The human pyridoxal kinase, a plausible target for
ginkgotoxin from Ginkgo biloba
Uta Ka
¨
stner
1
, Christian Hallmen
2
, Michael Wiese
2
, Eckhard Leistner
1
and Christel Drewke
1
1 Institut fu
¨
r Pharmazeutische Biologie, Universita
¨
t Bonn, Germany
2 Pharmazeutisches Institut, Pharmazeutische Chemie, Endenich, Bonn, Germany
4¢-O-methylpyridoxine (MPN, ginkgotoxin; Fig. 1) is a
neurotoxic compound that causes severe neuronal dis-
orders in mammals after ingestion. Symptoms of this
poisoning called ‘gin-nan sitotoxism’ are mainly epilep-
tic convulsions, paralysis of the legs and loss of con-
sciousness [1]. There are even reports of death due to
overconsumption of Ginkgo seeds, which are the main
source of ginkgotoxin [1]. In addition to the seeds,
which accumulate the toxin, ginkgotoxin has also been
found in the leaves of Ginkgo biloba as well as in rem-
edies produced from leaf extracts [2,3]. These remedies

Keywords
Ginkgo biloba; ginkgotoxin; pyridoxal kinase;
c-aminobutyric acid; pyridoxal phosphate
Correspondence
C. Drewke, Institut fu
¨
r Pharmazeutische
Biologie, Rheinische Friedrich-Wilhelms-
Universita
¨
t Bonn, Nussallee 6, 53115 Bonn,
Germany
Fax: +49 228 733250
Tel. +49 228 732563
E-mail: [email protected]
Website: http://www.uni-bonn.de/pharmbio/
(Received 5 September 2006, revised
11 December 2006, accepted 15 December
2006)
doi:10.1111/j.1742-4658.2007.05654.x
Ginkgotoxin (4¢-O-methylpyridoxine) occurring in the seeds and leaves of
Ginkgo biloba, is an antivitamin structurally related to vitamin B
6
. Ingestion
of ginkgotoxin triggers epileptic convulsions and other neuronal symptoms.
Here we report on studies on the impact of B
6
antivitamins including ginkgo-
toxin on recombinant homogeneous human pyridoxal kinase (EC 2.7.1.35).
It is shown that ginkgotoxin serves as an alternate substrate for this enzyme

2
OH
N
OH
R
CH
2
O P
N
OH CH
2
O P
CHO
N
OH CH
2
OH
CHO
N
CH
2
OHOH
CHO
PM or PN PMP or PNP
R = CH
2
NH
2
or CH
2

Fig. 2. Reactions potentially affected by B
6
antivitamins (ginkgotoxin, deoxypyridoxine). (1,4), Pyridoxal kinase; (2), pyridoxine ⁄ pyridoxamine
phosphate oxidase; (3), pyridoxal phosphatase; (5), glutamate decarboxylase.
Fig. 1. B
6
antivitamins ginkgotoxin (phosphate) and 4¢-deoxypyridoxine (phosphate).
U. Ka
¨
stner et al. Influence of ginkgotoxin on human pyridoxal kinase
FEBS Journal 274 (2007) 1036–1045 ª 2007 The Authors Journal compilation ª 2007 FEBS 1037
cases of intoxication with ginkgotoxin, the decreased
availability of c-aminobutyric acid was assumed to be
caused by an inhibitory effect of the toxin on one or
both of the GAD isoenzymes due to the structural fea-
tures of ginkgotoxin [12].
To clarify these assumptions, the influence of gin-
kgotoxin on both GAD isoenzymes was examined at
the enzymatic level [8]. Although a significant decrease
in GAD
65
activity, to  35% of initial values, was
observed when GAD was incubated with 4¢-O-methyl-
pyridoxine 5¢-phosphate (ginkgotoxin phosphate,
MPNP; Fig. 1), the concentration of ginkgotoxin phos-
phate at which this inhibition took place (IC
50
¼
2.7 mm) probably is too high to be reached under phy-
siological conditions. Thus, the GAD isoenzymes could

catalysed by pyridoxal kinase (Fig. 2A) [7,14]. As a
consequence, there is a requirement for ubiquitous
expression of the kinase in mammalian tissues [14,15].
In the case of inhibition of the enzyme by ginkgotoxin
(phosphate) the availability of cofactor not only for
GAD, but also for all other PLP-dependent reactions
involved in neurotransmitter and amino acid metabo-
lism would be decreased resulting in a total physiologi-
cal imbalance of metabolism, very likely accompanied
by diverse pathological symptoms. Thus, human pyrid-
oxal kinase (PKH) obviously plays a crucial role in the
regulation of PLP homoeostasis. To elucidate the role
of this important enzyme and the mode of action of
ginkgotoxin in more detail, we studied recombinant
PKH as a possible target for ginkgotoxin.
Results
Properties of PKH
PKH catalyses the conversion of PL, PN and PM
(Fig. 2A) to the respective 5¢-phosphate esters using
ATP as a cofactor [15,16]. To examine the mode of
action of ginkgotoxin on PKH, we overexpressed and
purified the enzyme leading to a homogeneous protein
of the expected molecular mass as proven by
SDS ⁄ PAGE (data not shown) and by MALDI-TOF
spectroscopy (see Supplementary material).
PKH was characterized with respect to its biochemi-
cal properties. The enzymatic activity was stable at
)20 °C for  21 days. The enzyme was only active in
the presence of ATP. In contrast, no activity could be
detected with GTP.

mental procedures). The kinetic data given in Table 1
reveal that among all vitamin B
6
derivatives tested, the
antivitamin DPN is the substrate with the highest maxi-
mum velocity (2.62 · 10
)6
nmolÆmg
)1
Æmin
)1
) and the
highest turnover number (k
cat
¼ 1.535 s
)1
), whereas the
Influence of ginkgotoxin on human pyridoxal kinase U. Ka
¨
stner et al.
1038 FEBS Journal 274 (2007) 1036–1045 ª 2007 The Authors Journal compilation ª 2007 FEBS
antivitamin ginkgotoxin exhibits the highest affinity
(lowest K
m
¼ 4.95 · 10
)6
m) towards the PKH enzyme
(Table 1). This shows that the substituent at C-4¢ of the
pyridine ring system plays an important role in deter-
mining the kinetic features of PKH. However, the data

0.460 s
)1
) (Fig. 3A). Coincubation of PL and ginkgo-
toxin, both at a concentration of 0.025 mm, however,
reversed the velocity of phosphorylation of both sub-
strates with PL being an almost inactive substrate
within 14 min of start of the reaction (Fig. 3B). Coin-
cubation of both substrates with a reduced concentra-
tion of ginkgotoxin (0.0125 mm), compared with PL
(0.025 mm) still gave initially a faster phosphorylation
of ginkgotoxin than of PL (Fig. 3C). Note that in this
experiment a shorter incubation time (< 2 min) was
not possible for technical reasons (Experimental proce-
dures).
In these experiments the incubation mixtures were
analysed by HPLC. Supporting evidence for a relat-
ively fast conversion of ginkgotoxin was obtained
using the optical assay which detects PLP alone. When
PL was kept constant (0.05 mm), and increasing con-
centrations of ginkgotoxin (0–0.25 mm) were added,
formation of PLP by PKH was more and more
delayed depending on the concentration of ginkgotoxin
(Fig. 4A). However, after an initial ‘lag-phase’, PLP is
formed with the same velocity as in the control with-
out ginkgotoxin. With high concentrations of ginkgo-
toxin (0.25 mm) formation of PLP is suppressed
completely during the incubation period (Fig. 4A).
For comparison, we performed the same assay with
increasing concentrations of DPN [5] (Fig. 1). The
presence of DPN revealed a clear decrease in PLP for-

0.0125 m
M. Reactions were monitored using HPLC. The data are the mean of two independent experiments.
Table 1. Kinetic data of different B
6
vitamers and antivitamins accepted as substrates by PKH.
Compound Km(
M) V
max
(nmolÆmg
)1
Æmin
)1
) k
cat
(s
)1
) k
cat
⁄ K
m
(mol
)1
Æs
)1
) K
i
(M)
Pyridoxal 5.87 ± 0.28 · 10
)5
1.70 ± 0.36 · 10

0.460 9.30 · 10
4
4.14 · 10
)7
U. Ka
¨
stner et al. Influence of ginkgotoxin on human pyridoxal kinase
FEBS Journal 274 (2007) 1036–1045 ª 2007 The Authors Journal compilation ª 2007 FEBS 1039
Evidently, PLP formation by pyridoxal kinase follows
different kinetics in the presence of ginkgotoxin when
compared with DPN.
The inhibitor constants (Table 1) for both antivita-
mins as determined using the optical method revealed
a significantly lower K
i
for ginkgotoxin (4.14 · 10
)7
m)
than DPN (5.74 · 10
)5
m). It should be noted that the
K
i
value for ginkgotoxin was determined for the initial
linear range of the plot (Fig. 4A).
Enzymatic assays with a constant concentration of
ginkgotoxin (0.2 mm) and variable concentrations of
PL (0.05–1.00 mm) demonstrated that the inhibitory
effect of ginkgotoxin on PKH can be alleviated. With
increasing concentrations of PL, PLP was formed

regions of the substrate and cofactor binding domain
of PKH were modelled (see Experimental section).
0
20
40
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
time [min]
formation of pyridoxal-5'-phosphate
[nmol/ml]
without MPN
MPN 0.010 m
M
MPN 0.015 mM
MPN 0.020 mM
MPN 0.025 mM
MPN 0.035 mM

0
10
20
30
without DPN
DPN 0.01 mM
DPN 0.05 mM
DPN 0.10 mM
DPN 0.20 mM
DPN 0.30 mM
C
0
20
40
60
time [min]
0246810
formation of pyridoxal-5'-phosphate
[nmol/ml]
without DPN
PL 0.05 m
M
PL 0.10 mM
PL 0.05 mM
PL 0.04 mM
PL 0.03 mM
PL 0.02 mM
PL 0.01 mM
D
Fig. 4. Reversible inhibition of PKH by ginkgotoxin and 4¢-deoxypyridoxine during formation of pyridoxal phosphate. (A) Inhibition by increas-

brain. Decreased GAD activity in turn leads to an
imbalance between excitatory and inhibitory neuro-
transmission, which may result in epileptic convulsions
[18]. In this context, PKH appears to be a plausible
target for the antivitamin ginkgotoxin, which was
shown to trigger epileptic seizures in mammalia [1].
To date, the effect of ginkgotoxin on pyridoxal kin-
ase had been studied only with a partially purified
homogenate from mouse brain [5]. Detailed experi-
ments on the mode of inhibition of the human enzyme
by ginkgotoxin are lacking. This study shows for the
first time an enzymatic conversion of ginkgotoxin to
ginkgotoxin phosphate and of DPN to DPNP by
homogeneous human pyridoxal kinase. Because all
three physiological B
6
vitamers are also converted by
the purified enzyme in our assay, the term ‘pyridoxal
kinase’ may be changed to ‘PN ⁄ PL ⁄ PM kinase’ in
agreement with the term suggested for the PN, PL and
PM converting enzyme in Escherichia coli [19].
The conversion of ginkgotoxin and DPN to ginkgo-
toxin phosphate and DPNP, respectively, demonstrates
that both antivitamins, like the vitamers, are substrates
of the enzyme. Thus, the kinase acts on two competing
substrates, when PL and one of the two antivitamins
are simultaneously employed in the enzymatic assay.
This is evident from Fig. 4B,D, which show that the
inhibitory effect of ginkgotoxin and DPN on PLP for-
mation can be alleviated by increasing amounts of PL.

mol
)1
Æs
)1
) are phos-
phorylated more efficiently than PL (1.70 · 10
4
mol
)1
Æs
)1
) and PM (6.28 · 10
3
mol
)1
Æs
)1
) by PKH. This
reflects the antivitamin character of both compounds
and explains a depletion of cofactor formation in the
organism in the presence of the antivitamins.
The kinetic data (Table 1) are a reflection of the
structural features of ginkgotoxin, PL and PKH. LogP
values determined for PL ()1.182) and ginkgotoxin
()0.299) show a higher lipophilicity for ginkgotoxin in
comparison with the natural substrate. This is in agree-
ment with the molecular structure of ginkgotoxin,
which because of its 4¢-O-methyl group is more hydro-
phobic in comparison with the B
6

patient). According to another publication, several
other cases of seizure associated with Ginkgo have
been reported [24]. The authors of these case studies
assume that the occurrence of the epileptic convulsions
was due to the ingestion of Ginkgo remedies. However,
they neither specify the respective remedies, nor do
they mention their composition. Therefore, the possi-
bilities of overdosage or interactions with other medi-
cations cannot be excluded. However, Ginkgo biloba
extracts have been reported to have a proconvulsive
activity on chinchilla rabbits [25], and it should also be
mentioned that the action of GABAergic antiepileptica
was negatively influenced, when seizures were induced
in mice by various toxins [26]. Thus, the question ari-
ses whether the occurrence of seizures really could be
due to the presence of ginkgotoxin when Ginkgo rem-
edies are ingested, at least by predisposed patients. An
explanation may be derived from a comparison of the
vitamin B
6
concentration in human blood and the gin-
kgotoxin concentration in human blood after ingestion
of Ginkgo remedies. The maximum daily intake of gin-
kgotoxin in remedies based on Ginkgo extract was cal-
culated to 58.62 lg [2]. Accordingly, the maximum
concentration of ginkgotoxin in human plasma calcula-
ted for 6 and 4 L of blood should be in a range of
 53.33–80.24 nm, provided that the toxin is distri-
buted exclusively in the blood. This is in the same
order of magnitude for vitamin B

B
–
m
B
–
) gal dcm (DE3)] was then transformed with the
resulting recombinant vector pET11a-PKH. The recombin-
ant strain BL21 (D3) (pET11a-PKH) was grown in Luria–
Bertani medium containing penicillin G (100 lgÆmL
)1
)at
37 °C until D
600
¼ 0.5. Isopropyl thio-b-d-galactoside was
added to a final concentration of 1.0 mm, and the culture
was incubated with shaking for 24 h. Protein expression was
confirmed by SDS ⁄ PAGE [29]. The cell pellet derived from
1 L of culture was resuspended in 10 mL of column buffer
(20 mL of 1 m Tris ⁄ HCl, pH 7.4; 11.7 g NaCl, 2 mL of
0.5 m EDTA ad 1 L) and frozen overnight at )20 °C. The
frozen bacterial cells were thawed in a cold water bath before
ultrasonic treatment (Branson Sonifier, Danbury, MA, 10·,
10 s, 50% output at stage 5). After sedimentation of the cell
debris (30 min, 9000 g,4°C) the supernatant was treated as
described below.
Purification of PKH
Cell-free protein extract derived from 6 L of culture was
adjusted to 100 mm KCl and further successively subjected
to affinity chromatography (matrix: pyridoxyl–EAH–Seph-
arose

many). The matrix was prepared as follows: 10 mg of sinapi-
nic acid are vortexed in 1 mL of 60 : 40 (v ⁄ v) water ⁄
acetonitrile containing 0.1% trifluoroacetic acid. The method
was performed under reflection in the positive mode.
Determination of the logP values for ginkgotoxin
and PL
The determination of the logP values was performed
according to OECD guidelines for the testing of chemicals,
partition coefficient (n-octanol ⁄ water): Shake Flask Method
(adopted by the council on 27 July 1995).
Enzyme incubation
All enzyme incubations carried out to determine activity of
pyridoxal kinase were performed in 70 mm potassium phos-
phate buffer. The pH was adjusted to the optimum
(pH 6.2) as determined for the enzyme. The total volume of
the incubation samples was 1 mL containing 10 lL each of
a10mm ZnCl
2
stock solution and a 0.1 m ATP stock solu-
tion. The general incubation temperature was 37 °C. To
determine the temperature and pH optima, reactions were
performed in a range from 20 to 60 °C and from pH 5.0 to
8.0, respectively.
To determine K
m
values, 5 lg of purified PKH were incu-
bated for 3 min with PL (0.005–0.5 mm), for 1 min with PN
(0.005–0.1 mm), for 1 min with PM (0.05–0.4 mm), for
1 min with DPN (0.015–0.1 mm) and for 30 s with ginkgo-
toxin (0.005–0.025 mm), respectively. Incubation was ter-

with a resolution of 2.5 A
˚
. The 3D structure of ginkgotoxin
was generated using the sybyl sketch module (Tripos Inc.,
St. Louis, MO), with subsequent force field minimization
(MMFF94s force field; MMFF94 charges; termination cri-
terion: gradient 0.005 kcalÆ(mol*A)
)1
. Docking experiments
were performed using gold 3.0.1 (Cambridge Crystallo-
graphic Data Centre, Cambridge, UK). The active site was
defined as all protein atoms within 10 A
˚
distance from the
sidechain oxygen of Ser12. The number of GA runs was set
to 30, otherwise the standard default settings were used. In
the initial docking results the 4¢-O-methyl group of ginkgo-
toxin was located near Gly20. Comparing this result with
the crystal structure of PL cocrystalized with sheep pyrid-
oxal kinase [17] (PDB-ID: 1RFU), ginkgotoxin was turned
through 180 °. These first results make no sense, because in
this way it is not possible to phosphorylate the hydroxyl
group at position 5 of MPN as experimentally observed
(see below). Therefore we set distance constraints for the
docking algorithm. The distance between the carbon of the
2-methyl group and the C
b
-atom of Val41 was constrained
to be between 5 and 7 A
˚

¨
stner et al. Influence of ginkgotoxin on human pyridoxal kinase
FEBS Journal 274 (2007) 1036–1045 ª 2007 The Authors Journal compilation ª 2007 FEBS 1043
4 Caesar W (1994) In ‘Ginkgo’, Urbaum und
Arzneipflanze-Mythos, Dichtung und Kunst (Schmid M,
& Schmoll H, eds), pp. 23–31. Wissenschaftliche Ver-
lagsgesellschaft, Stuttgart, Germany.
5 Mizuno N, Kawakami K & Morita E (1980) Competi-
tive inhibition between 4¢-substituted pyridoxine ana-
logues and pyridoxal for pyridoxal kinase from mouse
brain. J Nutr Sci Vitaminol 26, 535–543.
6 Coburn SP, Slominski A, Mahuren JD, Wortsman J,
Hessle L & Millan JL (2003) Cutaneous metabolism of
vitamin B-6. J Invest Dermatol 120 (2), 292–300.
7 Ngo EO, LePage GR, Thanassi JW, Meisler N & Nut-
ter LM (1998) Absence of pyridoxine-5¢-phosphate oxid-
ase (PNPO) activity in neoplastic cells. Isolation,
characterization, and expression of PNPO cDNA. Bio-
chemistry 37, 7741–7748.
8 Buss K, Drewke C, Lohmann S, Piwonska A & Leistner
E (2001) Properties and interaction of heterologously
expressed glutamate decarboxylase isoenzymes GAD
(65kDa) and GAD (67kDa) from human brain with gink-
gotoxin and its 5¢-phosphate. J Med Chem 44, 3166–3174.
9 Bu DF, Erlander MG, Hitz BC, Tillakaratne NJK,
Kaufman DL, WagnerMcPherson CB, Evans GA &
Tobin AJ (1992) Two human glutamate decarboxylases,
65- kDa and 67- kDa GAD, are each encoded by a
single gene. Proc Natl Acad Sci USA 89, 2115–2119.
10 Nitsch C & Okada Y (1976) Differential decrease of

phate. Cellular physiology and biochemistry. Int J Exp
Cell Physiol Biochem Pharmacol 8 (3), 117–129.
19 Yang Y, Zhao G & Winkler ME (1996) Identification
of the pdxK gene that encodes pyridoxine (vitamin B6)
kinase in Escherichia coli K-12. FEMS Microbiol Lett
141, 89–95.
20 Willsta
¨
tter R, Kuhn R, Lind O & Memmen F (1927)
U
¨
ber Hemmung der Leberesterase durch Ketocarbon-
sa
¨
ureester. Hoppe-Seyler’s Z Physiol Chem 167, 303–309.
21 Dixon M & Webb EC (1979) Enzymes, 3rd edn. 1979,
Academic Press, New York, NY.
22 Sticher O (1992) Ginkgo biloba – Analytik und Zuberei-
tungsformen. Pharm Unserer Zeit 6, 253–265.
23 Granger AS (2001) Ginkgo biloba precipitating epileptic
seizures. Age Ageing 3, 523–525.
24 Gregory PJ (2001) Seizure associated with Ginkgo
biloba? Ann Intern Med 134 (4), 344.
25 Pilija V, Ivetic V, Mihalj M, Draganic-Gajic S & Popo-
vic M (2004) Effects of Ginkgo biloba extract on an
experimental model of epilepsy. Med Pregl 57 (11–12),
541–544.
26 Anshu M, Pillai K & Zakir H (1996) Influence of
Ginkgo biloba on the effect of anticonvulsants. Indian
J Pharmacol 28, 84–87.

60 min (b) with pyridoxal kinase (5 lg) at 37°C, as
analyzed by HPLC. (B) Formation of ginkgotoxin
(retention time: 15 min) from ginkgotoxin 5¢-phosphate
(retention time: 7.5 min) after treatment for 5 min (a)
and 60 min (b) with alkaline phosphase (Merck,
Darmstadt, Germany; 100 U) at 37°C, as analyzed by
HPLC.
This material is available as part of the online article
from http://www.blackwell-synergy.com
Please note: Blackwell Publishing is not responsible
for the content or functionality of any supplementary
material supplied by the authors. Any queries (other
than missing material) should be directed to the corres-
ponding author for the article.
U. Ka
¨
stner et al. Influence of ginkgotoxin on human pyridoxal kinase
FEBS Journal 274 (2007) 1036–1045 ª 2007 The Authors Journal compilation ª 2007 FEBS 1045