Aurantosides and rubrosides are orange-red pigments isolated from sponges of the genera Theonella, Homophymia and Siliquariaspongia, all belonging to the order Lithistida [1,2,3,4,5]. These complex molecules include three different structural moieties: a mono- or dichlorinated long (C₁₈ to C₂₈) conjugated polyene chain (terminating with an oxygenated five-membered ring in the rubroside subfamily), a tetramate ring, and an N-glycosidic moiety made up by one to three monosaccharide units, which are, invariably, pentose or deoxypentose sugars.

Compared to non-ribosomial cyclic and linear peptides and to polyketide macrolides, the most widely investigated secondary metabolites from sponges of the Lithistida order, aurantosides have been largely overlooked, also in terms of their biomedical potential. The reasons for this observation can be mostly found in the difficulties associated to their isolation in the pure form and to their relative instability, which is obviously related to the presence of the polyene moiety.

In the frame of our ongoing investigation of Indonesian invertebrates for bioactive secondary metabolites [6,7,8], we have recently analyzed a specimen of Theonella swinhoei (Theonellidae, Lithistida) and reported the isolation of dehydroconicasterol and of aurantoic acid (1) [9]. This latter compound closely resembles the polyene moiety of aurantosides G–I [4], although the biosynthetic relationship between aurantoic acid and aurantosides has not been established. Chemical analysis of the most polar fractions of the organic extract obtained from the same organism revealed that they were practically devoid of polar polypeptides or macrolides, while they contained consistent amounts of aurantosides. In this paper we describe the isolation and the structure elucidation of the three known aurantosides G–I (2–4) and of the new aurantoside J (5) from these fractions (see Scheme 1). In addition, we report the results of a detailed investigation of the antifungal activity of the isolated aurantosides, which revealed a strict dependency from the glycosylation pattern.

A specimen of the sponge T. swinhoei was collected by hand in the area of the Bunaken Marine Park of Manado (North Sulawesi, Indonesia) and kept frozen until sequentially extracted with MeOH and CH₂Cl₂ by soaking the sliced sponge tissues (560 g wet wt). The extracts were combined, concentrated and partitioned between EtOAc and H₂O. Then, the water-soluble fraction was partitioned against n-BuOH and the so obtained polar organic extract (3.8 g) was chromatographed on a C₁₈ reversed-phase silica gel column, using a gradient solvent system going from H₂O/MeOH 7:3 to H₂O/MeOH 1:9. Repeated RP18 HPLC purifications (H₂O/CH₃CN mixtures, with 0.05% TFA, as eluents) afforded aurantosides G (2, 25.5 mg), H (3, 12.8 mg), I (4, 15.6 mg) and the new aurantoside J (5, 8.3 mg) in the pure form. The structures of the known aurantosides G–I (2–4) were identified on the basis of the comparison of their spectral data with those reported in the literature [4]. It is worth noting that all the aurantosides isolated during this work showed the same length and constitution of the polyene side chain, while aurantosides with longer and/or dichlorinated polyene chain (as aurantosides A–F) were not detected in our sample.

Aurantoside J (5) was isolated as an orange-red amorphous solid. Its ESI mass spectrum (positive ions) showed pseudomolecular ion peaks at m/z 517 and 519 [M + Na]⁺ in the ratio of about 3:1, suggesting the presence of a chlorine atom. High resolution measurements (see Experimental Section) assigned the molecular formula C₂₃H₂₇ClN₂O₈ to 5, in accordance with the presence of eleven unsaturation degrees.

The ¹H NMR spectrum (700 MHz, CD₃OD) of 5 revealed the presence of a polyene unit (nine multiplets from δ_H 6.30 to 7.63), of an allylic methyl group (δ_H 2.21), and of a tetramic acid glycoside, classifying compound 5 as a metabolite of the aurantoside family. All the proton signals were associated to those of the directly attached carbon atoms through the 2D HSQC experiment and the so obtained sets of ¹H and ¹³C NMR data of 5 evidenced a close resemblance with parallel data reported for aurantoside G (2) [4], which also shared the same molecular formula of 5.

Combined inspection of COSY and TOCSY spectra allowed a deconvolution of the proton signals within the polyene moiety and the consequent assignment of the proton and carbon resonances, as well as of proton-proton coupling constants, within this spin system. The obtained data appeared in excellent agreement with those reported for aurantoside G, suggesting that 2 and 5 should share this part of their structures, including the all-trans geometry of the disubstituted double bonds. The Z geometry of the terminal, chlorine-bearing double bond, was secured on the basis of the ROESY cross-peak between the methyl singlet at δ 2.21 (Me-18) and the doublet at δ 6.30 (H-16).

The identification of the stereostructure of the tetramate moiety, as well as its attachment to the polyene chain, was largely based on the comparison with literature data. The HMBC spectrum (Figure 1) showed cross-peaks of the methine resonating at δ 4.18 (H-4) with C-1, C-2, C-3 and with the amide carbon at 174.1, while the doublet at δ 7.23 (H-8) showed cross-peaks with C-7 and C-2. This pattern of HMBC correlations and the consequent assignment of ¹³C NMR resonances were in perfect agreement with values reported for other aurantosides. In accordance with the behavior reported for aurantosides and for other tetramic acid derivatives [10], the keto-enol tautomerization of the adjoining tetramic acid portion of the molecule caused the appearance of a couple of small methine resonances (ratio 1:3 compared to other methine signals) in the sp² region of the ¹H NMR spectrum of 5, which are most likely attributable to H-8 and H-9 on the minor tautomer.

The spin system of a pentose monosaccharide was easily identified from the COSY spectrum. The HMBC correlation of the anomeric proton (δ_H 5.01, δ_C 82.9) with C-1 and C-4, as well as its relatively high-field resonance, defined its attachment to the nitrogen atom of the tetramate ring, while the HMBC correlation H-1′/C-5′ indicated its pyranose form. The large value of the ³J scalar coupling constants J_{H-2′/H-3′} and J_{H-3′/H-4′} (=10.2 Hz) and the small value of J_{H-1′/H-2′} (=2.6 Hz) allowed the identification of this sugar as a α-xylopyranose and strongly suggested the conformation depicted in Scheme 1. Of course, the predominance of different conformations in different solvents cannot be excluded. Given the small amount of compound available (largely used for evaluation of cytotoxic and antifungal activities, see below), the absolute configuration of the xylopyranosyl unit in 5 was not determined. However, a D configuration could be confidently assigned to this sugar, considering that the xylose units present in all the aurantosides isolated to date, including 2–4, invariably proved to belong to the D series.

In conclusion, aurantoside J (5) is a new secondary metabolite whose structure differs from that of the known aurantoside G (2) for the configuration at the anomeric carbon C-1′ of the xylose unit. In this regard, it is worth noting that aurantoside J represents the first member of this class of compounds to show an N-glycosidic linkage with the α-configuration.

All the members of the aurantoside family had been evaluated for their cytotoxic activity and data available in the literature [1,2,3,4] clearly suggest a close relationship between the length of the polyene chain and the toxicity against human cell lines: an increase in the length of the chain is associated to an increase of the cytotoxic activity. Accordingly, aurantosides G–I (C₁₈) and aurantosides A–B (C₂₀) are not cytotoxic, aurantosides D and E (C₂₂) are moderately cytotoxic, while aurantoside F (C₂₄) is strongly cytotoxic. In perfect agreement with this trend, the new aurantoside J (5) proved to be devoid of any cytotoxic activity (IC₅₀ > 70 μM) against several human cell lines (data not shown).

Since the antifungal potential of aurantosides has remained practically undisclosed (with the single exception of the good antifungal activity against Aspergillus fumigatus reported for aurantosides A and B³), and given the increasing need for new active agents to face fungal infections, we decided to evaluate aurantosides G–J (2–5) for their antifungal activity against Fusarium solani and four Candida species. These strains have been selected since, together with Zygomycetes, they represent the main responsible of invasive fungal infections (IFIs), an increasingly important clinical dilemma, which engenders high rates of morbidity and mortality, particularly in immuno-compromised populations. In the management of patients with haematological malignancies or transplant recipients, an increase in the incidence of fungal diseases due to opportunistic infections of genera Candida and Fusarium is commonly observed [11]. Unfortunately, despite an antifungal treatment with available agents, the mortality rate of patients with IFIs remains extremely high. Candida infections are the most frequent cause of IFIs worldwide and, in USA, Candida species are the fourth most common cause of nosocomial blood stream infection [12]. Among the 50 species of Fusarium identified to date, F. solani is responsible of 50% of pathogenic infections in humans [13], which are life-threatening in immunocompromised hosts: recognized risk factors are skin lesions, use of corticosteroids, prolonged neutropenia and hematological malignancy [14].

The antifungal activity of aurantosides G–J (2–5) has been expressed as the Minimum Inhibitory Concentrations (MICs) and Minimum Fungicidal Concentrations (MFCs) for 90% and 50% of isolates. The MICs and MFCs of the four aurantosides against 15 clinical isolates of fungi (12 isolates for the four Candida species C. albicans, C. parapsilosis, C. glabrata, C. tropicalis and 3 isolates for F. solani) are shown in Table 1.

Only aurantosides G (2) and I (4) showed a detectable antifungal activity, but the activity of aurantoside G was only moderate-poor, with a MIC₉₀ of 4–16 μg/mL against all the strains. In contrast, aurantoside I exhibited a very good antifungal effect against all the tested strains, especially against C. albicans, C. glabrata and C. tropicalis. The activity of aurantoside I against F. solani is also remarkable, given the high pathogenicity of this strain and the need of potent inhibitory agents. For the majority of isolates, MFCs of the test compounds were either the same as or one to two dilutions higher than the MICs.

The mechanism of the antifungal action of aurantoside I (4) has not been investigated; however, it could be related to the mechanism of other polyene antifungal agents as nystatin. Since all the compounds tested in this study are homogeneous in terms of the C₁₈ polyene chain, the role of the sugar moiety in the modulation of the activity can be highlighted. Clearly, a trisaccharide moiety, present in aurantoside I and in the previously reported aurantosides A–B, is needed for the antifungal activity of this class of compounds.

In conclusion, the chemical investigation of an Indonesian specimen of Theonella swinhoei has yielded four aurantosides (G–J), one of which, aurantoside J (5), is a new compound exhibiting the unprecedented N-α-glycosidic linkage between the pentose and the tetramate units. The four isolated aurantosides have been tested against five fungal strains (four Candida and one Fusarium) responsible of invasive infections in immuno-compromised patients. The non-cytotoxic aurantoside I (4) was the sole compound to show an excellent potency against all the tested strains, thus disclosing two essential requirements to optimize the antifungal activity in this class of compounds: (i) a C₁₈- to C₂₀-polyene chain (to minimize the cytotoxic activity against human cell lines); and (ii) a trisaccharide chain attached at the tetramate ring.