall materials of the phytopharmaceuticals were removed manually before use. Extraction step 1: Extraction of ginkgo terpene trilactones from ginkgo leaf samples and commercial supplements. Leaf samples (0.5 g) and phytopharmaceuticals (0.2 g) were weighed into 35mL polypropylene centrifuge tubes (Nalgene, Fisher Scientific), 6 mL of boiling ddH2O was added to the tubes, heated in a water bath (100 °C) for 5 min, and the supernatant was obtained after centrifuging at 2000g for 15 min. This process was repeated once. In the third extraction, the boiling ddH2O was replaced with a boiling Na2HPO4 solution (0.1% w/w, pH 8.0). The residue subjected to extraction for 15 min and centrifuged to obtain the supernatant. This process was also repeated once. All supernatants were pooled, pH adjusted to 4.5, and made up to 25 mL. Extraction step 2: Partitioning of the ginkgo terpene trilactones into organic solvents. The pH of aqueous samples from step one was adjusted to 3.0, 3.5, 4.0, 4.5, 5.0, 5.5, 6.0, and 6.5. Two 10-mL aliquots from each pH level were sequentially partitioned with ethyl acetate three times (10, 5, and 5 mL) by vortexing for 2 min and centrifuged at 2000g for 15 min to separate the phases. The organic phases were quantitatively collected into a 25-mL glass centrifuge tube and dried at 45 °C under N2 gas in a Turbovap evaporator (Zymark Co., Hopkinton, MA). The residue was redissolved with 400 µL of methanol and submitted for HPLC analysis. Benzyl alcohol (final concentration of 0.25-0.5 mg/mL) was added to the samples as internal standard, and a 10-µL aliquot was injected onto the HPLC column. Other organic solvents, namely, dichloromethane, ethyl ether, and hexane, were also used for partitioning to investigate the solubility of ginkgo terpene trilactones in different organic solvents. Enzymic Polymer Degradation Treatments. Leaf powders (0.5 g) were incubated with three carbohydrases, cellulase, hemicellulase, and pectinase (1% w/w), in 12 mL of potassium phthalate buffer (0.025 M, pH 5.5) in a water bath at 35 °C for 24 and 36 h and then centrifuged at 2000g for 15 min to separate the supernatant and residue. The residue was washed twice with 5 4334
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
Figure 2. HPLC chromatogram of a leaf sample. (A) Dichloromethane partitioning without internal standard; (B) ethyl acetate partitioning with internal standard.
mL of Na2HPO4 (0.1%, w/w, pH 8), and the supernatant combined, pH adjusted to 4.5, and made up to a volume of 25 mL. Resulting supernatant was then treated in the same manner as in extraction step 2 using ethyl acetate and subjected to HPLC and LC/MS analyses. Recovery Experiments. Known amounts of ginkgolides (GA, GB) and bilobalide were dispensed to leaf samples and subjected to extraction steps 1 and partition with ethyl acetate to determine recovery rate. Three different quantities of starting materials were used, and all assays were run in triplicate. Quantification. Both external and internal standards were used for quantification of BB, GC, GA, and GB. The content of GJ was calculated with a mean response factor obtained for GC external standards due to commercial unavailability of authentic terpene trilactones. Benzyl alcohol was used as an internal standard. For the internal standard procedure, the response factors (RF) and the amount of terpene trilactones in the samples were calculated using the eqs 1 and 2.
RF ) C (G)A (Is)/[C (Is)A (G)]
(1)
concentration (mg/mL) ) [A (S)/A (Is)] × RF × C (Is) (2)
Figure 3. Impact of pH on extraction of terpene trilactones from ginkgo leaf sample. (A) Serial aqueous extraction; (B) ethyl acetate partitioning.
where C (G) is the concentration of the ginkgo compounds in reference solution (mg/mL), C (Is) is the internal standard concentration in reference solution, A (G) is the peak area of the ginkgo compounds in reference solution, A (Is) is the peak area of the internal standard in reference solution, and A (S) is the peak area of the test samples. RESULTS AND DISCUSSION Ginkgo Terpene Trilactone Ionization and Extraction. Based on the study of ionization constants of ginkgolide B,10 ginkgolide B existed in a neutral molecule form at pH 6.0 or lower, which possessed higher affinity to organic solvent. On the contrary, ginkgolide B underwent ionization and became a charged ion at pH 8.0 or higher. The charged ion confirmed a higher affinity for water and ion-exchange characteristic. All studied ginkgo terpene trilactones have chemical structures similar to ginkgolide B (Figure 1) and hypothetically comply with this ionization theory. Bilobalide, compared to other terpene trilactones, was very unstable and had a rapid degradation rate when the pH was over 8.0. In this study, we extracted the materials twice with boiling water for 5 min to remove the bulk amount of bilobalide, followed by extracting twice with Na2HPO4 (pH 8.0, 0.1%, w/w) for 15 min to remove remaining terpene trilactones, especially GB. The resulting aqueous phase was then partitioned with ethyl acetate or van Beek’s cleanup procedure. We found that the results were in agreement with those obtained for the reference method (Table 1, columns 3-5). The results demonstrated that all ginkgo terpene trilactones were completely removed from the leaf and commercial samples with our serial aqueous extraction procedure. However, if the water extraction step was performed only once, the amount of bilobalide extracted was significantly lower than that in the extracts prepared using the reference method.8 We found that 67-78% of BB was extracted from leaf samples, while only 42% of BB was extracted from higher concentration ginkgo phytopharmaceticals (Table 1, columns 6 and 7). Solvent Selectivity. Four different organic solvents, hexane, dichloromethane, ethyl acetate, and ethyl ether, were employed to investigate the extractability of ginkgo terpene trilactones from
aqueous solution. The target compounds were completely insoluble in hexane (data not shown). In comparison with the reference method, dichloromethane partition only recovered 5060% of BB from leaf samples or ginkgo phytopharmaceticals, while almost no GJ and GC were recovered from leaf samples, and about 5.6-8.4% of GJ or GC was recovered from ginkgo phytopharmaceticals (Table 1 column 8). However, the amount of GA and GB recovered using dichloromethane partition was not significantly different from those obtained from the reference method or ethyl acetate partition. An HPLC chromatogram of the dichloromethane and ethyl acetate partition are shown in Figure 2. Compared to the results obtained from ethyl acetate partitioning, the amount of BB obtained from ethyl ether was not different. However, ethyl ether partitioning afforded only 32% of GC, 80% of GA and GB, and no GJ. Effect of pH and Sodium Phosphate Concentration on Extractions. The impact of pH on serial aqueous extraction and ethyl acetate partitioning was investigated. In the serial aqueous extraction, the optimal pH range of initial Na2HPO4 solution was 6.5-8.0 for BB and GA, and 7.5-8.0 for GJ, GC, and GB. The extractable amount of BB sharply declined from pH 8.0 to 8.5 and was reduced by 27% at pH 8.5 (Figure 3A). In the ethyl acetate partitioning, the optimal pH range was 3.5-5.0 for BB and GA and 3.0-5.0 for all other terpene trilactones. BB was one of the most pH-sensitive terpene trilactones and became difficult to recover with ethyl acetate when a pH higher than 5.0 was employed. We found that about 13.5 and 12.0% of BB was lost at pH 5.5 and pH 3.0, respectively (Figure 3B). The effect of Na2HPO4 concentration on the extraction of all target compounds was also investigated in this study. The results indicated that a concentration of 0.1% of Na2HPO4 was optimal for all ginkgolides and bilobalide. Recovery Rate and Accuracy. Known amounts of BB, GA, and GB were dispensed to samples and were extracted with the studied method to determine recovery rate. Obtained recovery rates were 97.5-100% with a RSD of 1.2-2.8% (Table 2). The limit of detection for the method was 0.05 µg for BB, GJ, GC, and GA and 0.1 µg for GB, and the linear range was 0.1-12 µg. The Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4335
Table 2. Compound Recoverya and Method Reproducibility amount (µg) sample 1 2 3
a
BB GA GB BB GA GB BB GA GB
added
initial
0.20
0.587 0.201 0.101 0.587 0.201 0.101 0.587 0.201 0.101
0.40 0.60
recovery amount (µg) 0.203 0.240 0.204 0.403 0.388 0.389 0.578 0.606 0.577
0.195 0.192 0.202 0.388 0.398 0.388 0.582 0.579 0.582
0.202 0.197 0.199 0.405 0.402 0.404 0.589 0.602 0.589
0.203 0.201 0.189 0.399 0.397 0.405 0.603 0.587 0.603
mean recovery (%)
RSD (%)
100 99.7 99.2 97.5 99.3 99.4 98.4 98.4 98.5
2.1 2.5 2.8 2.1 1.2 1.4 2.1 1.9 2.1
0.196 0.203 0.197 0.389 0.401 0.403 0.603 0.589 0.604
A known amount of terpene trilactones (BB, GA, GB) was spiked into ginkgo leaf samples before serial extractions.
Table 3. Response Factors of GA, GB, GC, and BB from Internal Standards compd BB
GA
std (µg/mL)
response factors
0.100 0.200 0.400 0.800 0.100 0.200 0.400 0.800
1.65 1.62 1.55 1.46 1.56 1.50 1.37 1.26
compd GB
GC
std (µg/mL)
response factors
0.100 0.200 0.400 0.800 0.100 0.200 0.400 0.800
2.97 1.76 1.69 1.55 1.69 1.54 1.50 1.48
response factors of GA, GB, GC, and BB obtained from the internal standard method are shown in Table 3. All resolved target peaks were qualitatively identified by the SSI-LC/MS method. The characteristic sonic spray mass spectral signals for all terpene trilectones were adducts of M + H, M + NH4, and M + Na under our experimental conditions. For instance, the mass spectrum of GJ showed signals at m/z of 424.93 (M + H)•+, 441.91 (M + NH4)•+, and 446.87 (M + Na)•+ (Figure 4). Evaluation of Enzymic Polymer Degradation. To farther investigate the accuracy of our serial aqueous extraction procedure, the enzymic polymer degradation method was employed. The results obtained by SSI-LC/MS of carbohydrase-treated samples agreed with those from the serial aqueous extraction procedure, which indicated that target compounds were not degraded during the high-termperature extraction and were completely extracted in extraction step 1. However, we found that the amount of GJ and BB quantified by HPLC-RI of the carbohydrase-treated samples was significantly higher than that of our extraction procedure. This suggested a low selectivity of the carbohydrase degradation method toward the target components that resulted in a large amount of conextracted materials interfering with GJ and BB resolution. Therefore, the carbohydrase treatment is not recommended as an extraction method for ginkgo terpene trilactione. CONCLUSION Ginkgo terpene trilactones can be effectively extracted from natural sources and the pharmaceutical products using the serial
4336 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
Figure 4. Characteristic mass spectrum of ginkgolide J (GJ) isolated from ginkgo leaf sample [m/z 424.93 (M + H)•+, 441.91 (M + NH4)•+, and 446.87 (M + Na)•+].
aqueous extraction technique, which is quick, effective, and highly selective. All terpene trilactones present in aqueous extract can be completely recovered by ethyl acetate at pH 4.5. HPLC detection limit obtained for our method represents a 10-fold improvement over previously reported limits. This study demonstrated that extraction of ginkgo terpene trilactones and their analogues based on the theory of terpene trilactone ionization in conjunction with HPLC is a viable technique for routine analysis of ginkgo terpene trilactones in natural and commercial samples.
Received for review February 3, 2004. Accepted May 11, 2004. AC049809A
