Soft corals have proven to be a rich source of terpenoids [1]. We previously have isolated a series of bioactive cembrane- [2,3,4] and norcembrane- [5,6,7,8] diterpenoids from the Formosan soft corals of the genus Sinularia. Previous chemical investigations on the soft coral Sinularia crassa have led to the isolation of structurally unique steroids and cembranoids, from which some have been shown to exhibit anti-inflammatory [9,10] and 5α-reductase inhibitory activities [11], respectively. Our continuing chemical investigation on S. crassa has led to the isolation of one new sterol, crassarosterol A (1), and four new steroidal glycosides, crassarosterosides A–D (2–5) (Figure 1). The structures of 1–5 have been established by extensive spectroscopic analysis, including 2D NMR (¹H–¹H COSY, HMQC, HMBC, and NOESY), chemical methods, and HPLC. The effect of 1–4 on the expression of the pro-inflammatory inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2) proteins in lipopolysaccharide (LPS)-stimulated RAW264.7 macrophage cells and the cytotoxicity of compounds 1–5 against a panel of cancer cell lines including human liver carcinoma (HepG2 and HepG3), human breast carcinoma (MCF-7 and MDA-MB-231), and human lung carcinoma (A-549) were evaluated in order to discover bioactive natural products.

The HRESIMS of crassarosterol A (1) exhibited a pseudomolecular ion peak at m/z 453.3342 [M + Na]⁺, consistent with a molecular formula of C₂₈H₄₆O₃ and requiring six degrees of unsaturation. The ¹³C NMR and DEPT spectroscopic data (Table 1) displayed 28 carbon signals, including five methyls, nine methylenes, ten methines, and four quaternary carbons. The IR spectrum revealed the presence of hydroxy groups (3389 cm^–1). The carbon resonances at δ 141.2 (C), 121.3 (CH), 156.9 (C), and 106.3 (CH₂) suggested the presence of two double bonds. The above data coupled with the characteristic ¹H NMR signals for methyl groups at δ 0.92 (3H, s), 1.18 (3H, s), 1.04 (3H, d, J = 6.8 Hz), and 1.03 (6H, d, J = 7.2 Hz) and signals for olefinic protons at δ_H 5.41 (1H, d, J = 5.6 Hz), 4.76 (1H, s), and 4.70 (1H, s) (Table 2) suggested 1 to be a member of the 24-methylenecholesterol class [12,13]. The ¹H−¹H COSY correlations allowed the assignment of four separated spin systems (Figure 2). The presence of an sp² methylene substituent at C-24 was further confirmed by the HMBC correlations from H₂-28 to C-23, C-24, and C-25 (Figure 2). Likewise, the steroidal nucleus was confirmed by the HMBC correlations from H₃-18 to C-12, C-13, C-14, and C-17 and H₃-19 to C-1, C-5, C-9, and C-10. The NOE correlations between H₃-19/H-1β, H-3/H-1α, and H₃-19/H-11 suggested the α and β orientations for H-3 and H-11, respectively. The absence of an NOE correlation between H₃-18/H-17 and the presence of the correlation between H-17/H-16, H-16/H-14, and H-14/H-9 suggested the α orientation for H-9, H-14, H-16, and H-17. Moreover, the β orientation for H-20 was evidenced from the NOE correlations between H₃-18/H-20, H₃-18/H-12a, and H₃-21/H-12a. The absolute configuration of 1 was determined by the application of Mosher’s method [14]. Analysis of the ¹H NMR data of esters 1a and 1b resulted in the determination of a 3S configuration (Figure 3).

Analysis of the HRESIMS and ¹³C NMR spectroscopic data of crassarosteroside A (2) suggested a molecular formula of C₃₆H₅₈O₈. The IR spectrum of 2 showed the presence of hydroxy (3461 cm⁻¹) and carbonyl (1741 cm⁻¹) groups. The latter was identified as an acetoxy group according to the carbon resonances at δ 170.9 (C) and 21.2 (CH₃) (Table 1). The ¹H NMR spectroscopic data of 2 showed characteristic methyl signals at δ 0.92 (3H, s), 1.18 (3H, s), 1.04 (3H, d, J = 6.8 Hz), 1.03 (6H, d, J = 7.0 Hz), 5.41 (1H, d, J = 5.6 Hz), 4.76 (1H, s), and 4.70 (1H, s) (Table 2), revealing that 2 has the same 24-methylenecholesterol skeleton as that of 1. By excluding the steroidal moiety and the acetoxy group, the remaining six carbons [δ 97.4 (CH), 74.0 (CH), 66.9 (CH), 70.9 (CH), 65.8 (CH), and 16.0 (CH₃)] were ascribed to the presence of a 6’-deoxyhexose residue. The sugar residue was deduced as an α-fucopyranose on the basis of 2D NMR analysis (Figure 2) and the coupling constants of ³J_{H-1’,H-2’} (4.0 Hz), ³J_{H-2’,H-3’} (10.0 Hz), ³J_{H-3’,H-4’} (3.5 Hz), and ³J_{H-4’,H-5’} (<1 Hz) (Table 2) [12,15,16]. The acetoxy group attached at C-2’ of the α-fucose residue was evidenced from the downfield chemical shift of H-2’ (δ 5.07), which was confirmed by the HMBC correlations from this proton to the acetate carbonyl carbon. The HMBC correlation from H-1’ to C-3 revealed that the α-fucose residue was attached at C-3 of the steroidal aglycone. The absolute configuration of the sugar moiety in 2 was determined by reversed phase HPLC analysis of its o-tolylthiocarbamate [17]. The liberated fucose from acid hydrolysis of 2 was treated with L-cysteine methyl ester followed by reaction with o-tolylisothiocyanate to afford the corresponding o-tolylthiocarbamate derivative. The retention time of the liberated sugar derivative by HPLC analysis was found to be consistent with that of a standard L-fucose derivative.

Crassarosteroside B (3) gave the same molecular formula as that of 2 based on the analysis of the HRESIMS and ¹³C NMR spectroscopic data (Table 1). The NMR spectroscopic data of 3 were similar to those of 2, with some exceptions for those of the sugar residue. An HMBC correlation from the anomeric proton at δ 5.12 (H-1’) to the carbon signal at δ 77.8 (C-3) connected the fucose residue to C-3 of the steroidal aglycone. The downfield proton chemical shift at δ 4.87 (1H, dd, J = 10.5, 4.0 Hz) was ascribed to the presence of an acetoxy group at C-3’ (Table 2). The detailed 2D NMR analysis confirmed the above elucidation. Likewise, the L-fucose residue was deduced according to RP HPLC analysis of the corresponding o-tolylthiocarbamate as described above.

Crassarosteroside C (4) was assigned the same molecular formula as those of 2 and 3. A comparison of NMR spectroscopic data of 4 with those of 2 and 3 revealed that an acetoxy group should be located at C-4’ of the fucose residue (Table 1 and Table 2). This was evidenced by the ¹H NMR shift of H-4’ at 5.20 (1H, d, J = 2.8 Hz). In the same manner, RP HPLC anslysis of the corresponding o-tolylthiocarbamate derived from the hydrolyte of 4 allowed the determination of the L-fucose moiety.

The HRESIMS and ¹³C NMR spectroscopic data of crassarosteroside D (5) established a molecular formula of C₃₇H₆₀O₇. The presence of an acetoxy group was evidenced by the ¹H NMR signal at δ 2.17 (3H, s) (Table 1) and ¹³C NMR signals at δ 171.3 (C) and 20.8 (CH₃) (Table 2) as well as the IR absorption band at 1737 cm⁻¹. The NMR spectroscopic data for the sugar moiety of 5 were quite similar to those of 4, suggesting that they shared the same 4’-O-acetylfucose residue. Except for the sugar moiety, the remaining 29 carbon signals as well as the characteristic methyl signals at δ 1.00 (3H, s), 1.05 (3H, s), 1.38 (3H, s), 0.86 (3H, d, J = 6.4 Hz), 0.89 (3H, d, J = 6.4 Hz), 0.76 (3H, d, J = 6.8 Hz), and 0.78 (3H, d, J = 6.4 Hz) (Table 2) revealed that the aglycone of 5 should have a C₂₉ steroidal skeleton [18]. The 23,24-dimethyl-20-hydroxy side chain was deduced by the ¹H−¹H COSY correlations from H₂-22 to H₃-29 through H-23, from H-24 to both H₃-26 and H₃-27, and from H-24 to H₃-28 as well as the HMBC correlations from H₃-21 to C-17, C-20, and C-22 and H₃-29 to C-22, C-23, and C-24. This rare steroidal side chain is the same as that of sarcophytosterol isolated previously by us from the soft coral Lobophytum sarcophytoides [18]. The NMR spectroscopic data for the aglycone moiety of 5 are almost the same as those of sarcophytosterol, except for some minor variations in ¹H and ¹³C chemical shifts at C-2, C-3, and C-4 between both compounds. This is due to the attachment of the sugar residue at C-3 of the steroidal aglycone. Similarly, HPLC analysis of the relevant o-tolylthiocarbamate derived from the hydrolysis of 5 suggested the presence of L-fucose.

The absolute configuration of sterol 1 has been established by Mosher’s method in the present work. On the basis of biogenesis, the steroidal moieties of the glycosides 2–5 should possess the same absolute configurations as shown in the formulae. Cytotoxicity of steroids 1–5 against HepG2, HepG3, MCF-7, MDA-MB-231, and A-549 cancer cell lines was evaluated. The results showed that 1 exhibited cytotoxicity toward HepG2 cancer cell line with an IC₅₀ value of 14.9 µM, while 4 also showed cytotoxicity toward HepG2 and HepG3 cell lines with IC₅₀ values of 17.6 and 18.9 µM, respectively. The other compounds were found to be inactive (IC₅₀ > 20 μM) toward the above cancer cell lines after 72 h exposure. The effect of steroids 1–4 on the expression of pro-inflammatory iNOS and COX-2 proteins in RAW264.7 macrophage cells stimulated with LPS was also evaluated using immunoblot analysis. At a concentration of 10 µM (Figure 4), steroid 2 was found to significantly reduce the level of iNOS protein to 12.9 ± 4.3% and 4 could reduce the iNOS espression to 50.1 ± 6.3%. However, these compounds could not effectively reduce the level of COX-2. On the contrary, 1–3 were found to stimulate the expression of COX-2.

Prior investigation of the genus S inularia reported on several steroidal glycosides, however, all of these were found to possess a 24-methylene substituted side chain [22,23]. 5 is the first example of steroidal glycoside with a 23,24-dimethyl substitited side chain from soft coral of the genus S inularia. Our biological data reveal that 2′-O-acetyl-L-fucose functionality plays an important role toward the inhibition of the pro-inflammatory iNOS. Steroidal glycoside 2 could be a useful anti-inflammatory agent, while steroids 1 and 4 have shown inhibitory activity toward the selected human liver cancer cells.