Quantification of a botanical negative marker without an identical standard: Ginkgotoxin in Ginkgo biloba

Article abstract of DOI:10.1021/np400874z

A new strategy for the analysis of natural products uses a combination of quantitative ¹H NMR (qHNMR) and adsorbent-free countercurrent separation (CS) methodology to establish a quantification method for ginkgotoxin (4′-O-methylpyridoxine) in Ginkgo biloba preparations. The target analyte was concentrated in a one-step CS process using the ChMWat +2 solvent system (CHCl₃-MeOH-H₂O, 10:5:5) and subsequently assayed by qHNMR. While commercial G. biloba seeds contained 59 μg of ginkgotoxin per seed, the compound was below the limit of detection (9 ppm) in a typical leaf extract. Due to the enrichment potential and loss-free operation of CS, the combination of CS and qHNMR is a generally suitable approach for threshold assays aimed at quantifying target compounds such as botanical negative markers at the low ppm level. As the proof of principle is demonstrated for relatively small CS capacities (20 mL, 1:40 loading) and modest NMR sensitivity (n = 16, 400 MHz, 5 mm RT probe), the approach can be adapted to quantification at the ppb level. The procedure enables the quantification of a botanical negative marker in the absence of identical reference material, which otherwise is a prerequisite for LC-based assays.

Full text of DOI:10.1021/np400874z

Article
pubs.acs.org/jnp
Quantification of a Botanical Negative Marker without an Identical
Standard: Ginkgotoxin in Ginkgo biloba
Yang Liu,† Shao-Nong Chen,† James B. McAlpine,†,‡ Larry L. Klein,‡ J. Brent Friesen,†,§ David C. Lankin,†
and Guido F. Pauli*,†,‡
†Department of Medicinal Chemistry and Pharmacognosy, College of Pharmacy, University of Illinois at Chicago, Chicago, Illinois
60612, United States
‡Institute for Tuberculosis Research, College of Pharmacy, University of Illinois at Chicago, Chicago, Illinois 60612, United States
§Department of Physical Sciences, Rosary College of Arts and Sciences, Dominican University, River Forest, Illinois 60305, United
States
S
* Supporting Information
ABSTRACT: A new strategy for the analysis of natural
products uses a combination of quantitative 1H NMR
(qHNMR) and adsorbent-free countercurrent separation
(CS) methodology to establish a quantification method for
ginkgotoxin (4′-O-methylpyridoxine) in Ginkgo biloba prep-
arations. The target analyte was concentrated in a one-step CS
process using the ChMWat +2 solvent system (CHCl3−MeOH−H2O, 10:5:5) and subsequently assayed by qHNMR. While
commercial G. biloba seeds contained 59 μg of ginkgotoxin per seed, the compound was below the limit of detection (9 ppm) in
a typical leaf extract. Due to the enrichment potential and loss-free operation of CS, the combination of CS and qHNMR is a
generally suitable approach for threshold assays aimed at quantifying target compounds such as botanical negative markers at the
low ppm level. As the proof of principle is demonstrated for relatively small CS capacities (20 mL, 1:40 loading) and modest
NMR sensitivity (n = 16, 400 MHz, 5 mm RT probe), the approach can be adapted to quantification at the ppb level. The
procedure enables the quantification of a botanical negative marker in the absence of identical reference material, which
otherwise is a prerequisite for LC-based assays.
Ginkgo biloba L., also known as “Yin Xing” in China and
“Ginnan” in Japan, has documented use in China and Japan for
the treatment of cough, bronchial asthma, irritable bladder, and
alcohol abuse for over one thousand years.1 Its seeds are used
as food or dietary supplements to improve memory, reduce
neuronal disorders, and increase blood flow.1,2 Its leaf extract is
known to have antioxidant activity and anti-inflammatory
effects and to stimulate neurotransmission.3
Ginkgotoxin (4′-O-methylpyridoxine, 1; Scheme 1) is
considered the major toxic compound in ginkgo seeds, and it
may be present in the leaves as well (see below). It behaves as
an antagonist of vitamin B6 (pyridoxine, 2), acting as a
pseudosubstrate of pyridoxal kinase.4,5 Clinical reports indicate
that overconsumption of G. biloba seeds has induced seizures or
even caused death.6 The National Institutes of Health has
warned that roasted seeds could also be dangerous.7
The WHO has estimated that 80% of people worldwide rely
on herbal medicines for the treatment of health problems. Sales
of G. biloba-related products placed this botanical among the
top-10 selling herbal products in the U.S. in 2007.8 Virtually all
of the dietary supplement products sold in the U.S. and Europe
are extracts of the leaves. Although the presence of 1 in seeds is
well established, the presence in leaves is controversial. Using
RP-HPLC/UV, Wada et al.9 were unable to detect the presence
of 1 and, thus, concluded that there was no toxic potential in
leaf extracts. Later, Arenz et al.10 found that the concentration
in leaves varied over time and at its peak exceeded even that in
seeds.
Wada and Haga11 have summarized adverse reactions to
ginkgo seeds reported from 1930 to 1996 with more than 100
cases, most of which involved the occurrence of convulsions.
Children under 6 years of age accounted for 74% of these cases,
and mortality was up to 27%. Considering more recently
reported cases,6,12−14 it is clear that the high-incidence
population was that of infants and of the aged. However, the
dose/concentration threshold for 1-induced seizures in patients
is still unclear.15 The presence and effect of 1 and its
relationship to other bioactive markers of G. biloba have also
been addressed in recent publications.1,16
Although 1 was first reported from G. biloba in 1985,5
virtually no effective and convenient standard detection
methods have been established to date. Considering botanical
safety in general, and the challenge of identifying and
quantifying a low-level negative marker (i.e., an undesirable
compound in a botanical extract, such as a toxin), there is a
need to establish methods for the accurate analysis of 1. Herein,
we report a method based on a combination of countercurrent
Special Issue: Special Issue in Honor of Otto Sticher
Received: October 16, 2013
© XXXX American Chemical Society and
American Society of Pharmacognosy
A
dx.doi.org/10.1021/np400874z | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Scheme 1. Route for the Microwave-Assisted Synthesis of Ginkgotoxin (1) from Pyridoxine (2)
Figure 1. Purity evaluation by qHNMR (under identical quantitative experimental conditions; 400 MHz, in CDCl3) of synthetic 1 after different
purification: TLC (a) yielded 96% pure material, while NP-silica gel column chromatography resulted in a much lower purity of 81%. Purification by
HSCCC gave the highest purity (98%). Only the resonance of H-6 exhibited a slight δ shift (0.08 ppm), which was attributed to pH changes due to
the various degrees and patterns of impurities.
Figure 2. Example of the determination of the CS K value of synthetic 1 by qHNMR, directly from the upper and lower phases of a shake flask
experiment with the ChMWat +2 solvent system (dried and redissolved in CDCl3). The K values were calculated by (eq 2), as follows: Aupper phase
=
(2.26/1 + 4.46/2 + 4.45/2 + 6.49/3 + 7.11/3)/5 = 2.25; Alower phase = (1.81/1 + 3.52/2 + 3.60/2 + 5.15/3 + 5.68/3)/5 = 1.79; K value = 2.25/1.79 =
1.26.
B
dx.doi.org/10.1021/np400874z | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 3. Enrichment and identification of ginkgotoxin (1) in G. biloba seeds. Quantitative (400 MHz, CDCl3) 1H NMR analysis of (a) the extract
of a single seed; (b) the targeted HSCCC fraction corresponding to a K value of 1.62 (ChMWat, 10:5:5); (c) synthetic 1.
separation (CS)17 and NMR techniques. This approach can be
used to detect (limit of detection [LOD]: 9 ppm) and quantify
(limit of quantification [LOQ]: 30 ppm) 1 in methanolic
gingko extracts. While we found 1 in seeds at 59 ppm, the
content in a commercial ginkgo leaf sample was below the
LOD.
Since the establishment of the ginkgo lactones and
RESULTS AND DISCUSSION
■
Synthesis of Authentic Reference Standard. Despite
the prospect of establishing an analytical method that does not
rely on the authentic compound as a calibrant, the availability of
an authentic standard remained indispensable for establishing
an analytical method with high specificity for the analyte, in this
case, 1. Because commercial samples of 1 were unavailable, 1
(bi)flavonoids as positive markers (unique botanical
was synthesized in-house. The method of synthesis was a
markers),18−21 HPLC has been the mainstay of quality control
modification of Harris’ method30 using p-toluenesulfonic acid
of ginkgo preparations, and accordingly LC has dominated the
for activation of pyridoxine via the intermediate o-quinone
analytical methodology applied to ginkgo. The development of
methide. While this reaction is per se reversible, excess MeOH
alternative, multitargeted analytical methods for the quality
pushes it to the product. The synthetic route is outlined in
control of G. biloba preparation has been the subject of
Scheme 1, and the yield of 1 was 56%. In order to assess the
efficiency of CS and solid-phase-based preparative purification
methods, preparative-TLC and silica gel column chromatog-
previous reports.22−24 The present study focuses on the low-
level detection of the toxic negative marker, 1, with subsequent
quantification performed by an off-line high-speed counter-
raphy (both using 5% MeOH in EtOAc) were compared with
1
current chromatography (HSCCC) and quantitative H NMR
HSCCC using ChMWat (10:5:5) using qHNMR, as
(qHNMR) analysis, respectively. Compared with solid-phase
summarized in Figure 1. The pH characteristics of the samples
and/or residual impurities induced a 0.08 ppm downfield shift
chromatographic methods, the particular advantage of HSCCC
as a liquid-only CS method is the avoidance of absorption or
degradation losses, as the sample recovery is in practice close to
100%.25 Moreover, NMR spectroscopy is a nondestructive and
near-universal detection method and can simultaneously
perform both qualitative and quantitative analyses.23,26,27
Hence, qHNMR was used initially for the determination of
the distribution constant (K value, K-by-NMR)23 for the
selection of an HSCCC solvent system (Figure 2).
Furthermore, the present study uses qHNMR for the
quantification of 1 in G. biloba extracts. The qHNMR approach
utilizes external calibration28 and does not rely on the
availability of high-purity 1 as an identical calibrant,28,29 as
would be the case for LC-based quantification methods. This
approach can be used more generally not only to detect and
establish threshold assays for botanical negative markers, but
also to quantify minor components or unwanted additives in
plant extracts, herbal medicines, food, and other complex
nature-derived commercial preparations. The lack of a
requirement for identical calibrants is advantageous for natural
products and other rare chemical entities, such as 1, which are
not readily available commercially.
of only the H-6 (s, 1H), resonating at δ 7.773 in the HSCCC
purification product, which showed the highest qHNMR purity
(98%, Figures 1 and 3). Although the effect of pH on the
pyridine derivative, 1, was a primary consideration, the chemical
shift effect was not observed in the sample neutralized by
NaHCO3 (data not shown), indicating that residual traces of,
for example, reactants and/or other impurities are able to affect
the exact chemical shift of H-6.
1
The H NMR spectrum in Figure 3c can function as a
reference standard for identity of the target marker, 1, and the
specificity of the subsequently developed qHNMR method.
The synthesized 1 was also used for determination of the best
CS solvent system to be used for the HSCCC procedure. Due
to the low concentration of 1 in G. biloba extracts, synthesized
1 was also employed for piloting the enrichment of 1 in assay
samples.
Targeted Purification by High-Speed Counter-Current
Chromatography. The generally useful estimation of solvent
systems (GUESS)31 method represents a rapid and practical
method to link CS K values and TLC retention characteristics
(Rf values). By using GUESS, a ChMWat system was chosen
C
dx.doi.org/10.1021/np400874z | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Table 1. K Values of the Target Compound, Ginkgotoxin (1), in the Shaking Trials Using Two CS Solvent System (SS) Families
Based on CHCl3 (CHCl3−MeOH−H2O, syn. ChMWat) and EtOAc (n-Hex−EtOAc−MeOH−H2O, syn. HEMWat), Which in
the GUESS System31 Correspond to the CHCl3- and EtOAc-Based TLC SSs, SSC and SSE, Respectively
TLC
HSCCC
trial no.
solvent system
volume ratio
Rf value
solvent system
volume ratio
K value
1
2
3
4
5
SSE 5
SSE 7
SSE 9
SSC 1
SSC 6
5:5
7:3
<0.05
<0.05
<0.05
0.31
HEMWat 0
HEMWat +4
HEMWat +7
ChMWat −3
ChMWat +2
5:5:5:5
3:7:4:6
1:9:1:9
10:0:10
10:5:5
0.06
0.33
0.29
8.03
1.26
9:1
100:0:0
85:15:0.5
0.47
that would deliver 1 to the “sweet spot” of the chromatogram.
Based on previous studies,31 compounds with Rf values
between 0.29 and 0.71 (optimal 0.5) will have K values
between 0.4 and 2.5 (optimal = 1) and be eluted in the sweet
spot of an HSCCC run. Table 1 summarizes all partition trial
results. Using TLC solvent systems (SSs) based on CHCl3−
MeOH−H2O (SSC, Table 1) as guidance, the Rf value of 1 in
the SSC6 is 0.47, which is around an optimal value (0.5). The
partition trials were also analyzed using the K-by-NMR
approach,23 to give a K value of 1 in ChMWat +2 of 1.26,
comfortably in the CS sweet spot between 0.4 and 2.5. The
systems HEMWat 0, +4, and +7 as well as ChMWat −3 were
also tested. A partition trial using ChMWat +2 is shown in
Figure 2 as an example of a K-by-NMR determination. The K
value was calculated by eq 2 (Experimental Section).
of HSCCC as a means of chemical subtraction is clearly
demonstrated in Figure 3, which shows the difficulty in
quantifying 1 in a seed extract sample (Figure 3a), compared
with the ready quantification after a single-step HSCCC
enrichment (Figure 3b). Using the qHNMR results, the CS
enrichment factor achieved for 1 was 282, which in our
experience is a remarkably high value. The molarity-based
quantification of 1 was achieved using caffeine as external
calibrant (EC) and the residual solvent signal (CHCl3 in
CDCl3) for a combined external/internal calibration (ECIC;
Table 2; see also ref 28 for details about qHNMR calibration).
Table 2. Integral Values of the qHNMR Quantification of 1
in G. biloba Seed Extract Using the Combined Internal/
External Calibration with Caffeine and the Residual Solvent
Signal
HSCCC has previously been employed to effect chemical
subtraction of a single target compound from a crude plant
extract.32 The method can work in a single step and with high
selectivity. Accordingly, the recommended working range for
the HSCCC chemical subtraction is 1< K < 5. Thus, a K value
of 1.26 indicates that the compound is an applicable target for
this procedure. Indeed, this K value was used to perform a pilot
HSCCC run and determine the congruence between predicted
elution volume and the actual fraction(s) containing 1. Under
experimental conditions, 1 was eluted in a fraction with a K
value centered at 1.47. Considering potential pH effects in
caffeine
normalized integral
1 in HSCCC fraction 6 of seed extract
δ
δ
normalized integral
a
b
b
b
b
a
b
b
b
b
b
1
2
3
4
5
7.260
7.501
3.992
3.588
3.412
1.00
1
2
3
4
5
6
7.260
7.947
4.908
4.622
3.537
2.490
1.00
1.04 (1H)
3.22 (3H)
3.23 (3H)
3.31 (3H)
0.10 (1H)
0.20 (2H)
0.20 (2H)
0.35 (3H)
0.31 (3H)
a
b
Caffeine. Residual solvent signal.
1
crude extract samples (see also the above discussion of H
NMR data), this matched the predicted elution behavior well.
According to this result, as long as the content of 1 is over the
limit of detection, enrichment of 1 could be performed in a
single HSCCC run. This also complemented the outcome of
the trial of three different preparative purification methods for
the synthetic material of 1 (Figure 1).
All data were obtained based on NMR peak area (integration).
From the qHNMR analysis, the method specificity for 1
(structural ID) is clearly confirmed, and the content of 1
contained in the investigated commercial seeds was shown to
be 59 μg per seed. This matches well with a previous report,15
according to which one G. biloba seed contains about 80 μg of
1. The commercial leaf sample used in the present study
showed no detectable qHNMR signal in the extract (Figure 4)
nor in the combined HSCCC fractions that represented the
target K value range from 1.4 to 1.7. Meanwhile, both spiked
and unspiked samples were extracted using the USP standard
method and analyzed by qHNMR. Results from the spiked
sample indicate that 1, if contained in ginkgo leaf extract, does
indeed elute in the target HSCCC fractions. However, 1 could
not be detected in a ginkgo leaf extract prepared by using the
USP standard method, which eliminated the possibility that the
extraction method has a significant effect on the present
method. Hence, it can be concluded that the content of 1 in the
investigated sample of ginkgo leaves was less than 9 ppm,
corresponding to the LOD.
The K value of a compound in any two-phase SS is a physical
constant, which is specific to that compound and that solvent
system.17 One additional consideration is that particularly
natural products extracts typically contain complex chemical
matrixes, which together with varying sample concentrations
(see Figure 5 in ref 33) and pH effects can affect the actual
partition behavior in a CS experiment via, for example, solvent
strength modification of solutes and analyte−analyte inter-
action. However, the K value of 1 in an HSCCC enrichment
procedure with G. biloba extracts should still be around 1.47.
This was confirmed by an HSCCC separation of seed extract
(Figure 3), in which 1 was indeed enriched in a pooled fraction
centered at a K value of 1.62. On the basis of these results, the
K range from 1.4 to 1.7 was chosen as a target for the collection
of HSCCC fractions of the leaf extract for subsequent
quantification by qHNMR.
Ginkgotoxin Levels in Ginkgo Preparations. A review
of the reported cases of convulsion induced by 1 found that the
range of serum concentrations in affected patients who typically
consumed three daily doses of 1−2 mL of ginkgo preparation
Quantitative 1H NMR Analysis. CDCl3 was selected as the
solvent for all qHNMR analyses. Using this method, the power
D
dx.doi.org/10.1021/np400874z | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 4. Results of the qHNMR analyses (400 MHz, in CDCl3) of a commercial G. biloba leaf sample: (a) crude methanolic extract; (b) enriched
HSCCC fraction of (a), corresponding to the target K value range of 1.4 to 1.7; (c) G. biloba extract prepared using the USP and spiked with 2 mg of
1 and subject to the same HSCCC enrichment as in (b); (d) method for the same extract as in (c), but unspiked and also subjected to the HSCC
enrichment as for (b). Shown are the entire spectra (1) and representative expansions (2) for the signals of 1. The chemical shifts of 1 are bracketed
in red.
was between 90 and 484 ng/mL.2 Accordingly, a reasonable
approximation of a safety threshold for 1 that would likely
avoid lethality can be calculated as follows:
qHNMR sensitivity (ii), there are numerous options: possibly
the foremost is to increase the number of scans (ns) from the
ns = 16 used presently, keeping in mind that the sensitivity
increase is proportional to the square root of the ns increase.
Other options include the use of higher magnetic fields (e.g.,
500 MHz to 1 GHz) and cryoprobe and/or microcryoprobe
technology.35 In summary, for example, the use of ns = 256 on
a 600 MHz 5 mm cryoprobe-equipped instrument will result in
a 10- to 20-fold or better overall sensitivity increase. Together
with the potentially increased CS loading capacity, this will
enhance the overall sensitivity by >100-fold to an LOD of 0.3
μg, equivalent to detecting 0.09 ppm (LOD) of 1 in analytical
liquid ginkgo preparation. This shows that the method can be
adapted even to practically unimportant levels of 1 in
commercial liquid ginkgo preparations if needed.
(1)
m = cV
where m is the mass of intake of 1 by the patient, c is the serum
concentration, and V is the blood volume in an adult (averaging
around 5 L). Although the threshold level for 1 required to
induce seizures is still unclear, a blood level of 90 ng/mL could
potentially be reached by the consumption of as few as 8 ginkgo
seeds or an equivalent preparation thereof. In a survey of
commercial liquid ginkgo preparations sold in Europe, the
detected amounts of 1 ranged from 0.01 to 9.77 μg/mL.10 The
survey further showed that the content in seeds and leaves
depends on the time of the year of harvest and that widely used
ginkgo preparations are not necessarily ginkgotoxin-free
products.
The present study employed a 20 mL HSCCC instrument
equipped with a 500 μL sample loop, which was used to inject
the extract from ∼1 g of crude ginkgo preparation. The
qHNMR spectrum used for the quantification of 1 in the
ginkgo seed sample exhibited an average signal-to-noise (S/N)
value of the target signals of 39:1, and this was achieved with
the basic configuration of a 400 MHz NMR instrument
equipped with a 5 mm RT probe and acquiring 16 transients
(ns = 16). Considering the result of our recent qHNMR
validation study,26 the LOD of the present off-line HSCCC−
qHNMR method at 9 ppm corresponds to a qHNMR S/N of
3:1. Accordingly, as the LOQ is 3.3 times the LOD, this basic
qHNMR procedure already works quantitatively at 30 ppm and
above. This method can readily be adapted to accommodate a
much lower level for both LOD and LOQ, and this can be
achieved by two straightforward means: (i) increase of HSCCC
sample loading; (ii) increase of qHNMR sensitivity. Regarding
(i), the use of abundant CS instruments with 200−300 mL
total volume will enlarge the sample loading capability by 10- to
15-fold compared to the present setup. This takes advantage of
the fully predictable scale-up capability of CS.34 With respect to
CONCLUSION
■
To summarize, the present study established an off-line
HSCCC−qHNMR detection method for determining the
ginkgotoxin (1) content in ginkgo preparations. The key
aspects of this methods are as follows: (i) using a modified
procedure, an authentic standard of 1 was synthesized and
characterized; (ii) The GUESS method was applied to rapidly
select an efficient CS solvent system, and as a proof of principle
for the suitability of the GUESS prediction, the K value of 1 was
rapidly measured using the K-by-NMR approach. The
compound was then successfully enriched by HSCCC of an
extract of ginkgo seeds, which are known to contain 1; (iii)
quantification of 1 was performed by qHNMR in CDCl3
solution, using the spectroscopic characteristics of synthetic 1
as reference for identity and assay specificity, and by applying
combined external/internal calibration with high-purity caffeine
as EC and the residual solvent signal as IC.
This methodology could have broader impact in various
fields: generally and with regard to (i), it facilitates access to
authentic reference material of 1 by synthesis and subsequent
CS purification, which could be useful for LC quality control
E
dx.doi.org/10.1021/np400874z | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
area of the target compound in the lower phase. An example is given in
Figure 2.
purposes. Considering the recent availability of automated CS
instrumentation, the enrichment−detection approach (above
points ii and iii) could potentially be employed for exploiting
threshold levels of 1 in body fluids related to seizures and for
the determination of 1 content in commercial products. In
more generalized terms, this approach can potentially be
transferred to other low-level metabolites and can be used to
detect ppm levels of any target compound from a natural or
commercial source. Finally, the case of 1 shows the enrichment
potential of CS and its power for chemical subtraction32 and
targeted isolation of toxins and other bioactive principles of
interest.
High-Speed Counter-Current Chromatography. CS was
performed using a TBE-20A HSCCC (20 mL, Tauto Biotech,
China), equipped with a 500 μL sample loop, pump, UV detector,
thermostatic circulator, and data processing system (Cherry One,
Cherry Instruments, Chicago, IL, USA). The procedure was as follows:
100 mg of extract was dissolved in 500 μL of ChMWat (10:5:5) made
up from equal parts of the two phases. The solution was loaded into
the injection loop. Then, the column was rotated at 800 rpm and filled
with UP as the stationary phase. The LP was pumped as the mobile
phase in a head-to-tail direction at a flow rate of 0.5 mL/min, and the
column was rotated at 2000 rpm.36 After the hydrodynamic
equilibrium between the two phases was reached, i.e., the eluate had
changed from stationary to mobile phase, the sample solution was
injected through a six-port valve. Throughout the separation, the K
values were monitored by the Cherry One system. Outflow was
collected by a Foxy Jr. fraction collector (Teledyne Isco, Lincoln, NE,
USA) and matched to these K values.
EXPERIMENTAL SECTION
■
Materials and Reagents. Whole G. biloba seeds (commercial
name: White Nut) were purchased in a local grocery store of
Chinatown, Chicago (origin of material: Hongchanglong, China).
Powdered G. biloba leaves were sourced from Mountain Rose Herbs
(lot no.: M10579, Eugene, OR, USA). Both samples were stored at
−20 °C until analyzed. Pyridoxine hydrochloride, caffeine (>99%), and
CDCl3 (99.8 atom % D) were obtained from Sigma-Aldrich Inc. (St.
Louis, MO, USA). All solvents were ACS grade, purchased from
Pharmco-AAPER (Crookfield, CT, USA), and were re-distilled before
use.
General Instrumentation. HRESIMS was performed on a Waters
Synapt mass spectrometer (Waters, USA). NMR measurements were
performed on a Bruker DPX-400 spectrometer (Karlsruhe, Germany)
in 5 mm NMR tubes (Norell, Landisville, NJ, USA), using Bruker
standard pulse sequences at 25 °C (298 K).
Quantitative 1H NMR (qHNMR) Analysis. The samples were
dissolved in 600 μL of CDCl3 using an analytical syringe (Valco
Instruments, Baton Rouge, LA, USA). A total of 16 scans (ns) were
acquired, collecting 64 k of time domain data and using a 30 degree
excitation pulse as well as a relaxation delay (D1) of 30 s. The spectra
were analyzed using MestReNova v6.2.1−7569 (Mestrelab Research,
Santiago de Compostela, Spain) software. Line resolution was
improved by applying a Gaussian−Lorentzian window functions
(GB 1.0, LB −0.3) and zero-filling to 256 k prior to Fourier
transformation of the FID. Baseline correction used a fifth-order
polynomial function, and phase correction was done manually.
Extraction. The outer shells of the seeds were peeled off, and the
meat was dried at 37 °C, then powdered by trituration, and stored at
−20 °C. Extraction was performed according to ref 10 briefly as
follows: 100 g of the powder was transferred to a 1 L flask, 500 mL of
freshly distilled MeOH was added, and the mixture was shaken
horizontally at low speed for 1 h. The extract was filtered, and each
extraction was performed three times. The combined extract solutions
were concentrated on a rotary evaporator at 30 °C under vacuum. The
concentrated sample was stored at −20 °C. Leaf extractions were
performed either in the way mentioned above or by the USP standard
method.
Synthesis of Ginkgotoxin (syn. 4′-O-methylpyridoxine, 1).
The method of Harris et al.30 was modified in order to produce 1.
Pyridoxine hydrochloride (100 mg) and p-toluenesulfonic acid (4 mg)
were combined in MeOH (1 mL) and heated at 110 °C (sealed) for 3
h in a microwave. The CEM Explorer 48/72/96 automated microwave
synthesizer was used for microwave-heated and sealed reactions,
controlled by an external computer loaded with the Synergy
application software (version 1.1). The solution was cooled, the
mixture filtered, and the solvent evaporated in vacuo. Aliquots of the
crude residue were purified by preparative-TLC (5% MeOH in
EtOAc), silica column (5% MeOH in EtOAc), and HSCCC
(ChMWat, 10:5:5, v/v/v). The structural characterization of 1 was
ASSOCIATED CONTENT
■
S
* Supporting Information
1
1
NMR spectra (1D H and 13C NMR and 2D H,13C-HMBC)
for 1 in DMSO-d6. This material is available free of charge via
the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: gfp@uic.edu. Tel: (312) 355-1949. Fax: (312) 355-
2693.
■
Notes
1
1
based on NMR (1D H and 13C NMR and 2D H,13C-HMBC in
DMSO-d6; see Supporting Information) and HRESIMS. Related
physical properties are listed below.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Ginkgotoxin (1): 1H NMR (400 MHz, CDCl3) δ 7.773 (s, 1H, H-
6), 4.895 (s, 2H, H-5), 4.569 (s, 2H, H-4), 3.518 (s, 3H, H-4), 2.442
(s, 3H, H-2); HRESIMS m/z 184.08954 [MH]+ (calcd for C9H13NO3,
183.08956).
The vital support of the team at Cherry Instruments, Chicago
(IL), in particular by Mr. S. Pro and W. Friedel, is gratefully
acknowledged. The authors are also grateful to Dr. C.
Erdelmeier, of Dr. Wilmar Schwabe Pharmaceuticals, Karlsruhe
(Germany), for his kind support.
Selection of Solvent Systems for Countercurrent Separa-
tion. Five biphasic solvent systems from the HEMWat and ChMWat
SS families31 (Table 1) were assayed using partition shaking trials. To
confirm an optimal K value for the HSCCC experiment, the following
procedure was used: in a separatory funnel, 5 mg of 1 was added to 10
mL of a mixture containing equal volumes of the upper and lower
phases of the two-phase SS, and the mixture was shaken adequately to
equilibrate the sample. The two phases, UP and LP, were separated
and evaporated.
DEDICATION
■
Dedicated to Prof. Dr. Otto Sticher of ETH-Zurich, Zurich,
Switzerland, for his pioneering work in pharmacognosy and
phytochemistry.
REFERENCES
■
Kvalues = Aupper phase/Alower phase
(1) Leistner, E.; Drewke, C. J. Nat. Prod. 2010, 73, 86−92.
(2) Wada, K. Jpn. J. Toxicol. 2005, 18, 11−16.
(3) Yoshikawa, T.; Naito, Y.; Kondo, M. Antioxid. Redox Signaling
1999, 1, 469−480.
(2)
where Aupper phase is the average NMR peak area of the target
compound in the upper phase and Alower phase is the average NMR peak
F
dx.doi.org/10.1021/np400874z | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
(4) Jones, D. C.; Alphey, M. S.; Wyllie, S.; Fairlamb, A. H. Mol.
Microbiol. 2012, 86, 51−64.
(5) Wada, K.; Ishigaki, S.; Ueda, K.; Sakata, M.; Haga, M. Chem.
Pharm. Bull. 1985, 33, 3555−3557.
(6) Gregory, P. J. Ann. Intern. Med. 2001, 134, 344.
(7) U.S. National Library of Medicine Ginkgo. http://www.nlm.nih.
gov/medlineplus/druginfo/natural/333.html (11/28/2013).
(8) Bent, S. J. Gen. Int. Med. 2008, 23, 854−859.
(9) Wada, K.; Sasaki, K.; Miura, K.; Yagi, M.; Kubota, Y.; Matsumoto,
T.; Haga, M. Biol. Pharm. Bull. 1993, 16, 210−212.
(10) Arenz, A.; Klein, M.; Fiehe, K.; Gross, J.; Drewke, C.;
Hemscheidt, T.; Leistner, E. Planta Med. 1996, 62, 548−551.
(11) Wada, K.; Haga, M. Ginkgo biloba, a Global Treasure: from
Biology to Medicine; Springer: Tokyo, 1997.
(12) Granger, A. S. Age Ageing 2001, 30, 523−525.
(13) Kupiec, T.; Raj, V. J. Anal. Toxicol. 2005, 29, 755−758.
(14) Hasegawa, S.; Oda, Y.; Ichiyama, T.; Hori, Y.; Furukawa, S.
Pediatr. Neurol. 2006, 35, 275−276.
(15) Kajiyama, Y.; Fujii, K.; Takeuchi, H.; Manabe, Y. Pediatrics 2002,
109, 325−327.
(16) Salamon, N.; Gurgui, C.; Leistner, E.; Drewke, C. Planta Med.
2009, 75, 563−567.
(17) Pauli, G. F.; Pro, S.; Friesen, J. B. J. Nat. Prod. 2008, 71, 1489−
1508.
(18) Hasler, A.; Gross, G. A.; Meier, B.; Sticher, O. Phytochemistry
1992, 31, 1391−1394.
(19) Hasler, A.; Sticher, O.; Meier, B. J. Chromatogr. 1990, 508, 236−
240.
(20) Hasler, A.; Sticher, O.; Meier, B. J. Chromatogr. 1992, 605, 41−
48.
(21) Sticher, O. Planta Med. 1993, 59, 2−11.
(22) Napolitano, J. G.; Godecke, T.; Rodriguez Brasco, M. F.; Jaki, B.
̈
U.; Chen, S.-N.; Lankin, D. C.; Pauli, G. F. J. Nat. Prod. 2012, 75,
238−248.
(23) Qiu, F.; Friesen, J. B.; McAlpine, J. B.; Pauli, G. F. J. Chromatogr.
A 2012, 1242, 26−34.
(24) Napolitano, J. G.; Lankin, D. C.; Chen, S.-N.; Pauli, G. F. Magn.
Reson. Chem. 2012, 50, 569−575.
(25) Ito, Y. Nature 1987, 326, 419−420.
(26) Pauli, G. F.; Jaki, B. U.; Lankin, D. C. J. Nat. Prod. 2005, 68,
133−149.
(27) Pauli, G. F.; Jaki, B. U.; Godecke, T.; Lankin, D. C. J. Nat. Prod.
̈
2012, 75, 834−851.
(28) Godecke, T.; Napolitano, J. G.; Rodriguez Brasco, M. F.; Chen,
S.-N.; Jaki, B. U.; Lankin, D. C.; Pauli, G. F. Phytochem. Anal. 2013, 24,
581−597.
̈
(29) Schoenberger, T. Anal. Bioanal. Chem. 2012, 403, 247−254.
(30) Harris, S. A. J. Am. Chem. Soc. 1940, 62, 3203−3205.
(31) Friesen, J. B.; Pauli, G. F. J. Liq. Chromatogr. Relat. Technol.
2005, 28, 2877−2806.
(32) Chen, S.-N.; Turner, A.; Jaki, B.; Nikolic, D.; van Breemen, R.
B.; Friesen, J. B.; Pauli, G. F. J. Pharm. Biomed. Anal. 2008, 46, 692−
698.
(33) Friesen, B.; Pauli, G. F. J. Chromatogr. A 2009, 1216, 4225−
4231.
(34) Sutherland, I. A.; Hewitson, P.; Ignatova, S. J. Chromatogr. A
2009, 1216, 4201−4205.
(35) Hilton, B. D.; Martin, G. E. J. Nat. Prod. 2010, 73, 1456−1469.
(36) Ito, Y. J. Chromatogr. A 2005, 1065, 145−168.
G
dx.doi.org/10.1021/np400874z | J. Nat. Prod. XXXX, XXX, XXX−XXX