Cytotoxic and Antibacterial Compounds from the Coral-Derived Fungus Aspergillus tritici SP2-8-1

Three novel compounds, 4-methyl-candidusin A (1), aspetritone A (2) and aspetritone B (3), were obtained from the culture of a coral-derived fungus Aspergillus tritici SP2-8-1, together with fifteen known compounds (4–18). Their structures, including absolute configurations, were assigned based on NMR, MS, and time-dependent density functional theory (TD-DFT) ECD calculations. Compounds 2 and 5 exhibited better activities against methicillin-resistant strains of S. aureus (MRSA) ATCC 43300 and MRSA CGMCC 1.12409 than the positive control chloramphenicol. Compound 5 displayed stronger anti-MRSA and lower cytotoxic activities than 2, and showed stronger antibacterial activities against strains of Vibrio vulnificus, Vibrio rotiferianus, and Vibrio campbellii than the other compounds. Compounds 2 and 10 exhibited significantly stronger cytotoxic activities against human cancer cell lines HeLa, A549, and Hep G2 than the other compounds. Preliminary structure–activity relationship studies indicated that prenylation of terphenyllin or candidusin and the tetrahydrobenzene moiety in anthraquinone derivatives may influence their bioactivity.


Introduction
To date, approximately 70,000 species of fungi have been characterized [1]. Among them, about 1500 species of marine-derived fungi were mentioned, mainly from coastal ecosystems [1]. In recent years, the fungal sources of novel metabolites have broadened from saprophytic terrestrial strains to marine habitats and living plants with their endophytes [2]. Specifically, metabolites isolated from species of the genus Aspergillus have continually attracted the interest of pharmacologists due to their broad array of biological activities and their structural diversity. A. tritici, A. campestris, A. taichungensis, and A. candidus, which are members of the Aspergillus section Candidi, are known to be the prolific producers of bioactive secondary metabolites, including terphenyllin, candidusins, and anthraquinones [3].

Results and Discussion
4-methyl-candidusin A (1) was obtained as a colorless amorphous solid. Its molecular formula was established as C21H18O6 by high-resolution electrospray ionization mass spectroscopy  (Table 1). In comparison with the previously reported candidusin A [4], the lack of a phenolic OH unit and the appearance of a methoxyl group in 1 were observed, confirmed by evidence of a 14 amu increase in the molecular weight of 1. In addition, they shared the same substructures of rings B and C, with the main differences located on ring A. In combining the correlations of 1 H-1 H COSY and heteronuclear multiple bond correlation (HMBC) spectra ( Figure 2) with the "no splitting" of H-2 and H-5, we assigned the structure of compound 1 as 4-methyl-candidusin A (1).

Results and Discussion
4-methyl-candidusin A (1) was obtained as a colorless amorphous solid. Its molecular formula was established as C 21 [4], the lack of a phenolic OH unit and the appearance of a methoxyl group in 1 were observed, confirmed by evidence of a 14 amu increase in the molecular weight of 1. In addition, they shared the same substructures of rings B and C, with the main differences located on ring A. In combining the correlations of 1 H-1 H COSY and heteronuclear multiple bond correlation (HMBC) spectra ( Figure 2) with the "no splitting" of H-2 and H-5, we assigned the structure of compound 1 as 4-methyl-candidusin A (1).     3 -7)]. In comparison with the published data on bostrycin [5,6], both the 1 H NMR and 13 C NMR were similar, suggesting that compound 2 was a bostrycin derivative. Analysis of 1D NMR, 1 H-1 H COSY, heteronuclear single quantum correlation (HSQC), and HMBC data revealed the presence of one 1,2-dihydroxy-3-methylbutane unit and one pentasubstituted naphthoquinone moiety. In the HMBC spectrum, correlations of H-1 with C-9 and C-13, and of H-4 with C-10 and C-14, indicated 1,2-dihydroxy-3-methylbutane was connected to the naphthoquinone by linkage of C-1 with C-13 and of C-4 with C-14 ( Figure 2). The phenolic OH was attached to C-10 by HMBC correlations of δ H [12.18 (s, OH-10)] with C-10, C-11, and C-14. The aromatic proton δ H [7.65 (s, H-9)] showed HMBC correlations with C-1, C-8, C-11, and C-14, suggesting C-9 was unsubstituted and the two methoxy groups were attached to C-6 and C-7. Therefore, the planar structure of compound 2 was identified as 3, 9-deoxy-7-methoxybostrycin and named aspetritone A (2).
The relative configuration of 2 was elucidated based on NOESY spectra ( Figure 3). The strong NOESY correlations of H-1 with H-3 and of H-2 with Hax-4 indicated that both H-1 and H-3 faced to the same side of the tetrahydrobenzene ring and H-2 oriented to the opposite side. Therefore, two possible isomers of (1S, 2S, 3R)-2 and (1R, 2R, 3S)-2 were proposed, and their ECD spectra were calculated by time-dependent density functional theory (TD-DFT). The experimental ECD spectrum of 2 was in good agreement with the calculated ECD spectrum of (1S, 2S, 3R)-2 (Figure 4), and the axial-axial coupling constants of 3 JHax-1, H-2 (7.03) and 3 JHax-3, Hax-4 (11.46) indicated a half-chair form of the tetrahydrobenzene ring with all of OH-1, OH-2, and CH3-3 in equatorial positions. In combining the NOESY correlations with the proton coupling constants, the absolute configuration of 2 was established as (1S, 2S, 3R)-3, 9-deoxy-7-methoxybostrycin (2).   Aspetritone B (3) was obtained as a yellow amorphous solid, and its molecular formula was determined as C 17 -7)]. In comparison with the published data of prisconnatanone A [7][8][9], both the 1 H NMR and 13 C NMR were similar, suggesting that compound 2 was a tetrahydroanthraquinone derivative. Analysis of 1D NMR, 1 H-1 H COSY, HSQC, and HMBC data revealed the presence of one 2,3-dihydroxy-3-methylbutane unit and one pentasubstituted naphthoquinone moiety. In HMBC spectra, correlations of H-1 with C-9 and C-14, and H-4 with C-10 and C-13, indicated that 2,3-dihydroxy-3-methylbutane was connected to the naphthoquinone by linkage of C-1 with C-13 and of C-4 with C-14.
The relative configuration of 3 was elucidated based on NOESY spectra ( Figure 3). The strong NOESY correlations of H ax -1 with OH-3 and of H-2 with CH 3 -3 indicated cis-configuration of OH-2 and OH-3.

Fungal Material
Strain SP2-8-1 of A. tritici was isolated from the coral Galaxea fascicularis collected at Port Dickson, Malaysia, and was identified by ITS sequence homology (100% similarity with A. tritici CBS 266.81 with Genbank Accession No. KP987088.1 (max score 972, e value 0.0, query cover 100%)). The fungal strain was inoculated into a 15 mL centrifuge tube containing 3 mL of potato dextrose medium and cultured at 28 • C at 150 rpm for 3 days. Total genomic DNA was extracted as described by Lai et al. [17]. The internal transcribed spacer (ITS) region of rDNA was amplified by PCR using primers ITS1 and a final extension at 72 • C for 7 min. The ITS1-5.8S-ITS2 rDNA sequence of the fungus has been submitted to GenBank with the accession number MF716581. A voucher specimen was deposited at the Third Institute of Oceanography, SOA, China. The working strain was prepared on potato dextrose agar slants and stored at 4 • C.

Antibacterial Assay
Antibacterial activities against MRSA (ATCC 43300, CGMCC 1.12409), V. rotiferianus (MCCC E385), V. vulnificus (MCCC E1758), and V. campbellii (MCCC E333) were tested by continuous dilution in 96-well plates using resazurin as a surrogate indicator. Blue resazurin was reduced by metabolically active bacteria to pink resorufin. A mid-logarithmic-phase tested strain was added at a starting inoculum of 5 × 10 5 CFU/mL to the plate containing tested compound (final concentration ranging from 250 to 0.98 µg/mL in two-fold dilution) plus 10% resazurin solution (6.75 mg/mL in sterile water). The foil covered plate was incubated for 24 h with shaking at 37 • C. After that, by observing the blue-to-pink color change, the MIC value was determined to be the lowest concentration that did not induce the color change [18][19][20].

Cytotoxicity Assay
Hela (cervical cancer cell), Hep G2 (human liver cancer cell), and A549 (adenocarcinomic human alveolar basal epithelial cell) cells were maintained in DMEM, MEM, and F-12K medium respectively, and supplied with 10% FBS, 100 U/mL of penicillin, and 100 mg/mL of streptomycin [21]. Cells were grown in a humidified chamber with 5% CO 2 at 37 • C. For cytotoxicity assays, cells were seeded at a density of 5000 cells per well in 96-well plates, grown at 37 • C for 12 h, and then treated with tested compound at five different concentrations (100 µL medium/well). The cytotoxicity was measured by Cell Counting Kit-8 (CCK-8) (DOJINDO) at 48 h post-treatment, following the manufacturer's instructions.
In brief, 10 µL of CCK-8 solution was added to each well of the 96-well plates. After incubation at 37 • C for 2 h, the absorbance at 450 nm was measured using a SpectraMAX M5 microplate reader. Wells with only culture medium and CCK-8 solution were used to determine the background, and cells treated with DMSO were included as the negative controls [21].

ECD Calculation
Conformational analysis was initially performed using Confab [23] with the MMFF94 force field for all configurations. Room-temperature equilibrium populations were calculated according to the Boltzmann distribution law Equation (1). The conformers with Boltzmann-populations of over 1% were chosen for ECD calculations. The energies and populations of all dominative conformers were provided in Table S1.
N i is the number of conformer i with energy E i and degeneracy g i at temperature T, and k B is the Boltzmann constant. The theoretical calculation was carried out using Gaussian 09. First, the chosen conformer was optimized at PM6 using the semi-empirical theory method, and then optimized at B3LYP/6-311G** in methanol using the conductor-like polarizable continuum model (CPCM) ( Table S2). The theoretical calculation of ECD was conducted in methanol using TD-DFT at the same theory level. Rotatory strengths for a total of 50 excited states were calculated. The ECD spectrum is simulated in SpecDis [24] by overlapping Gaussian functions for each transition.