New Dibenzo-α-pyrone Derivatives with α-Glucosidase Inhibitory Activities from the Marine-Derived Fungus Alternaria alternata

Three new dibenzo-α-pyrone derivatives, alternolides A–C (1–3), and seven known congeners (4–10) were isolated from the marine-derived fungus of Alternaria alternata LW37 assisted by the one strain-many compounds (OSMAC) strategy. The structures of 1–3 were established by extensive spectroscopic analyses, and their absolute configurations were determined by modified Snatzke′s method and electronic circular dichroism (ECD) calculations. Compounds 6 and 7 showed good 1,1-diphenyl-2-picrylhydrazyl (DPPH) antioxidant scavenging activities with IC50 values of 83.94 ± 4.14 and 23.60 ± 1.23 µM, respectively. Additionally, 2, 3 and 7 exhibited inhibitory effects against α-glucosidase with IC50 values of 725.85 ± 4.75, 451.25 ± 6.95 and 6.27 ± 0.68 µM, respectively. The enzyme kinetics study indicated 2 and 3 were mixed-type inhibitors of α-glucosidase with Ki values of 347.0 and 108.5 µM, respectively. Furthermore, the interactions of 2, 3 and 7 with α-glucosidase were investigated by molecular docking.

More than 70% of the Earth's surface is covered by oceans, and numerous marinederived fungi have been isolated and identified in oceanic sediments, sponges, algae, More than 70% of the Earth's surface is covered by oceans, and numerous marinederived fungi have been isolated and identified in oceanic sediments, sponges, algae, etc. [25]. Marine-derived fungi survive extreme conditions such as absence of light, low levels of oxygen and intensely high pressures, which may result in unique biological metabolic pathways, and were considered to be a rich source of structurally diverse and biologically active metabolites for drug discovery [26][27][28]. Alternaria species have a widespread distribution in nature, acting as plant (include marine algae) pathogens, endophytes and saprophytes [29][30][31][32][33]. A. alternata is an extremely common and cosmopolitan species occur in many types of plant, soil and marine environments [34,35]. The fungal genus Alternaria can produce diverse secondary metabolites including dibenzo-α-pyrones [2], terpenoids [36] and polyketides [37,38], which show a broad range of biological activities such as antibacterial [2], anti-inflammatory [36], acetylcholinesterase inhibitory [37] and cytotoxic activities [38]. The secondary metabolites from marine-derived Alternaria sp. are also endowed with unique structures and varieties of bioactivities, such as the anti-inflammatory agent tricycloalternarene A possessing the unique fusion of an oxaspiro [5.5]nonane and a cyclohexenone ring [39], the cytotoxic agent altertoxin VII featured by a perylenequinone skeleton [40], and the antibacterial agent alternaramide with cyclic pentadepsipeptide skeleton [41].
As part of our ongoing search for bioactive metabolites from the marine-derived fungi [42,43], A. alternata LW37, a fungus isolated from a deep-sea sediment sample collected at a depth of 2623 m in the Southwest Indian Ridge in November 2014, was screened for chemical investigation. The fungus A. alternata LW37 was then cultured in six different media (Table S1) guided by the OSMAC strategy [44]. Analysis of HPLC fingerprints ( Figure S1) showed that the metabolic profile of this fungus on rice is more productive than those on other media. The fungus was cultured on rice for large-scale fermentation. Chemical investigation of the EtOAc extract of the fungus A. alternata LW37 led to the isolation of three new dibenzo-α-pyrones derivatives, alternolides A-C (1-3), and seven known compounds 4−10 ( Figure 1). The isolated compounds were evaluated for their cytotoxic, antioxidant and α-glucosidase inhibitory activities. Details of the isolation, structure elucidation and biological evaluation of these metabolites are reported herein.

Results
Alternolide A (1) was isolated as a yellow oil, and its molecular formula was established as C14H16O6 based on the high-resolution electrospray ionization mass spectrometry (HRESIMS) data at m/z 281.1026 [M + H] + (calcd for C14H17O6 281.1020), indicating 7 degrees of unsaturation. The infrared (IR) absorptions at 3375, 1722, 1629 and 1463 cm −1 sug-

Results
Alternolide A (1) was isolated as a yellow oil, and its molecular formula was established as C 14 43.4 and 28.5), one oxygenated tertiary carbon (δ C 84.7), one methine carbon (δ C 43.5), and one methyl carbon (δ C 20.9). These data accounted for all 1 H and 13 C NMR resonances except for four unobserved exchangeable protons and suggested that 1 was a tricyclic compound with one phenyl subunit. The 1 H-1 H correlation spectroscopy (COSY) spectrum ( Figure S7 (Figure 2) from H-1 to C-2 , C-3 and C-7 , from H 2 -3 to C-1 , C-2 and C-7 , from H-4 and H-6 to C-2 , and from H 3 -7 to C-1 , C-2 and C-3 established the cyclohexane moiety. Other HMBC correlations from aromatic proton H-4 to C-2, C-3, C-5, C-6 and the ester carbonyl carbon C-7 and from H-6 to C-2, C-4, C-5 and C-7 indicated the presence of a 1,2,3,5-tetrasubstituted benzene ring with the ester carbonyl carbon C-7 located at C-2. In addition, key HMBC correlations from H-6 to C-1 and from H-1 to C-1, C-2 and C-6 led to the connection of the tetrasubstituted benzene ring to the cyclohexane moiety via the C-1-C-1 single bond. The four hydroxyl groups were located at C-3, C-5, C-4 and C-5 , respectively, which was supported by the chemical shift values for C-3 (δ C 165.6), C-5 (δ C 166.6), C-4 (δ C 70.0) and C-5 (δ C 72.2). Furthermore, considering one remaining degree of unsaturation and the 13 C NMR chemical shifts of C-7 (δ C 170.6) and C-2 (δ C 84.7), both carbons were connected to the same oxygen atom to form lactone moiety, completing the 6H-benzo[c]chromen-6-one core skeleton. Thus, the planar structure of 1 was determined as depicted ( Figure 1).
The relative configuration of 1 was determined by analysis of the 1 H-1 H coupling constants and nuclear Overhauser effect spectroscopy (NOESY) data (    The absolute configuration of the 4′,5′-diol moiety in 1 was established by the Mo2(OAc)4-induced ECD experiment developed by Santzke [45,46]. As shown in Figure  4, the positive Cotton effect at 310 nm observed in the Mo2(OAc)4-induced ECD spectrum of 1 indicated the 4′S and 5′R configurations. Therefore, the absolute configuration of 1 was assigned as 1′R,2′S,4′S,5′R. This inference was further supported by comparison of the experimental and calculated ECD spectra ( Figure 5). The simulated ECD spectra of (1′R,2′S,4′S,5′R)-1 (1a) and (1′S,2′R,4′R,5′S)-1 (1b) were generated by the time-dependent density functional theory (TDDFT), and the experimental ECD spectra of 1 were in good agreement with the calculated ECD spectrum for 1a. Thus, the structure of 1 was then demonstrated as depicted.   The absolute configuration of the 4′,5′-diol moiety in 1 was established by the Mo2(OAc)4-induced ECD experiment developed by Santzke [45,46]. As shown in Figure  4, the positive Cotton effect at 310 nm observed in the Mo2(OAc)4-induced ECD spectrum of 1 indicated the 4′S and 5′R configurations. Therefore, the absolute configuration of 1 was assigned as 1′R,2′S,4′S,5′R. This inference was further supported by comparison of the experimental and calculated ECD spectra ( Figure 5). The simulated ECD spectra of (1′R,2′S,4′S,5′R)-1 (1a) and (1′S,2′R,4′R,5′S)-1 (1b) were generated by the time-dependent density functional theory (TDDFT), and the experimental ECD spectra of 1 were in good agreement with the calculated ECD spectrum for 1a. Thus, the structure of 1 was then demonstrated as depicted. The absolute configuration of the 4 ,5 -diol moiety in 1 was established by the Mo 2 (OAc) 4induced ECD experiment developed by Santzke [45,46]. As shown in Figure 4, the positive Cotton effect at 310 nm observed in the Mo 2 (OAc) 4 -induced ECD spectrum of 1 indicated the 4 S and 5 R configurations. Therefore, the absolute configuration of 1 was assigned as 1 R,2 S,4 S,5 R. This inference was further supported by comparison of the experimental and calculated ECD spectra ( Figure 5). The simulated ECD spectra of (1 R,2 S,4 S,5 R)-1 (1a) and (1 S,2 R,4 R,5 S)-1 (1b) were generated by the time-dependent density functional theory (TDDFT), and the experimental ECD spectra of 1 were in good agreement with the calculated ECD spectrum for 1a. Thus, the structure of 1 was then demonstrated as depicted.      Alternolide B (2) was also obtained as a yellow oil. The molecular formula was determined as C14H14O6 (eight degrees of unsaturation) by HRESIMS (m/z 279.0872 [M + H] + ), which is two mass units less than that of 1. The 1 H and 13 C NMR data (Table 1)  Alternolide B (2) was also obtained as a yellow oil. The molecular formula was determined as C 14 H 14 O 6 (eight degrees of unsaturation) by HRESIMS (m/z 279.0872 [M + H] + ), which is two mass units less than that of 1. The 1 H and 13 C NMR data (Table 1) of 2 were similar to those of 1, with the exception of the absence of one methine (δ H/C 3.14/43.5, C-1 ) and one methylene (δ H/C 1.71; 2.23/28.5, C-6 ) signal and the presence of the additionally trisubstituted double bond signals (δ C 135.2, C-1 ; δ H/C 6.16/129.7, C-6 ). This was further supported by the HMBC correlations from H-3 , H-5 and H 3 -7 to C-1 , and from H-6 to C-1, C-2 and C-5 ( Figure 2), as well as the 1 H-1 H COSY correlations (Figure 2) of H 2 -3 /H-4 /H-5 /H-6 . Consequently, the gross structure of 2 was established ( Figure 1).
The relative configuration of 2 was also determined by 1 H-1 H coupling constants (Table 1)  that these protons were on the same face of the cyclohexene ring, while the H-4 was on the opposite face of the cyclohexene ring. Thus, the relative configuration was established.
The absolute configurations of C-1 , C-4 and C-5 in 3 were also deduced by comparison of the experimental spectrum of 3 with the calculated ECD spectra for a pair of enantiomers, (1 S,4 S,5 S)-3 (3a) and (1 R,4 R,5 R)-3 (3b). The calculated ECD spectrum of (1 S,4 S,5 S)-3 (3a) showed good agreement with the experimental curve ( Figure 5), which supported the absolute configuration as being 1 S,4 S,5 S. Thus, the completed structure of 3 was elucidated as depicted ( Figure 1).
Compound 9 was identified as 1-deoxyrubralactone by comparison of 1 H and 13 C NMR spectroscopic data and optical rotation with those reported previously in the literature [47]. However, its absolute configuration had never been reported before. Through a comparison of the experimental spectrum of 9 with the calculated ECD spectra for the enantiomers (1S)-9 (9a) and (1R)-9 (9b), we observed that the calculated ECD spectrum of 9a showed good agreement with the experimental one ( Figure 5). Thus, the absolute configuration of 9 was determined as 1S (Figure 1).
Compounds 1-3 were tested for their cytotoxic activities against B16 (mouse melanoma cells), MCF-7 (human breast carcinoma cells) and HepG2 (human hepatocellular carcinoma cells). However, these compounds did not show detectable inhibitory effects on the cell lines tested at 50 µM. Additionally, all of the isolated compounds were tested for their antioxidative activity against DPPH and α-glucosidase inhibitory activities. Compounds 6 and 7 showed good DPPH antioxidant scavenging activities with IC 50 values of 83.94 ± 4.14 and 23.60 ± 1.23 µM, respectively, whereas the corresponding positive control, ascorbic acid, showed an IC 50 value of 23.70 ± 1.03 µM. α-Glucosidase inhibition assay results showed that compounds 2, 3, 7, 8 and 9 exerted α-glucosidase inhibitory activities with inhibition rates of 36.62%, 49.24%, 93.70%, 37.29% and 53.95%, respectively, at a concentration of 400 µM ( Figure 6). Compounds 2 and 3 exhibited inhibition on α-glucosidase with IC 50 values of 725.85 ± 4.75 and 451.25 ± 6.95 µM, respectively, while compound 7 showed significant inhibitory activity with an IC 50 value of 6.27 ± 0.68 µM (the positive control, acarbose, showed an IC 50 value of 1.59 ± 1.37 µM). Acarbose is one of the three α-glucosidase inhibitors in clinics for the treatment of diabetes. In order to gain a better understanding of the α-glucosidase inhibition patterns of 2 and 3, Lineweaver−Burk plots were applied. In the Lineweaver−Burk plots ( Figure 7A,D), both K m and V max values of compounds 2 and 3 decreased with increasing concentration, and the lines of 2 and 3 intersected at the third quadrants. These results suggested that compounds 2 and 3 were mixed-type inhibitors against α-glucosidase, indicating that they were able to bind either the free α-glucosidase or the α-glucosidase-substrate complex. By the secondary plots ( Figure 7B,C,E,F) of the slope and intercept versus concentrations, their K is values (the inhibition constant of the enzyme−substrate complex) were calculated as 982.5 and 513.5 µM, respectively, and K i values (the inhibition constant of the free enzyme) were 347.0 and 108.5 µM, respectively. The K i values were smaller than their K is values, indicating the priority in binding with the free enzyme. In order to gain a better understanding of the α-glucosidase inhibition patterns of 2 and 3, Lineweaver−Burk plots were applied. In the Lineweaver−Burk plots ( Figure 7A,D), both Km and Vmax values of compounds 2 and 3 decreased with increasing concentration, and the lines of 2 and 3 intersected at the third quadrants. These results suggested that compounds 2 and 3 were mixed-type inhibitors against α-glucosidase, indicating that they were able to bind either the free α-glucosidase or the α-glucosidase-substrate complex. By the secondary plots ( Figure 7B,C,E,F)    To investigate the molecular interactions between compounds (2, 3, and 7) and αglucosidase, a molecular docking study was performed using the program AutoDock Vina 1.1.2. The binding modes predicted for compounds 2, 3, and 7 are shown in Figure 8. Compound 2 formed three hydrogen bonds with the Asp1157, His1584 and Thr1586 residues, and 3 formed four hydrogen bonds with the Asp1157, Asp1420, His1584 and Thr1586 residues. Compound 7 formed six hydrogen bonds with Asp1157, Asp1279, Asp1420, Arg1510 and Thr1586 residues ( Figure 8C). The docking results of 2 and 3 revealed that different relative configurations of 4 ,5 -diol unit caused change in the binding mode. It can be argued that the 5 -OH with the absolute configuration S, forming a hydrogen bond with Asp1420, can enhance the α-glucosidase inhibition activity of this class of dibenzo-αpyrones. This conclusion is consistent with the experimental results for enzyme activity. residues, and 3 formed four hydrogen bonds with the Asp1157, Asp1420, His1584 and Thr1586 residues. Compound 7 formed six hydrogen bonds with Asp1157, Asp1279, Asp1420, Arg1510 and Thr1586 residues ( Figure 8C). The docking results of 2 and 3 revealed that different relative configurations of 4′,5′-diol unit caused change in the binding mode. It can be argued that the 5′-OH with the absolute configuration S, forming a hydrogen bond with Asp1420, can enhance the α-glucosidase inhibition activity of this class of dibenzo-α-pyrones. This conclusion is consistent with the experimental results for enzyme activity.

Strain and Fermentation
The fungal strain A. alternata LW37 was isolated from a deep-sea sediment sample collected at a depth of 2623 m in the Southwest Indian Ridge in November 2014. Phylogenetic analyses ( Figure S2) based on LSU, SSU, ITS and RPB2 sequences and morphological

Strain and Fermentation
The fungal strain A. alternata LW37 was isolated from a deep-sea sediment sample collected at a depth of 2623 m in the Southwest Indian Ridge in November 2014. Phylogenetic analyses ( Figure S2) based on LSU, SSU, ITS and RPB2 sequences and morphological features ( Figure S3) indicated that LW37 should be identified as the known species A. alternata, deposited in GenBank as accessions OP316895 (ITS), OP326732 (LSU), OP326733 (SSU) and OP326734 (RPB2), and in the culture collection at the Institute of Microbiology, Chinese Academy of Sciences, Beijing.
The strain was cultured on potato dextrose agar (PDA) plates at 25 • C for 5 d. Additionally, the plugs of agar, supporting mycelial growth, were cut from solid culture medium and transferred aseptically to 250 mL Erlenmeyer flasks, each containing 50 mL liquid medium (0.4% glucose, 1% malt extract and 0.4% yeast extract). Flask cultures were incubated at 28 • C on a rotary shaker at 170 rpm for 5 d to obtain the seed culture. Later, a large-scale fermentation of A. alternata LW37 was performed in solid rice medium using 500 mL × 40 conical flasks for 30 d at 28 • C, and each flask contained 100 g of rice, 110 mL water and 1 mL of the seed culture.

Extraction and Isolation
The fermented rice material was extracted repeatedly with EtOAc (3 × 5.0 L), and the organic solvent was evaporated to dryness to afford the crude extract (35.0 g). The extract was fractionated by silica gel CC using petroleum ether (PE)/EtOAc (

Absolute Configuration of the 4 ,5 -Diol Moiety in 1
Snatzke's method was used to determine the absolute configuration of the 4 ,5 -diol moiety in 1. Dissolve 0.3 mg of 1 and 0.36 mg of Mo 2 (OAc) 4 in dry DMSO to produce a solution at a compound concentration of 0.8 mg/mL. After mixing, the first ECD was recorded immediately, and the additional induced ECD spectra were recorded every 5 min until reaching the stationary state. The inherent ECD spectrum was subtracted. The absolute configuration of the 4 ,5 -diol for compound was demonstrated by the sign at around 310 nm in the observed ECD spectrum.

ECD Calculation
Conformational analysis of compounds 1-3 within an energy window of 3.0 kcal/mol was performed by using the OPLS3 molecular mechanics force field. The conformers were then further optimized with the software package Gaussian 09 [51] at the B3LYP/6-311G(d,p) level, and the harmonic vibrational frequencies were also calculated to confirm their stability. Then, the 60 lowest electronic transitions for the obtained conformers in vacuum were calculated using time-dependent density functional theory (TD-DFT) methods at the B3LYP/6-311G(d,p) level. ECD spectra of the conformers were simulated using a Gaussian function. The overall theoretical ECD spectra were obtained according to the Boltzmann weighting of each conformer.

Bioassays for Cytotoxic Activity
The cytotoxicity evaluations were performed according to the previously described protocol [42].

Antioxidant Assay
The DPPH antioxidant scavenging assay was performed according to the previously reported method [52]. Briefly, 50 µL of DPPH (0.34 mmol/L in EtOH) and 50 µL of a series of solutions (12.5, 25, 50, 100, and 200 µM) of the test compounds 1−10 were mixed in the wells of 96-well plates. Each mixture was incubated at 37 • C for 30 min in a dark environment. The absorbance was read at 517 nm using a microplate reader, employing distilled water as a blank for baseline correction. All experiments were performed in triplicate, and ascorbic acid was used as a positive control.

Bioassays for α-Glucosidase Inhibition Assay
The α-glucosidase inhibitory activity assay was measured as described in previous reports [43,53]. Briefly, 50 µL of 0.5 U/mL α-glucosidase and 25 µL of a series of solutions (0.1, 0.2, 0.4, 0.8 and 1.6 mM) of the test compounds 1−10 were added into 96-well plates. After incubation at 37 • C for 10 min, 25 µL of 25 mM p-NPG was added and further incubated at 37 • C for 10 min. The absorbances were determined at 405 nm on an automatic microplate reader, and acarbose was used as a positive control.

Enzyme Kinetics of α-Glucosidase Inhibition Assay
The inhibition types of compounds 2 and 3 on α-glucosidase were determined by Lineweaver−Burk plots according to a previous report [30]. The α-glucosidase inhibition kinetics were determined with selected concentrations of p-NPG (1.5625, 3.125, 6.25, 12.5 and 25 mM) under different concentrations of 2 and 3 (200, 400 and 800 µM) by keeping the enzyme concentration at 0.5 U/mL. The inhibition constant was determined by the second plots of the apparent K m /V m or 1/V m versus the concentration of the inhibitor.

Molecular Docking Assay
The molecular docking method was used to predict the possible binding sites of 2, 3 and 7 with α-glucosidase [43]. The crystallographic structure of α-glucosidase from yeast (PDB ID: 3TOP) was obtained from the Protein Data Bank. Then, Chemdraw (20.0) and Chem3D (20.0) were used to obtain the chemical and MM2 energy-minimized 3D structures of compounds 2, 3 and 7. AutoDock Vina (1.1.2) was used to prepare the ligand and receptor and subsequent docking. Finally, pymol (2.4.0) was applied to visualize the interaction process for receptor and ligand.

Conclusions
In conclusion, three new dibenzo-α-pyrone derivatives, alternolides A-C (1-3), along with seven known compounds (4−10) were isolated from the crude extract of the marinederived fungus A. alternata LW37 guided by OSMAC strategy. The structures of 1-3 were elucidated on the basis of spectroscopic data, modified Snatzke s method and ECD calcula-tions. Furthermore, we first reported the absolute configuration of 1-deoxyrubralactone (9). As for the bioactivities, the new compounds alternolides B and C were tested as mixed-type inhibitors against α-glucosidase with IC 50 values of 725.85 ± 4.75 and 451.25 ± 6.95 µM, respectively. Unprecedentedly, we perceived that alternariol 1 -hydroxy-9-methyl ether (7) has promising α-glucosidase inhibition activity with an IC 50 value of 6.27 ± 0.68 µM. Meanwhile, the molecular docking assay was used to determine the binding models of 2, 3 and 7 with α-glucosidase. Based on the differences between the absolute configurations, experimental results of enzyme activity and molecular docking results of 2 and 3, we speculated that the absolute configuration of 5 -OH had an effect on the α-glucosidase inhibitory activity of this kind of dibenzo-α-pyrone. This study not only provided a deeper insight into the chemical diversities and bioactivities of dibenzo-α-pyrones, but also demonstrated that marine-derived fungi represent promising producers of natural products with bioactivities for use in drug discovery and development.