Triterpenoids from Ainsliaea latifolia and Their Cyclooxyenase-2 (COX-2) Inhibitory Activities

Abstract Eight new triterpenoids were isolated from Ainsliaea latifolia. The structures of these compounds were elucidated by interpretation of spectroscopic data, including HRESIMS and NMR data. Compounds 4–6 are identified as rare trinorcucurbitane or tetranorcucurbitane triterpenoids. The absolute configurations of compounds 1 and 2 were confirmed by Snatzke’s method. All compounds were evaluated for their inhibition against cyclooxyenase-2 (COX-2), in which compound 4 showed significant inhibitory effect against COX-2 with IC50 value of 3.98 ± 0.32 μM, comparable to that of positive control NS-398 (IC50 4.14 ± 0.28 μM). Graphic Abstract Electronic supplementary material The online version of this article (10.1007/s13659-019-00228-x) contains supplementary material, which is available to authorized users.

anorexia [7]. In Ainsliaea species, sesquiterpenoids are usually considered as characteristic chemical constituents. However, in our study of the chemical constituents from A. latifolia, eight new triterpenoids (1-8) and one known triterpenoid (9) were isolated and identified from the whole plants of A. latifolia. Herein, we described the isolation and structural elucidation of compounds 1-8, as well as their inhibition against cyclooxyenase-2 (COX-2).

Results and Discussion
The CHCl 3 -soluble of the EtOH-H 2 O (80:20, v/v) extract of A. latifolia was purified by repeated column chromatography (CC) over silica gel, Sephadex LH-20, and semi-preparative HPLC to yield eight new and one known compounds. By comparison of their NMR and MS data with the published references, the known compound 9 was then identified as one triterpenoid cucurbita-5,23-diene-3β,25-diol (9) [8]. The structures of eight new triterpenoids were determined by analysis of HRESIMS and NMR spectroscopic data (Fig. 1).
Compound 1 was isolated as white solid. Its molecular formula (C 30 H 50 O 3 ), ascertained via high resolution ESI-MS analysis, indicated six degrees of unsaturation. The 1 H NMR spectrum of 1 (Table 1) exhibited signals for three olefinic protons at δ H 5.59 (2H), 5.42 (1H, m), two oxygenated methine groups at δ H 3.83 (1H, d, J = 7.1 Hz), 3.47 (1H, brt, J = 2.5 Hz), eight methyl groups (δ H 1.20, 1.14, 1.13, 1.02, 1.00, 0.92, 0.87, 0.81). The 13 C NMR spectrum revealed the presence of thirty carbon signals including four olefinic carbons at δ C 141.2, 141.3, 125.7 and 121.4, three oxygenated carbons at δ C 79.7, 76.6 and 72.9, and eight methyl carbons at δ C 28.0, 27. 2, 26.3, 25.4, 23.7, 20.4, 17.8 and 15.7. The other carbon signals were assigned to seven methylenes, four methines, and four quaternary carbons. A comparison of these carbon resonances with those of the related cucurbitane-type triterpenoids suggested that 1 possessed the same cucurbitane skeleton, and the differences between the spectroscopic data of 1 and those of known compound 9 were primarily the observation of an oxymethine and the absence of a methylene. In the 1 H-1 H COSY spectrum of 1, two mutual coupling olefinic protons exhibited the correlations with H-20 and the oxygenated methine proton at δ H 3.83 (Fig. 2), respectively, ascribing a double bond to C-22 and C-23 positions. The HMBC correlation (Fig. 2) of CH 3 -21 with the olefinic carbon at δ C 141.3 confirmed the above deduction. Also, the observation of HMBC correlations from CH 3 -26 and CH 3 -27 to C-24 (δ C 79.7) and the oxygenated quaternary carbon at δ C 72.9 supported the hydroxyl substituents at C-24 and C-25 positions. The absolute configuration of C-24 in 1 was assigned using the Snatzke's method [9,10]. Metal complex of compound 1 in DMSO gave a significant induced CD spectrum (ICD) (Fig. 4), in which the positive cotton effect observed at 315 nm permitted the assignment of a 24S configuration for 1. The relative configurations of other stereocenters of 1 were established to be identical to those of known compound 9 due to NOESY experiment (Fig. 3). Thus, the structure of compound 1 was identified as cucurbita-5, 22-diene-3β, 24S, 25-triol.  (Table 1) of 2 were very similar to 1, except that two olefinic protons of 1 were replaced by two methylenes in 2. Therefore, the structure of 2 was determined to be a hydrogenated derivative of 1 at C-22/C-23 double bond. The assignment was confirmed by the 1 H-1 H COSY correlations of CH 3 -21/H-20/CH 2 -22/CH 2 -23/H-24 and key HMBC correlations of the oxygenated methine proton at δ H 3.31 (H-24) with C-22 and C-23, and of CH 3 -26 and CH 3 -27 with C-24 (δ C 79.6). Similarly, the absolute configuration of C-24 in 2 was confirmed using the Snatzke's method [9,10]. The positive Cotton effect observed at 310 nm ( Fig. 4) permitted the assignment of a 24S configuration for 2. Thus, the structure of compound 2 was identified as cucurbita-5-ene-3β,24S,25-triol.
Compound 3 was isolated as white solid. Its molecular formula (C 30 H 50 O 3 ), ascertained via high resolution ESI-MS analysis, indicated six degrees of unsaturation. Detailed analysis of the NMR (Table 1) and MS spectra led to the conclusion that the only difference between 3 and known compound 9 was that there is an epoxide group between C-5 (δ C 66.8, s) and C-6 (δ C 53.2, d) in 3 instead of a double bond between C-5 (δ C 141.2, s) and C-6 (δ C 121.4, d) in 9. The epoxide group was elucidated by HMBC correlations of H-1, H-3, H-7, CH 3 -29 and CH 3 -30 with C-5, and of H-8 and H-10 with C-6, as well as the 1 H-1 H COSY correlations of H-6/H-7. The NOESY correlations of H-6/CH 3 -29 indicated the epoxy ring of 3 was in β-orientation. Thus, the structure of compound 3 was identified as cucurbita-5β,6βepoxy-23-ene-3β, 25-diol.  unsaturation. In the 1 H NMR spectrum (Table 1), the signals of five tertiary methyl groups (δ H 1.14, 1.03, 0.92, 0.86, 0.81) and one secondary methyl group (δ H 0.91, 3H, d, J = 5.3 Hz) were observed. The 13 C NMR spectrum of 4 showed signals for 27 carbons due to six methyl groups, two olefinic carbons, ten methylenes (including an oxygenated one), five methines (including an oxygenated one), and four quaternary carbons. Detailed comparison of the 13 C NMR spectrum of 4 with that of 2 displayed similarities in rings A-D, except for the absence of the signals for C-25, 26, 27. These evidences revealed that compound 4 is a rare 25,26,27-trinorcucurbitane triterpenoid. This can be confirmed via the 1 H-1 H COSY correlations of H 3 -21/H-20/ H 2 -22/H 2 -23/H 2 -24. Thus, the structure of compound 4 was identified as 25,26,27-trinorcucurbita-5-ene-3β,24-diol.
Compound 5 was isolated as white solid. Its molecular formula (C 27 H 44 O 3 ), ascertained via high resolution ESI-MS analysis, indicated six degrees of unsaturation. Analysis of the 1 H and 13 C NMR spectroscopic data of 5 (Table 2) indicated a structural similarity with compound 4, except that compound 5 has a carboxyl (δ C 178.8, C-24) instead of hydroxyl methyl signals in 4. The deduction was confirmed via the HMBC correlations from H-22, H-23 to the carboxyl carbon (C-24). The relative configurations of 5 were evidenced to be identical to those of 4 by analysis of NOESY spectrum. Thus, the structure of compound 5 was identified as 25,26,27-trinorcucurbita-5-ene-3β-ol-24-acid (   (Fig. 2). The relative configurations of 7 were assigned as shown in Fig. 3 by analysis of the NOESY spectrum (Fig. 3). Thus, the structure of compound 7 was identified as 18R-D:A-friedoeuph-20-ene-22-ol-3-one.  [12]. Thus, the structure of compound 8 was identified as 18R-D:A-friedoeuph-22(30)-en-21S-ol-3-one.
All compounds were evaluated for their COX-2 inhibitory activities with NS-398 as a positive control. The results (  1-6 and 9 share similar or even the same rings A, B, C, D, and the major difference is their side chains. Therefore, it seems that the side chain is the main factor to influence the inhibitions of compounds 1-6 and 9 against COX-2.

Conclusion
In conclusion, this research led to the isolation of eight new triterpenoids and one known triterpenoid from the A. latifolia, in which compounds 4-6 are rare trinorcucurbitane or tetranorcucurbitane triterpenoids. It is the first report of cucurbitane-type triterpenoids from the genus Ainsliaea.
Interestingly, compound 4 showed potent inhibition against COX-2 with IC 50 values of 3.98 ± 0.32 μM. These results imply, except for sesquiterpenoids, triterpenoids may be another type of important chemical constituents being responsible for anti-inflammation in Ainsliaea species. Therefore, more attention should be paid to structural novel triterpenoids of Ainsliaea plants.

General Experimental Procedures
Optical rotations were measured on a PerkineElmer 341 polarimeter. 1 H and 13 C NMR spectra were recorded on Bruker Avance-500 spectrometers. ESI-MS were measured on an Agilent LC/MSD Trap XCT spectrometer, and HRESIMS were performed on an Agilent 6520 Accurate-MS Q-TOF LC/MS system. A preparative column (ZOR-BAX-ODS GSA10250AP1301, C18, 5 μm, 250 × 10 mm) was used for semi-preparative HPLC (Shimadzu LC-2010A HT). TLC analysis was run on HSGF 254 silica gel plates

Plant Material
The

Determination of the Absolute Configuration of C-24 in Compounds 1 and 2
According to the published literature [9,10], a mixture of compound 1 (1.1 mg) and Mo 2 (OAc) 4 (1.2 mg) was prepared for CD measurement. The mixture was kept for 30 min to form a stable chiral metal complex, the CD spectrum of which was then recorded. The observed sign of the diagnostic ICD (induced CD spectrum) curve at around 315 nm was correlated with the absolute configuration of C-24 in compound 1. Compound 2 was also dealt with the same method as 1.