Eremophilane Sesquiterpenes from a Deep Marine-Derived Fungus, Aspergillus sp. SCSIOW2, Cultivated in the Presence of Epigenetic Modifying Agents

Chemical epigenetic manipulation was applied to a deep marine-derived fungus, Aspergillus sp. SCSIOW2, resulting in significant changes of the secondary metabolites. Three new eremophilane-type sesquiterpenes, dihydrobipolaroxin B (2), dihydrobipolaroxin C (3), and dihydrobipolaroxin D (4), along with one known analogue, dihydrobipolaroxin (1), were isolated from the culture treated with a combination of histone deacetylase inhibitor (suberohydroxamic acid) and DNA methyltransferase inhibitor (5-azacytidine). 1–4 were not produced in the untreated cultures. 2 and 3 might be artificial because 1 could form 2 and 3 spontaneously in water by intracellular acetalization reaction. The absolute configurations of 1 and 2 were assigned based on ECD spectroscopy combined with time-dependent density functional theory calculations. All four compounds exhibited moderate nitric oxide inhibitory activities without cytotoxic effects.


Introduction
Marine fungi have attracted increasing attention as a source of structurally novel and biologically active secondary metabolites [1,2]. However, with the recent completion of fungal genomes, it has become clear that the number of gene clusters that encode secondary metabolites greatly outnumbers the characterized compounds from these organisms [3]. This observation suggests that many gene clusters generally remain unexpressed under a variety of laboratory culture conditions. Epigenetic modifying agents, primarily histone-modifying and DNA methylation-modifying agents, have been introduced as promising tools for manipulating the silent fungal genes for the purpose of discovering novel structures [4][5][6]. This approach has already been successfully applied to several marine-derived fungi to discover novel natural products using chemical epigenetic modifying agents, such as the histone deacetylase inhibitors suberohydroxamic acid (SBHA) [7,8] and sodium butyrate [9] and the DNA methyltransferase inhibitor 5-azacytidine (5-AZA) [10,11]. Moreover, studies on the effect of the concomitant addition of an HDAC inhibitor and a DNA methyltransferase inhibitor on fungal secondary metabolite production have been conducted. A marked increase was observed in the secondary metabolites produced by Isaria tenuipes and Gibellula formosana cultivated in the presence of both SBHA and RG108 (a DNA methyltransferase (DMNT) inhibitor) compared with those of fungi cultivated without inhibitors or with either SBHA or RG-108 [12,13]. In our efforts to identify with either SBHA or RG-108 [12,13]. In our efforts to identify novel structures and bioactive metabolites from deep sea-derived (over 1000 m) fungi, we found that the EtOAc extracts of Aspergillus sp. SCSIOW2 exhibited strong potency for inhibiting nitric oxide production without cytotoxic effects. A bioassay-guided chemical investigation resulted in the isolation of a novel cyclic dipeptide, cyclo-(L-N-MeTyr-anthranilic acid), which we have named 14-hydroxy-cyclopeptine [14]. To enhance the chemical diversity from this strain, the chemical epigenetic induction method was applied to the fermentation. We found that the cultivation of Aspergillus sp. SCSIOW2 with a combination of 1 mM SBHA, a competitive histone deacetylase (HDAC) inhibitor, and 1 mM 5-azacytidine (5-AZA), a DNMT inhibitor, resulted in a marked increase in the production of secondary metabolites compared with those of fungi cultivated without inhibitors, as indicated by LCMS and TLC analyses ( Figure S1A-D). Therefore, the EtOAc extract of this culture was scaled-up and further separated using column chromatography and semi-preparative HPLC, resulting in the isolation of four eremophilane-type sesquiterpenes, dihydrobipolaroxin (1), dihydrobipolaroxin B (2), dihydrobipolaroxin C (3), and dihydrobipolaroxin D (4) (Figure 1). Compounds 1-4 were not produced in the untreated cultures ( Figure S1). 1 is a known compound that was previously reported from Bipolaris cynodontis, a fungal pathogen of Bermuda grass [15], however, there is no 13 C-NMR data and the absolute stereochemistry was not resolved. Here we reported the chemical epigenetic induction, production, isolation, and nitric oxide production inhibitory activities of these four compounds from a deep marine-derived fungus, Aspergillus sp. SCSIOW2.

Results
Compound 1 was isolated as a white amorphous powder. The structure of 1 was solved based on comprehensive NMR and HRMS analysis. Using HRESIMS, a molecular ion was measured at 265.1435 [M + H] + (calcd. for C15H21O4, 265.1440), indicating a molecular formula of C15H20O4 with six degrees of unsaturation. The UV spectrum (MeOH) presented a single absorption maximum at λmax (log ε) 281 (4.37) nm. The 1 H and 13 C-NMR, DEPT, and HMQC spectroscopic data revealed a carbonyl carbon, three olefinic methines, an olefinic quaternary carbon, an exo-methylene group, an oxy-quaternary carbon, an oxymethine, an oxymethylene, an aliphatic methylene, two methyl groups, and three hydroxyl groups. The planar structure of 1 was determined to be dihydrobipolaroxin based on careful 1 H-1 H COSY and HMBC analyses (Table 1 and Figure 2) and comparison with the literature data [8]. Because the stereochemistry was not previously reported for this compound, NOESY correlations were utilized to determine the relative configuration of 1.
The axial methyl Me-14 exhibited correlations with 7-OH, Me-15, and H-6β. H-4 exhibited cross peaks with H-3 and H-6α; Me-15 exhibited cross peaks with Me-14 and H-6β; and Me-15 also exhibited cross peaks with H-3 and 3-OH with similar peak intensities; therefore, the α orientations

Results
Compound 1 was isolated as a white amorphous powder. with six degrees of unsaturation. The UV spectrum (MeOH) presented a single absorption maximum at λ max (log ε) 281 (4.37) nm. The 1 H-and 13 C-NMR, DEPT, and HMQC spectroscopic data revealed a carbonyl carbon, three olefinic methines, an olefinic quaternary carbon, an exo-methylene group, an oxy-quaternary carbon, an oxymethine, an oxymethylene, an aliphatic methylene, two methyl groups, and three hydroxyl groups. The planar structure of 1 was determined to be dihydrobipolaroxin based on careful 1 H-1 H COSY and HMBC analyses (Table 1 and Figure 2) and comparison with the literature data [8].
Because the stereochemistry was not previously reported for this compound, NOESY correlations were utilized to determine the relative configuration of 1. The axial methyl Me-14 exhibited correlations with 7-OH, Me-15, and H-6β. H-4 exhibited cross peaks with H-3 and H-6α; Me-15 exhibited cross peaks with Me-14 and H-6β; and Me-15 also exhibited cross peaks with H-3 and 3-OH with similar peak intensities; therefore, the α orientations of Me-14, Me-15, and 3-OH were established, while Me-15 was equatorial. Furthermore, one of the exo-methylene protons (H-13a) that resonated at δ H 5.18 showed cross peaks with H-6β, while 7-OH presented correlations with Me-14 in the NOESY spectra. Thus, 7-OH was determined to be in a β orientation (Table 1, Figure 3). The relative configuration of 1 was thus established. A crystal for an XRD structure of 1 was unachievable due to the limited amount and the instability of the structure. The ECD spectrum of 1 was then recorded and compared with those calculated for each enantiomer using the time-dependent density functional theory (TDDFT) method. After conformation space analysis, 10 conformers were found for 1 (Table S1 shows the equilibrium populations of 10 stable conformations in methanol at the B3LYP/aug-cc-pVDZ level). Consequently, the calculated ECD spectrum for the enantiomer 3S4R5R7R showed a perfect fit with the experimental plot of 1, which exhibited one negative and three positive Cotton effects at 219, 244, 279, and 356 nm, respectively ( Figure 4). However, the calculated ECD spectrum of the 3R4S5S7S enantiomer was opposite to the experimental ECD data, with one positive and three negative Cotton effects at 210, 240, 282, and 356 nm, respectively ( Figure 4). Hence, the stereochemistry of 1 was determined to be 3S4R5R7R.  Figure 3). The relative configuration of 1 was thus established. A crystal for an XRD structure of 1 was unachievable due to the limited amount and the instability of the structure. The ECD spectrum of 1 was then recorded and compared with those calculated for each enantiomer using the time-dependent density functional theory (TDDFT) method. After conformation space analysis, 10 conformers were found for 1 (Table S1 shows the equilibrium populations of 10 stable conformations in methanol at the B3LYP/aug-cc-pVDZ level). Consequently, the calculated ECD spectrum for the enantiomer 3S4R5R7R showed a perfect fit with the experimental plot of 1, which exhibited one negative and three positive Cotton effects at 219, 244, 279, and 356 nm, respectively ( Figure 4). However, the calculated ECD spectrum of the 3R4S5S7S enantiomer was opposite to the experimental ECD data, with one positive and three negative Cotton effects at 210, 240, 282, and 356 nm, respectively ( Figure 4). Hence, the stereochemistry of 1 was determined to be 3S4R5R7R.     Figure 3). The relative configuration of 1 was thus established. A crystal for an XRD structure of 1 was unachievable due to the limited amount and the instability of the structure. The ECD spectrum of 1 was then recorded and compared with those calculated for each enantiomer using the time-dependent density functional theory (TDDFT) method. After conformation space analysis, 10 conformers were found for 1 (Table S1 shows the equilibrium populations of 10 stable conformations in methanol at the B3LYP/aug-cc-pVDZ level). Consequently, the calculated ECD spectrum for the enantiomer 3S4R5R7R showed a perfect fit with the experimental plot of 1, which exhibited one negative and three positive Cotton effects at 219, 244, 279, and 356 nm, respectively ( Figure 4). However, the calculated ECD spectrum of the 3R4S5S7S enantiomer was opposite to the experimental ECD data, with one positive and three negative Cotton effects at 210, 240, 282, and 356 nm, respectively ( Figure 4). Hence, the stereochemistry of 1 was determined to be 3S4R5R7R.     Figure 3). The relative configuration of 1 was thus established. A crystal for an XRD structure of 1 was unachievable due to the limited amount and the instability of the structure. The ECD spectrum of 1 was then recorded and compared with those calculated for each enantiomer using the time-dependent density functional theory (TDDFT) method. After conformation space analysis, 10 conformers were found for 1 (Table S1 shows the equilibrium populations of 10 stable conformations in methanol at the B3LYP/aug-cc-pVDZ level). Consequently, the calculated ECD spectrum for the enantiomer 3S4R5R7R showed a perfect fit with the experimental plot of 1, which exhibited one negative and three positive Cotton effects at 219, 244, 279, and 356 nm, respectively ( Figure 4). However, the calculated ECD spectrum of the 3R4S5S7S enantiomer was opposite to the experimental ECD data, with one positive and three negative Cotton effects at 210, 240, 282, and 356 nm, respectively ( Figure 4). Hence, the stereochemistry of 1 was determined to be 3S4R5R7R.       13 C-NMR were almost identical to those of 1, except for three major differences: the absence of the resonance for the carbonyl moiety (C-8), which had been replaced by an acetal carbon (δ C 101.4, q); the oxymethylene group (δ H 3.82 d 4.2, H-12) changed to an AB quartet at δ H 4.37 (d, 13.8) and δ H 4.24 (d, 13.8), indicating a restrictive partial structure; and an additional methoxy group (δ H 3.27 s, δ C 48.1) ( Table 2). These data strongly suggested an acetal group rather than a carbonyl group at C-8. The acetal structure at C-8 was further confirmed by HMBC correlations from the 8-methoxy and 12-H to the acetal carbon (C-8) (Table 2, Figure 2). Finally, careful 1 H-1 H COSY, HMQC, and HMBC analyses confirmed the structure as 2 (Table 2, Figure 2). The NOESY correlation between 8-methoxy and 14-Me indicated the β-orientation of the C-8 methoxy group. The NOE cross peak between H-12α and H-6α also supported the β-orientation of the C-8 methoxy because the distance between these two protons would be too far for an α-orientation. The other NOESY correlations were almost identical to those of 1, indicating the same relative configurations (Table 2, Figure 3). Determining the absolute configuration of 2 was also attempted by ECD analysis. Only one conformer was found through confirmation space analysis because of the rigid structure of 2. The calculated ECD spectrum for the enantiomer 3S4R5R7R8S exhibited a good fit with the experimental plot of 2, which presented one major positive Cotton effect at 238 nm, whereas the calculated ECD spectrum of the 3R4S5S7S8R enantiomer was opposite to the experimental ECD data, with one negative Cotton effect at 230 nm ( Figure 5). Thus, the 3S4R5R7R8S of 2 was determined.   13 C-NMR spectra exhibited two sets of signals belonging to two bipolaroxin-type sesquiterpenes, respectively (Table 3). One set of signals is similar with those of 1, possessing an 8′-carbonyl carbon (δC 196.7) and a free rotational 12′-oxymethylene group (δH 4.08 s, δC 61.4); the other set is almost identical with those of 2, possessing a rigid five-membered ring and a C-8 acetal unit, except for the lack of a methoxy group. Careful 2D NMR analysis confirmed these two partial structures ( Table 3). The HMBC correlation from the 12′ proton to the 8-acetal carbon unambiguously connected these two partial structures, and the structure was then characterized as 3. NOE correlations of the two partial eremophilanes were almost identical with those of 1 and 2, suggesting the same relative configurations. The only NOE correlation between these two partial structures is H-12′ with H-9, which did not provide us with much information about the stereochemistry of the C-8 linkage.   13 C-NMR spectra exhibited two sets of signals belonging to two bipolaroxin-type sesquiterpenes, respectively (Table 3). One set of signals is similar with those of 1, possessing an 8 1 -carbonyl carbon (δ C 196.7) and a free rotational 12 1 -oxymethylene group (δ H 4.08 s, δ C 61.4); the other set is almost identical with those of 2, possessing a rigid five-membered ring and a C-8 acetal unit, except for the lack of a methoxy group. Careful 2D NMR analysis confirmed these two partial structures ( Table 3). The HMBC correlation from the 12 1 proton to the 8-acetal carbon unambiguously connected these two partial structures, and the structure was then characterized as 3. NOE correlations of the two partial eremophilanes were almost identical with those of 1 and 2, suggesting the same relative configurations. The only NOE correlation between these two partial structures is H-12 1 with H-9, which did not provide us with much information about the stereochemistry of the C-8 linkage. However, the NOE cross peak between H-12α and H-6α convincingly indicated the β-orientation of the C-8 linkage. 1 was found to be unstable in MeOH:H 2 O (9:1) solution when attempting crystallization one week after storage at room temperature, approximately 50% of 1 changed to 2 and 3 ( Figure S2). Based on structures of 1-3, the conversion was considered to be a typical acetalization reaction. The first step was nucleophilic addition of the α,β-unsaturated alcohol (12-OH) to the C-8 ketone; the second step was the addition of a methoxy group or 12-OH from another molecule to form 2 or 3, respectively.  13 C-NMR spectra were quite similar to those of 3, including two sets of bipolaroxin-type signals ( Table 3). The major differences between 4 and 3 were the absence of two oxymethine groups, which were replaced by two carbonyl carbons (δ C 200.3, 200.0) ( Table 3). HMBC correlations between one carbonyl carbon at δ C 200.3 with H1, H4, and Me-15 and the other one at δ C 200.0 with H-1 1 , H-4 1 , and Me-15 1 confirmed that the carbonyl carbons were located at the C-3 or C-3 1 position. 3 was determined to be a dimer of 1, connected with an acetal linkage at C-8; however, it was determined that 4 was a mixture of two compounds by careful 2D-NMR analyses (Table 3). One compound (A form) is similar to 1, containing a free rotational 12 1 -OH. The other compound (B form) is similar to 2, containing a rigid five-membered ring with a hemiacetal structure. The A and B forms of 4 both contained 3 (or 3 1 )-carbonyl groups. NOESY correlations suggested the same relative configurations of 4 with those of 1, 2, and 3. In the A form, H-4 1 exhibited a cross peak with H-6 1 , Me-15 1 exhibited cross peaks with Me-14 1 , H-6 β 1 , and OH-7 1 . Therefore, the α orientations of Me-14, Me-15, and OH-7 were established, while Me-15 was equatorial. In the B form, Me-14 exhibited a cross peak with OH-8, indicating the β orientation of the 8-hydroxyl group. The other NOE correlations were almost identical with those of the A form, suggesting the same relative configurations. The 1 H-NMR integral area showed that the two distinct sets of signals can be observed in an approximately 3:5 (A:B, Figure S21) ratio in DMSO-d 6 . On standing in MeOD-d 4 , the equilibration changed to be 1:4 (A:B, Figure S27). The attempt to isolate the pure form of 4 failed due to the quick equilibration during the evaporation.
Great efforts have been conducted to purify 1, too. HPLC analysis of purified 1 showed one clean peak ( Figure S1A), however, it still contained minor impurities based on NMR spectra ( Figures S3  and S4). The impurities might be generated from the same intramolecular acetalization to form an equilibria of compounds because those minor signals were very similar to 2 ( Figures S3, S4, S9 and S10). However, we could not confirm the minor structure due to the low intensity of those signals.
We examined the inhibitory effects of 1-4 on the production of NO induced by LPS/INF-γ. 1-4 all exhibited NO production inhibitory activity with a dose dependent manner ( Figure 6). The cytotoxic effects of 1-4 were measured using an MTT assay. None of the compounds exhibited any cytotoxic effects in the tested dose range (100-12.5 µM, Figure 7). It should be noticed that 1 contained minor impurities. The impurities could possibly be an equilibria form of 1 generated from intramolecular acetalization, which we could not confirm in this paper.
unsaturation. The 1 H-and 13 C-NMR spectra were quite similar to those of 3, including two sets of bipolaroxin-type signals ( Table 3). The major differences between 4 and 3 were the absence of two oxymethine groups, which were replaced by two carbonyl carbons (δC 200.3, 200.0) ( Table 3). HMBC correlations between one carbonyl carbon at δC 200.3 with H1, H4, and Me-15 and the other one at δC 200.0 with H-1′, H-4′, and Me-15′ confirmed that the carbonyl carbons were located at the C-3 or C-3′ position. 3 was determined to be a dimer of 1, connected with an acetal linkage at C-8; however, it was determined that 4 was a mixture of two compounds by careful 2D NMR analyses (Table 3). One compound (A form) is similar to 1, containing a free rotational 12′-OH. The other compound (B form) is similar to 2, containing a rigid five-membered ring with a hemiacetal structure. The A and B forms of 4 both contained 3 (or 3′)-carbonyl groups. NOESY correlations suggested the same relative configurations of 4 with those of 1, 2, and 3. In the A form, H-4′ exhibited a cross peak with H-6′, Me-15′ exhibited cross peaks with Me-14′, H-6 β′, and OH-7′. Therefore, the α orientations of Me-14, Me-15, and OH-7 were established, while Me-15 was equatorial. In the B form, Me-14 exhibited a cross peak with OH-8, indicating the β orientation of the 8-hydroxyl group. The other NOE correlations were almost identical with those of the A form, suggesting the same relative configurations. The 1 H NMR integral area showed that the two distinct sets of signals can be observed in an approximately 3:5 (A:B, Figure S21) ratio in DMSO-d6. On standing in MeOD-d4, the equilibration changed to be 1:4 (A:B, Figure S27). The attempt to isolate the pure form of 4 failed due to the quick equilibration during the evaporation.
Great efforts have been conducted to purify 1, too. HPLC analysis of purified 1 showed one clean peak ( Figure S1A), however, it still contained minor impurities based on NMR spectra ( Figures S3 and S4). The impurities might be generated from the same intramolecular acetalization to form an equilibria of compounds because those minor signals were very similar to 2 (Figures S3, S4, S9 and S10). However, we could not confirm the minor structure due to the low intensity of those signals.
We examined the inhibitory effects of 1-4 on the production of NO induced by LPS/INF-γ. 1-4 all exhibited NO production inhibitory activity with a dose dependent manner ( Figure 6). The cytotoxic effects of 1-4 were measured using an MTT assay. None of the compounds exhibited any cytotoxic effects in the tested dose range (100-12.5 μM, Figure 7). It should be noticed that 1 contained minor impurities. The impurities could possibly be an equilibria form of 1 generated from intramolecular acetalization, which we could not confirm in this paper.  using a t test to determine any significant differences. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001 compared to control group with LPS/IFN-γ.

Discussion
Eremophilane type sesquiterpenes belong to bicyclic sesquiterpenes, containing only two complete isoprene subunits, are present in about 20 genera of the Asteraceae family, and can be regarded as the chemotaxonomic markers sources of Ligularia, Senecio, Cacalia, and Petasites genera [16]. They are also widely found from fungi species. These compounds are of interest because many of them showed a wide range of pharmaceutical functions including phytotoxic, cytotoxic, anti-bacterial, anti-microbial, anti-inflammatory, anti-allergic, and so on [16]. Macrophages play major roles in the immune responses by releasing various factors such as pro-inflammatory cytokines and nitrogen species. Excessive production of nitric oxide (NO) appears to associate with many chronic or acute diseases related to inflammation such as rheumatoid arthritis, cancer and even Alzheimer's disease [17,18]. Inducible nitric oxide synthase (iNOS) is a key enzyme in the macrophage inflammatory response, which is the source of NO that is potently induced in response to proinflammatory stimuli. Activation interferon (IFN)-γ increases iNOS promoter activity and expression in response to lipopolysaccharide (LPS) in vitro [19,20]. Thus, we used LPS and IFN-γ activated Macrophage cells to evaluate NO production inhibitory activities. The cell viability was measured by MTT assay. Measurement of mitochondrial metabolic rate using MTT assay to indirectly reflect viable cell numbers has been widely applied. However, metabolic activity may be changed by different conditions or chemical treatments which can cause considerable variation in results reported from these assays [21]. Based on our knowledge, seven eremophilane type sesquiterpenes have been reported for their NO production inhibitory activities with the highest activity of 0.55 μg/mL (IC50), and 6-hydroxy group was considered critical for increasing the ability to inhibit NO production [22,23]. In our study, all four compounds (1-4) exhibited nitric oxide inhibitory activities in a dose dependent manner without cytotoxic effects. Although the activities were not as strong as reported components, these results showed that eremophilanes might have therapeutic benefits against various types of diseases by NO production inhibition potency.

General Experimental Procedures
Optical rotations were determined on a Jasco P-1020 polarimeter (Jasco, Tokyo, Japan). UV data were recorded on a Perkin Elmer Lambda 25 UV/Vis spectrometer (PerkinElmer, Boston, MA,

Discussion
Eremophilane type sesquiterpenes belong to bicyclic sesquiterpenes, containing only two complete isoprene subunits, are present in about 20 genera of the Asteraceae family, and can be regarded as the chemotaxonomic markers sources of Ligularia, Senecio, Cacalia, and Petasites genera [16]. They are also widely found from fungi species. These compounds are of interest because many of them showed a wide range of pharmaceutical functions including phytotoxic, cytotoxic, anti-bacterial, anti-microbial, anti-inflammatory, anti-allergic, and so on [16]. Macrophages play major roles in the immune responses by releasing various factors such as pro-inflammatory cytokines and nitrogen species. Excessive production of nitric oxide (NO) appears to associate with many chronic or acute diseases related to inflammation such as rheumatoid arthritis, cancer and even Alzheimer's disease [17,18]. Inducible nitric oxide synthase (iNOS) is a key enzyme in the macrophage inflammatory response, which is the source of NO that is potently induced in response to proinflammatory stimuli. Activation interferon (IFN)-γ increases iNOS promoter activity and expression in response to lipopolysaccharide (LPS) in vitro [19,20]. Thus, we used LPS and IFN-γ activated Macrophage cells to evaluate NO production inhibitory activities. The cell viability was measured by MTT assay. Measurement of mitochondrial metabolic rate using MTT assay to indirectly reflect viable cell numbers has been widely applied. However, metabolic activity may be changed by different conditions or chemical treatments which can cause considerable variation in results reported from these assays [21]. Based on our knowledge, seven eremophilane type sesquiterpenes have been reported for their NO production inhibitory activities with the highest activity of 0.55 µg/mL (IC 50 ), and 6-hydroxy group was considered critical for increasing the ability to inhibit NO production [22,23]. In our study, all four compounds (1-4) exhibited nitric oxide inhibitory activities in a dose dependent manner without cytotoxic effects. Although the activities were not as strong as reported components, these results showed that eremophilanes might have therapeutic benefits against various types of diseases by NO production inhibition potency.

General Experimental Procedures
Optical rotations were determined on a Jasco P-1020 polarimeter (Jasco, Tokyo, Japan). UV data were recorded on a Perkin Elmer Lambda 25 UV/Vis spectrometer (PerkinElmer, Boston, MA, USA). IR data were recorded using a Nicolet Avatar 330 FT-IR spectrometer (Thermo Scientific, Waltham, MA, USA). NMR spectra were acquired on a Bruker ASCEND 600 MHz NMR magnet system (Bruker, Ettlingen, Germany) using TMS as the internal standard. HR-ESIMS was performed using an AB SCIEX TOF/TOF™ 5800 system (AB Sciex, Redwood City, CA, USA). CD spectra were recorded on a Jasco J-815 CD Spectrometer (Jasco, Tokyo, Japan). Column chromatography was conducted using silica gel (

Fermentation, Extraction, and Isolation
Both seed and production media have the same constituents (2.0% glucose, 1.0% peptone, 0.5% yeast extract, 3.0% sea salt, with the pH adjusted to 7.5). Aspergillus sp. SCSIOW2 was cultured in 250 mL Erlenmeyer flasks containing 50 mL of seed medium. After growing at 28˝C, 220 rpm for two days, the cellular material was placed in a sterile Falcon tube and mixed by vortexing for several minutes to create a uniform fungal cell/spore suspension. Aliquots (5 mL) of seed cultures were inoculated into 160 mL of production medium in 1000 mL Erlenmeyer flasks.

Quantum Chemical ECD Calculations
In this study, we used the ECD calculation protocol proposed by Nugroho and Morita [24]. The initial 3D structures of the molecules were prepared using Chem3D and minimized with the MMFF94S force field implemented in Chem3D. After the initial structure was further optimized with XedMin in default mode, the conformation space was sampled using XedeX with an energy window of 5 kcal¨mol´1 above the ground state and RMSD 0.8 to remove duplicated conformers [25]. Then, each conformer was optimized and verified as true minima of the potential energy surface using Gaussian 09 with the DFT method at the B3LYP/aug-cc-pVDZ level [26]. The polarizable continuum model (IEFPCM) was used to take the solvent effects of methanol into account. The optimized conformers were further used to perform a TDDFT calculation at the B3LYP/aug-cc-pVDZ level with the polarizable-conductor calculation model (IEFPCM, methanol as the solvent). In each TDDFT calculation, the 100 lowest electronic transitions were calculated for each conformer. The ECD spectra and overall ECD spectra (weighted by Boltzmann statistics) and comparison of the experimental and calculated spectra were performed using the software SpecDis [27,28].

Cell Culture Condition
Mouse RAW264.7 macrophage cells purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) , were cultured in RPMI 1640 medium supplemented with 10 % heat-inactivated fetal bovine serum (Gibco) at 37˝C in a humidified incubator with 5 % CO 2 and 95 % air. The medium was routinely changed every two days. RAW 264.7 cells were passaged by trypsinization until they attained confluence.

Nitric Oxide Inhibitory Assay
The cells were plated in a 96-well plate at a density of 1.0ˆ10 5 cells/well and grown for 2 h to allow the cells to attach to the plate. The tested samples were dissolved in DMSO, and then two-fold serial dilutions were performed in RPMI-1640 (Thermo Fisher Scientific, Carlsbad, CA, USA). The final concentrations of samples 1-4 in the culture medium were 100, 50, 25, and 12.5 µM. The samples were added to the culture simultaneously with both Escherichia coli LPS (1.5 µg/mL, Sigma, St. Louis, MO, USA) and recombinant mouse IFN-γ (10 ng/mL, PeproTech, Rocky Hill, NJ, USA). Then, the cells were incubated at 37˝C for approximately 24 h and subsequently chilled on ice. Subsequently, 100 µL of the culture supernatant was placed in duplicate in the wells of 96-well plates. To quantify nitrite, 50 µL of Griess reagent (1% sulfanilamide in 5% H 2 PO 4 and 0.1% N-1-naphthyletylenediamide dihydrochloride) was added to each well. After 10 min, the reaction products were colorimetrically quantified at 550 nm using a microplate reader (BIO-RAD, Hercules, CA, USA). The concentrations of nitrite were calculated by using a standard calibration curve [29,30].

MTT Assay
Cell viability was determined using an MTT assay. After 24 h of incubation with or without test samples, the old medium was replaced with 100 µL of fresh culture medium, and 10 µL of the 12 mM MTT stock solution was then added to each well. The cells were cultured at 37˝C for 4 h, the supernatant was removed, and then 50 µL of DMSO was added to each well and mixed thoroughly. The cells were further incubated at 37˝C for 10 min. The reaction products were quantified at 540 nm using a microplate reader (BIO-RAD). The untreated cells were considered as having 100% of viable cells. Results are expressed as percentage of viable cells when compared with the control group [31].

Statistical Analysis
All results were expressed as mean˘SD. Statistical comparison was conducted using Student's t test. The results were considered to be significant when p < 0.05.

Conclusions
In conclusion, the marine-derived fungal strain Aspergillus sp. SCSIOW2 was effectively induced by epigenetic modifying agents to produce novel compounds. Three new eremophilane-type sesquiterpenes, dihydrobipolaroxin B (2), dihydrobipolaroxin C (3), and dihydrobipolaroxin D (4), along with one known analogue, dihydrobipolaroxin (1), were isolated from the culture treated with a combination of SBHA and 5-AZA. The eremophilane sesquiterpenes were produced by Aspergillus sp. SCSIOW2 only when treated with chemical epigenetic modifiers. This result suggests that the combination of HDAC and DNMT inhibitors can be used to identify diverse natural products hidden in silent biosynthetic pathways from marine fungi. The structures of all compounds were determined on the basis of spectroscopic data. Compound 1 was previously reported from Bipolaris cynodontis, a fungal pathogen of Bermuda grass, as a minor constituent, along with a major compound, bipolaroxin [15].