Ascidians have been an extremely rich source of sulfur-containing molecules which, on the other hand, are quite unusual in marine organisms. A number of sulfides/polysulfides, sulphur heterocycles, sulfoxides, and alkyl sulfates have been isolated from marine ascidians [1,2]. In particular, these latter compounds, although occasionally reported from marine source [3,4,5,6], have shown to be often present in remarkable amounts in solitary ascidians of the families Ascididae and Pyuridae, as well in colonial Polyclinidae species [7,8,9,10,11,12,13,14]. They often have quite simple structures, mostly of polyketide derivation even though in some frameworks an isoprenoid origin is clearly recognized; almost all the isolated compounds are endowed with cytotoxic and/or antiproliferative activity. In the course of our program on discovery of cytotoxic metabolites from Mediterranean ascidians [15,16], we have analyzed the chemical composition of methanol extracts of two species, Aplidium elegans (Monniot & Monniot, 1983) and Ciona edwardsii (Roule, 1884), both collected in the Bay of Naples, Italy. This investigation yielded, in addition to the known sulfated terpenoids 4 and 5 [10,14], some analogues, compounds 1 and 2 from A. elegans, and compound 3 from C. edwardsii (Figure 1). The present paper deals with the isolation and structure elucidation of the three new compounds as well as with their cytotoxic activity measured in vitro on J774 (murine macrophage) and C6 (rat astrocytic glioma) cells.

Fresh thawed tissues of both A. elegans and C. edwardsii, collected in the bay of Naples, immediately frozen, and kept at −28 °C until extraction, were exhaustively extracted with methanol and, subsequently, with chloroform. For each ascidian sample, the extracts were combined and concentrated; the resulting aqueous suspensions were then partitioned between water and butanol. The butanol extract of A. elegans was fractionated by MPLC over a reversed phase C-18 column using stepwise elution with aqueous MeOH. The fractions eluted with 50% aqueous MeOH were further separated and purified by reversed-phase HPLC eluting with 70% aqueous MeOH containing 0.1% TFA, thus affording compounds 1 and 2 in the pure state. The butanol extract of C. edwardsii was fractionated by reversed-phase MPLC under the same conditions as described for that of A. elegans. Further separation and purification of both fractions eluted with 70% and 50% aqueous MeOH by repeated reversed-phase HPLC yielded compounds 3–5 in the pure state.

A first survey of the 1D NMR spectra (CD₃OD) of 1 and the comparison with those of the already known compound 4 readily allowed us to hypothesize that 1 was the 15-hydroxy analogous of 4. The proton spectrum of 1 lacked indeed the signals due to the isopropyl terminus portion of 4, whereas it contained a quite deshielded methyl signal (δ 1.14), resonating as a singlet and integrating for six protons. Likewise, the ¹³C-NMR spectrum of 1 contained a quaternary carbon resonance (δ 70.8) attributable to an oxygen-bearing carbon while the methine signal at δ 29.1 due to C-15 in 4 was absent. Chemical shifts and coupling patterns of the remaining signals of 1, assigned by aid of COSY, HSQC, and HMBC experiments were very similar to those of 4. Mass data analysis confirmed the assumption made. ESI mass spectrum (negative ion mode) contained an ion peak at m/z 511; an intense fragment at m/z 453 was present and it was interpreted as the result of an α cleavage at C-15. High resolution analysis on the pseudomolecular ion peak gave m/z 511.1987, which was consistent with the molecular formula C₂₀H₄₀NaO₉S₂ corresponding to [M−Na]⁻ (calculated value: 511.2006). The location of the hydroxyl group at C-15 was unambiguously established through 2D-NMR experiments which allowed us also to assign all the proton and carbon resonances of compound 1 (Table 1). In particular, diagnostic ³J_C-H long-range couplings were observed in the HMBC spectrum between the signal at δ 1.14 (s, 6H, CH₃-16/20) and both the oxygenated quaternary carbon at δ 70.8 (C-15) and the methylene carbon at δ 44.4 (C-14).

The negative HRESI mass spectrum (negative ions) of compound 2 displayed an ion peak at m/z 497.1825 corresponding to [M−Na]⁻ (calculated value: m/z 497.1849); the molecular formula of 2 was thus established as C₁₉H₃₈NaO₉S₂. The ¹H- and ¹³C-NMR spectra of compound 2 displayed a close resemblance to those of 1, and the observed differences were due to the initial segment of the linear skeleton of both compounds (C1-C3/C16 in 2 and C1-C4/C17 in 1). The multiplet at δ 4.02 due to 2H-1 in compound 1 was replaced by an ABX system at δ 3.80 (1H, dd, J = 9.1 and 6.6 Hz, H-1a) and 3.86 (1H, dd, J = 9.1 and 6.6 Hz, H-1b). COSY connectivities, which allowed to delineate the large spin system C1-C13 through CH₃-16, CH₃-17 and CH₂-18, as well as HMBC information, clearly evidenced 2 as the 1-nor-derivative of 1 or, alternatively, the 15-hydroxy analogue of 5 (Table 1).

The NMR features (CD₃OD) of compound 3 appeared almost identical to those of 5, except for the chemical shift of the C-18 methylene protons as well as that of the relevant carbon, both showing a significant upfield shift (δ_H: 3.94 in 5vs. 3.44 in 3; δ_C: 73.9 in 5 vs. 65.5 in 3). The negative ion HRESI mass spectrum of 3 displayed an ion peak at m/z 379.2522 which was consistent with the molecular formula C₁₉H₃₉O₅S⁻, corresponding to [M−Na]⁻(calculated value 379.2513). These few data readily allowed us to deduce that 3 was the 18-desulfated analogue of 5; this conclusion was fully corroborated by 2D NMR spectra analysis which also led to the full assignment of all NMR resonances in 3 (Table 1).

The cytotoxic effect of compounds 1–3 was evaluated on J774A.1 (BALB/c murine macrophages) and C6 (rat glioma) cell lines in vitro. Compounds 1 and 2 induced a concentration-dependent mortality on J774A.1 (Figure 2) whereas both drugs were inactive on C6 cells, with a LC₅₀ > of 300 µM (data not shown). When comparing optical density values of control and treated cells, the cytotoxic effect of compound 1 and 2 was significant at the highest concentrations (p < 0.05, compound 1 at 100 µM; p < 0.05 and p < 0.01, compound 2 at 30 and 100 µM, respectively). The LC₅₀ value of compound 2 on J774A.1 cell line was 45.12 µM, while that of compound 1 was > of 100 µM (Figure 2). The compound 3 was ineffective on both J774A.1 and C6 cells (data not shown). On the basis on these although preliminary data, a sulfate group in R2 and/or a hydroxyl group in R1 seems to be essential for the cytotoxic activity, since they are absent in compound 3 which proved inactive. Nevertheless, previous reports showed that compounds 4 and 5, possessing exclusively the sulfate group in R2, are still active [10,14] therefore, we can conclude that only the latter functionality has an obligatory role for the cytotoxic activity showed by compounds 1 and 2 against the macrophagic cell line. The slightly different potency of compounds 1 and 2 could be ascribed to the different length chain; this hypothesis is supported by cytotoxicity data reported in literature for compounds 4 and 5, where compound 5, characterized by a shorter C-chain, resulted more active than its higher analogue 4 [10,14].

The structures of compounds 1–3, isolated from the Mediterranean ascidians Aplidium elegans and Ciona edwardsii, have been elucidated using mass spectrometry and NMR experiments and their in vitro cytotoxic effects have been evaluated on J774A.1 (BALB/c murine macrophages) and C6 (rat glioma) cell lines. A moderate but selective cytotoxic effect on J774A.1 cell line has been evidenced for compounds 1 and 2; the inactivity of 3, as well as further pharmacological data available in the literature [10,14], indicated that the hydroxyl group does not confer per se cytotoxic activity, while the contribution of the sulfate is of pivotal importance. A slight influence of alkyl chain length on the potency of the active compounds has been also evidenced.