Nine New and Five Known Polyketides Derived from a Deep Sea-Sourced Aspergillus sp. 16-02-1

Nine new C9 polyketides, named aspiketolactonol (1), aspilactonols A–F (2–7), aspyronol (9) and epiaspinonediol (11), were isolated together with five known polyketides, (S)-2-(2′-hydroxyethyl)-4-methyl-γ-butyrolactone (8), dihydroaspyrone (10), aspinotriol A (12), aspinotriol B (13) and chaetoquadrin F (14), from the secondary metabolites of an Aspergillus sp. 16-02-1 that was isolated from a deep-sea sediment sample. Structures of the new compounds, including their absolute configurations, were determined by spectroscopic methods, especially the 2D NMR, circular dichroism (CD), Mo2-induced CD and Mosher’s 1H NMR analyses. Compound 8 was isolated from natural sources for the first time, and the possible biosynthetic pathways for 1–14 were also proposed and discussed. Compounds 1–14 inhibited human cancer cell lines, K562, HL-60, HeLa and BGC-823, to varying extents.

During the ongoing search for new bioactive natural products from marine-sourced fungi, we have evaluated cytotoxicity and antifungal activities for 16 fungal strains from deep-sea habitats, and found that an Aspergillus sp. 16-02-1 produced metabolites with both cytotoxic and antifungal activities. The strain Aspergillus sp. 16-02-1 was isolated from a deep-sea sediment sample that was collected at a Lau Basin hydrothermal vent (depth 2255 m, temperature 114 °C ) in southwest Pacific. We previously reported 8 known metabolites from this strain by a liquid fermentation [17]. In a continuation, we re-fermented this strain using solid-substrate fermentation medium and obtained nine new (1-7, 9 and 11) and five known (8,10, and 12-14) polyketides shown in Figure 1. We report herein the isolation, structure elucidation, and cytotoxicity evaluation of these compounds in detail.

Structure Determination of New Compounds
Aspiketolactonol (1) 2938,2906,1422 and 1360 cm −1 ), α,β-unsaturated γ-lactone (1752 and 1656 cm −1 ) [ 21,22] and keto carbonyl (1722 cm −1 ) groups. The olefinic proton and carbon signals at the lower field (δ H 7.43 and δ C 149.4) [22] of 1 H and 13 C NMR spectra (Tables 1 and 2) and the ester carbonyl carbon signal at δ C 173.5 in the 13 C NMR spectrum (Table 2) supported the presence of the α,β-unsaturated γ-lactone moiety. The carbonyl carbon signal at δ C 203.6 in the 13 C NMR spectrum further supported the presence of one keto carbonyl in 1. Interpretation of the 1 H-1 H COSY, HMQC and HMBC data (Table S1 in the Supplementary Information) established the planar structure. The structural part related to the 1 H spin system, C-7-C-4(via quaternary sp 2 C-3)-C-8, was deduced from the 1 H-1 H COSY and HMQC data, including the allylic coupling between H-4/H 2 -8, which was confirmed by the HMBC correlations of H-4, H-5 and H 2 -8 with C-3. The acetyl group that consisted of the C-9 keto carbonyl and the C-10 methyl was linked to C-8 by the HMBCs on H 2 -8/C-9, H 3 -10/C-8 and H 3 -10/C-9. The ester carbonyl carbon (C-2, δ C 173.5) was linked to C-3 by the HMBC correlations of H-4 and H 2 -8 with C-2. Then, C-2 was further linked to C-5 by an ester linkage to form the α,β-unsaturated γ-lactone ring according to the IR absorption at 1752 cm −1 .
The absolute configuration at C-5 was assigned to be S by the positive π→π* Cotton effect at 225 nm and the negative n→π* Cotton effect around 239 nm ( Figure 2) [26,27]. The absolute configuration at C-9 was assigned by the modified Mosher's method [30,31].  Both they showed end UV absorptions, and similar to 1 and 2, their IR spectra revealed the presence of α,β-unsaturated γ-lactone moieties (around 1740 and 1652 cm −1 ) [21,22]. This was supported by the lower field olefinic proton and carbon signals (around δ H 7.55, H-4; around δ C 150.4, C-4) [22] in the 1 H and 13 C NMR spectra (Tables 1 and 2) and the ester carbonyl (δ C 174.5, C-2) and the sp 2 carbon (around δ C 137, C-3) signals in the 13 C NMR spectra ( Table 2). Their IR spectra also indicated the presence of OH and CH 3 /CH 2 groups, and the strong OH signals around 3384 cm −1 further revealed more OH groups in 3 and 4 than in 1 and 2 (the IR spectra in the Supplementary Information). Interpretation of the 1 H-1 H COSY and HMQC data (Tables S3 and S4 in (Tables S3 and S4 in the Supplementary Information). The C-2 carbonyl carbons in 3 and 4 were linked to C-5 and C-3 by the HMBC correlations between H-5/C-2, H-4/C-2 and H-8/C-2 to form the α,β-unsaturated γ-lactone rings with the ester linkage between C-5 and C-2.
The absolute configuration at C-5 in 3 and 4 was assigned both to be R by the positive π→π* Cotton effect around 226 nm in the CD of 3 and 4 ( Figure 2) [26,27]. Because the coupling of H-5 and H-6 (4.6 Hz for 3 and 5.0 Hz for 4) indicated the erythro relative configuration of 5,6-diols in 3 and 4 [23][24][25], the absolute configuration at C-6 in 3 and 4 was also assigned both to be S. Thus, 3 and 4 should be stereoisomers at the vicinal diol methine carbons C-8/C-9. There are many reports recorded that the coupling of vicinal diol methine protons is generally lager than 6 Hz in threo isomers but smaller than 5 Hz in erythro isomers of the relevant compounds with a vicinal diol unit similar to the 8,9-diols in 3 and 4 [32][33][34][35][36]. The coupling of H-8 and H-9 (4.6 Hz for 3 and 4.9 Hz for 4) indicated the erythro relative configuration of the 8,9-diols in 3 and 4 [32][33][34][35][36]. The absolute configuration of the erythro- 8,9-diols in 3 and 4 was determined by the dimolybdenum induced CD (ICD) analysis. In the ICDs by the Snatzke's method using dimolybdenum tetraacetate (Mo 2 (OAc) 4 ) in DMSO [37,38], the Mo 2 -complex of 3 gave negative CD bands II (near 400 nm) and IV (around 329 nm), while the Mo 2 -comlex of 4 gave the positive bands II and IV (Figure 4). By the Snatzke's helicity rule, the sign of O-C-C-O torsional angle in the favored conformation of the chiral Mo 2 -complexes determines the sign of the bands II and IV [37,38]. We have demonstrated that in the erythro-diols closely resembled 3 and 4, the conformation with an antiperiplanar orientation of the OH and methyl groups, O-C-C-CH 3 , is favored conformation of the Mo 2 -complexes [39], as shown for 3 and 4 in Figure 4. Therefore, the absolute configuration at C-8 and C-9 in 3 and 4 could be assigned to be 8S,9R for 3 and 8R,9S for 4 on the basis of their band II and IV signs (Figure 4), respectively.  (Tables 1 and 2). These NMR data indicated that 5 was an O-methylated. Analyses of the 1 H-1 H COSY, HMQC and HMBC spectra ( Table S5 in the Supplementary Information) established its planar structure. The α,β-unsaturated γ-lactone ring was confirmed by the HMBC correlations between H-5/C-2, H-4/C-2 and H-8/C-2. The O-methyl group was located at C-8 by the HMBC correlations of the O-methyl protons with C-8. The absolute configuration at C-5 in 5 was assigned to be R by the positive π→π* Cotton effect at 231.5 nm in the CD spectrum ( Figure 2) [ 26,27]. The coupling of H-5/H-6 (4.5 Hz) indicated the erythro relative configuration of 5,6-diol in 5 [23][24][25], and thus the absolute configuration at C-6 was assigned to be S. The R absolute configuration of C-9 was determined by the modified Mosher's method [30,31] on the basis of the Δδ (δ S -δ R ) values from the (S)-and (R)-MTRA esters (Figure 3). Since the coupling of H-8/H-9 (4.7 Hz) indicated the erythro relative configuration of 8,9- showed the end absorption, and the IR spectrum showed absorptions due to the OH, CH 3 /CH 2 and α,β-unsaturated γ-lactone groups (Experimental Section). The 1 H and 13 C NMR spectra of 6/7 in CD 3 OD showed signals similar to 3-5, but they were characterized by the appearance of methylene signals instead of the signals from an oxygenated methine in 3-5 and an O-methyl group in 5 (Tables 1   -1 II IV III and 2). These NMR data indicated the same skeletal structures in 3-7. The appearance of the proton H-4 and the carbon signals except for the C-10 signal as pairs in an approximate 1:1 ratio indicated that 6/7 was a 1:1 mixture of stereoisomers. The planar structure of 6/7 was deduced by the 1 H-1 H COSY, HMQC and HMBC data ( Table S6 in the Supplementary Information), coupled with the IR absorptions at 1748 and 1651 cm −1 from an α,β-unsaturated γ-lactone ring [21,22]. The absolute configuration at C-5 in 6/7 was assigned both to be R by the strong positive π→π* Cotton effect at 228.5 nm in the CD spectrum ( Figure 2) [ 26,27]. Since the coupling of H-5/H-6 (4.4 Hz) indicated the erythro relative configuration of 5,6-diols in 6/7 [23][24][25], the absolute configuration at C-6 was assigned to be S for both 6/7. Thus, 6/7 was a 1:1 mixture of epimers at C-9. Although a Doctor's Thesis has recorded the same planar structure of 6/7, its stereochemistry was not elucidated [40]. We therefore named 6/7 as aspilactonols E/F as new compounds.
Aspyronol (9) Table 3) resembled those of dihydroaspyrone (10) except additional signals from a methoxy and an oxygenated methine groups were detected instead of the methylene signals in 10. There were also slight changes in several 1 H and 13 C signals. These NMR data suggested that 9 was a methoxylated derivative of 10, and this was confirmed by analysis of the 1 H-1 H COSY, HMQC and HMBC spectra (Table S7 in the Supplementary Information) to complete the planar structure. The carbon chain related to the proton spin system, C-7-C-4(via quaternary sp 2 C-3)-C-8-C-10, was deduced by interpretation of the 1 H-1 H COSY and HMQC data ( Table S7 in the Supplementary  Information). The allylic coupling between H-4 and H-8 suggested the connectivity of C-4 and C-8 via a quaternary sp 2 carbon C-3, and C-3 was assigned by the HMBC correlations between H-8/C-3 and H-9/C-3. The OCH 3 group was located at C-8 by the HMBCs of the methoxy protons with the carbon C-8. The carbonyl carbon C-2 was linked to C-3 by the HMBCs on H-4/C-2 and H-8/C-2. The ester linkage of the C-2 carbonyl was then linked to C-6 to form a δ-lactone ring by the coupling of H-5/H-6 (9.4 Hz), which requires C-5/C-6 fixed in a six-membered ring with trans orientated H-5/H-6. This was supported by the chemical shift of C-6, δ C 80.0 in 9 and δ C 79.6 in 10. The R absolute configuration at C-9 was determined by the modified Mosher's method [30,31] on the basis of the Δδ (δ S -δ R ) values from the (S)-and (R)-MTRA esters of 9 (Figure 3), and the absolute configuration at C-8 was assigned to be S because the coupling of H-8/H-9 (4.6 Hz) indicated the erythro relative configuration of 8,9-diol in 9 [32][33][34][35][36]. The absolute configuration at C-5 and C-6 in 9 was determined to be 5S,6R, the same as 10, according to the negative signs of the Cotton effects around 260 nm both from the chiral α,β-unsaturated δ-lactone units in 9 and 10 ( Figure 2) [28]. This was also supported by the co-generation of 9 and 10 by the same Aspergillus sp. 16-02-1 strain from a biogenetic consideration.  [19]. In contrast, 11 showed the opposite optical rotation. Thus, the epiaspinonediol (11) was determined to be the epimer of aspinonediol at C-7 with the 7R absolute configuration.

Inhibitory Effects of 1-14 on Several Human Cancer Cell Lines
Antitumor activities of 1-14 were tested by the MTT method using the human cancer K562, HL-60, HeLa

Discussions
Chemical investigation of a deep sea-sourced Aspergillus sp. 16-02-1 has resulted in the elucidation of 14 secondary metabolites 1-14, including nine new (1-7, 9 and 11) and five known (8, 10, and 12-14) compounds, shown in Figure 1. Although compound 8 has been chemically prepared [18], it is the first time to report 8 from natural sources in present study. Structures of the new compounds, including their absolute configurations, were determined by extensive spectroscopic methods, especially the 2D NMR, CD, ICD and Mosher's 1 H NMR analyses. The determination of the absolute configuration of α,β-unsaturated γ-lactone ring in 1-7 mainly relied on the CD data. In most cases, a chiral α,β-unsaturated γ-lactone ring gave both π→π* and n→π* Cotton effects with the opposite sign [26,27] in 200-235 nm and 235-270 nm regions, respectively [26][27][28][29]. However, usually the n→π* Cotton effect is weak and sometimes could not be observed [26,27] or even appeared with the same sign of the π→π* Cotton effect [26]. The same is true of the case of 1-8. As shown in Figure 2, 2 and 8 gave both opposite n→π* and π→π* Cotton effects in similar magnitude and 4 also showed a weak n→π* Cotton effect opposite to the π→π* transition. However, the others did not give opposite n→π* Cotton effect and rather they showed a weak CD curve with the same sign of the π→π* Cotton effect in the n→π* transition region (Figure 2). Since the easy influence of the n→π* Cotton effect by the external asymmetry [26] and the decisive role of the π→π* Cotton effect in absolute configuration assignment has been known [26,27], the absolute configurations of the α,β-unsaturated γ-lactone rings in 1-7 could be assigned by the sign of their π→π* Cotton effects shown in Figure 2.
Relating to the above mentioned metabolites, plausible biosynthetic pathways for 1-7 and 9-13 are proposed in Scheme 1. Reduction of either one of the two epoxides in I-5, the precursor of aspinonene [47], coupled with hydration at either site of the other epoxide ring followed by dehydration, would give I-6 and compounds 12 and 13, which further underwent oxidation at C-2 to produce compound 11 and aspinonediol, the epimer of 11 at C-7 (Scheme 1). Aspyrone and asperlactone are proposed to be intermediates for 9-10 and 1/3-7. Reduction of the epoxide in aspyrone would give 10, and hydration at C-8 of the epoxide, followed by methylation, would afford 9 (Scheme 1). Reduction or hydration of the epoxide in asperlactone and further modification of the products by methylation, reduction or dehydration followed by keto-enol tautomerization, and oxidation/reduction reactions would produce 1 and 3-7, as shown in Scheme 1. The γ-lactone 2 seems likely to be produced from the intermediate I-7. Reduction of the epoxides in I-7 coupled with double bond rearrangement, followed by lactonization of the product I-8, would give 2 (Scheme 1).

Scheme 1. Plausible biosynthetic pathways of 1-7 and 9-13.
In view of the structural features of 1, 3-4 and 6-7, one of these compounds may probably form from some or other during extraction with EtOAc at the slightly acidic conditions by the acid-catalyzed chemical reactions artificially. In order to confirm whether this occurred in truth, each 0.3 mg of the crude compound samples was dissolved in 0.2 mL water-saturated EtOAc in a 0.5 mL Eppendorf tube, capped the tube and kept at room temperature for 6 days, and then further treated at 50 °C for 16 h. These treatment conditions simulated the extraction conditions (whole extraction was achieved within 4 days with a total of 12 h evaporating times at the temperature lower than 40 °C ). Then, the aqueous EtOAc was removed by blowing inside of the tube with nitrogen gas to dryness. The residue was dissolved in MeOH and then subjected to HPLC analysis. No any one of them was detected in other compound samples by the HPLC analysis ( Figure S1 in the Supplementary Information), confirming that none of these compounds are artificial product formed from the others in the extraction conditions.
Differing from 1-7/9-13, they are all pentaketide derivatives as shown in Scheme 1, compound 8 seems likely to be a tetraketide derivative. A plausible pathway for the 8 biosynthesis is proposed in  In present MTT assay, 9 and 11 showed stronger inhibitory effect than the others on human cancer K562 and/or HL-60 cell lines. Compound 9 inhibited the HL-60 cells with the IR% value of 67.2% at 100 μg/mL, while 11 inhibited the K562 and HL-60 cells with the IR% values of 79.7% and 72.5% at 100 μg/mL, respectively. Both compounds also inhibited to some extents the K562 (9: an IR% of 27.9% at 100 μg/mL) and the BGC-823 (11: an IR% of 21.8% at 100 μg/mL) cells. These data suggested that the new compounds 9 and 11 showed somewhat selective inhibitory effect on the HL-60 cells and the K562 and HL-60 cells, respectively, although the inhibitory effect of both compounds by the IC 50 of 9 (52.1 μg/mL or 241.2 μM on HL-60) and 11 (44.3 μg/mL or 260.6 μM on K562 and 32.8 μg/mL or 192.9 μM on HL-60) is not so exciting. On the other hand, except three other new compounds 3-4 weakly inhibited the K562 and HL-60 cells with the IR% values larger than 20% at 100 μg/mL, all the others showed very weak effect on some of the tested four cancer cell lines with the IR% values lower than 20% at 100 μg/mL, as shown in Section 2.3. Aspyrone and asperlactone have been identified for some biological activities, including the remarkable insect growth regulator activity of asperlactone against Tribolium castaneum and Nezara viridula [48], the nematicidal activity of aspyrone on Pratylenchus penetrans [49], and the antifungal and antibacterial activities of both compounds on several fungal and bacterial strains [50]. Antibacterial activities were also reported for several chlorine containing derivatives from aspyrone and asperlactone by opening of the epoxy ring therein [22,51]. Although aspinonene and mono (S)-and (R)-MTRA esters of dihydroaspyrone (10) at C-9 have been reported to show a low cytotoxicity on mouse lymphocytic leukemia cells [19], there are few of reports recorded the antitumor activities for the compounds structurally closely related to 1-14.
The present bioassay results, as well as the report in the literature [19], seem to suggest that the branched C 9 polyketides structurally related to 1-14 were likely worthy for further extended studies to obtain antitumor agents with more strong activity and higher selectivity.

Producing Strain, Fermentation and Extraction
The producing fungal strain 16-02-1 was isolated from a deep-sea sediment, DY19-4-TVG11, which was collected at a Lau Basin hydrothermal vent (depth 2255 m, temperature 114 °C ) in the southwest Pacific (south latitude 20.9280°, west longitude 176.2401°) during round-the-world ocean research of Dayangyihao in May 2007. This strain was identified as a species of the genus Aspergillus by sequence analysis of the ITS region of the rDNA and by morphological characteristics.
For fermentation of the 16-02-1 strain, a spore suspension was prepared using fresh spores by the method that we have previously reported [39,52] at first. Then, the fermentation was carried out in sixteen of 500 mL Erlenmeyer flasks, each containing 80 g of rice. Distilled water (120 mL) was added to each flask, and the contents were soaked overnight before autoclaving at 121 °C for 30 min. After cooling to room temperature, each flask was inoculated with 200 μL of the 16-02-1 spore suspension and incubated at 28 °C for 36 days. The fermented material was extracted repeatedly with EtOAc (3 × 6 L), and the organic solvent was evaporated under reduced pressure to obtain an EtOAc extract (21.9 g). The EtOAc extract inhibited K562 cells with an IR% of 76.8% at 100 μg/mL.