3722 J. Agric. Food Chem., Vol. 52, No. 12, 2004
Li et al.
Sample Preparation. For extraction, 1 g of dry Ginkgo leaves was
extracted three times with 30 mL of 70% MeOH by sonication for 30
min. The three extracts were combined and evaporated to dryness using
a rotary evaporator. The crude extract was dissolved in 30 mL of 5%
HCl, heated to reflux for 30 min (23), and then partitioned three times
with cyclohexane (30 mL). The remaining water layer was partitioned
by methylethyl ketone (30 mL × 3 times). The methylethyl ketone
fractions were evaporated and dissolved in 0.6 mL of a mixture of
methanol-d4 and benzene-d6 (65:35). All experiments were performed
in triplicate. The commercial Ginkgo extracts (50 mg) and Ginkgo
product (1 tablet, about 265 mg) were dissolved in 30 mL of 5% HCl
for hydrolysis and processed as described above.
clean up or derivatization steps still remain problematic in the
reported methods. In addition, the above methods require pure
reference compounds of terpene trilactones, which are difficult
to obtain commercially due to their limited purity and high price.
Consequently, Van Beek et al. established a more attractive
procedure for the quantitative determination of terpene trilac-
tones using nuclear magnetic resonance (NMR), after a chro-
matographic preparation step (18).
The flavonoids of Ginkgo leaves include great varieties of
flavonol glycosides based on kaempferol, quercetin, and isorham-
netin and are found as mono-, di-, and triglycosides as well as
their cinnamic acid esters. Thus, the flavonoid profile of Ginkgo
is very complex (19), and analysis during pharmaceutical quality
control is rather tedious. In addition, most of the reference
compounds are not commercially available. In the pharmaceuti-
cal industry, Ginkgo extracts have long been standardized on
flavonol glycosides. A simple HPLC method was established
for the analysis of flavonol aglycones after hydrolysis of the
glycosides in Ginkgo and sample clean up (20).
Recovery Test. The recovery test of reference compounds was
according to Choi’s method (24). One gram of filter paper disks (5892
white ribbon ashless, Schleicher & Schuell, GmbH, Cassel, Germnay)
was cut into a 1 cm diameter and placed in the extraction vessel. Each
standard of bilobalide; ginkgolides A, B, C, and J; kaempferol;
quercetin; and rutin (3.0 mg) was spiked into the filter paper disks.
Then, the spiked samples were dried in a vacuum oven at 40 °C for 24
h. These cellulose papers were extracted, hydrolyzed, processed as
1
described above, and quantified by the H NMR method.
There are two methods reported in the literature, which can
analyze terpene trilactones and flavonoids simultaneously. One
is the GC-MS method (15), and the other is the HPLC-ELSD
method (11). The GC-MS method needs numerous preparation
and derivatization steps, and the chromatographic running time
was more than 40 min. The numerous steps and the poor
reproducibility due to the output of the light source and/or
detector response stability remain problematic in the HPLC-
ELSD method. At present, it appears that there is no suitable
method for the simultaneous analysis of both active components
of Ginkgo, terpene trilactones, and flavonols. Thus, to resolve
the analytical problems of Ginkgo terpene trilactones and
simultaneously analyze flavonol glycosides, we investigated a
reliable analytical method using 1H NMR spectrometry after a
simple hydrolysis step as an alternative to the conventional
analyses. The method was applied to the quantitative analysis
of 10 commercial Ginkgo products containing terpene trilactones
and flavonols and three samples of Ginkgo leaves collected from
different regions.
RESULTS AND DISCUSSION
In the 1H NMR quantification of terpene trilactones and
flavonol glycosides in Ginkgo leaves and commercial prepara-
tions, it would be desirable to quantify each individual terpene
trilactone and flavonol by means of the integral of a specific
1
proton signal. However, the H NMR peaks of other Ginkgo
constituents may interfere with the target signals of the active
compounds. In previous studies, tedious cleaning or preparation
steps, changing NMR solvents, controlling the pH value, or the
addition of a shift reagent were evaluated to solve this problem.
Van Beek et al (18). reported a one column clean up method,
and Choi et al (24). changed the NMR solvents to determine
ginkgolide A, B, and C and bilobalide contents. Choi’s method
resolved the problems of low recoveries and time-consuming
column preparation steps. However, poor solubility in the
solvent system (acetone-d6-benzene-d6 50:50) and degradation
of the internal standard (phloroglucinol) remained problematic.
To analyze terpene trilactones and flavonol glycosides efficiently
and simultaneously, we selected a suitable solvent system with
a high solubility and a more stable internal standard.
MATERIALS AND METHODS
In previous reports (18, 24), the H-12 protons of bilobalide
and ginkgolides appeared as well-separated singlets with little
interference. Because of these special characteristics, the H-12
signal was selected as the target signal for determining terpene
trilactones in the present study. In the previous report, it was
found that these terpene trilactones were extraordinarily stable
even in boiling HNO3 (25). Utilizing this unique stability, we
can hydrolyze the Ginkgo extracts to simultaneously analyze
the flavonol aglycones. The many varieties of flavonol glyco-
sides in Ginkgo can be reduced in number by hydrolysis to the
corresponding aglycones (Figure 1). Each aglycone has its own
characteristic H-2′ signal (or H-2′/6′ for kaempferol) in the 1H
NMR spectrum due to the different substitution patterns. In
addition, this aglycone signal appears between 7.8 and 8.8 ppm,
which generally is a noncrowded spectroscopic region and thus
shows relatively little interference with other compounds.
Therefore, the H-2′ signal (or H-2′/6′ for kaempferol) was
selected to be the target signal to quantify the main aglycones:
kaempferol, quercetin, and isorhamnetin.
Chemicals and Instrument. HPLC grade methanol, cyclohexane,
and methylethyl ketone were purchased from E. Merck. Methanol-d4
(99.8%), dimethylsufoxide-d6 (99.9%), pyridine-d5 (99.5%), acetone-
d6 (99.9%), benzene-d6 (99.6%), and toluene-d8 (99.5%) were obtained
from Aldrich. 1,3,5-Trimethoxybenzene was prepared by methylation
of phloroglucinol (Sigma) (21). The reference compounds (bilobalide;
ginkgolides A, B, C, and J; kaempferol; quercetin; and isorhamnetin)
were isolated from G. biloba leaves in a prior study (22). The purity
of all of these reference compounds was established by NMR and
HPLC. 1H NMR spectra were recorded in various solvent systems using
a Varian UNITY plus 400 MHz spectrometer. For each sample, 100
scans were recorded with the following parameters: 0.187 Hz/point;
spectra width, 3600 Hz; pulse width, 4.0 µs; relaxation delay, 1 s; and
acquiring time, 2.67 µs. For quantitative analysis, the peak area was
used and the start and end points of the integration of each peak were
selected manually.
Commercial Ginkgo Extracts, Ginkgo Products, and Ginkgo
Leaves. Ginkgo extracts were purchased from Chemax (Ellis Bridge,
Ahmedabed, India), Chart (Paterson, NJ), Tokiwa PhytoChemical
(Sakura-shi, Chiba, Japan), YBS (Tokyo, Japan), CONBA (Lanxi,
Zhejiang, China), Indena (Milan, Italy), J C Bright M. (Vancouver,
BC, Canada), Ningbo (Ningbo, Zhejiang, China), and USA NutraSource
(Eugene, OR). The Ginkgo product (Cebonin) was sold by Nang Kuang
Pharmaceutical Company (Tainan, Taiwan). Two samples of Ginkgo
leaves were purchased from a Taiwan market, and one sample was
collected at Chi-Tou, Taiwan.
To improve solubility, polar deuterated solvents such as
methanol-d4, dimethyl sulfoxide-d6, and pyridine-d5 were used.
To more readily resolve the target signals and reduce inter-
ference from other Ginkgo constituents, the addition of benzene-
d6 or toluene-d8 to the selected high polar solvent was evaluated.