A Series of Natural Flavonoids as Thrombin Inhibitors: Structure-activity relationships

Abstrak

A series of natural flavonoids has been evaluated as potential inhibitors of thrombin using the optimized
method of thrombin time. Myricetin and quercetin have shown to be the best thrombin inhibitors tested.
In order to investigate the thrombin recognition of the most active and selective compounds, a molecular
modeling study has been performed using available Protein Data Bank (PDB) structures as receptor
models for docking experiments. Structure-activity relationships of flavonoids (SARs) on thrombin would
facilitate the design of chemical compounds with higher potency to serve as potential thrombin inhibitors,
and provide information for the exploitation and utilization of flavonoids as thrombin inhibitors for
thrombotic disease treatment.

Introduction

Thrombotic diseases are major causes of mortality and morbidity in the industrial world. In thrombotic
diseases, thrombin acts as a multifunctional serine protease which is generated in response to vascular
injury and catalyzes the proteolytic cleavage of the soluble plasma-protein fibrinogen to form insoluble
fibrin leading to clot formation. Thrombin also serves as a potent platelet agonist and amplifies its own
generation by feedback activation of several steps in the coagulation cascade. Because thrombin plays a
pivotal role in thrombogenesis, some thrombin inhibitors such as heparin, hirudin, bivalirudin and
argatroban are used in the treatment of thrombotic diseases and have obtained great efficiency [13].

Traditional Chinese medicines of rich flavonoids such as Carthamus tinctorius L. [4], Abelmoschus manihot
L. [5], Ginkgo biloba L. [6] have been used to treat thrombotic diseases in clinic for many years. Due to an
increasing public interest in alternative medicine and disease prevention, the use of herbal preparations
containing high doses of flavonoids for health maintenance has become very popular, and can raise
potential interactions with conventional drug therapies. Bearing in mind the above considerations, natural
products like flavonoids (e.g. flavones and flavanones) may be promising lead compounds as thrombin
inhibitors.

Flavonoids are a group of polyphenolic compounds, diverse in chemical structure and characteristics
present in plants, regularly consumed foods (e.g. vegetables and fruits), olive oil, and beverages like tea
and wine. They are prominent plant secondary metabolites and usually subdivided, according to their
chemical structure, into several subclasses including flavonols, flavones, flavanones, catechins,
anthocyanidins, isoflavones, dihydroflavonols, and chalcones [79]. These different flavonoids could
exhibit a wide range of biological activities [1020], including antibacterial, antiviral, anti-inflammatory,
antioxidant, antitumor properties.

A number of studies have been conducted to elucidate the structure-activity relationships (SARs) of
flavonoids for their biological activities mentioned above. However, SARs of flavonoids that inhibited
thrombin activity had not been discussed in detail. In present studies, thrombin inhibition activity of a
series of natural flavonoids with the optimized method of thrombin time had been examined in vitro [21].
Furthermore, research into SARs of flavonoids on thrombin inhibition activity provided new experimental
data for the future exploitation and utilization of flavonoids as inhibiting thrombin medicines.
Materials and chemicals

Galangin (1), kaempferol (2), quercetin (7), hyperin (8), rutin (9), isorhamnetin (11), isohamnetin-3-o-
nehesperridin (12), typhaneoside (14), myricetin (15), myricitrin (16), apigenin (17), acacetin (18), luteolin
(19), baicalein (20), baicalin (21), naringenin (22), naringin (23), hesperetin (24), hesperidin (25),
dihydromyricetin (26) puerarin (29), and epicatechin (30) were purchased from National Institute for the
Control of Pharmaceutical and Biological Products, China. Kaempferol-3-o-glucose (3), kaempferol-3-o-
glucose (2-1)rhamnose (4), kaempferol-3-o-(2"-di-E-pcoumaroyl)-rhamnoside (5), and kaempferol-3-o -
(2",4"-di-E-pcoumaroyl)-rhamnoside (6) quercetin-3-o-rhamnose(1-2)glucose(6-1)rhamnose (10),
isorhamnetin-3-o-glucose(6-1)rhamnose (13), amentoflavone (27), hinokiflavone (28), with 98% purity
was provided by Jiangsu Key Laboratory for TCM Formulae Research, Nanjing University of Chinese
Medicine, China. The compounds were dissolved in dimethyl sulfoxide (DMSO, Sigma, St.Louis, MO, USA).
Bovine thrombin was purchased from sigma.

Flavonoids are classified according to their chemical structure. The major flavonoid classes include
flavonols, flavones, flavanones, catechins (or flavanols), anthocyanidins, isoflavones,
dihydroflavonols,chalcones and biflavones. The set of 30 flavonoids used in this study (indicated as
compounds 130) included three main classes of flavonoids: flavonoles (16 compounds), flavones (5
compounds), flavanones (5 compounds), isoflavones (1 compound), catechins (1 compound), and
biflavones (2 compounds) (structures shown in Table 1).

Optimizing the method of thrombin time

Blood collection

Male New Zealand white rabbits, weighing 2-2.5 kg were obtained from the experimental animal center
of Nanjing university of Chinese Medicine and were approved by Animal Ethics Committee of Nanjing
University of Chinese Medicine. Rabbits were anesthetized with pentobarbital and blood was drawn from
the common carotid artery. Blood was collected into plastic tubes with 3.8% sodium citrate (citrate/blood:
1/9, v/v) for plasma anticoagulation. Plasma was separated from blood by centrifugation at 3000 rpm for
10 min.

Standard curve established

Thrombin time (TT) was determined by a coagulometer (Model LG-PABER-I, Steellex Co., China). Shortly,
after adding thrombin solution, coagulometer was started and TT was recorded. To establish the standard
curve of TT and thrombin concentration, TT was determined by incubating 40 l plasma for 3 min at 37
C, followed by addition of 40 l thrombin solution (different concentrations in Thris-HCl butter PH 7.4)
and 20 l solvent for 3 min at 37 C.

Precision test

The precision was evaluated by calculating intra-day and inter-day variations at the intermediate
concentration of standard solution. Three different concentrations of thrombin (3U, 6U, 12U) were
prepared. For the intra-day variability test, the standard solution was analyzed five times within one day;
while for the inter-day variability test, the standard solution was examined in quintuple on five
consecutive days.
Recovery test

The recovery test was used to evaluate the accuracy of the method. Log (TT prolongation) was calculated
with the optimized method of TT. It was converted into thrombin concentration by the regression
equation recorded as the found concentration. The recovery was the ratio of found concentration and
added concentration. The mean recovery was calculated on five assays.

TT, PT, APTT, FIB determination

TT was examined with the above method and converted into thrombin concentration by the regression
equation recorded.

Prothrombin time (PT) and activated partial thromboplastin time (APTT) and fibrinogen content (FIB) were
examined with commercial kits following the modified manufacturer's instructions. PT was determined
by incubating 40 l plasma solution for 3 min at 37 C, followed by addition of 40 l thromboplastin agent
and 20 l sample. APTT was determined by incubating 10 l sample solution and 50 l plasma with 50 l
APTT-activating agent for 3 min at 37 C, followed by addition of 50 l CaCl2. FIB was determined by
incubating 10 l plasma with 90 l imidazole buffer for 3 min at 37 C, followed by addition of 50 l FIB
agent and 10 l sample solution.

Molecular docking [22,23]

The X-ray crystal structures of thrombin (protein database code 2R2M) and 2-(2-Chloro-6-fluorophenyl)
acetamides (I-50) as the potent thrombin inhibitor [24] were used as the basis of the docking experiment.

Compounds were drawn by using ISIS-Draw and then exported formats for the mol. The two-dimensional
structures were converted to three dimensions by small molecule-line structure transformation services.
The three dimensions were added hydrogen atoms and charges schemes with Open Babel and exported
in mol2 format for further study of molecular docking.

Molegro Virtual Docker (MVD) is based on a differential evolution algorithm; the solution of the algorithm
takes into account the sum of the intermolecular interaction energy between the ligand and the protein,
and the intramolecular interaction energy of the ligand. It could be accurately forecast the active sites of
proteinmolecules based ligands. MVD is a precise semi-flexible molecular docking program. By inceaseing
the qualifications, the recognition accuracy of bonding modles is enhanced. Compared with the other
dock softwares, the accuracy of MVD (http://www.molegro.com/products.php) is higher. (MVD:87%,
Glide:82%, Surflex:75%, FlexX:58%).

The crystal structure of 2R2M was used as the basis of the docking experiments. The active sites exploited
in docking studies were defined as a subset region of 10.0 radius from the centroid of the ligand I-50.
Before screening, the docking protocol was validated. I-50 was docked into the binding pocket to obtain
the docked pose and the RMSD (Root Mean Square Deviation) of all atoms between these two
conformations was atb 1.00 indicating that the parameters for docking simulation were good in
reproducing the X-ray crystal structure.

Generally, between 10 and 100 individual docking simulations were performed, and every possible
docking model was analyzed by MolDock (MD) scores and hydrogen bonds. After multiple docking
simulations, consistent docking models among the best-fitted models were used for the development of
the hypotheses.
Statistical analysis

Statistical calculations were carried out with Microsoft Excel 2003.

Results were expressed as the meansSD of six independent experiments.

Results

Building standard curve

Thrombin time of thrombin 15 U was about 7-8 seconds as the control group [25]. Then TT of control
group was recorded as TT0 (s). TT of other different thrombin concentrations was recorded as TT1 (s). TT
prolongation (%)= (TT1-TT0)/TT0 100. Log (TT prolongation) was calculated.

Some studies had reported that the standard curve of plasma concentration of heparin and log (TT
prolongation) was used in clinic to monitor the plasma concentration of heparin. It was also used to
determine the plasma concentration of heparin in Pharmacopoeia of the People's Republic of China [26
29].

As a result, the method was optimized. Standard curves of thrombin concentration (C) and TT, C and TT
prolongation, C and log (TT prolongation), log C and TT, log C and TT prolongation and log C and log (TT
prolongation) were built, respectively and compared. Correlation coefficients (r) were 0.8380, 0.8459,
0.9927, 0.9628, 0.9660, and 0.9673, respectively. The result showed that the standard curve of thrombin
concentration and log (TT prolongation) was better than the others. Regression equation between log (TT
prolongation) and thrombin concentrations was built. Y was log (TT prolongation) and X was the thrombin
concentration (U) in Fig. 1.

Precision and recovery

Intra-day precisions of R.S.D. were 4.99%, 5.67%, 6.58%, respectively. Inter-day precisions of R.S.D. were
4.24%, 10.12%, 9.62%. The result was shown in Table 2. The recovery was the ratio of found concentration
and added concentration. Recoveries were 106.3%, 104.8%, 101.6%, respectively. The result was shown
in Table 3.

Activities on TT

In this study, 30 compounds (130) belong to different subgroups of flavonoids were investigated for
thrombin inhibition activity, and their compound names and molecular structures were shown in Table 1.

Thrombin time of different compound was recorded as TT (s), then TT prolongation (%) and log (TT
prolongation) were calculated. Log (TT prolongation) was converted into thrombin concentration by the
building standard curve. Inhibition ratio of each compound on thrombin inhibition activity was calculated.
Some compounds examined in this study were expressed as 50% inhibitory concentration (IC50).

The flavonoids with IC50b0.05 mM are stong inhibitors and this group includes myricetin (15) with
IC50=0.006 mM and quercetin (7) with IC50=0.035 mM. The flavonoids with IC50 with ranging from 0.05
mM to 0.1 mM are moderate inhibitors and this group includes luteolin (19) with IC50=0.052 mM,
baicalein (20) with IC50=0.060 mM, hinokiflavone (28) with IC50=0.071 mM, kaempferol-3-o-(2",4"-di-E-
pcoumaroyl)- rhamnoside (6) with IC50=0.052 mM, and kaempferol-3-o-(2"-di-Epcoumaroyl)-rhamnoside
(5) with IC50=0.083 mM. Whereas, those with IC50N0.1 are weak inhibitors and this group includes
kaempferol (2) with IC50=0.109 mM, apigenin (17) with IC50 =0.180 mM, acacetin (18) with IC50=0.140
mM and isorhamnetin (11) with IC50=0.246 mM. These results were shown in Table 4.

Activities on PT, APTT and FIB

In this study, 30 compounds (130) belong to different subgroups of flavonoids were investigated for PT,
APTT and FIB. But no compounds had activities on PT and FIB. Compounds 11 and 14-19 have activity on
APTT. But there were no relationship between structure and activity. The results were shown in Table 5.

Docking results

In the thrombin inhibition tests, myricetin (15) and quercetin (7) exhibited the better inhibitory activities
than other compounds. They were selected for the subsequent molecular docking experiment. In order
to analyze the binding modes of myricetin (15) and quercetin (7) within the thrombin, the most stable
theoretical complexes were graphically inspected reporting similar interaction and binding modes. There
were three pockets (S1, S2, S3) in the 2R2M as suggested by molecular modeling augured well for
anticipating in vitro activity as well [30,31].

The average MD score value of ligand I-50 was -144.47 kJ/mol. The average MD score value of myricetin
(15) (-111.55 kJ/mol) and quercetin (7) (-89.72 kJ/mol) were also reported in Table 3. According to
interaction energy data, myricetin (15) showed a more productive recognition with respect to the
quercetin (7) into the thrombin. As displayed in Fig. 2, ligand I-50 forms eight hydrogen bonds with active
site residues of 2R2M in this proposed binding mode. Myricetin (15) and quercetin (7) forms three
hydrogen bonds with active site residues of 2R2M, respectively. Ligand I-50 interacted with active site
residues Ala230, Asp229, Cys231, Gly258, Ser235, His79, Ile209, Leu132, Ser256, Trp257, Trp86, and Tyr83
of 2R2M. Myricetin (15) interacted with active site residues Ala230, Cys231, Glu232, Gly258, Lys88,
Ser235, Ser256, Trp257, and Trp86 of 2R2M. Quercetin (7) interacted with active site residues Ala230,
His79, Lys88, Gly258, Ser256, and Trp86 of 2R2M (see in Table 6). As shown in Fig. 3, ligand I-50 as the
potent thrombin inhibitor interacted with S1, S2, and S3 pockets, and myricetin (15) and quercetin (7)
mainly interacted with S3 pocket.

Discussion

In this study, 30 compounds (130) belong to different subgroups of flavonoids were investigated for TT,
PT, APTT and FIB. TT had been examined with the optimized method of thrombin time in vitro [21]. TT
was converted into thrombin inhibition activity by the standard curve. The relationship between thrombin
inhibition activity and the chemical structure were found. Myricetin (15) and quercetin (7) were selected
for docking experiments for the molecular modeling investigation. But no compounds had activities on PT
and FIB. Compounds (11) and (14-19) had activity on APTT. But there were no obvious relationship
between structure and activity.

Galangin (1), kaempferol (2), quercetin (7), and myricetin (15) belong to the flavonols class, which have
an OH group at the C3 position. Galangin (1) has no OH group in the B-ring, kaempferol (2) has a single
OH group (4 position) in the B-ring, quercetin (7) has two OH groups (3, 4 position) in the B-ring, and
myricetin (15) has three OH groups (3, 4, 5 position) in the B-ring. The thrombin inhibition activity of
these flavonols is as follows: myricetin (15)N quercetin (7)N kaempferol (2)N galangin (1). The result shows
that in flavonols class, the thrombin inhibition activity of flavonol with tri-hydroxyl groups N di-hydroxyl
groups, N mono-hydroxyl group N none-hydroxyl group. It confirms that more OH groups in the B-ring will
increase thrombin inhibition activity.

Apigenin (17), luteolin (19), and baicalein (20) belong to the flavone class of compounds, which have no
OH group at the C3 position. Apigenin (17) has a single OH group (4 position) and luteolin (19) has two
OH groups (3, 4 position) in the B-ring. Thrombin inhibition activity of luteolin (19) is better than apigenin
(17). Hence it seems that more OH groups in the B-ring will improve thrombin inhibition activity. With no
OH group in the B-ring and three OH groups (5, 6, 7 position) in the A-ring, baicalein (20) also exhibits
strong thrombin inhibition activity, which showed that more OH groups in the A-ring will increase the
activity.

When OH group in the B-ring is replaced by OCH3 groups, thrombin inhibition activity reduces.
Isorhamnetin (11), which has one OCH3 group (3 position) and one OH group (4 position) in the B-ring,
has weaker thrombin inhibition activity compared with quercetin (7) (3, position OH). The same as
acacetin (18) which has a single OCH3 group (4 position) in the B-ring also has weaker thrombin inhibition
activity than apigenin (17) (4 position OH).

In the isoflavone class, the B-ring is connected at the C3 position to the C-ring. Puerarin (29) belongs to
this group, but it doesn't have thrombin inhibition activity. It indicates that B-ring connected at the C2
position to the C-ring has a very important role in the thrombin inhibition activity.

However, flavanones have a single bond between the C-2 and C-3 positions in the C-ring. Naringenin (22),
hesperitin (24) and dihydromyricetin (26) have no thrombin inhibition activity. Epicatechin (30) belongs
to flavanols class of compounds, which lack the C(2)=C(3) bond and C(4)=O in the C-ring, thrombin
inhibition activity is also none. These results confirm that the presence of C(2)=C(3) and C(4)=O bonds are
very important in showing thrombin inhibition activity.

In the biflavone class, compared with amentoflavone (27), hinokiflavone (28) exhibits stronger thrombin
inhibition activity because of different connection locations.

Flavonoids glycosides exhibit no thrombin inhibition activity because of the presence of glycosylation
groups, which indicates that glycosylation groups significantly disfavor the effect. Myricetin (15) has
strong thrombin inhibition activity, while myricitrin (16) doesn't have the activity. Kaempferol-3-o-glucose
(3) and kaempferol-3-o-glucose(2-1)rhamnose (4) with kaempferol (2) as the nucleus, hyperin (8), rutin
(9) and quercetin-3-o-rhamnose(1-2) glucose(6-1)rhamnose (10) with quercetin (7) as the nucleus also
have no thrombin inhibition activity. However, if the glycosylation groups connected to the other groups,
such as pcoumaroyl, would have thrombin inhibition activity. With the number of pcoumaroyl increasing,
thrombin inhibition activity is enhanced.

Myricetin (15) and quercetin (7) were selected for docking experiments for the molecular modeling
investigation. Although myricetin (15) and quercetin (7) both forms three hydrogen bonds with active site
residues, the average MD score value of myricetin (15) is higher than quercetin (7) Myricetin (15)
interacted with more active site residues than quercetin (7) Maybe thrombin inhibition activity of
myricetin (15) became stronger than quercetin (7) due to its 5-OH group in the B-ring which could have
interacted with the S3 pocket in the thrombin. It had been validated that myricetin (15) interacted with
thrombin by capillary zone electrophresis [5] which had been reported.
In this study, we examined the effects of 30 natural flavonoids from different types on thrombin inhibition
activity. The results showed that eleven flavonoids produced dose-dependently inhibitory activities. They
were as follows: myricetin (15), quercetin (7), kaempferol (2), isorhamnetin (11), kaempferol-3-o-(2",4"-
di-E-pcoumaroyl)-rhamnoside (5) and kaempferol-3-o-(2"-di-E-pcoumaroyl)- rhamnoside (6) in the
flavonols class; baicalein (20) , luteolin (19), apigenin (17) and acacetin (18) in the flavone class, and
hinokiflavone (28) in the biflavones class. Preliminary SARs analysis showed that a hydroxyl group at C-3
played a key role in the inhibitory activity and more OH groups in the B-ring could increase thrombin
inhibition activity. Furthermore, it was found that the presence of C(2)=C(3) and C(4)=O bonds were very
important in showing the activity. It was also important to note that the presence of a glycosylation groups
greatly reduced the thrombin inhibition activity. To our knowledge, this is the first time anyone has
studied the SARs of several types of flavonoids as they relate to thrombin inhibition activity. Therefore,
the findings of this study provide important information relating the Fig. 3. I-50, myricetin and quercetin
docked into the pockets of 2R2M. L. Liu et al. / Thrombosis Research 126 (2010) e365e378 e377chemical
structure of flavonoids to thrombin inhibition activity. This understanding would facilitate the design of
chemical compounds with higher potency to serve as potential thrombin inhibitors, and provide
information for the exploitation and utilization of flavonoids as thrombin inhibitors for thrombotic disease
treatment.

Conflict of interest statement

The authors reported no potential conflicts of interest.

Acknowledgments

This research was financially supported by Key Research Project in Basic Science of Jiangsu College and
University (06KJA36022, 07KJA36024), 2009 Program for New Century Excellent Talents by the Ministry
of Education (NCET-09-0163), National Natural Science Foundation of China (30873235), Natural Science
Foundation of Jiangsu Province, China (BK2008455), 2006 and 2007 Project for Supporting Jiangsu
Provincial Talents in Six Fields (06-C-020, 07-C- 010), 2009 Program for Excellent Scientific and
Technological Innovation Team of Jiangsu Higher Education.