Four 14(13 → 12)-Abeolanostane Triterpenoids with 6/6/5/6-Fused Ring System from the Roots of Kadsura coccinea

Four new rearranged 6/6/5/6-fused lanostane-type triterpenoids, kadcoccitanes A–D (1–4), were isolated from the roots of Kadsura coccinea, and their structures were mainly elucidated by comprehensive analysis of their spectroscopic data. Additionally, the structure of 1 was ambiguously verified by single-crystal X-ray diffraction, while the structure of 2, which features a novel 8,16-epoxy motif, was validated by quantum chemical calculation of NMR parameters and ECD spectrum. Moreover, 1 and 4 were found to exhibited anticoagulant activity, while 3 and 4 were found to possess anti-platelet aggregation activity. Electronic supplementary material The online version of this article (10.1007/s13659-019-0203-4) contains supplementary material, which is available to authorized users.

In the aforementioned research, the secondary metabolites of K. coccinea collected from different regions manifested significant differences. So, it was interesting to consider whether the secondary metabolites from different parts of the same plant differed or not. Thus, a phytochemical study on the roots of K. coccinea cultivated in Jingzhou Miao and Dong Autonomous County in Hunan Province was carried out. As a result, four new 14(13 → 12)-abeolanostane triterpenoids, kadcoccitanes A-D (1-4) were obtained (Fig. 1). Considering the folk medicinal value of K. coccinea, some of the compounds were tested for their antiplatelet aggregation and anticoagulant activities. Herein, the isolation, structure elucidation, and bioactivity screening of these compounds are described.

Results and Discussion
Kadcoccitane A (1) was obtained as colorless acicular crystals and possessed a molecular formula of C 30 H 46 O 5 , which was determined by the HRESIMS ions at m/z 509.3251 [M + Na] + , calcd 509.3237), demonstrating eight degrees of unsaturation. The 1 H-NMR data of 1 (Table 1) indicated the existence of five singlet methyls (δ H 1.06, 1.15, 1.25, 1.39 and 2.14), one doublet methyl (δ H 1.16) and two olefinic protons (δ H 5.29 and 6.07). Its 13 C NMR and DEPT spectra revealed 30 carbon resonances (Table 2), including six methyls, nine methylenes (one hydroxymethyl carbon), seven methines (two olefinic carbons), and eight quaternary carbons (one carboxyl, one carbonyl and two olefinic carbons). Subtracting the four degrees of unsaturation generated by carboxyl, carbonyl, and olefinic groups, the remaining four degrees of unsaturation manifested that 1 had a tetracyclic structure.
Careful analysis of the HMBC and 1 H-1 H COSY spectra ( Fig. 3) of 2 demonstrated that it shared the same structure of side chain (C-20-C-27) as that of compound 1, as well as a similar A/B ring system except that C-8 (δ C 92.8) was oxygenated, as revealed by the 1 H-1 H COSY correlations of H-5/H 2 -6/H 2 -7, together with the HMBC correlations from H 2 -6 and H 2 -7 to C-8. Additionally, the HMBC correlations from olefinic H-11 (δ H 5.94) to C-12 (δ C 88.2), from H 3 -28 to C-15 (δ C 78.1), and from H-15 (δ H 4.33) to C-16 (δ C 80.0) demonstrated that C-12, C-15, and C-16 were all oxygenated. Besides, the correlations from H 3 -18 and H-16 (δ H 4.70) to C-13 (δ C 137.7) and C-17 (δ C 136.8) suggested the presence of the C-13/C-17 double bond. Most importantly, the correlation of H-16/C-8, the intensity of which was nearly as strong as that of the H-15/C-8 correlation, was observed in the HMBC spectrum. Hence, in consideration of the 4-bond distance between H-16 and C-8 if following the H-16/C-16/C-15/C-14/C-8 path, an epoxy was tentatively constructed between C-8 and C-16.
As for the stereochemistry of 2, the H-15/H 3 -28α correlation in the ROESY spectrum of 2 ( Fig. 3) indicated that H-15 adopted α-orientation. Moreover, the fact that H-15 and H-16 both existed as singlets in the 1 H-NMR spectrum, together with the absence of the 1 H-1 H COSY correlation of H-15/H-16, suggested that the dihedral angle of H-15/C-15/C-16/H-16 was around 90°, thus demanding H-16 to be α-oriented. Noteworthily, though there existed two possibilities for the orientation of HO-12 theoretically, the rigidity of the C/D ring system denied the existence of the HO-12β isomer. Moreover, the stereochemistry of C-20 couldn't be determined presently since it was located on the flexible side chain. Thus, the two possible C-20 stereoisomers of 2, (5R*,8S*,10S*,1 2R*,14R*,15S*,16R*,20R*)-2 (2a) and (5R*,8S*,10S*,1 2R*,14R*,15S*,16R*,20S*)-2 (2b) (Fig. S71) were subjected to quantum calculations of NMR chemical shifts  (Tables S2 and S3). As a result, the calculated NMR shifts of 2b were found to be in better agreement with their experimental counterparts, as indicated by parameters including R 2 , MAE, CMAE, as well as the DP4 + probability (Table 3). Then, quantum chemical calculation of spin-spin coupling constants (SSCC) of 2b was run at B97-2/pcJ-1 level with IEFPCM solvent model in chloroform (Table S5) , one doublet methyl (δ H 1.11) and one olefinic proton (δ H 6.12) were observed from its 1 H NMR spectrum ( Table 1). The 13 C NMR combined with DEPT spectra displayed 30 carbon signals ( Table 2) which comprised six methyls, nine methylenes (one hydroxymethyl carbon), six methines (one olefinic and one oxygen-bearing carbons), and nine quaternary carbons (one carboxyl, one carbonyl and three olefinic carbons). Judging from the total degrees of unsaturation and those which was occupied by carbon-carbon and carbon-oxygen double bonds, a tetracyclic scaffold of 3 was conjectured to exist.
The structure of 4 also differed from that of 1 in two aspects. Firstly, the carbonyl at C-3 of compound 4 was reduced to hydroxy, which could be determined by the HMBC correlations from H 2 -1, H 2 -2, H 3 -29 and H 3 -30 to  C-3 (δ C 78.3) (Fig. 6). Moreover, the HMBC correlations from H 3 -18 to C-12, C-13 (δ C 76.7) and C-17 manifested that the hydroxymethyl of C-18 was replaced by a methyl (Fig. 6).

The ROESY correlations of H-3/H-5α and H-8β/H-18/H-20β demonstrated the spatial α-orientation of H-3 and
β-orientation of H 3 -18 (Fig. 6). The phytochemical investigations on K. coccinea indicated that the secondary metabolites from the roots generally possess higher degrees of oxidation than those obtained from its stems, which may be mainly attributed to the abundant oxidases existing in the roots. However, to arrive at more definite conclusion, more in-depth research still needs to be undertaken. In addition, compounds 1, 3 and 4 were screened for the bioactivity against platelet aggregation induced by colloid. The result suggested that the inhibition ratios of 3 and 4 were 12.4 ± 12.5% and 19.4 ± 14.4% (p < 0.05), respectively, under the concentration of 100 µM.
Compounds 1 and 4 were tested for anticoagulant activity. The result showed that their IC 50 values were 37.8 and 31.5 µM, respectively.

General Experimental Procedures
1D and 2D NMR spectra were recorded on Bruker AV III 500 MHz spectrometer using TMS as the internal standard. Chemical shifts (δ) are expressed in ppm. HRESIMS was performed on an API QSTAR Pulsar i spectrometer. Melting point was recorded on an RDY-1B micro melting point apparatus. UV spectra were obtained on a Shimadzu UV-2401PC spectrophotometer. Optical rotations were measured in MeOH with a JASCO P-1020 polarimeter or an Autopol VI, Serial #91058. IR spectra were obtained on a Bruker Tensor-27 FT-IR spectrometer using KBr pellets. Column chromatography (CC) was performed with silica gel (80-100 or 100-200 mesh; Qingdao Marine Chemical, Inc., Qingdao, People´s Republic of China). Analytical HPLC was performed on an Agilent 1260 liquid chromatograph with a Zorbax SB-C18 (4.6 mm × 250 mm) column. Semipreparative HPLC was performed on an Agilent 1200 liquid chromatograph with a Zorbax SB-C18 (9.4 mm × 250 mm) column or a COSMOSIL Cholester (10ID × 250 mm) column. Fractions were monitored by thin layer chromatography, spots were visualized by UV light (254 nm and 365 nm) and by heating silica gel plates sprayed with 10% H 2 SO 4 in EtOH. All solvents used in column chromatography were distilled.

Extraction and Isolation
The air-dried and powdered roots (15.0 Kg) of K. coccinea were soaked with industrial alcohol (25 L) for four times, 3 days each time at room temperature. The total extracting solution was sequentially condensed to a certain extent under the condition of heating (40 °C) and reduced pressure in vacuo. Then the concentrate was suspended in water (1/1, v/v). The remaining alcohol was continuously evaporated under the same condition. After that, the sample was extracted with EtOAc and n-BuOH (1/2, v/v) for four times, respectively.
The EtOAc extract (1.8 Kg) was subjected to a silica gel column chromatography eluted with CHCl 3 /Me 2 CO gradiently (1:0 to 0:1) to obtain seven fractions A-G Fraction D (68 g) was chromatographed by RP-C 18 Tables 1 and 2. 1 H and 13 C NMR data in CDCl 3 , see Table S1.

Computational Method
_ENREF_1Conformational analysis of 2a and 2b was initially performed in Spartan'16 (Wavenfunction, Irvine, CA, USA, 2016) using the Monte Carlo algorithm and Merck molecular force field (MMFF). To avoid losing relevant conformations during the conformational search stage, the "set torsions" function was used to give all rotatable bonds a six fold sampling, as well as to allow the atoms on the aliphatic ring to flip. Maximum 20000 conformers were examined for each diastereoisomer, and those obtained conformers within 20 kcal/mol were kept (1000 ones for each isomer).
These conformers were subjected to semiempirical geometry optimization using the GFN2-xTB method [12] implemented in the XTB code (version 6.1) in order to obtain conformers better correlating with DFT calculations. Subsequently, XTB geometries with a difference of distance geometry within 0.25 were clustered. Then, clustered XTB geometries within an energy window of 8 kcal/mol were subjected to a DFT energy calculation at M06-2X/def2-SVP level of theory with DFT-D3 dispersion correction [13] using Gaussian 09 program [14], and those conformers within an energy window of 5 kcal/mol were kept. The completion of step was hugely aided by the Molclus program [15] (and its "isostat" module).
The above screened conformers were subjected to DFT geometry optimization at M06-2X-D3/def2-SVP level of theory. Frequency analyses of all optimized conformers were undertaken at the same level of theory to ensure they were true local minima on the potential energy surface. Then, energies of all optimized conformers were evaluated at M06-2X-D3/def2-TZVP level of theory. Gibbs free energies of each conformers were calculated by adding "Thermal correction to Gibbs Free Energy" obtained by frequency analysis to electronic energies obtained at M06-2X-D3/def2-TZVP level of theory. Room-temperature (298.15 K) equilibrium populations were calculated according to Boltzmann distribution law: where P i is the population of the ith conformer; n i the number of molecules in ith conformer; ΔG is the relative Gibbs free energy (kcal/mol); T is the temperature, usually the room temperature (298.15 K); R is the ideal gas constant (0.0019858995); Q is the partition function. Those conformers accounting for over 98% population were subjected to subsequent calculations.
NMR shielding constants were calculated with the GIAO method at mPW1PW91-SCRF/6-31 + G(d,p) level with IEF-PCM solvent model in chloroform solvent. The shielding constants obtained were converted into chemical shifts by referencing to TMS at 0 ppm (δ cal = σ TMS − σ cal ), where the σ TMS was the shielding constant of TMS calculated at the same level. For each possible candidate, the parameters a and b of the linear regression δ cal = aδ exp + b; the correlation coefficient, R 2 ; the mean absolute error (MAE) defined as Σ n |δ cal − δ exp |/n; the corrected mean absolute error, CMAE, defined as Σ n |δ corr − δ exp |/n, where δ corr = (δ cal − b)/a, were calculated [16,17]. The DP4 + probabilities of each possible candidate were calculated with the EXCEL spreadsheet provided by Sarotti et al. [18].
TDDFT ECD calculations were run at CAM-B3LYP/ def2-SVP level of theory in MeOH with IEFPCM solvent model. For each conformer, 30 excited states were calculated. The calculated ECD curves were generated using Multiwfn 3.6 software [20].

Anti-platelet Aggregation Induced by Colloid
The experimental blood was extracted from the ears of Japanese white rabbits and conserved in vacuum bloodcollection tubes with sodium citrate (whole blood: sodium citrate = 9:1). The blood-collection tubes were turned upside down gently to ensure homogeneous mixing of blood and anticoagulant and then centrifuged (200×g, 10 min). The supernatant was thus collected as platelet rich plasma (PRP). The residual blood was sequentially centrifuged (2400×g, 20 min) and supernatant was collected as platelet poor plasma (PPP). The platelet count of PRP was adjusted to 500 × 10 9 L −1 based on PPP [21][22][23]. The test samples were weighed accurately and dissolved with DMSO to 10 mM. The reference substance, Ticagrelor, was prepared to 0.5 mg/ mL with DMSO. The colloid and ADP were prepared to 1 mg/mL and 1 mmol/L, respectively, and conserved at specified temperature according to the instruction.
Two cuvettes, one with stir bar and the other without stir bar, were put in the heater of the platelet aggregation apparatus, heated at 37 °C for 10 min. Then 250 μL of PRP and 2.5 μL of test sample were added into the cuvette with stir bar. 250 μL of PPP and 2.5 μL of DMSO were added into the cuvette without stir bar. After 5 min of heating, the two cuvettes were put at the test positions of PRP and PPP, respectively. After adjusting the baseline of recording curve, the inducer (1 μL of colloid) was added into the cuvettes. The curve of platelet aggregation was recorded and the maximum aggregation rate were accordingly calculated.
The inhibition rate of the test samples against rabbit platelet aggregation induced by colloid could also be calculated. The formula is as follows: R = (A − B) * 100%/A. R: Inhibition rate, A: The maximum aggregation of solvent, B: The maximum aggregation of test sample (or reference substance).

Anticoagulant Activity
The test compounds were diluted with DMSO to 10 mM and then diluted with 0.02 M Tris-HCl (pH 7.4) with 5% Tween 80 to 1 Mm. The positive control was Low Molecular Weight Heparin (LMWH) and the blank control was 0.02 M Tris-HCl (pH 7.4) with 5% Tween 80 and 10% DMSO. The solution of the test sample or reference substance was added into the cuvette preheated at 37 °C and then the control plasma was also added into it. After heating at 37 °C for 2 min, the TT (thrombin time) reagent preheated at 37 °C was also added into it. The clotting time was recorded afterwards.