Evaluation of novel factor Xa inhibitors from Oxya chinensis sinuosa with anti-platelet aggregation activity

The edible grasshopper Oxya chinensis sinuosa is consumed worldwide for its various medicinal effects. The purpose of this study was to investigate potential bioactive antithrombotic and antiplatelet compounds from O. chinensis sinuosa. Five N-acetyldopamine dimers (1–5) were isolated from O. chinensis sinuosa and compounds 1 and 2 were identified as new chemicals with chiral centers at H-2 and H-3 of the benzo-1,4-dioxane structure. Compounds 1–4 were found to have both FXa and platelet aggregation inhibitory activities. These compounds inhibited the catalytic activity of FXa toward its synthetic substrate, S-2222, by noncompetitive inhibition, and inhibited platelet aggregation induced by ADP and U46619. Furthermore, compounds 1–4 showed enhanced antithrombotic effects, which were assessed using in vivo models of pulmonary embolism and arterial thrombosis. The isolated compounds also showed anticoagulant effects in mice. However, compounds 1–4 did not prolong bleeding time in mice, as shown by tail clipping. N-Acetyldopamine dimers, including two new stereoisomers 1 and 2, are novel antithrombotic compounds showing both FXa inhibition and antiplatelet aggregation activity with a low bleeding risk. Collectively, these results suggest that compounds 1–4 could serve as candidates and provide scaffolds for development of new antithrombotic drugs.

The ESIMS data of compound    Table 3. The results of a chromogenic substrate assay demonstrate that compounds 1-4 are a potent inhibitor of human FXa with a K i = 2.96, 4.13, 4.45, and 4.72 μM, respectively ( Table 4). The selectivity of compounds 1-4 for FXa compared with a selection of other human enzymes was shown in Table 4. Compounds 1-4 are highly selective for FXa and demonstrates a selectivity ratio (based on the respective K i values) of >10,000 for FXa versus thrombin, trypsin, elastase, plasmin, protein Ca, streptokinase, tPA, and urokinase.

Effect of the Isolated Compounds on Clotting Time Ex Vivo and In Vivo.
To evaluate the effects of compounds 1-5 on coagulation parameters ex vivo, we measured the influence of the isolated compounds on activated partial thromboplastin time (aPTT) and prothrombin time (PT). Pre-administration of compounds 1-4 intravenously to mice significantly increased aPTT in a dose-dependent manner at doses ranging from 38 to 193 µg/kg (compounds 1-4, Fig. 4A). However, compound 5 did not affect these measures (data not shown). The average circulating blood volume for mice is 72 mL/kg 21 . Because the average weight of mice used in this study was 27 g and the average blood volume was 2 mL, the dose of compounds 1-5 (38.6, 77.2, or 193.1 µg/kg) equaled a peripheral blood concentration of approximately 1, 2, and 5 µM, respectively. However, PT was not significantly higher in mice treated with compounds 1-5 than that of mice treated with vehicle only (data not shown). Compounds 1-4 significantly prolonged blood clotting time in a dose-dependent manner as shown in the in vivo clotting time experiments. These data indicate that compounds 1-4 but not compound 5 have significant, dose-dependent anticoagulant effects in vivo (Fig. 4B).

Effect of Compounds 1-5 on Platelet Aggregation.
To study the effects of compounds 1-5 on platelet aggregation, we used mouse platelet-rich plasma (PRP) induced with various agonists as an in vitro model. The results show that compounds 1-4 selectively inhibit platelet aggregation induced by different agonists. As shown in Fig. 5, compounds 1-4 inhibited platelet aggregation induced by ADP (10 µM, Fig. 5A) and U46619 (a stable thromboxane A2 analogue/aggregation agonist, 6 µM, Fig. 5B) in a dose-dependent manner. However, the compounds did not inhibit thrombin-induced platelet aggregation (Fig. 5C). In addition, compound 5 did not affect the aggregation of platelets, regardless of induction (data not shown).    Table 4. Enzyme kinetics and selectivity of isolated compounds against different human enzymes. a K i is represented by the mean ± SD (n = 5), μM. b Ratio = K i enzyme/K i Factor Xa. n.d., not determined.

Effects of Compounds 1-4 on FXa Activity and
Lineweaver-Burk plots in Fig. 6A (inset) show noncompetitive inhibition of FXa by compound 1. We observed a decrease in V max , with no effect on the K m of FXa toward S-2222 in the presence of compound 1. The K i value for FXa inhibition toward S-2222 by compound 1 was determined to be 2.96 µM. We also observed a slope of 0.401 for the control without compound 2, and slopes of 0.442 and 0.465 for compound 2 at concentrations of 2 and 5 µM, respectively. Lineweaver-Burk plots ( Fig. 6B (inset)) show a noncompetitive inhibition FXa by compound 2. We observed a decrease in V max , with no effect on the K m of FXa toward S-2222 in the presence of compound 2. The K i value for S-2222-associated inhibition of FXa by compound 2 was determined to be 4.13 µM. Inhibition activity plots were obtained for different concentrations of compound 3, showing a slope of 0.388 for the control without compound 3, and slopes of 0.408 and 0.464 for compound 3 at concentrations of 2 and 5 µM, respectively. Lineweaver-Burk plots ( Fig. 6C (inset)) show noncompetitive inhibition of FXa by compound 3. We observed a decrease in V max , with no effect on the K m of FXa toward S-2222 in the presence of compound 3; the K i was determined to be 4.45 µM. We also observed a slope of 0.363 for the control without compound 4, and slopes of 0.376 and 0.398 for compound 4 at concentrations of 2 and 5 µM, respectively. Lineweaver-Burk plots (Fig. 6D (inset)) show a noncompetitive inhibition of FXa by compound 4. We observed a decrease in V max , with no effect on the K m of FXa toward S-2222 in the presence of compound 4. The K i value for S-2222-associated inhibition of FXa by compound 4 was determined to be 4.72 µM. Production of FXa by FVIIa is dependent on tissue factor (TF) expression in tumor necrosis factor-α (TNF-α)-stimulated HUVECs 22 . Thus, we investigated the effects of compounds 1-4 on the production of FXa by FVIIa. HUVECs were stimulated with TNF-α to induce TF expression, causing an 11.9-fold increase in the rate of FX production by FVIIa (109.4 ± 9.2 nM) over that of non-stimulated HUVECs (9.2 ± 0.8 nM). These effects were abrogated by anti-TF IgG (15.9 ± 2.3 nM; Fig. 6E). In addition, pre-incubation with compounds 1-4 dose-dependently inhibited FX production by FVIIa, with compound 1 showing the strongest inhibition (Fig. 6E). We next determined whether each compound compete with tissue factor pathway inhibitor (TFPI) in the inhibition of FXa. As shown in Fig. 6F, inhibitory effect of TFPI on FXa activity was further increased in the presence of each compound. Noting that compounds 1-4 inhibited the catalytic activity of FXa by a noncompetitive inhibition model (Fig. 6A-D), compounds 1-4 may bind to other sites on FXa that are different from the active site of FXa. , or 4 (black box), blood was collected from the mice and platelet-poor plasma (PPP) was obtained by centrifugation at 2,000 × g for 10 min at room temperature to test ex vivo activated partial thromboplastin time (aPTT). (B) Each group received a daily intravenous injection of the indicated compound for four consecutive days. Fifteen minutes after the last administration, blood samples were collected and in vivo aPTT was measured. (C) Fifteen minutes after administration of each compound or aspirin (4.5 or 9 mg/kg for 30 min) or the vehicle, tail tips (3 mm long) were cut from each mouse, and the remaining tail was immediately immersed into saline at 37 °C. Accumulated bleeding times (including periods of re-bleeding) were recorded. D = 0.2% DMSO used as the vehicle control. Data are presented as means ± SD of three independent experiments. *p < 0.05 vs. vehicle alone, analyzed by one-way ANOVA, followed by Tukey multiple comparison testing.

In Vivo Effects of Compounds 1-4 in Models of Arterial and Pulmonary Thrombosis.
To examine whether compounds 1-4 have antithrombotic and antiplatelet effects in vivo, each compound was challenged in a ferric chloride (FeCl 3 )-induced carotid artery thrombosis model 23 . Tirofiban, a clinical anti-thrombosis drug that acts as a selective glycoprotein IIb/IIIa inhibitor, was used as a positive control. Time to thrombus formation and thrombi size are summarized in Fig. 7. The data show that endothelial injury after FeCl 3 treatment in control mice leads to growth of large thrombi at 9.3 ± 0.8 min, and tirofiban significantly slowed the growth of large thrombi to 61.1 ± 5.1 min. Compounds 1-4 significantly slowed thrombi growth (Fig. 7A). We also examined the effects of each compound on thrombus size 60 min after FeCl 3 -induced endothelial injury (Fig. 7B). The results show that compounds 1-4 reduce FeCl 3 -induced thrombus formation. The results from the in vivo pulmonary thrombosis model are shown in Fig. 7C. A mixture of collagen and epinephrine that was injected intravenously into mice induced massive pulmonary thromboses, causing acute paralysis and sudden death (90% mortality). Mortality was significantly lower in mice co-treated with compounds 1-4 than that of mice treated with the mixture of collagen and epinephrine only (Fig. 7C).

In Vivo Effects of Compounds 1-4 on Bleeding Time. To assess bleeding risk incurred by compounds
1-4, we measured bleeding time in mice treated with each compound (up to 30 µM/mouse) by using a tail-cutting assay; the concentration used was three-fold higher than the doses used for the in vivo anti-thrombotic studies. Aspirin-treated mice (9 mg/kg) served as positive controls. As shown in Fig. 4C, a slight prolongation of bleeding time was observed in mice treated with 30 µM (14.3 ± 1.2, 13.5 ± 1.1, and 12.3 ± 1.3 min for compounds 1, 2, and 3, respectively, mean ± SD, n = 10). However, the bleeding time in mice treated with compounds 1-4 was shorter than that observed in mice treated with aspirin (18.2 ± 1.6 min, mean ± SD, n = 10). The dose required to avoid thrombus formation and to protect against paralysis or death in mice was determined to be 5 µM per mouse (Fig. 7C). At double this concentration (i.e., 10 µM), bleeding time was not significantly longer in mice treated with compounds 1-4 than that of mice treated with vehicle (13.5 ± 1.3, 12.5 ± 1.2, and 12.1 ± 1.1 min for compounds 1, 2, and 3, respectively, mean ± SD, n = 10), suggesting that compounds 1-4 confer a low bleeding risk.

Discussion
Although several anticoagulants such as fondaparinux, warfarin, low-molecular-weight heparins (LMWHs), and unfractionated heparin are effective in the prevention and treatment of thrombotic diseases, these drugs also have undesirable effects 24,25 . In the past decades, numerous research efforts were made to identify novel anticoagulants with improved efficacy and safety. After extensive investigation, FXa became a promising target for the development of potent and selective anticoagulants, especially since FXa is situated at the beginning of the intrinsic and extrinsic coagulation pathways 6,7 .
In this study, compounds 1-5 were isolated from O. chinensis sinuosa, and their antithrombotic and antiplatelet activity was determined. Compounds 1 and 2 were isolated for the first time from a natural source, while compounds 3-5 have been reported previously 18,20 . The absolute configuration of the 1,4-dioxane ring in the N-acetyldopamine dimers was found to be (2 R, 3 S) or (2 S, 3 R). In contrast to the reported N-acetyldopamine dimers, compounds 1 and 2 are specific diastereomers with 2 R and 3 R absolute configuration. N-acetyldopamine dimers have been reported to have diverse bioactivities, such as 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging activity and anti-inflammatory effects related to nuclear factor-κB (NF-κB), inducible nitric oxide synthase (iNOS), interleukin (IL)-6, TNF-α, and cyclooxygenase (COX)-2 in LPS-stimulated RAW264.7 cells 20 . In addition, tyrosinase and COX-2 inhibitory activities of N-acetyldopamine dimers have been reported 26 . However, this is the first report of the antithrombotic activity of N-acetyldopamine dimers. Compounds 1 and 2, with 2 R, 3 R absolute configuration on the 1,4-dioxane ring of the N-acetyldopamine dimer, displayed stronger antithrombotic activity than compounds 3 and 4 with 2 R, 3 S absolute structure. Compound 5, which bears one double bond in the N-acetylamino-2-ethyl moiety, did not show antithrombotic activity. This result suggests that flexibility of the N-acetylamino-2-ethyl group on the dibenzo-1,4-dioxane structure is essential to the activity. Compounds 1-4 showed FXa inhibitory and anti-platelet aggregation activities and showed potent antithrombotic activity using in vivo models of arterial and pulmonary thrombosis. Interestingly, these compounds did not prolong bleeding times in mice at effective antithrombotic doses, indicating a favorable efficacy to bleeding ratio.
FXa is a trypsin-like serine protease that plays a pivotal role in the blood coagulation cascade 2 . During activation of the coagulation cascade, FXa forms a complex with calcium ions and FVa on the platelet membrane and converts prothrombin to thrombin 2 . Noting that direct thrombin inhibitors have a narrow therapeutic index, we hypothesized that selective inhibition of FXa is a potential and promising strategy for the design of new antithrombotic agents, especially since FXa is positioned at the beginning of the common extrinsic and intrinsic coagulation pathways 6,7 . Many FXa inhibitors isolated from animals have molecular weights in the 1-29 kDa range [27][28][29][30] , but the molecular weights of compounds 1-4 (molecular weight of 386.15) are smaller than any other known FXa inhibitors. Therefore, compounds 1-4 would have advantages, such as low immunogenicity and low production costs, compared to relatively large antithrombotic proteins or peptides. Platelet aggregation assays and clotting time measurements are the most commonly used methods to determine the efficacy of new anti-thrombotic drugs 31 . The anticoagulant effects of compounds 1-4 were verified with both an aPTT and PT ex vivo assay. Compounds 1-4 are also selective platelet aggregation inhibitors, as shown by inhibition of ADP-and U46619-induced, but not thrombin-induced, platelet aggregation. These data confirm that compounds 1-4 do not inhibit the activity or production of thrombin under these conditions (data not shown).
Compounds 1-4 showed similar anti-thrombotic efficacy to rivaroxaban (a direct FXa inhibitor) such as increased blood clotting time, delayed thrombogenesis and thrombogenic time, and inhibition of the production and activity of FXa. Although rivaroxaban did not affect platelet aggregation induced by ADP, thrombin, or U46619 32 , compounds 1-4 effectively and concentration-dependently inhibited ADP-or U46619-induced platelet aggregation. Further, rivaroxaban prolonged the generation of thrombin and reduced the thrombin burst produced in the propagation phase 32 whereas compounds 1-4 did not affect the generation and activity of thrombin. Furthermore, the bleeding times were not significantly affected by rivaroxaban at antithrombotic doses below the ED 50 required for antithrombotic efficacy in the bleeding time models 32,33 . At higher doses of rivaroxaban in the rat tail-bleeding time model, bleeding times were dose-dependently prolonged 32,33 . However, the mouse tail-bleeding time was not affected by compounds 1-4. Therefore, as each compound selectively inhibited FXa amidolytic activity without affecting the function of thrombin in the circulation system, it should have a minimal (C) After each compound was injected intravenously, a mixture of collagen (C, 500 µg/kg) and epinephrine (E, 50 µg/kg) was injected into the tail vein of mice to induce acute thrombosis 6 h later. Afterward, mice (20 mice per group) were carefully examined for 15 min to determine whether the mouse was paralyzed, dead, or had recovered from the acute thrombosis challenge. D = 0.2% DMSO used as the vehicle control. *p < 0.05 vs. DMSO, analyzed via one-way ANOVA, followed by Tukey multiple comparison testing. effect on normal hemostatic responses and regulatory processes, indicating that compounds 1-4 have favorable benefits compared to rivaroxaban. Furthermore, a noncompetitive inhibitory mechanism of compounds 1-4 on FXa activity was proved by enhanced inhibitory effects of TFPI on FXa activity in the presence of each compound.
A characteristic of the FeCl 3 -induced thrombosis mouse model is the formation of thrombi and large platelet aggregates that are surrounded by erythrocytes and fibrin 23,34 . FXa binds to clots during clot formation and contributes to the procoagulant growth of thrombi 35,36 . When associated with thrombi, FXa is resistant to inhibition by thrombin-dependent anticoagulants 35 . In this study, compounds 1-4 suppressed the formation of stable occlusive thrombi that quickly form in the FeCl 3 -injury model (Fig. 7). Therefore, the inhibition of clot-associated FXa by direct (i.e., thrombin-independent) FXa inhibitors seems more effective than by indirect (i.e., thrombin-dependent) FXa inhibitors for the prevention of thrombosis.
(Sungnam, Republic of Korea) and were used after a 12-day acclimatization period. The mice were housed at five per polycarbonate cage under a controlled temperature (20-25 °C) and humidity (40-45% relative humidity) with a 12:12 h light:dark cycle. They received normal rodent pellet diet and water ad libitum during acclimatization. They were treated in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by Kyungpook National University, Republic of Korea (IRB No. KNU 2016-54).
Ex vivo coagulation assay. Fifteen min after the administration of each compound or saline, blood was sampled, and plasma was obtained by centrifuging at 2,000 × g for 10 min at room temperature in order to measure prothrombin time (PT) and activated partial thromboplastin time (aPTT). The aPTT and PT were determined using a Thrombotimer (Behnk Elektronik, Norderstedt, Germany), according to the manufacturer's instructions and as described previously 37 . Briefly, platelet-poor plasma (PPP, 100 μL) was mixed with aPTT assay reagent (100 μL) for 1 min at 37 °C, followed by the addition of 20 mM CaCl 2 (100 μL). The clotting times were recorded. For the PT assays, PT assay reagent (200 μL) was incubated for 10 min at 37 °C. And, PPP (100 μL) was mixed with PT assay reagent (200 μL) for 3 min at 37 °C and the clotting time was recorded. All experimental protocols (KNUH 2012-01-010) were approved by the Institutional Review Board of Kyungpook National University Hospitals (Daegu, Republic of Korea).
Bleeding time. Tail bleeding times were measured using the method described by Dejana et al. 37 . Briefly, C57BL/6 mice were fasted overnight prior to the experiments. One hour after the i.v. administration of each compound, the tails of the mice were transected 2 mm from their tips. Bleeding time was defined as the time elapsed until bleeding stopped. Bleeding times that exceeding 15 min were recorded as 15 min.

Inhibition of amidolytic activity of FXa. Each compound with or without TFPI was dissolved in 50 mM
Tris-HCl buffer (pH 7.4) containing 7.5 mM EDTA and 150 mM NaCl. Following a 2-min incubation at 37 °C, FXa solution (150 μL, 1 U/mL) was added, followed by incubation at 37 °C for 1 min. S-2222 (an FXa substrate, 150 μL, 1.5 mM) solution was subsequently added, and the absorbance at 405 nm was monitored for 20 min using a spectrophotometer (TECAN, Männedorf, Switzerland). The absorbance-time curve and the slope of the curve (V i ) were used to represent the activity of the enzyme. Distilled water served as a control (the slope of curve V 0 ). The inhibitory effect was calculated according to equation (1): where V 0 represents the slope of the vehicle and V i represents the slope of the samples.

Determination of the inhibitory constant for FXa and enzyme inhibition.
The inhibitor (I) constants (K i ) were determined for the inhibition of a series of human enzymes (FXa, thrombin, trypsin, plasmin, protein Ca, streptokinase, tPA, and urokinase) by each compound. Chromogenic substrate assays were performed using a Labsystems IEMS (Cergy Pontoise, France) microtiter plate reader. In each assay, the compound was tested at a minimum of seven concentrations in duplicate to obtain an inhibition curve. Assays were performed according to the following general procedure. In a 96-well microtiter plate, 25 μL of compound, inhibitor solution or buffer was added to 50 μL of substrate. A volume of 25 μL of enzyme solution was added just before the plate was placed in the microtiter plate reader for 1 h at 37 °C. The hydrolysis of the substrate yields p-nitroaniline, which was continuously monitored spectrophotometrically at 405 nm. Data were collected and the initial rate of substrate hydrolysis [V o (mOD/minute)] was calculated. Following Michaelis-Menten kinetics, the affinity of the enzyme for the substrate, in the absence of the inhibitor (K m ) and in the presence of an inhibitor (K p ), was determined to be the negative inverse x-intercept of Lineweaver-Burk plots. The dissociation constant for inhibition (K i ) was calculated using the following equation: K i = (K m ·[I])/(K p − K m ) as described by Williams and Morrison 40 .
Cell culture. Primary HUVECs were obtained from Cambrex Bio Science (Charles City, IA, USA) and were maintained as described previously 41,42 . Briefly, cells were cultured in EBM-2 basal media supplemented with growth supplements (Cambrex Bio Science, Charles City, IA, USA) at 37 °C under a 5% CO 2 atmosphere until confluent. All experiments were performed with HUVECs at passage 3-5.
Production of factor Xa on the surface of HUVECs. The TNF-α-stimulated (10 ng/mL for 6 h in serum-free medium), confluent monolayer of HUVECs (preincubated with the indicated concentrations of each compound for 10 min) in a 96-well culture plate was incubated with FVIIa (10 nM) in buffer B (buffer A [10 mM HEPES, pH 7.45, 150 mM NaCl, 4 mM KCl, and 11 mM glucose] with 5 mg/mL bovine serum albumin [BSA] and 5 mM CaCl 2 ) for 5 min at 37 °C in the presence or absence of anti-TF IgG (25 μg/mL). FX (175 nM) was subsequently added to the cells in a final reaction volume of 100 μL, and the cells were incubated for 15 min. The reaction was stopped by the addition of buffer A containing 10 mM EDTA, and the FXa generated was measured using a chromogenic substrate. The changes in absorbance at 405 nm over 2 min were monitored using a microplate reader (Tecan Austria GmbH, Grödig, Austria). The initial color development rates were converted into FXa concentrations using a standard curve prepared with known dilutions of purified human FXa. Arterial thrombosis animal model. The FeCl 3 -induced thrombosis mouse model was established as previously described 23 . Male C57BL/6 mice were fasted overnight and were administered each indicated compound in DMSO by intravenous injection. Mice were then anesthetized with 3% isoflurane (Forane ® , Choongwae Pharma. Corp., Seoul, Korea) and injected intravenously with 0.1 mL of 0.1% rhodamine 6 G (Sigma). A testicular artery (200 μm in diameter) was carefully exposed and a cotton thread (0.2 mm in diameter) saturated with 0.25 M FeCl 3 was applied to the adventitial surface. After 5 min, the cotton thread was removed, and the wound was flushed with saline solution. Thrombus formation was monitored at 35 °C by 3D imaging as previously described 43 . The size and time of thrombus formation were monitored, and the findings were categorized as follows: score 0 indicated no thrombus; 1 indicated a small thrombus (50 μm × 75 μm); 2 indicated a medium-sized thrombus (100 μm × 150 μm); and 3 indicated a large thrombus (200 μm × 300 μm). The time from FeCl 3 -mediated endothelial injury to occlusion of the testicular artery by a large thrombus was also recorded.
Acute thrombosis induced by a combination of collagen and epinephrine in mice. Male C57BL/6 mice were fasted overnight and divided into groups of 10 animals. Each compound, suspended in DMSO, was administered to mice intravenously. A mixture of collagen (500 μg/kg) plus epinephrine (50 μg/kg) was injected into the tail vein of mice to induce acute thrombosis 6 h later. Each mouse was carefully examined for 15 min to determine whether the mouse was paralyzed, dead, or had recovered from the acute thrombotic challenge. For statistical analysis, five separate experiments were performed.
Statistical analysis. The results were expressed as means ± standard deviations (SD) of at least three independent experiments performed in duplicate. P < 0.05 was considered statistically significant and was determined using SPSS software (version 14.0, SPSS Science, Chicago, IL, USA). Statistical differences were determined by one-way analysis of variance (ANOVA) and Tukey's post-test.