Antiplatelet Effect and Selective Binding to Cyclooxygenase (COX) by Molecular Docking Analysis of Flavonoids and Lignans

The known flavonoids ginkgetin (1), taiwanhomoflavone A (2), taiwanhomoflavone B (3), and taiwanhomoflavone C (4) and eight known lignans: justicidin B (9), justicidin C (10), justicidin D (11), chinensinaphthol methyl ether (12), procumphthalide A (13), procumbenoside A (15), and ciliatosides A (16) and B (17) were isolated from Cephalotaxus wilsoniana and Justicia species, respectively. The antiplatelet effects of the above constituents on human platelet-rich plasma (PRP) were evaluated. Of the compounds tested on human PRP, compounds 1, 4, 9, and 11 showed inhibition of secondary aggregation induced by adrenaline. Compound 1 had an inhibitory effect on cyclooxygenase-1 (COX-1). Molecular docking studies revealed that 1 and the related compounds apigenin (5), cycloheterophyllin (6), broussoflavone F (7), and quercetin (8) were docked near the gate of active site of COX-1. It indicated that the antiplatelet effect of 1, 4, 9, and 11 is partially owed to suppression of COX-1 activity and reduced thromboxane formation. Flavonoids, 1, 5, 6, 7, and 8 may block the gate of the active site of COX-1 and interfere the conversion of arachidonic acid to prostaglandin (PG) H2 in the COX-1 active site.


Results and Discussion
Platelet aggregation is an important pathogenic factor in the development of atherosclerosis and associated thrombosis in human [7]. A rational approach for the discovery of antithrombotic drugs is to search for inhibitors of platelet aggregation. In previous papers, several prenylflavonoids exhibited strong inhibition of platelet aggregation induced by AA and their antiplatelet effects are partially due to an inhibitory effect on COX-1 [10,11]. To study the structure-antiplatelet activity relationships of various flavonoids and lignans, the antiplatelet effects of 1-4, 9-13, and 15-17 on platelet aggregation induced by adrenaline (5 µM) in human PRP were studied. Lignans, 12 and 15-16 (data not shown) showed no antiplatelet effects in the test. While 1, 4, and 13 (each at 300 µM) and 9-11 (each at 100 µM) had significant antiplatelet effects on the platelet aggregation of human PRP induced by adrenaline (Table 1 and Figures 3 and 4).
*P < 0.05; **P < 0.01; ***P < 0.001: compared with the control value.  The antiplatelet effects of 1, 4, 9, 11, and 13 appeared to be concentration dependent, with IC 50 values of 293.6 ± 18.4, 259.0 ± 22.3, 104.8 ± 25.3, 106.5 ± 39.4, and 255.1 ± 86.2 µM, respectively. Compound 12 significantly inhibited rabbit platelet aggregation induced by AA [6], but did not significantly prevent secondary aggregation induced by adrenaline on human PRP. This could be due to the higher binding capacity of plasma for this compound. The positive control, aspirin (50 µM), strongly inhibited platelet aggregation induced by adrenaline on human PRP. In adrenaline induced platelet aggregation all compounds tested in Table 1 prevented secondary aggregation (i. e. 1 and 11 in Figures 5 and 6, respectively).  This could be also due to higher binding capacity of plasma for these compounds.
It is indicated that the antiplatelet effects were probably mediated through the suppression of cyclooxygenase-1 (COX-1) activity and reduced thromboxane formation or owing to the inhibition of thromboxane synthase, leading to reduced thromboxane formation [8,9].
Compounds 1 and 4 showed antiplatelet effect on platelet aggregation of human PRP induced by adrenaline. It indicated 3', 8'' linked biflavonoids such as 1 showed stronger antiplatelet effect on platelet aggregation of human PRP induced by adrenaline while a methyl group substituted at C-6 such as 2 decreased antiplatelet effect on platelet aggregation of human PRP induced by adrenaline. Previously we had reported that cilinaphthalide B, justicidin A, and taiwanin E methyl ether (14) were shown to have an antiplatelet effects on human PRP [5]. Among them 14 showed a strong inhibitory effect on platelet aggregation induced by adrenaline in a concentration dependent manner with an IC 50 value of 27.6 µM [5]. From Figure 4, it is clearly indicated that 1-aryl-2,3-naphthalide lignan type compounds such 14 had a stronger antiplatelet effect than that of 4-aryl-2,3-naphthalide lignan type ones such as 11 on human PRP.
When the activity of fatty acid COX-1 from ram vesicular glands was measured in the presence of the selected compounds. Compounds 1 and 14 weakly inhibited the enzyme (  The antiplatelet activity of 1 and 14 is partially due to suppression of COX-1 activity and reduced thromboxane formation. In crystallization experiments, it was reported that all COX-1 inhibitors interacted with the putative catalytic amino acid residue Tyr 385 and formed hydrogen bonds with Arg 120 and Tyr 355 [12]. When mapping 1 and related flavonoids 5-8 into a COX-1 model (Figures 7-11), all these compounds docked near the gate of COX-1. It could be observed that the oxygen or phenolic oxygen atom of flavonoids accept hydrogen bond from Arg 120 or Tyr 355. The scoring values given by DockScore [14] are given in Table 3. Compounds 7 and 8 indicated good qualitative agreement between experimental (Table 2) and predicted (Table 3)

Conclusions
In the present study, we had further found biflavonoid, 1, tetraflavonoid, 4, and lignans, 9 and 11 showed inhibition of secondary aggregation induced by adrenaline in human PRP. It indicates that the antiplatelet effects of these compounds are partially due to an inhibitory effect on thromboxane formation [7][8][9]. Additionally, compounds 1, 5, 6, 7, and 8 showed inhibitory effect on COX-1 activity (Table 3) [10,13]. It indicated a mode of action including the interference in arachidonic acid binding channel might be possible.

Platelet aggregation
Platelet aggregation in human PRP assays were performed by the method described previously [7].

COX-1 activity
The reaction mixture (0.1 M Tris-HCl, PH 8.0, 5 mM tryptophan, 8 mM hematin, test drugs, and 10 µg/mL of ram seminal vesicles COX was incubated for 3 min at 30 °C. The reaction was initiated by adding 100 µM AA. The velocity of oxygen consumption in the reaction mixture was monitored continuously with a Clark-type oxygen electrode using a YSI biological oxygen monitor (Model 5300) [17].

Molecular docking study
Docking experiments were performed using DS modeling 1.2 (Accelrys Inc.). The coordinates for the x-ray crystal structure of the enzyme COX-1 was obtained from the RCSB Protein Data Bank and hydrogens were added. The ligand molecules were constructed using DS Modeling 1.1 and energy-minimized. The docking experiment on COX-1 was carried out by superimposing the energy-minimized ligand on FLP in the PDB file 1CQE after which FLP was detected. The chemistry at HARvard Molecular Mechanics force field (CHARMm) was employed for all docking purposes. The optical binding orientation of the ligand-enzyme site obtained after docking was further minimized for 1000 iterations using the conjugate gradient method until a convergence of 0.001 Kcal/mol-Å. [18] Data analysis Data are presented as means ± s. e. m. One-way analysis was used for multiple comparison, and if there are significant variation between the treatment groups and the inhibitor-treated groups, they were then compared with control group by student's t-test. Values of P < 0.05 were considered statistically significant.