Cytotoxic Drimane-Type Sesquiterpenes from Co-Culture of the Marine-Derived Fungi Aspergillus carneus KMM 4638 and Beauveria felina (=Isaria felina) KMM 4639

Chemical investigation of a coculture of the marine-derived fungi Beauveria felina KMM 4639 and Aspergillus carneus KMM 4638 led to the identification of three new drimane-type sesquiterpenes, asperflavinoids B, D and E (2, 4, 5), and nine previously reported related compounds. The structures of these compounds were established using spectroscopic methods and by comparison with known analogues. We also investigated the cytotoxic activity of the isolated compounds against several cancer and normal cell lines. Asperflavinoid C (3) and ustusolate E (9) exerted a significant effect on human breast cancer MCF-7 cell viability, with IC50 values of 10 µM, and induced in caspase-dependent apoptosis and arrest of the MCF-7 cell cycle in the G2/M phase in these cells.


Introduction
Recent studies show that one of the factors causing the induction of the biosynthesis of new bioactive metabolites from marine fungi is coculture with other microorganisms, including fungi. The essence of this approach is to simulate, to a certain extent, the natural microbial complex, whereby microorganisms produce bioactive secondary metabolites necessary for survival in a competitive environment. Interspecies microbial control can have a significant impact on secondary metabolites produced for habitat protection or as chemical signals [1]. Mixed cultivation can activate silent genes that are inactive in monocultures [2]. Joint cultivation can result in an increase in the yields of previously isolated bioactive compounds, obtaining their derivatives, as well as the synthesis of novel metabolites of various chemical classes. For example, the new meroterpenoid derivative chermebilaene A, with potent inhibitory activities against Ceratobasidium cornigerum (MIC 0.5 µg/mL), and Edwardsiella tarda (MIC 0.25 µg/mL) were characterized from a coculture of the marinederived fungal isolates of Penicillium bilaiae MA-267 and Penicillium chermesinum EN-480 [3]. The new alkaloid aluminumneohydroxyaspergillin was isolated from a coculture of the marine-derived fungi Aspergillus sclerotiorum and Penicillium citrinum and showed significant and selective cytotoxicity against the human histiocytic lymphoma U937 cell line (IC 50 = 4.2 µM) [4].
Marine-sediment-derived fungus Beauveria felina KMM 4639 was previously identified as a producer of highly oxygenated chromene derivatives oxirapentyns B-K, pyran from a coculture of the marine-derived fungi Aspergillus sclerotiorum and Penicillium citrinum and showed significant and selective cytotoxicity against the human histiocytic lymphoma U937 cell line (IC50 = 4.2 μM) [4].
Marine-sediment-derived fungus Beauveria felina KMM 4639 was previously identified as a producer of highly oxygenated chromene derivatives oxirapentyns B-K, pyran polyketide, isariketide A, and benzofuran acremine S [5][6][7]. It was suggested that this strain includes a complex of oxygenases to produce such metabolites.
Coculture of this fungal strain with Aspergillus sulphureus KMM 4640 resulted in the isolation of a number of new prenylated indole alkaloids and a new diorcinol J [8,9]. Oxidative prenylated compounds were not produced by a monoculture of A. sulphureus but were usual for the B. felina strain. Thus, prenylation and further oxygenation may be the result of B. felina's prenyltransferase and oxygenase action.

Isolation and Identification of Compounds
Compound 1 was isolated as an amorphous solid, and the HRESIMS provided a molecular formula of C 23 H 36 O 7 , as confirmed by 13 C NMR data. The signals of 1D NMR spectra of compound 1 (Tables 1 and 2 and Figures S15 and S16), as well as HMBC and COSY correlations (Figure 2), undoubtedly established the planar structure of compound 1. The structure of compound 1 and the chemical shifts in its 1 H NMR spectrum registered in DMSO-d 6 ( Figure S21) fully correspond to those of recently reported asperflavinoid A [12].   tra of compound 1 (Tables 1 and 2 and Figures S15 and S16), as well as HMBC and COSY correlations (Figure 2), undoubtedly established the planar structure of compound 1. The structure of compound 1 and the chemical shifts in its 1 H NMR spectrum registered in DMSO-d6 ( Figure S21) fully correspond to those of recently reported asperflavinoid A [12]. The relative stereochemistry of the side chain of compound 1 was determined using its acetonide derivative (1a). The small coupling constant (J6′,7′ = 6.0 Hz) and dissimilar magnetic environment of acetonide methyls (Δ = 0.12 ppm) ( Figure S53) indicate an erythro configuration of the diol group at positions C-6′ and C-7′ [19]. The absolute configurations of the chiral centers of the drimane core in compound 1 were defined as 5S,6R,9S,10S based on biogenetic considerations. An analysis of the literature data indicates that the configurations of the C-5, C-9, and C-10 asymmetric centers remain unchanged, enabling the use of X-ray diffraction data [20,21], as well as the CD spectral data [22] of known drimanes in projection to asperflavinoid A.
The splitting of a number of 1 H and 13 C NMR signals (assigned to the octadiene side chain) indicated that compound 1 is a mixture of two compounds in a 1:1 ratio. A similar case was previously reported for asperflavinoid A; therefore, compound 1 is obviously a mixture of stereoisomers at the hydroxy groups in the side chain: 5S,6R,9S,10S,6′R,7′S and 5S,6R,9S,10S,6′S,7′R.
Xu et al. reported asperflavinoid A as a mixture of three stereoisomers at positions C-6 and C-7 (without specifying which ones), which, upon separation, quickly reverted to the original mixture [12]. This was not confirmed by our experimental data or by the data reported other authors who described drimanes with the same side chain [22,23]. Although we were unable to separate the two stereoisomers of compound 1, compound 11, which has an identical side chain, was separated into two individual diastereomers, asperienes C (11a) and D (11b). After two days, these compounds remained unchanged. The molecular formula of compound 2 was determined as C23H36O7 according to the HRESIMS peak at m/z 447.2347 [M + Na] + , in accordance with the 13 C NMR data. A close inspection of the 1 H and 13 C NMR data (Tables 1 and 2   The relative stereochemistry of the side chain of compound 1 was determined using its acetonide derivative (1a). The small coupling constant (J 6 ,7 = 6.0 Hz) and dissimilar magnetic environment of acetonide methyls (∆ = 0.12 ppm) ( Figure S53) indicate an erythro configuration of the diol group at positions C-6 and C-7 [19]. The absolute configurations of the chiral centers of the drimane core in compound 1 were defined as 5S,6R,9S,10S based on biogenetic considerations. An analysis of the literature data indicates that the configurations of the C-5, C-9, and C-10 asymmetric centers remain unchanged, enabling the use of X-ray diffraction data [20,21], as well as the CD spectral data [22] of known drimanes in projection to asperflavinoid A.
The splitting of a number of 1 H and 13 C NMR signals (assigned to the octadiene side chain) indicated that compound 1 is a mixture of two compounds in a 1:1 ratio. A similar case was previously reported for asperflavinoid A; therefore, compound 1 is obviously a mixture of stereoisomers at the hydroxy groups in the side chain: 5S,6R,9S,10S,6 R,7 S and 5S,6R,9S,10S,6 S,7 R.
Xu et al. reported asperflavinoid A as a mixture of three stereoisomers at positions C-6 and C-7 (without specifying which ones), which, upon separation, quickly reverted to the original mixture [12]. This was not confirmed by our experimental data or by the data reported other authors who described drimanes with the same side chain [22,23]. Although we were unable to separate the two stereoisomers of compound 1, compound 11, which has an identical side chain, was separated into two individual diastereomers, asperienes C (11a) and D (11b). After two days, these compounds remained unchanged.
The molecular formula of compound 2 was determined as C 23 H 36 O 7 according to the HRESIMS peak at m/z 447.2347 [M + Na] + , in accordance with the 13 C NMR data. A close inspection of the 1 H and 13 C NMR data (Tables 1 and 2  The acetonide derivative (2a) of compound 2 was prepared for further investigation of the stereochemistry at the diol position in fatty acid. The large coupling constant (J 6 ,7 = 8.0 Hz) and the almost identical magnetic environment of the acetonide methyls (∆ = 0.00 ppm) ( Figure S54) indicate a threo configuration of the diol group at positions C-6 and C-7 [19].
The HMBC and COSY correlations ( Figure 3 and Figure S33) proved the presence of a pereniporin A drimane moiety in compound 3.
The absolute configurations of the chiral centers of the drimane core in compound 3 were defined based on ROESY (Figure 3 and Figure S34) correlations H 3 -14 (δ H 1.01)/H-5 (δ H 2.11), H-6 (δ H 5.75), and H-11 (δ H 5.38)/H 3 -13 (δ H 1.24), as well as biogenetic considerations, such as 5S,6R,9S,10S,11R. Esterification of compound 3 with (S)-and (R)-MTPA-Cl led to destruction of the compound. Compound 3 was named asperflavinoid C. An unnamed compound with the same structure was previously reported in [13], but the report did not include any detailed NMR data that can be used for identification and comparison.
Although all isolated compounds are closely related, and many of them were previously described as fungal metabolites; only strobilactone A (8) was previously isolated from a monoculture of the fungus A. carneus KMM 4638 [10,11]. Other drimane derivatives may be products of the influence of coculture with B. felina. Moreover, compounds 1, 2, 4, 5, and 10-12 are probably the results of the action of oxygenases from B. felina on an octatrienoic acid residue (Figure 4) [7], in accordance with the previously obtained results of the mixed cultivation of B. felina KMM 4639 with Aspergillus sulphureus KMM 4640 [9]. On the other hand, the hydroxylation of the double bond of the side chain may be non-enzymatic. Thus, further research is required to confirm these assumptions.

Biological Activity of Compounds
Isolated compounds 1-5 and 8-12 were tested for cytotoxic activity against a number of cell lines, including rat glioblastoma C6, human breast cancer MCF-7, human prostate cancer PC-3, and human lymphoma Raji, as well as normal rat cardiomyocyte H9c2 cell lines (Table 3). Compounds 6 and 7 were isolated in quantities insufficient for bioassays.

Biological Activity of Compounds
Isolated compounds 1-5 and 8-12 were tested for cytotoxic activity against a number of cell lines, including rat glioblastoma C6, human breast cancer MCF-7, human prostate cancer PC-3, and human lymphoma Raji, as well as normal rat cardiomyocyte H9c2 cell lines (Table 3). Compounds 6 and 7 were isolated in quantities insufficient for bioassays. The cells were incubated with the isolated compounds for 24 h, and the viability of cells was measured by MTT assay. All experiments were carried out in three independent replicates, and data are presented as mean ± standard error of the mean (SEM). MCF-7 cells were more sensitive to the toxic influence of the investigated compounds than other cancer cell lines tested in our experiments. Therefore, subsequent experiments were carried out using the MCF-7 cell line.
The prolonged incubation (48 h) of MCF-7 cells with compounds 1-5 and 8-12 at concentrations up to 10 µM resulted in an increase in the cytotoxic effect of the compounds. Nonetheless, only compounds 3 and 9 caused a decrease in MCF-7 cell viability of nearly 50%, IC 50 values calculated as 10.0 ± 0.8 µM and 10.1 ± 0.5 µM, respectively. Furthermore, the toxicity of all compounds against the normal cardiomyocyte H9c2 line was lower, with a viability of H9c2 cells of more than 80% when treated for 48 h ( Figure 5).
To further investigate the cytotoxic activity of the most effective compounds, we investigated the MCF-7 cell cycle, apoptosis level, and caspase activity after incubation with 3 and 9 at a concentration of 10 µM for 48 h.
We found that compounds 3 and 9 significantly changed the apoptotic profile of MCF-7 cells (Figure 6). Control MCF-7 cells comprised only 1.7% early and late apoptotic cells, and treating the cells with compounds 3 and 9 caused an increased the amount of apoptotic cells to 15.0% and 6.8%, respectively. Accordingly, the amount of living cells was decreased from 98.6% in the control population to 84.9% and 94.1% in cells treated with compounds 3 and 9, respectively. was lower, with a viability of H9c2 cells of more than 80% when treated for 48 h ( Figure  5). To further investigate the cytotoxic activity of the most effective compounds, we investigated the MCF-7 cell cycle, apoptosis level, and caspase activity after incubation with 3 and 9 at a concentration of 10 µ M for 48 h.
We found that compounds 3 and 9 significantly changed the apoptotic profile of MCF-7 cells ( Figure 6). Control MCF-7 cells comprised only 1.7% early and late apoptotic cells, and treating the cells with compounds 3 and 9 caused an increased the amount of apoptotic cells to 15.0% and 6.8%, respectively. Accordingly, the amount of living cells was decreased from 98.6% in the control population to 84.9% and 94.1% in cells treated with compounds 3 and 9, respectively. Changes in the apoptosis profile of MCF-7 cells after treatment with compounds 3 and 9 were accompanied by an increase in total caspase activity (Figure 7).
The influence of the compounds on the percentage of live MCF-7 cells (caspase−/7-AAD−), live MCF-7 cells with caspase activity (caspase+/7-AAD−), dead MCF-7 cells with caspase activity (caspase+/7-AAD−), and dead MCF-7 cells (caspase−/7-AAD+) was measured by a Muse ® multicaspase kit (Luminex, Austin, TX, USA). The kit utilizes a VAD-peptide derivatized with a fluorescent group called fluorescent-labeled inhibitor of caspases (FLICA) [24]. The peptide binds to activated caspases, resulting in a fluorescent signal proportional to the number of active caspases in the cell. Fluorescent dye 7-AAD was used as a dead cell marker in this assay.
Both compounds 3 and 9 significantly increased the percentage of living cells exhibiting caspase activity and 7-AAD-labeled cells exhibiting caspase activity to 12.3% and 12.4%, respectively, whereas control cell population comprised only 4.7% of cells exhibiting caspase activity. The effect of compounds 3 and 9 on caspase activity in MCF-7 cells was the same, in contract their influence on the cell apoptotic profile.
The influence of compounds 3 and 9 on the MCF-7 cell cycle is presented in Figure 8. Compounds 3 and 9 significantly reduced the percentage of cells in the G0/G1 phase by 30.9% and 33.3%, respectively. Moreover, we observed a decrease in the percentage of cells in the S phase due to the action of compounds 3 and 9 by 54.5% and 41.2%, respectively. Furthermore, the percentage of cells in the G2/M phase was significantly increased by 25.2% in the case of compound 3 and by 67.3% in the case of compound 9. Additionally, after treatment with compounds 3 and 9, the subG0 population of cells was dramatically increased by 6.4 and 4.6 times, respectively. The control cells comprised only 5.4% in the subG0 zone, whereas cells treated with compounds 3 and 9 comprised 34.6% and 24.7% subG0 cells, respectively. The decrease in cells in the G0/G1 and S phases and accumulation in subG0 phase were apparently caused by cell death after incubation All experiments were carried out in three independent replicates, and data are presented as mean ± standard error of the mean (SEM). * indicates significant differences; p < 0.05.
Changes in the apoptosis profile of MCF-7 cells after treatment with compounds 3 and 9 were accompanied by an increase in total caspase activity (Figure 7).
The influence of the compounds on the percentage of live MCF-7 cells (caspase−/7-AAD−), live MCF-7 cells with caspase activity (caspase+/7-AAD−), dead MCF-7 cells with caspase activity (caspase+/7-AAD−), and dead MCF-7 cells (caspase−/7-AAD+) was measured by a Muse ® multicaspase kit (Luminex, Austin, TX, USA). The kit utilizes a VADpeptide derivatized with a fluorescent group called fluorescent-labeled inhibitor of caspases (FLICA) [24]. The peptide binds to activated caspases, resulting in a fluorescent signal proportional to the number of active caspases in the cell. Fluorescent dye 7-AAD was used as a dead cell marker in this assay.
Both compounds 3 and 9 significantly increased the percentage of living cells exhibiting caspase activity and 7-AAD-labeled cells exhibiting caspase activity to 12.3% and 12.4%, respectively, whereas control cell population comprised only 4.7% of cells exhibiting caspase activity. The effect of compounds 3 and 9 on caspase activity in MCF-7 cells was the same, in contract their influence on the cell apoptotic profile. (d) summary graph, with early apoptotic and late apoptotic cells presented as "apoptotic cells". All experiments were carried out in three independent replicates, and data are presented as mean ± standard error of the mean (SEM). * indicates significant differences; p < 0.05. Thus, compounds 3 and 9 significantly decreased MCF-7 cell viability via the apoptosis caspase-dependent pathway, as well as cell cycle arrest in the G2/M phase. The effect of both compounds on MCF-7 cell viability was similar, but compound 3 was a stronger inducer of apoptosis than compound 9, whereas compound 9 arrested the MCF-7 cell cycle more than compound 3. These compounds are interesting targets for future investigations as anticancer agents, as they affected in vitro breast cancer MCF-7 cells but not normal cardiomyocyte H9c2 cells. The influence of compounds 3 and 9 on the MCF-7 cell cycle is presented in Figure 8. Compounds 3 and 9 significantly reduced the percentage of cells in the G0/G1 phase by 30.9% and 33.3%, respectively. Moreover, we observed a decrease in the percentage of cells in the S phase due to the action of compounds 3 and 9 by 54.5% and 41.2%, respectively. Furthermore, the percentage of cells in the G2/M phase was significantly increased by 25.2% in the case of compound 3 and by 67.3% in the case of compound 9. Additionally, after treatment with compounds 3 and 9, the subG0 population of cells was dramatically increased by 6.4 and 4.6 times, respectively. The control cells comprised only 5.4% in the subG0 zone, whereas cells treated with compounds 3 and 9 comprised 34.6% and 24.7% subG0 cells, respectively. The decrease in cells in the G0/G1 and S phases and accumulation in subG0 phase were apparently caused by cell death after incubation with compounds 3 and 9, whereas cell cycle arrest was observed in the G2/M phase after incubation with compounds 3 and 9. (d) summary graph. All experiments were carried out in three independent replicates, and data are presented as mean ± standard error of the mean (SEM). * indicates significant differences; p < 0.05.
The cytotoxic activity of some drimane sesquiterpenoids was previously reported by several research groups. Additionally, the cytotoxic effect of asperflavinoid A (1) against human hepatocarcinoma HepG2 and gastric cancer MKN-45 cells was previously reported, with IC 50 values of 84.4 and 63.2 µM, respectively [12]. Moreover, it was reported that compound 10 decreased the viability of murine leukemia P388 cells, with an IC 50 value of 8.7 µM [18], and compound 11(12) at a concentration of 120 µM decreased the viability of human leukemia MOLT-4 cells by 72% [25]. However, in our experiments, both of these compounds exhibited very weak toxicity against human leukemia Raji cells. Ustusolate E (9) was previously reported to significantly decrease the viability of human leukemia HL-60 cells, with an IC 50 of 8 µM, but was nontoxic against human lung cancer A549 cells [26]. Moreover, compound 9 decreased the viability of lymphoma L5178Y, PC 12, and HeLa cells, with IC 50 values of 1.6, 19.3, and 15.8 µM, respectively [17]. In our experiment, we did not observe a selective effect of compound 9 on human leukemia Raji cells in comparison with other investigated cell lines. Obviously, the chemical structure of isolated drimane sesquiterpenoids exerts a significant influence on their anticancer activity. Both of the most active isolated compounds, 3 and 9, have aldehyde functionalities in their side chains, which probably play a key role in the observed activity.
Mar. Drugs 2022, 20, x FOR PEER REVIEW 11 of 17 Thus, compounds 3 and 9 significantly decreased MCF-7 cell viability via the apoptosis caspase-dependent pathway, as well as cell cycle arrest in the G2/M phase. The effect of both compounds on MCF-7 cell viability was similar, but compound 3 was a stronger inducer of apoptosis than compound 9, whereas compound 9 arrested the MCF-7 cell cycle more than compound 3. These compounds are interesting targets for future investigations as anticancer agents, as they affected in vitro breast cancer MCF-7 cells but not normal cardiomyocyte H9c2 cells. The cytotoxic activity of some drimane sesquiterpenoids was previously reported by several research groups. Additionally, the cytotoxic effect of asperflavinoid A (1) against human hepatocarcinoma HepG2 and gastric cancer MKN-45 cells was previously reported, with IC50 values of 84.4 and 63.2 μM, respectively [12]. Moreover, it was reported that compound 10 decreased the viability of murine leukemia P388 cells, with an IC50 value of 8.7 µ M [18], and compound 11(12) at a concentration of 120 μM decreased the viability of human leukemia MOLT-4 cells by 72% [25]. However, in our experiments, both of these compounds exhibited very weak toxicity against human leukemia Raji cells. Ustusolate E (9) was previously reported to significantly decrease the viability of human leukemia HL-60 cells, with an IC50 of 8 µ M, but was nontoxic against human lung cancer A549 cells [26]. Moreover, compound 9 decreased the viability of lymphoma L5178Y, PC (d) summary graph. All experiments were carried out in three independent replicates, and data are presented as a mean ± standard error of the mean (SEM). * indicates significant differences; p < 0.05.
Compound 3 was previously patented as an inhibitor of PTP1B and SHP2 enzyme activity only in cell-free assays, with IC 50 values of 2.8 and 3.0 µg/mL, respectively. Compound 9 also was found to be an inhibitor of PTP1B and SHP2 enzymes, with IC 50 values of 50 and 40 µg/mL, respectively [13]. It has been suggested that inhibition of the PTP1B enzyme in breast cancer cells can result in delayed tumor formation via several pathways [27]. PTP1B inhibitors, such as docosahexaenoic acid [28] and cinnamaldehyde [29], affected human breast cancer MCF-7 cell viability. Inhibition of PTP1B results in decreased cell adhesion, loss of extracellular matrix attachment and apoptosis, AMPK activation, and autophagy [28].
Thus, PTP1B inhibition may be one of the pathways explaining the anticancer activity of compounds 3 and 9 toward MCF-7 cells in our experiments. Nevertheless, compounds 3 and 9 have a similar effect on MCF-7 cell viability, although their reported effect on PTP1B activity differed by ten times, perhaps indicating that other pathways also participate in the cytotoxic action of compounds 3 and 9. Our flow cytometry data evidence that apoptosis in MCF-7 cells treated with compounds 3 and 9 is activated via a caspase-dependent pathway. Moreover, MCF-7 cell cycle arrest was detected in the G2/M phase as a result of the action of compounds 3 and 9.
We observed that compounds 3 and 9 at a concentration of 10 µM for 48 h decreased MCF-7 cell viability by nearly 50% in MTT tests, whereas 7-AAD/annexin V straining resulted in an increase in the amount of apoptotic cells to 15.0% and 6.8%, respectively. MTT reagent is metabolized by various cellular enzymes, and inhibition of its activity results in a decrease in formazan production, which is reflected by a decrease in the viability of cells. Nonetheless, the decrease in formazan production could be a result of not only cell apoptosis but also a consequence of cell death via other mechanisms or inhibition of cell adhesion.

General Experimental Procedures
Optical rotations were measured on a Perkin-Elmer 343 polarimeter (Perkin Elmer, Waltham, MA, USA). UV spectra were recorded on a Shimadzu UV-1601PC spectrometer (Shimadzu Corporation, Kyoto, Japan) in methanol. CD spectra were measured with a Chirascan-Plus CD spectrometer (Leatherhead, UK) in methanol. NMR spectra were recorded in CDCl 3 , acetone-d 6 , and DMSO-d 6 on a Bruker DPX-300 (Bruker BioSpin GmbH, Rheinstetten, Germany), a Bruker Avance III-500 (Bruker BioSpin GmbH, Rheinstetten, Germany), and a Bruker Avance III-700 (Bruker BioSpin GmbH, Rheinstetten, Germany) spectrometer, using TMS as an internal standard. HRESIMS spectra were measured on a Maxis impact mass spectrometer (Bruker Daltonics GmbH, Rheinstetten, Germany). Microscopic examination and photography of fungal cultures were performed with an Olympus CX41 microscope equipped with an Olympus SC30 digital camera. Detailed examination of the ornamentation of the fungal conidia was performed using and EVO 40 scanning electron microscope (SEM).

Fungal Strain
The A. carneus fungal strain was isolated from superficial mycobiota of the brown alga Laminaria sachalinensis (Miyabe) collected on Kunashir Island and was identified based on morphological evaluation by Dr. Mikhail V. Pivkin from the Pacific Institute of Bioorganic Chemistry (PIBOC). The strain is stored in the Collection of Marine Microorganisms, PIBOC, Vladivostok, Russia, under the code KMM 4638.
The Beauveria felina fungal strain was isolated from marine sediments collected at a depth of 10 m (Van Phong Bay, the South China Sea, Vietnam) during 34th expedition of r/v "Akademik Oparin" and was identified based on morphological evaluation by Dr. Natalya N. Kirichuk from the Pacific Institute of Bioorganic Chemistry (PIBOC). The strain is stored in the Collection of Marine Microorganisms, PIBOC, Vladivostok, Russia, under the code KMM 4639.

Cultivation of Fungus
The fungi A. carneus and B. felina were cultivated separately at 22 • C for 7 days in Erlenmeyer flasks (500 mL), each containing 20 g of rice, 20 mg of yeast extract, 10 mg of KH 2 PO 4 , and 40 mL of natural sea water. B. felina mycelium was inoculated into 20 flasks with A. carneus culture. The fungus was inoculated with three pieces of mycelium (~0.5 × 0.5 mm) in each flask. Then, fungal cultures were cocultivated for 14 days.

Extraction and Isolation
At the end of the incubation period, the mycelium and medium were homogenized and extracted with EtOAc (4 L). The extract was concentrated to dryness. The residue was dissolved in 20% EtOH-H 2 O (1 L) and was extracted with n-hexane (200 mL × 3), EtOAc (200 mL × 3), and n-BuOH (150 mL × 2). After evaporation of the EtOAc layer, the residual material (6 g) was passed over silica columns (4 × 20 cm), which was eluted first with n-hexane (2 L), followed by a step gradient from 5% to 100% EtOAc in n-hexane (total volume 10 L). Fractions of 200 mL were collected and combined on the basis on TLC (silica gel, toluene-isopropanol 6:1, v/v).