Coumarin derivatives from Ainsliaea fragrans and their anticoagulant activity

Coumarin derivatives are an important class of C6–C3 plant metabolites that show a variety of bioactivities. Currently, most clinical anticoagulant agents are coumarins, such as warfarin, dicoumarol and acenocoumarol, and patients taking these drugs must be monitored for adverse reactions. In a search for safe and effective anticoagulant compounds from Chinese herbal medicine, a screening procedure on the whole plant of Ainsliaea fragrans was performed. The phytochemical investigation of this plant afforded five new coumarin derivatives, including a pair of natural 4-hydroxycoumarin enantiomers (1), a pair of coumarin enantiomers with a rare polycyclic pyrano[3-2c] carbon skeleton (2) and a 7-hydroxycoumarin derivative (3), together with 5 known biogenetically related compounds (4–8). Enantioseparation of 1 and 2 produced optically pure compounds 1a, 1b, 2a and 2b. The absolute configurations of the new compounds were confirmed by single-crystal X-ray diffraction analysis. In addition, we evaluated the anticoagulant activity of all isolates via activated partial thromboplastin time (APTT), thrombin time (TT) and prothrombin time (PT) assays in vitro and in vivo. Of note, compound 3 displayed potent anticoagulant activity and no significant hepatic or renal toxicity, which could make it a promising agent for further preclinical evaluation for preventing abnormal blood clotting.

) The 13 C NMR and DEPT spectra revealed the presence of 20 carbon resonances, including one conjugated ester, ten olefinic carbons, five methyls, and two methylenes. The presence of a 1, 2, 3-trisubstituted phenyl moiety was supported by the NMR information as follows: δ H 7.01 (H-6) to δ C 128.8 (C-6), δ H 7.34 (H-7) to δ C 132.2 (C-7), and δ H 7.24 (H-8) to δ C 116.1 (C-8). A characteristic single peak (δ H 2.78, δ C 24.9) suggested an aromatic methyl. The 1D NMR data of 1 were similar to those of cyclobrachycoumarin [22][23][24] , which indicated that 1 was a 5-methylcoumarin derivative. Of the 20 carbon resonances, ten were typical of a 5-methylcoumarin moiety, whereas the additional ten carbons consisted of two isoprene moieties, as shown by HMBC correlations from H-11 to C-12/C-20/C-19, from H 3 -20 to C-11/C-3/C-12, from H 3 -19 to C-11/C-12/C-13, and from H 3 -18/H 3 -17 to C-16/C-15. The two isoprene moieties were connected by C-13 and C-14, as shown by the 1  between H-14 (δ 2.26) and H-13 (δ 1.98). Furthermore, the 5-methylcoumarin moiety and the isoprene side-chain were connected at C-3, as shown by the HMBC correlations from H-11 to C-3/C-4 and from H 3 -20 to C-3/C-11. All of these signals suggested that 1 was a C-3 substituted 5-methylcoumarin. (Fig. 1) The relative configuration of the two ortho-position chiral carbon atoms (C-11 and C-12) could not be unambiguously determined by NOE correlations. After many attempts, a suitable crystal of 1 was obtained from a solvent system of CH 2 Cl 2 /MeOH/H 2 O (Fig. 2). However, the single-crystal X-ray diffraction experiment showed that compound 1 was a mixture of two enantiomers in a ratio of 62.7%:37.3%. Therefore, the planar structure of 1 was constructed as a natural C3-substituted 4-hydroxy coumarin. Chiral analysis and optical resolution of 1 were achieved by HPLC with a CHIRALPAK AS-H chiral column (n-hexane/i-PrOH, 95:5) at 0.8 ml/min, which afforded compounds 1a and 1b. The relative peak area ratio of 1a to 1b was approximately 2:1, consistent with the X-ray result (62.7%:37.3%). (Fig. 3) Similarly, a suitable crystal of 1a for the single-crystal X-ray diffraction experiment was obtained. Remarkably, the melting point of 1 was at 184 °C, but the melting points of 1a and 1b were increased to 191 °C and 190 °C, respectively. Finally, the absolute configuration of 1a was confirmed to be 11R, 12R, and the absolute configuration of 1b was confirmed as 11S, 12S, by combining the X-ray result of 1 and the CD spectrum of 1a (Fig. 4). Thus, the structures of 1a and 1b were assigned and named ainsliaeasin A1 and ainsliaeasin A2.
Compound 2 was obtained as colourless needles. Its formula C 20  , which indicated a 1, 2, 3-substituted phenyl moiety. A characteristic methyl singlet (δ H 2.69, δ C 23.1) implied that 2 was also a 5-methylcoumarin. The 13 C NMR of 2 showed 20 carbon resonances comprising a conjugated ester carbonyl, ten sp 2 carbons, two quaternary sp 3 , two sp 3 methine, one sp 3 methylene, and four methyls. The 1 H and 13 C NMR spectra of 2 were extremely similar to those of gerberlin B 25 , as further supported by the 2D NMR spectroscopic spectra, including 1 H-1 H COSY, HSQC, and HMBC spectra. However, the specific rotatory and CD spectra of 2 could not be detected. To confirm  the planar structure of 2, a single-crystal X-ray diffraction experiment was performed, which indicated that the space group of 2 was a mixture of two enantiomers.
Like 1, 2 was resolved to 2a and 2b using a chiral column (CHIRALPAK AD) under reverse phase conditions (Acetonitrile/H 2 O = 40/60) at 0.5 ml/min. The relative retention times for 2a and 2b were 20.8 min and 24.9 min, respectively. The relative peak areas of 2a and 2b were approximately 1:1, which was highly consistent with the X-ray result (ee 50%). In addition, the melting point of 2 was 208 °C, but the melting points of 2a and 2b increased to 214 °C and 213 °C, respectively. The successful X-ray diffraction with Cu-Kα, which resulted in a Flack parameter of 0.11 (15), allowed an unambiguous assignment of the absolute configuration of 2a as 12S, 13S. Based on the X-ray of 2 and CD spectrum of 2a, the absolute configuration of compound 2b was confirmed as 12R, 13R. Finally, the structures of 2a and 2b were assigned and named ainsliaeasin B1 and ainsliaeasin B2.
Compound 3 was obtained as colourless needles. Its molecular formula C 15    This observation was supported by the HMBC data, which showed correlations of H 3 -11 (δ H 4.06, s) to C-8 (δ C 132.6) and H-2′ (δ H 4.41, d) to C-3′ (δ C 72.5). The hydroxy group was also observed in the 1 H-1 H COSY spectrum from H-2′ to H-3′ (δ H 5.37, d). The absolute configuration of 3 was determined as 2′ S and 3′ R by X-ray diffraction with a Flack parameter of 0.09 (12). As a result, the structure of 3 was assigned and named ainsliaeasin C.
To date, only two polycyclic pyrano [3-2c] coumarins (gerberlin A and gerberlin B) have been isolated from Gerbera saxatilis 25 , so 2 is the third example of this carbon skeleton. In nature, chiral natural products are usually produced in optically pure forms; however, occasionally, both enantiomers are formed 27 . In this study, the enantiomers of 1 and 2 were isolated then enantioseparated to obtain compounds 1a, 1b, 2a, and 2b.
In vitro coagulation studies. All of the compounds isolated from this plant were analysed for their anticoagulant activities by monitoring the activated partial thromboplastin time (APTT), thrombin time (TT) and prothrombin time (PT). Coumarins usually interfere with the intrinsic coagulation process by inhibiting the vitamin K conversion cycle, but not with the extrinsic process 32 . Consistent with this finding, compound 4, without 3′ -hydroxyl, and compound 5, without 2′ -isopropyl, presented no anticoagulation effects. However, compound 3, with a hydroxyl group at the 3′ -position, showed moderate anticoagulant activity in vitro by significantly increasing the PT, exceeding the full scale of the instrument. The other compounds were inactive according to these three parameters. (Table 2) By comparing the structures of compounds 3, 4 and 5, it was concluded that the anticoagulant activity is closely related to the 3′ -hydroxy and 2′ -isopropyl moieties. Subsequently, compound 3 was found to exert anticoagulant activity at a minimum concentration of 1 mg/ml. (Table 2) In vivo coagulation studies. To estimate the putative in vivo efficacy, we performed studies in Wistar rats to measure anticoagulant activity. Initially, the dose and sampling time of warfarin were tested on rats at 1 mg/kg, 0.5 mg/kg, and 0.2 mg/kg after 1 day, 2 days, and 3 days.
To evaluate the in vivo anticoagulant activity of compounds 1-8, the APTT, PT, and TT were determined on the third and fifth days after administration. After treating for 3 days, compound 7 markedly prolonged the TT (P < 0.01), whereas compound 8 extended it to an insignificant degree (P > 0.05). However, compounds 1, 2, 3, 4, 5, and 6 did not change the APTT, PT or TT on the third day compared with the negative control group. (Table 3) To better characterize the anticoagulant profile, changes in the APTT, PT, and TT values were also determined on the fifth day. The PT and TT were prolonged in rats treated with compounds 3 and 4, whereas compounds 5 and 6 increased the TT value but not to a significant degree (P > 0.05). The TT values of group 7 and group 8 reached normal levels on the fifth day. (Table 3) The result showed that compound 3 exhibited anticoagulant activity at 1 mg/kg in Wistar rats.  Table 2. PT, APTT, and TT of normal human platelet-poor plasma. a Positive control b negative control (1% DMSO) c _ _ _._ exceeded the full scale of the instrument. All of the data are expressed as mean ± SD (n = 3). * P < 0.05, ** P < 0.01, compared with the negative control group.
Moreover, no death was found in rats treated with compound 3. However, one death caused by haemorrhage was observed in each of the groups treated with warfarin, compound 1 and compound 7 on the third day, and another death was observed in each of the groups treated with warfarin, compound 1, compound 4 and compound 8 on the fifth day. As a result, compound 3 was shown to be less toxic than the other compounds.
After the dissection of the rats on the seventh day, no significant liver or renal toxicity was observed in any of the groups of rats.

Conclusion
In this study, five new coumarins were isolated from Ainsliaea fragrans Champ. The chiral resolution of enantiomers (1 and 2) led to optically pure compounds 1a, 1b, 2a and 2b. Their planar structures and absolute configurations were determined by NMR, X-ray and CD analysis. Biologically, the anticoagulant activity of all of the isolates was evaluated. Compound 3 had anticoagulant activity both in vitro and in vivo. Additionally, compound 3 proved to be less toxic than warfarin and showed no significant liver or kidney toxicity. However, it is important to note that the results of the current study are preliminary, pending confirmation of the anticoagulant activity in vitro and in vivo. Further research is necessary to evaluate the action and mechanism of action of compound 3.

Methods
General. The melting point (uncorrected) was determined on an apparatus made by Beijing TECH INSTRUMENT CO. LTD. Optical rotations were measured with a Perkin Elmer spectropolarimeter. The UV spectra were measured on a VARIAN SARY 50 spectrophotometer. The IR spectra were recorded using a BRUKER VERTEX 70 spectrometer. The NMR experiments were run on a Bruker AM-400 spectrometer. The HRFABMS data were obtained on a VG 7070-HF spectrometer. Column chromatographic separations were carried out using silica gel H60 and ODS as packing materials.   Extraction and isolation. The whole plant of Ainsliaea fragrans (20 kg) was percolated with 95% industrial ethanol at room temperature. The filtrate was concentrated in vacuo. The residue was partitioned with petroleum ether (200 g), EtOAc (120 g), and n-BuOH (150 g), successively. The petroleum ether portion was subjected to column chromatography over silica gel eluted with a petroleum-acetone gradient to afford five fractions (A-E). Fraction B was separated into three subfractions, B1, B2, and B3. Compound 1 (15.0 mg, t R 46.2 min) was separated from B2 by semipreparative HPLC (80:20 MeOH-H 2 O, 230 nm, 2.0 ml/min), while 1a and 1b were obtained from a chiral column (CHIRALPAK AS-H) under normal phase conditions (n-hexane : isopropanol = 9:1) at 0.5 ml/min using a UV detector at 230 nm, 254 nm and 210 nm; the relative retention times of 1a and 1b were 34.3 min and 25.5 min, respectively. Compound 5 (50 g) was recrystallized from fraction C, while compound 6 (15 g) was recrystallized from fraction D.
The n-BuOH portion was separated using a silica gel column eluted with a CH 3 Cl-MeOH gradient to afford four fractions to obtain 7 (22 mg) and 8 (180 mg).   Anticoagulant activity assay in vitro 33 . Assays were performed for each sample using an automatic coagulative instrument (Sysmex CA-7000 System) according to the instructions provided by the biological reagent provider (Siemens Healthcare Diagnostics Products Gmbh).

Single-Crystal X-ray Diffraction Analysis and Crystallographic
Fresh whole blood (50 ml), collected in sodium citrate coagulation test tubes, was donated by the first author of this paper, which was approved by the ethics committee of Puai Hospital. After centrifugation (3000 rpm, 8 min), the supernatants (270 μL) were divided into containers. Compounds 1-8 were diluted for use with normal saline (with 1% DMSO) at 1 mg/ml. Then, all of the isolates (30 μL) were added into the plasma sequentially. After incubation in a 37 °C thermostatic water bath for 10 minutes, all of the samples were analysed with the automatic coagulative instrument, which had been supplied with activated partial Thromborel S, Throbin, Dade Actin Activated Cephaloplastin Reagent (Siemens Healthcare Diagnostics Products Gmbh) and calcium-chloride solutions. All of the tests were performed in an automated environment. 34 . The methods were carried out in Wistar rats accordance the European Community guidelines for the use of experimental animals and all experimental protocols were approved by the ethics committee of Puai Hospital. Rats were kept in polyethylene cages with wood shavings as bedding and maintained in a temperature controlled room at 20 ± 1 °C with a 12/12 h lighting schedule (lights on at 08:00 h, off at 20:00 h) and a relative humidity of 50% for at least 2 weeks prior to use.

Anticoagulant activity assay in vivo
All of the experiments were performed using adult male Wistar rats (250-300 g, body wt, Institute of Laboratory Animals of Sichuan Academy of Medical Sciences, SCXK 2013-24.) The animals were grouped and housed with seven per cage/group. The warfarin (0.2 mg/kg) (WUHAN XIANGHESHUNDA FINE CHEMICAL CO. LTD, XH20150206) and the tested compounds (1 mg/kg) were dissolved in sodium carboxymethyl cellulose and administered to animals by gavage for three days. Citrated blood was collected from the eye socket on the third day and the fifth day.
Platelet-poor plasma was prepared by centrifugation for measuring the APTT, PT, and TT on a semi-automatic biochemical analyser (MC-4000, Germany). All of the data are expressed in relative fold values, compared with the values obtained with the control group. The data were tested for statistical significance by nonparametric two-tailed Mann-Whitney test using the SPSS 17.0 software. A value of P < 0.05 was considered significant.