New structures, chemotaxonomic significance and COX-2 inhibitory activities of cassane-type diterpenoids from the seeds of Caesalpinia minax

Abstract

Eleven new cassane-type diterpenoids (1–11), along with 12 known compounds were isolated from the seeds of Caesalpinia minax Hance. Their structures were determined by extensive spectroscopic methods in combination with X-ray diffraction analysis. All the compounds were evaluated for COX-2 inhibitory activity, and displayed different levels of inhibition except for 14 whose C and D rings were connected through a spiro-atom. Among them, compound 23, a furanoditerpenoid lactone with a hydroxyl group at C-1 displayed the most potent inhibitory activity with an inhibition ratio of 82.0% at 4.0 μM, which was stronger than its analog 16 with an additional hydroxyl group at C-2. Further detailed testing showed that both compounds 23 and 16 dose-dependently inhibited the COX-2 enzyme with IC₅₀ values of 2.4 ± 0.1 and 3.2 ± 0.2 μM, respectively. Molecular docking analysis showed significant hydrogen bonding and π–π interactions with the enzyme, and revealed the docking scores 36.3108 and 28.6678 for 23 and 16, respectively, which are consistent with their IC₅₀ values. Cytotoxic assay showed that all these diterpenoids only exhibit weak activities with IC₅₀ values over 50 μM on normal cells. Thus, these diterpenoids might be potential anti-inflammatory agents targeting the COX-2 enzyme with low toxicity.

---

Introduction

Cassane-type diterpenoids are characterized by a molecular skeleton comprised of three fused cyclohexane rings and a furan ring (or a butenolide).¹ These diterpenoids showed various interesting pharmacological activities, such as antimalarial,² antianaphylaxis,³ antinociceptive,⁴ antiinflammatory, antianalgesic,⁵ anticancer,⁶ antiviral,⁷ antimalarial,⁸ and plant growth regulation.⁹

Plants belonging to genus Caesalpinia are rich sources of cassane-type diterpenoids. C. minax Hance, a member of genus Caesalpinia, is a famous medicinal plant which widely distributed in tropical and subtropical regions. This plant is called“ku-shi-lian” in Chinese folk remedy, and its seeds were traditionally used for treating fever, common cold and dysentery; while in recently years, this herb is popularly used for the treatment of prostatitis.¹⁰ A preliminary study on the seeds collected in Guangdong province led to the isolation of caesalpinimin A, a novel rearranged furanoditerpene with an unprecedented carbon skeleton.¹¹ For the purpose of further search for bioactive components responsible for the clinical application of this herb, a systematic phytochemical study was undertaken on the EtOAc extract of the seeds of C. minax which led to the isolation of 11 new cassane-type diterpenoids (1–11), as well as 12 known compounds, norcaesalpinin E (12),² neocaesalpin B (13),¹² spirocaesalmin (14),¹³ 1-deacetoxy-1-oxocaesalmin C (15),¹⁴ caesalmins A, C, D, E, and F (16–20),¹⁵ and bonducellpins A, C, and D (21–23) (Fig. 1).¹⁶ The structure of these compounds were elucidated by extensive spectroscopic analysis. The three dimensional structures of compounds 1 and 13 were further confirmed by single crystal X-ray diffraction analysis. In addition, all of the compounds were evaluated for cytotoxic effects and COX-2 inhibitory activity. None of these compounds showed cytotoxic effect with IC₅₀ > 50 μM. In contrast, all compounds demonstrated different level of inhibitory activities on COX-2. Among them compound 23, a furanoditerpenoid lactone with a hydroxyl group at C-1, demonstrated the most potent inhibitory activities.

Fig. 1 Chemical structures of compounds 1–23.

Results and discussion

Compounds 1–23 were isolated from the EtOAc fraction of a 95% ethanol extract of the seeds of C. minax by column chromatography (CC) and semi-preparative HPLC purification.

Compound 1 was isolated as colorless crystals and its molecular formula was determined as C₂₃H₃₀O₈ from its HRESIMS data (m/z 435.2017 [M + H]⁺). The IR absorptions at 3567 and 1739 cm⁻¹ indicated the presence of hydroxyl and carbonyl groups, respectively. The ¹H-NMR spectrum showed the presence of three tertiary methyls at δ_H 1.12 (s, H₃-18), 1.16 (s, H₃-19) and 1.33 (s, H₃-20), two acetoxyl methyls at δ_H 2.04 (s) and 2.11 (s), and three oxymethines at δ_H 4.87 (t, J = 2.4 Hz, H-1), 5.42 (d, J = 9.0 Hz, H-6), and 4.24 (dd, J = 9.0, 10.2 Hz, H-7). The low-field doublets at δ_H 6.62 (d, J = 1.8 Hz, H-15) and 7.23 (d, J = 2.1 Hz, H-16) revealed the presence of a 1,2-disubstituted furan ring. Analysis of ¹³C NMR and DEPT spectra implied that 1 possessed 23 carbon atoms, composed of eight quaternary carbons, seven methines, three methylenes and five methyls. The carbon signals at δ_C 171.7, 172.5 and 197.9 indicated the existence of three carbonyls (including two ester carbonyls and one ketone carbonyl). The presence of a furan ring was confirmed by the carbon signals at δ_C 107.0, 120.9, 145.1, and 168.3. In addition, four carbon atoms bearing oxygenated functionalities at δ_C 76.2 (C-1), 79.9 (C-5), 77.4 (C-6) and 71.5 (C-7) were observed in the mid-field region. These data suggested that 1 possesses a cassane-furanoditerpenoid skeleton.¹⁵

Full assignments of all the ¹H and ¹³C-NMR signals (Tables S1 and S2†) of 1 were achieved by extensive analysis of the ¹H-¹H COSY, HSQC and HMBC data (Fig. 2). The ¹H-¹H COSY spectrum showed correlations H-1 ↔ H-2 ↔ H-3, H-6 ↔ H-7 ↔ H-8 ↔ H-9 ↔ H-11 and H-15 ↔ H-16 which were demonstrated in bold face in Fig. 2. In the HMBC spectrum, the correlations of H-1 (δ_H 4.82) with C-5 (δ_C 79.9) and the carbonyl carbon at δ_C 171.7, and those of H-6 (δ_H 5.42) with C-10 (δ_C 46.3) and the carbonyl carbon at δ_C 172.5 suggested that two acetoxyl groups were attached to C-1 and C-6, respectively. The HMBC correlations of H-7 (δ_H 4.24) with C-6 (δ_C 77.4) and C-8 (δ_C 51.9) indicated that the hydroxyl group was attached to C-7. Moreover, the position of ketone carbonyl at C-14 was determined by the HMBC correlations of H-7 (δ_H 4.24), H-8 (δ_H 2.63, m), and H-15 (δ_H 6.62) with C-14 (δ_C 197.9).

Fig. 2 Key ¹H-¹H COSY and HMBC correlations of compound 1 and 10.

The relative configuration of 1 was established by ROESY experiment (Fig. 3). The ROESY spectrum showed correlations between H-1 (δ_H 4.87) and H₃-20 (δ_H 1.33), H-1 and H₃-19 (δ_H 1.16), H₃-20 and H-6 (δ_H 5.42), and H-6 and H-8 (δ_H 2.63), indicating that those protons were in the same β-orientation. While NOE correlation between H-7 (δ_H 4.24) and H-9 (δ_H 3.11) revealed those protons were α-oriented. However, the configuration of 5-hydroxyl group could not be determined by ROESY spectrum. Fortunately, the single crystal of 1 was obtained from methanol solution, and final X-ray analysis revealed a 5α-orientation (Fig. 4). It was noteworthy that a suitable low Flack parameter 0.1(2) in the final refinement permitted the determination of the absolute configurations of 1 as 1S, 5R, 6S, 7R, 8S, 9S, and 10S. Thus, compound 1 was identified as 14-oxo-1α,6α-diacetoxy-5α,7β-dihydroxy-17-norvoucapan, and was accorded the trivial name caesalmin I.

Fig. 3 Key NOE correlations of compounds 1 and 10.

Fig. 4 Ortep plot of single crystal X-ray structure of 1.

Compound 2 was isolated as a white powder. Its molecular formula was identical to that of 1 by HR-ESI-MS at m/z 457.1848 [M + Na]⁺. The ¹H and ¹³C-NMR data of 2 are very similar to those of 1 (Tables S1 and S2†), except for those of H-6 and H-7, indicating 2 was an isomer of 1 with respect to the location of hydroxyl and the acetoxyl group. The HMBC correlations of H-6 (δ_H 3.94) with C-5 (δ_C 80.2) and C-7 (δ_C 74.8) and those of H-7 (δ_H 5.48) with C-14 and the carbonyl at 172.9 revealed the location of hydroxyl group at C-6 (δ_C 75.7) and acetoxyl group at C-7. The relative configurations of 2 were determined by the results of ROESY experiments, which was the same as those of 1. Thus, the structure of 2 was identified as 14-oxo-1α,7β-diacetoxy-5α,6α-dihydroxy-17-norvoucapan, and was accorded the trivial name caesalmin J.

Compound 3 was obtained as a white powder. The HRESIMS mass spectrum of 3 showed a quasimolecular ion [M + Na]⁺ at m/z 441.1890, corresponding to a molecular formula C₂₃H₃₀O₇ with nine degrees of unsaturation. All the ¹H and ¹³C NMR data of 3 (Tables S1 and S2†) was assigned unambiguously by an extensive analysis of ¹H-¹H COSY, HSQC and HMBC spectra, which were similar to those of 2, except for the absence of an oxymethine and the presence of a methylene (δ_H 2.29 and 1.58; δ_C 31.7). The ¹H-¹H COSY correlation between the methylene and H-7 (δ_H 5.59) implied 3 was the 6-dehydroxy derivative of 2. Thus, the structure of 3 was established as 14-oxo-1α,7β-diacetoxy-5α-hydroxy-17-norvoucapan, and was named caesalmin K.

Compound 4 was isolated as a white powder. Assignment of the molecular formula C₂₂H₃₀O₆ was based on HRESIMS (m/z 391.2216 [M + H]⁺). The ¹H NMR spectrum displayed three tertiary methyls at δ_H 1.19 (s, H₃-20), 1.28 (s, H₃-19) and 1.30 (H₃-18), one acetyl methyl at δ_H 2.03 (s), and three oxymethines at δ_H 4.82 (t, J = 2.1 Hz, H-1), 4.04 (t, J = 9.0 Hz, H-7), and 3.84 (d, J = 9.0 Hz, H-6). Two mutually coupled olefinic protons at δ_H 6.46 (d, J = 1.8 Hz, H-15) and 7.27(d, J = 1.8 Hz, H-16) indicated the presence of a 1,2-disubstituted furan ring. In addition to the signals for one acetoxyl group (δ_C 171.1 and 21.1), the ¹³C-NMR and DEPT spectrum of 4 showed 20 carbon signals including three methyl carbons (δ_C 31.8, 25.6 and 17.6), four olefinic carbon atoms of the furan ring (δ_C 152.3, 121.0, 107.50 and 142.9), two exocyclic double-bonded carbon atoms (δ_C 141.6 and 106.8), three methylene carbons, four oxygenated carbon atoms (δ_C 77.1, 80.0, 77.5 and 75.2), two other methine carbons, and two quaternary carbons. The full assignment and connectives were determined by ¹H-¹H COSY, HSQC and HMBC spectra (Tables S1 and S2†). The HMBC correlations of H-17 (δ_H 5.20 and 5.32, 1H each, br s) with C-8 (δ_C 44.6), C-13 (δ_C 121.0), C-14 (δ_C 141.6) and C-15 (δ_C 107.5) indicated that the exocyclic double bond conjugated with the furan ring and located between C-14 (δ_C 141.6) and C-17 (δ_C 106.8). In addition, the location of the acetoxyl group was determined to be C-1 (δ_C 77.1) by the HMBC correlations of H-1 (δ_H 4.82) with C-5 (δ_C 80.0) and the acetyl carbonyl at δ_C 171.1, and the hydroxyl groups were inferred to locate at C-6 (δ_C 77.5) and C-7 (δ_C 75.2) respectively on the basis of the HMBC correlations of H-6 (δ_H 3.84) with C-8 (δ_C44.6) and C-10 (δ_C 45.5), as well as H-7 (δ_H 4.04, t, J = 9.3 Hz) with C-5 (δ_C 80.0) and C-9 (δ_C 39.4). The relative stereochemistry of compound 4 was assigned by the NOE correlations. Therefore, the structure of 4 was established as 1α-acetoxy-5α,6α,7β-trihydroxyvoucapan-14(17)-ene, and was accorded the trivial name caesalmin L.

Compound 5 was isolated as an amorphous solid. Its molecular formula was determined to be C₂₂H₃₀O₅ by HRESIMS data (m/z 375.2168 [M + H]⁺). The NMR spectra of 5 were very similar to those of 4, except for the absence of one oxygenated methine and the presence of a methylene (δ_H 2.15 and 1.67, 1H each; δ_C 34.9). The location of the hydroxy group was deduced to be at C-7 (δ_C 67.9) on the basis of the HMBC correlations of H-7 (δ_H 4.45) with C-9 (δ_C 38.5) and C-14 (δ_C 140.4). The relative configuration of 5 was the same as that of 4 by as revealed by ROESY experiment. Thus, the structure of 5 was established as 1α-acetoxy-5α,7β-dihydroxyvoucapan-14(17)-ene, and was accorded the trivial name caesalmin M.

Compound 6 was obtained as an amorphous solid. The HR-ESI-MS of 6 showed a quasimolecular ion peak at m/z 443.2038 [M + Na]⁺, ascribable to the molecular formula C₂₃H₃₂O₇. The ¹H and ¹³C NMR spectra of 6 displayed three tertiary methyls, one acetyl methyl, four methylenes, three aliphatic methines, two oxymethines, a methoxycarbonyl group, and a 1,2-disubstituted furan ring. The methoxycarbonyl group was assigned to C-14 on the basis of the HMBC correlations of 17-OCH₃ (δ_H 3.72, s) with C-17 (δ_C 174.9) and the HMBC correlations of H-14 (δ_H 3.39) with C-17 (δ_C 174.9). In addition, the HMBC correlations of H-7 (δ_H 5.13) with C-5 (δ_C 79.8), C-14 (δ_C 46.2) and the ester carbonyl at δ_C 170.5 indicated the acetoxyl group was located at C-7 (δ_C 76.9). Furthermore, the location of the hydroxyl group was deduced to be at C-1 by the ¹H-¹H COSY correlation between H-1 (δ_H 3.72) and H-2 (δ_H 2.03 and 1.68) and the HMBC correlation of H₃-20 (δ_H 3.72, br s) with C-1 (δ_C 72.3). The relative stereochemistry of compound 6 was determined through analysis of its ROESY spectrum. H-14 (δ_H 3.39) had cross peaks with H-7 (δ_H 5.13) and H-9 (δ_H 2.92), suggesting they were α-oriented. Meanwhile the cross peak between H-1 with H₃-20 indicated that they were β-oriented. Therefore, the structure of 6 was established as methyl 1α,5α-dihydroxy-7β-acetoxyvoucapan-17-oate, and was accorded the trivial name caesalmin N.

Compound 7 was obtained as an amorphous solid. Its molecular formula C₂₃H₃₀O₇ was determined by HRESIMS data (m/z 419.2064 [M + H]⁺). The NMR data for 7 (Tables S1 and S2†) were similar to those of 6, except for the absence of one oxymethine and the presence of a ketone carbonyl signal δ_C 216.1. The HMBC correlations of H-2 (δ_H 2.08 and 2.96), H-3 (δ_H 1.63 and 2.06), and H₃-20 (δ_H 1.49) with the ketone carbonyl carbon at δ_C 216.1 revealed its location was C-1. Accordingly, the structure of 7 was identified as 1-keto derivative of 6, and was accorded the trivial name caesalmin O.

Compound 8 was obtained as colorless amorphous solid. Assignment of the molecular formula C₂₄H₃₄O₈ was based on HRESIMS data (m/z 473.2146, [M + Na]⁺). The ¹H and ¹³C-NMR spectra of 8 are similar to those of 1, except for the absence of the ketone group in 1 and the presence of one oxygenated quaternary carbon (δ_C 72.9) as well as one tertiary methyl (δ_H 1.49, 3H, s; δ_C 25.8). The HMBC correlations of H₃-17 (δ_H 1.49) with C-8 (δ_C 48.8), C-13 (δ_C 124.2) and the oxygenated quaternary carbon (δ_C 72.9) revealed the location of one methyl and one hydroxyl at C-14. The NOE correlations between H₃-17 and H-7 (δ_H 4.44), and H₃-17 and H-9 (δ_H 2.60), revealed H₃-17 was α-oriented. Thus, the structure of compound 8 was deduced as 1α, 6α-diacetoxy-5α,7β,14β-trihydroxy-vouacapan, and was accorded the trivial name caesalmin P.

Compound 9 was isolated as colorless powder. The molecular formula C₂₂H₃₀O₇ was based on HRESIMS data (m/z 429.1894 [M + Na]⁺). The NMR data of 9 were similar to those of 5, except for the absence of signals for an exocyclic double bond and the presence of an aldehyde group (δ_H 9.47, s; δ_C 201.5) and an oxygenated quaternary carbon (δ_C 76.2) in 9. In the HMBC spectrum, the correlations of H-17 (δ_H 9.47) with C-8 (δ_C 47.4), C-13 (δ_C 116.8) and C-14 (δ_C 76.2) indicated the location of the aldehyde and hydroxyl groups at C-14. The NOE correlations between H-17 (δ_H 9.47) and H-8 was observed, indicating the α-orientation for the hydroxyl group at C-14, and β-orientation for the aldehyde group. Accordingly, the structure of 9 was identified as 1α-acetoxy-5α,7β,14α-trihydroxyvoucapan-14β-al, and was accorded the trivial name caesalmin Q.

Compound 10 was obtained as colorless powder, whose molecular formula was established as C₂₄H₃₄O₈ by the positive ion peak at m/z 473.2146 [M + Na]⁺ in HRESIMS. The ¹H NMR spectra (Table S3†) displayed four methyl groups at δ_H 1.23 (d, J = 7.2 Hz, H₃-17), 1.04 (s, H₃-18), 1.05 (s, H₃-19), and 1.05 (s, H₃-20), two oxymethine at δ_H 4.98 (br s, H-1) and 5.41 (td, J = 10.5, 4.5, H-7), and one olefinic proton at δ_H 5.74 (s). The ¹³C NMR and DEPT spectra exhibited 24 carbon signals, including two acetoxyl groups (δ_C 169.8 and 21.2; 170.3 and 21.2), four methyls (δ_C 12.3, 17.0, 25.0 and 28.1), four methylenes (δ_C 22.6, 29.9, 31.2, and 37.9), six methines (δ_C 75.1, 70.6, 44.0, 32.7, 43.3, and 114.0), and six quaternary carbons (δ_C 38.3, 78.5, 43.3, 104.8, 170.3, and 171.7). The olefinic proton signal at H-15 (δ_H 5.74, s) and downfield carbon signals at δ_C 170.3 (C-13), 114.0 (C-15), and 171.7 (C-16) indicated the presence of an α,β-unsaturated butenolide moiety.¹⁷ The locations of two acetoxyl groups and hydroxyl group were determined to be at C-1, C-7 and C-12 on the basis of the HMBC correlations (Fig. 2). Moreover the ¹H-¹H COSY correlations of H₃-17 (δ_H 1.23) with H-14 (δ_H 2.76), and the HMBC correlations of H₃-17 (δ_H 1.23) with C-14 (δ_C 31.8), C-8 (δ_C 44.0) and C-13 (δ_C 170.3), indicated that the methyl group (δ_H 1.23 and δ_C 12.3) was located at C-14. The relative configuration of 10 was established by ROESY experiment (Fig. 3). The ROESY spectrum showed correlations between H-1 (δ_H 4.98) and H₃-20 (δ_H 1.05), H-1 and H₃-19 (δ_H 1.05), H₃-20 and H-6β (δ_H 5.42), and H-6β and H-8 (δ_H 1.77), indicating that those protons were in the same β-orientation. While NOE correlation between H-7 (δ_H 5.41) and H-9 (δ_H 3.14), and H-7 and H-17 (δ_H 1.23) revealed those protons were α-oriented. In addition, the configuration of 12-hydroxyl group was inferred to have the same stereochemistry as that of neocaesalpin B (13, Fig. 5) by comparison with their NMR spectrums. Accordingly, the structure of 10 was established as 1α,7β-diacetoxy-5α,12α-dihydroxy-14α-methylcassa-13(15)-en 16,12-olide, and was accorded the trivial name caesalmin R.

Fig. 5 Ortep plot of X-ray structure of 13.

Compound 11 was obtained as white powder. Assignment of the molecular formula C₂₇H₃₈O₁₀ was based on its HRESIMS data (m/z 545.2369 [M + Na]⁺). The ¹H and ¹³C NMR data (Table S3†) of compound 11 were similar to those of compound 10, showing the characteristic signals for an α,β-unsaturated γ-lactone moiety [δ_H 5.78 (d, J = 1.8 Hz, H-15); δ_C 163.2 (C-13), 117.5 (C-15), 168.6 (C-16)] and two acetoxyl groups at C-1 and C-7. Furthermore, the HMBC correlations of H-14 (δ_H 3.16) with C-17 (δ_C 170.8) and of the methoxy group at δ_H 3.78 with C-17 indicated the existence of a carbomethoxyl group at C-14. Moreover, obvious proton signals of an ethoxyl group [δ_H 3.16 (m, 1H), 3.43 (m, 1H); 1.16 (t, J = 7.2 Hz, 3H)] were observed in the ¹H NMR spectrum, which was determined to locate at C-12 by the ¹H-¹H COSY correlation between the methylene (δ_H 3.16, 3.43) and the methyl (δ_H 1.16), as well as the HMBC correlations of the methylene with C-12 (δ_C 106.5). The relative stereochemistry of compound 11 was determined through analysis of its NOESY spectrum. H-14 (δ_H 3.16) showed cross peaks with H-7 (δ_H 5.83), H-9 (δ_H 2.80) and 12-ethoxyl group, suggesting they were α-oriented; while the correlations between H-8 and CH₃-10 indicated they were β-oriented. Therefore, compound 11 was elucidated as 1α, 7β-diacetoxy-5α-hydroxy-12α-ethoxyl-14β-carboxymethylcassa-3(15)-en-16,12-olide, and was accorded the trivial name caesalmin S.

Other known compounds were identified by comparison of the physical and spectroscopic data with the reported values in literatures. During the isolation process, besides the common solvents, no base or acid was used. Thus compounds 1–23 should be natural secondary metabolites, which were consistent with the structural types reported in genus Caesalpinia.¹⁸ It is noteworthy that though the structure of 13 was reported before, its absolute configuration was not determined. Fortunately single crystals of 13 were obtained from the methanol solution. Final refinement from the Cukα data resulted in a Flack parameter 0.2(3) with large deviation, which indicated that only the relative configuration could be assigned (Fig. 5). However, considering that the methyl at C-10 is β-oriented for all the cassane furanoditerpenoids, the absolute configuration of 13 could be assigned as shown in Fig. 5.

Cassane furanoditerpenes mainly include the caesalpin-type and neocaesalpin-type, which are characterized by a molecular skeleton constructed from the fusion of three cyclohexane rings with a furan ring or an α,β-unsaturated lactone ring, respectively. Diterpenes of this structural class, are mostly distributed in the genera Caesalpinia¹⁸ and Pterodon.¹⁹ Currently, more than 100 cassane furanoditerpenes were isolated from these two genera. In most cases, furanoditerpenoids from the former genus possess a hydroxyl group at C-5; however, it was absent in the latter genus. This is the major difference between the furanoditerpenoids from these two genera. Our present study showed that though the furanoditerpenoids from Guangdong province are different from those collected from Guangxi province, however, all of them possess a C-5 hydroxyl group. Thus cassane furanoditerpenoid with a C-5 hydroxyl group might be a unique chemotaxonomic marker for genus Caesalpinia. It is noteworthy that the cassane furanoditerpenoid skeleton is quite stable under normal extraction and isolation conditions, and the acetyl, hydroxyl and ketone groups are common in both genera.^18,19 Thus new compounds isolated in the study should not be artifacts.

Considering the pronounced treatment effect of Caesalpinia minax on prostatitis,¹⁰ compounds 1–23 were evaluated for COX-2 inhibitory activity using an enzyme immunoassay (EIA) kit (catalog no. 56131, Cayman Chemical, Ann Arbor, MI). Most of the isolated compounds showed moderate inhibitory activity (Table 1) at a concentration of 4 μM. Among them, the furanoditerpenoid lactone 23 with a hydroxyl group at C-1 displayed the most potent inhibitory activity with an inhibition ratio of 82.0%, which was weaker than that of celecoxib, a known inhibitor of COX-2.²⁰ Compound 16, 2-hydroxyl analog of 23, showed a lower inhibition ratio 71.6%. Other compounds without the lactone functionality displaced inhibitions less than 70%. While compound 14, a structurally rearranged furanoditerpenoid with a spiro atom at C-12, was inactive. Due to the similar structures but different potency, we further tested compounds 16 and 23 at several concentration levels, which showed that both compounds could dose-dependently inhibit the COX-2 enzyme with IC₅₀ values of 3.2 ± 0.2 and 2.4 ± 0.1 μM (for dose-dependent curves, see ESI†), respectively.

Table 1 The inhibitory activity against COX-2 of compounds 1–23 at a concentration of 4.0 μM

Compounds	Inhibition ratio (%)	Compounds	Inhibition ratio (%)
1	67.8	13	66.4
2	48.3	14	0
3	51.4	15	27.8
4	54.7	16	71.6
5	34.2	17	41.9
6	24.1	18	40.6
7	61.8	19	61.9
8	65.4	20	44.8
9	52.2	21	42.6
10	49.4	22	65.2
11	30.0	23	82.0
12	28.4	Celecoxib	99.0

---

Molecular docking analysis was performed to compare the binding modes of compounds 23 and 16 with COX-2 and investigate the role of lactone functionality. The result indicates that both compounds 23 and 16 adopted similar binding mode with COX-2 (Fig. 6). Both of them can form two H-bonds with TYR354 and VAL492 of COX-2 as well as a π–π interaction with PHE487. However, the docking score for 23 and 16 were 36.3108 and 28.6678, respectively; indicating that 23 might has stronger van der Waals interactions with the enzyme and less hydrophobic repulsion as compared with those of 16. The docking scores of 23 and 16 are consistent with their IC₅₀ values. It is noteworthy that in both docking diagrams of 23 and 16, the lactone between C-7 and C-14 functioned as the hydrogen bond acceptor, confirming that the lactone functionality played an important role in the binding interactions as revealed by the enzymatic assay.

Fig. 6 Molecular docking of 23 and 16 with COX-2 (score for 23: 36.3108; score for 16: 28.6678). (A) Docking diagram of 23 with COX-2; (B) binding mode of 23 with COX-2; (C) docking diagram of 16 with COX-2; (D): binding mode of 16 with COX-2. Residues involved in hydrogen-bond, charge or polar interactions are represented by pink circles; residues involved in van der Waals interactions are represented by green circles; the solvent accessible surface of an interacting residue is represented by a blue halo around the residue, and the diameter of the circle is proportional to the solvent accessible surface; hydrogen-bond interactions with amino acid side-chains are represented by a blue dashed arrow directed towards the electron donor and a green dashed arrow directed towards the electron acceptor; Pi interactions are represented by an orange line with symbols indicating the interaction.

Normally, the anti-inflammatory agents should be non-toxic. To test if these compounds had cytotoxicity against normal cells, the inhibitory activities of these compounds on a normal cell line (Vero) were evaluated. The results showed that all these diterpenoids only exhibit weak inhibitory activities with IC₅₀ values over 50 μM. Thus, these diterpenoids might be potential anti-inflammatory agents targeting the COX-2 enzyme with low toxicity.

Experimental

General experimental procedures

Ultraviolet (UV) spectra were determined by V-550 UV/vis spectrophotometer. Infrared (IR) spectra were measured on a Jasco FT/IR-480 plus Fourier transform infrared spectrometer using KBr pellet. HR-ESI-MS data were obtained on an Agilent 6210 ESI/TOF mass spectrometer. Optical rotation was recorded in CH₃Cl₃ on Jasco P-1020 polarimeter at room temperature. Nuclear magnetic resonance (NMR) spectra were measured on Bruker AV-300 or AV-400 spectrometers. Thin-layer chromatography (TLC) analyses were carried out using pre-coated silica gel GF₂₅₄ plates (Qingdao Marine Chemical Plant, Qingdao, People's Republic of China). Column chromatographies were performed on silica gel (200–400 mesh, Qingdao Marine Chemical Plant, Qingdao, P. R. China), reverse-phase C₁₈ silica gel (Merck, Darmstadt, Germany) and Sephadex LH-20 (Pharmacia Biotec AB, Uppsala, Sweden). All solvents used in column chromatography and high-performance liquid chromatography (HPLC) were of analytical (Shanghai Chemical Plant, Shanghai, People's Republic of China) grade and chromatographic grade (Fisher Scientific, NJ, USA), respectively.

Plant material

The seeds of C. minax were collected and authenticated as described previously.¹¹

Extraction and isolation

The dried ground seeds (5 kg) were extracted with 95% ethanol. The extract was combined, filtered, and then concentrated under reduced pressure to afford a crude residue (664 g), which was suspended in water and partitioned with cyclohexane, EtOAc and n-BuOH to eventually result in cyclohexane fraction, EtOAc fraction, n-BuOH fraction and water soluble, respectively. The EtOAc fraction (65 g) was subjected to silica gel (200–300 mesh), eluted with cyclohexane-acetic ether (50 : 1, 40 : 1, 30 : 1, 20 : 1, 15 : 1, 10 : 1, 5 : 1, 3 : 1 and 1 : 1) to give 10 fractions (Fr. 1 to Fr. 10). Fr. 3 was subjected to column chromatography over silica gel (300–400 mesh, 1 × 60 cm) and eluted with gradient solvent system cyclohexane–acetic ether (30 : 1, 20 : 1, 15 : 1, 10 : 1, 5 : 1, 3 : 1) to give Fr. 3a–f. Fr. 3b was further purified by column chromatography over silica gel with cyclohexane–acetone (25 : 1, 15 : 1, 10 : 1, 5 : 1, 2 : 1), yielding compounds 6 (6.1 mg), 7 (3.2 mg), 14 (9.3 mg) and 15 (18.0 mg). Fr. 3e was purified by preparative HPLC using CH₃CN–H₂O (65 : 35) as the eluent to yield 12 (4.0 mg), 16 (3.4 mg) and 18 (12.0 mg). Fr. 4 was subjected to reverse-phase C₁₈ silica gel eluted with a gradient of MeOH–H₂O (20 : 80 to 90 : 10) to afford a mixture, which was then purified by preparative HPLC using CH₃CN–H₂O (55 : 45) as the eluent to yield 8 (3.6 mg), 2 (4.5 mg), 3 (5.4 mg) and 17 (7.2 mg). Fr. 5 was separated by Sephadex LH-20 eluting with chloroform–methanol (1 : 1), further purified by preparative HPLC using CH₃CN–H₂O (50 : 50) as the eluent to yield 4 (6.0 mg) and 1 (8.1 mg). Fr. 6 was chromatographed on reverse-phase C₁₈ silica gel eluted with MeOH–H₂O (20 : 80 to 90 : 10) to give Fr. 6a–e. Fr. 6c was subjected to column chromatography over silica gel (300–400 mesh, 1 × 60 cm) and eluted with gradient solvent system cyclohexane–acetic ether (15 : 1, 10 : 1, 5 : 1, 3 : 1, 1 : 1) to give 13 (5.3 mg), 19 (3.0 mg), 21 (4.0 mg) and 23 (6.0 mg). Fr. 6d was purified by preparative HPLC using CH₃CN–H₂O (45 : 55) as the eluent to yield 10 (5.3 mg) and 20 (4.1 g). Fr. 6e was separated via preparative HPLC with a CH₃CN–H₂O (45 : 55) as the eluent, and then purified by Sephadex LH-20 eluting with chloroform–methanol (1 : 1) to give 11 (6.0 mg) and 22 (3.0 mg). Fr. 7 was subjected to Sephadex LH-20 eluting with chloroform–methanol (1 : 1), and then purified by preparative HPLC using CH₃CN–H₂O (40 : 60) as the eluent to yield 5 (3.0 mg) and 9 (4.2 mg).

Caesalmin I (1). Colorless crystal; [α]²⁰_D +23.5° (c 0.1, CHCl₃); IR (KBr) ν_max: 3567, 1739 cm⁻¹; ESI-MS: m/z 457.5 [M + Na]⁺, 891.3 [2M + Na]⁺; HR-ESI-MS: m/z 435.2017 [M + H]⁺ (calcd for C₂₃H₃₁O₈: 435.2013); for ¹H and ¹³C NMR data, see Tables S1 and S2.†

Caesalmin J (2). White powder; [α]²⁰_D +9.0° (c 0.05, CHCl₃); IR (KBr) ν_max: 3575, 1728 cm⁻¹; ESI-MS: m/z 457.5 [M + Na]⁺, 891.3 [2M + Na]⁺; HR-ESI-MS: m/z 457.1848 [M + Na]⁺ (calcd for C₂₄H₃₀O₈Na: 457.1833); for ¹H and ¹³C NMR data, see Tables S1 and S2.†

Caesalmin K (3). White powder; [α]²⁰_D +10.5° (c 0.05, CHCl₃); IR (KBr) ν_max: 3576, 1739 cm⁻¹; ESI-MS: m/z 441.4 [M + Na]⁺, 859.2 [2M + Na]⁺; HR-ESI-MS: m/z 441.1890 [M + Na]⁺ (calcd for C₂₃H₃₀O₇Na: 441.1884); for ¹H and ¹³C NMR data, see Tables S1 and S2.†

Caesalmin L (4). White powder; [α]²⁰_D +14.7° (c 0.05, CHCl₃); IR (KBr) ν_max: 3558, 1734 cm⁻¹; ESI-MS: m/z 413.5 [M + Na]⁺, 803.3 [2M + Na]⁺; HR-ESI-MS: m/z 391.2216 [M + H]⁺ (calcd for C₂₂H₃₁O₆: 391.2115); for ¹H and ¹³C NMR data, see Tables S1 and S2.†

Caesalmin M (5). Amorphous solid; [α]²⁰_D +7.2° (c 0.05, CHCl₃); IR (KBr) ν_max: 3558, 1739 cm⁻¹; HR-ESI-MS: m/z 375.2168 [M + H]⁺ (calcd for C₂₂H₃₁O₅: 375.2166); for ¹H and ¹³C NMR data, see Tables S1 and S2.†

Caesalmin N (6). Amorphous solid; [α]²⁰_D +46.2° (c 0.1, CHCl₃); IR (KBr) ν_max: 3576, 1726 cm⁻¹; ESI-MS: m/z 443.3 [M + Na]⁺, 863.1 [2M + Na]⁺; HR-ESI-MS: m/z 443.2038 [M + Na]⁺ (calcd for C₂₃H₃₂O₇Na: 443.2046); for ¹H and ¹³C NMR data, see Tables S1 and S2.†

Caesalmin O (7). Amorphous solid; [α]²⁰_D +57.3° (c 0.1, CHCl₃); IR (KBr) ν_max: 3575, 1733 cm⁻¹; ESI-MS: m/z 441.4 [M + Na]⁺, 859.3 [2M + Na]⁺; HR-ESI-MS: m/z 419.2073 [M + H]⁺ (calcd for C₂₃H₃₁O₇: 419.2064); for ¹H and ¹³C NMR data, see Tables S1 and S2†.

Caesalmin P (8). Colorless amorphous solid; [α]²⁰_D +17.8° (c 0.05, CHCl₃); IR (KBr) ν_max: 3575, 1725 cm⁻¹; ESI-MS: m/z 473.5 [M + Na]⁺, 923.5 [2M + Na]⁺; HR-ESI-MS: m/z 473.2152 [M + Na]⁺ (calcd for C₂₄H₃₄O₈Na: 473.2146); For ¹H and ¹³C NMR data, see Tables S1 and S2†.

Caesalmin Q (9). White powder; [α]²⁰_D +8.6° (c 0.05, CHCl₃); IR (KBr) ν_max: 3596, 1728 cm⁻¹; HR-ESI-MS: m/z 429.1894 [M + Na]⁺ (calcd for C₂₂H₃₀O₇Na: 429.1884); for ¹H and ¹³C NMR data, see Tables S1 and S2.†

Caesalmin R (10). White powder; [α]²⁰_D −8.3° (c 0.05, CHCl₃); IR (KBr) ν_max: 3577, 1742 cm⁻¹; ESI-MS: m/z 473.5 [M + Na]⁺, 923.5 [2M + Na]⁺; HR-ESI-MS: m/z 473.2152 [M + Na]⁺ (calcd for C₂₄H₃₄O₈Na: 473.2146); for ¹H and ¹³C NMR data, see Table S3.†

Caesalmin S (11). White powder; [α]²⁰_D −9.9° (c 0.05, CHCl₃); IR (KBr) ν_max: 3577, 1741 cm⁻¹; ESI-MS: m/z 545.3 [M + Na]⁺, 1067.0 [2M + Na]⁺; HR-ESI-MS: m/z 545.2369 [M + Na]⁺ (calcd for C₂₇H₃₈O₁₀Na: 545.2357); for ¹H and ¹³C NMR data, see Table S3.†

X-ray analysis

The structures were solved by using direct methods (SHELXTL version 5.1) and refined by using full-matrix least-squares treatment on F². In the structure refinements, non-hydrogen atoms were refined anisotropically. Hydrogen atoms bonded to carbons were placed at geometrically ideal positions by using the ‘ride on’ method. Hydrogen atoms bonded to oxygen were located by employing the difference Fourier method and were included in the calculation of structure factors with isotropic temperature factors.

Caesalmin I (1). Colorless blocks, formula: C₂₃H₃₀O₈, formula weight: 434.47. Orthorhombic P2₁2₁2₁, T = 173 (2) K, a = 7.6369 (3), b = 9.8408 (4), c = 28.3369 (12) Å, V = 2129.61 (15) ų, Z = 4, d_x = 1.355 mg cm⁻³, F(000) = 928, absorption coefficient = 0.849, theta range for data collection: 3.12 to 60.81, final R indices [I > 2sigma(I)]: R₁ = 0.0335, wR₂ = 0.0853, R indices (all data): R₁ = 0.0370, wR₂ = 0.0891.

Neocaesalpin B (13). Colorless blocks, formula: C₂₄H₃₈O₉, formula weight: 470.51. Monoclinic P2₁ T = 173 (2) K, a = 6.7833(5), b = 12.2913(7), c = 14.6874(7) Å, β = 101.192(5), V = 1201.28(13)ų, Z = 2, d_x = 1.295 mg cm⁻³, F(000) = 504, absorption coefficient = 0.819, theta range for data collection: 3.07 to 62.98, final R indices [I > 2sigma(I)]: R₁ = 0.0433, wR₂ = 0.1198, R indices (all data): R₁ = 0.507, wR₂ = 0.1448.

COX-2 inhibitory activity assay

Compound 1–23 was evaluated for COX-2 inhibitory activity at 4 μM using an enzyme immunoassay kit (for detailed method, see ESI†).¹¹

Molecular docking

Ligand and protein structure preparation. The crystal structure of COX-2 (ID: 1CX2) was obtained from RSCB protein data bank. The water should be removed from the crystal structure except the one existing in active site. Meanwhile, the original ligand of 1CX2 was extracted from complex structure as the reference ligand. In addition, the charge of the receptor was calculated in MMFF94 force field without energy optimization. The 3D structure of 16 and 23 were sketched by Sybyl 8.0 (Tripos Inc., St. Louis, Missouri, USA). Energy optimization and calculation of charge were performed when preparation of these two ligand structures.

Docking analysis. The docking simulation process was accomplished by GOLD (Genetic Optimization for Ligand Docking), which is a genetic algorithm for docking flexible ligands into protein binding sites. In this study, the binding site residues of COX-2 were determined by the reference ligands in complex structure. Then compounds 16 and 23 were docked into the binding site according to the default parameters. The docking score is applied to quantify the interactions between ligands and receptor as well as identify the preferable docking conformation. The suitable docking conformation of compounds 16 and 23 were subsequently applied to further analyze the interactions with COX-2.

Cytotoxicity assay

The MTT assay for determination of the cytotoxicity was performed as described previously.²¹ Briefly, Vero cells were plated into 96-well plates at a density of 3 × 10⁵ cells per well. After 48 h of exposure, the cells were stained with MTT. Absorbance at 570 nm was measured on a multiplate reader.

Conclusions

In summary, eleven new (1–11) and 12 known cassane-type furanoditerpenoids were isolated from the seeds of Caesalpinia minax Hance. Compounds 1–9 were constructed from fusion of three cyclohexane rings and a furan ring; while compounds 10 and 11 was characterized by the tri-carbocyclic derivatives fused with an α,β-butenolide ring. All the isolated compounds were evaluated for COX-2 inhibitory activity and cytotoxic effects. Most of them displaced different level of inhibitory activities against the COX-2 enzyme. Among them, compound 23, a furanoditerpenoid lactone with a hydroxyl group at C-1 displayed the most potent inhibitory activity with an inhibition rate of 82.0% at 4 μM, which was stronger than an analog 16 with an additional hydroxyl group at C-2 (inhibition rate 71.6%). Further detailed testing showed that both compounds dose-dependently inhibited the COX-2 enzyme with IC₅₀ values of 3.2 ± 0.2 and 2.4 ± 0.1 μM, respectively. Molecular docking analysis showed significant hydrogen bonding and π–π interaction with the enzyme, and revealed the docking scores 36.3108 and 28.6678 for 23 and 16, respectively, which are consistent with their IC₅₀ values. Cytotoxic assay showed that all these diterpenoids only exhibit weak activities with IC₅₀ values over 50 μM on normal cells. Thus, these diterpenoids might be potential anti-inflammatory agents targeting the COX-2 enzyme with low toxicity.

Acknowledgements

This study was supported financially by Guangdong high level talent scheme (R.W.J.) and Ph.D. Programs Foundation of Ludong University (LY2015012).

Notes and references

1. R. W. Jiang, P. P. H. But, S. C. Ma, W. C. Ye, S. P. Chan and T. C. W. Mak, Tetrahedron Lett., 2001, 43, 2415–2418.
2. T. Z. Linn, S. A. Awale, Y. Tezuka, A. H. Banskota, S. K. Kalauni, F. Attamimi, J. Ueda, P. B. S. Asih, D. Syafruddin, K. Tanaka and S. Kadota, J. Nat. Prod., 2005, 68, 706–710.
3. K. Ogawa, I. Aoki and Y. Sashida, Phytochemistry, 1992, 31, 2897–2898.
4. I. D. G. Duarte, D. L. Ferreira-Alves, D. P. Piló-Veloso and M. Nakamura-Craig, J. Ethnopharmacol., 1996, 55, 13–18.
5. M. M. M. Rubinger, D. Pilo-Veloso, G. M. Stefani, D. L. F. Alves and C. R. Maltha, J. Braz. Chem. Soc., 1991, 2, 124–127.
6. G. X. Ma, X. D. Xu, L. Cao, J. Q. Yuan, J. S. Yang and L. Y. Ma, Planta Med., 2012, 78, 1363–1369.
7. R. W. Jiang, S. C. Ma, Z. D. He, X. S. Huang, P. P. H. But, H. Wang, S. P. Chan, V. E. C. Ooi, H. X. Xu and T. C. W. Mak, Bioorg. Med. Chem., 2002, 10, 2161–2170.
8. T. Z. Linn, S. Awale, Y. Tezuka, A. H. Banskota, S. K. Kalauni, F. Attamimi, J. Y. Ueda, P. B. Asih, D. Syafruddin, K. Tanaka and S. Kadota, J. Nat. Prod., 2005, 68, 706–710.
9. J. D. Antonio, C. A. B. Luiz, P. V. Dorila and L. F. A. Dalton, J. Nat. Prod., 1996, 59, 770–772.
10. M. J. Gong, Y. Y. Xie and Z. J. Zou, J. Integr. Tradit. West. Med., 2011, 6, 938–939.
11. J. L. Zhang, H. Y. Tian, N. H. Chen, X. Y. Bai, J. Li, R. R. Zhang, R. B. Wu and R. W. Jiang, RSC Adv., 2014, 4, 7440–7443.
12. T. Kinoshita, M. Kaneko, H. Noguchi and I. Kitagawa, Heterocycles, 1996, 43, 409–414.
13. R. W. Jiang, S. C. Ma, P. P. H. But and T. C. W. Mak, J. Chem. Soc., Perkin Trans. 1, 2001, 2920–2923.
14. S. K. Kalauni, S. Awale, Y. Tezuka, A. H. Banskota, T. Z. Linn and S. Kadota, Chem. Pharm. Bull., 2005, 53, 1300–1304.
15. R. W. Jiang, S. C. Ma, P. P. H. But and T. C. W. Mak, J. Nat. Prod., 2001, 64, 1266–1272.
16. S. R. Peter, W. F. Tinto, S. McLean, W. F. Reynolds and M. Yu, J. Nat. Prod., 1997, 60, 1219–1221.
17. T. Kinoshita, Y. Haga, S. Narimatsu, M. Shimada and Y. Goda, Chem. Pharm. Bull., 2005, 53, 717–720.
18. (a) M. Wu, Y. F. Wang, M. L. Zhang, C. H. Huo, M. Dong, Q. W. Shi and H. Kiyota, Chem. Biodiversity, 2011, 8, 1370–1399; (b) J. L. Zanin, B. A. de Carvalho, P. S. Martineli, M. H. dos Santos, J. H. Lago, P. Sartorelli, C. J. Viegas and M. G. Soares, Molecules, 2012, 17, 7887–7902.
19. (a) M. M. M. Rubinger, P. D. A. Castelo-Branco, S. Guilardi, E. M. R. Souza, M. T. D. P. Gambardella, E. E. L. Borges, D. L. Ferreira-Alves and D. Piló-Veloso, J. Braz. Chem. Soc., 2004, 15, 219–223; (b) H. Daiane, H. Mitsue and A. Antonio, Braz. J. Pharm. Sci., 2010, 46, 607–616.
20. C. Waskewich, R. D. Blumenthal, H. Li, R. Stein, D. M. Goldenberg and J. Burton, Cancer Res., 2002, 62, 2029–2033.
21. T. Mosmann, J. Immunol. Methods, 1983, 65, 55–63.

---

Footnote

† Electronic supplementary information (ESI) available: ¹H and ¹³C NMR data, HR-ESI-MS, 1D and 2D NMR spectra of compounds 1–11. CCDC 1033413 and 1033414. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5ra14221k

---