Six New Antimicrobial Metabolites from the Deep-Sea Sediment-Derived Fungus Aspergillus fumigatus SD-406

Six new metabolites, including a pair of inseparable mixtures of secofumitremorgins A (1a) and B (1b), which differed in the configuration of the nitrogen atom, 29-hydroxyfumiquinazoline C (6), 10R-15-methylpseurotin A (7), 1,4,23-trihydroxy-hopane-22,30-diol (10), and sphingofungin I (11), together with six known compounds (2–5 and 8–9), were isolated and identified from the deep-sea sediment-derived fungus Aspergillus fumigatus SD-406. Their structures were determined by detailed spectroscopic analysis of NMR and MS data, chiral HPLC analysis of the acidic hydrolysate, X-ray crystallographic analysis, J-based configuration analysis, and quantum chemical calculations of ECD, OR, and NMR (with DP4+ probability analysis). Among the compounds, 1a/1b represent a pair of novel scaffolds derived from indole diketopiperazine by cleavage of the amide bond following aromatization to give a pyridine ring. Compounds 1, 4, 6, 7, 10 and 11 showed inhibitory activities against pathogenic bacteria and plant pathogenic fungus, with MIC values ranging from 4 to 64 μg/mL.


Introduction
Deep-sea sediment has proven to be a treasure trove for structurally unique and biologically active secondary metabolites [1]. In the extreme environment of the deep-sea, microorganisms have gradually developed unique metabolic mechanisms out of adaption, thus possessing the great potential to produce natural products with significant biological properties, such as antimicrobial [2,3], cytotoxic [4], and antiviral [5] activities.

Structure Elucidation
The culture broth of fungus Aspergillus fumigatus SD-406 was extracted with EtOAc, and the crude extract was subjected to multiple chromatographic methods (a combination of column chromatography on Silica gel, Sephadex LH-20, and Lobar LiChroprep RP-18) to yield subfractions, which were further purified by preparative TLC and semipreparative HPLC to give compounds 1-11.
Compounds 1a/1b were isolated as a mixture present in a ratio of 1:0.7 (major:minor). Attempts to separate two isomers by various types of chiral columns along with different elution ratios, unfortunately failed. The mixture (1a/1b) was found to have the same molecular formula as C 23 (Table 1), revealing that their planar structures both possessed 4 methyls (including 2 methoxyls), 3 aliphatic methylenes, 6 methines (including 5 sp 2 hybridized and 1 connected to heteroatoms), and 10 quaternary carbons (including 2 carbonyls). Detailed analysis and comparison of the 1D and 2D NMR data ( Figures S4-S7 in the supplementary material) indicated that the planar structures of 1a/1b were similar to that of the known compound fumitremorgin C (2) [6]. However, signals for the methine groups at C-3 (δ H/C 5.98/51.1) and C-12 (δ H/C 4.16/56.9), methylene group at C-13 (δ H/C 3.51/22.0), and amino carbonyl at C-5 (δ C 165.9) in 2 were absent. Instead, signals for sp 2 hybridized quaternary carbons at δ C, major 138.5/δ C, minor 138.2 and δ C, major 141.9/δ C, minor 141.8, for sp 2 hybridized methines at δ H/C, major 8.33/113.3/δ H/C, minor 8.43/113.4, and for methyl ester groups at δ C, major 172.6/δ C, minor 172.9 and δ H/C, major 3.67/51.6/δ H/C, minor 3.46/51.4 were observed in the NMR spectra of 1a/1b (Table 1), respectively. The above observation suggested that compounds 1a and 1b might be the derivatives of cleavage at the amide bond between N-4 and C-5 of fumitremorgin C (2). Meanwhile, aromatization occurred to generate a pyridine ring. This deduction was further verified by the key HMBC correlations from H-13 to C-2, C-11 and C-15, from H-26 and H-7 to C-5, and from H-21 to C-3 ( Figure 2).
The main differences between the two sets of NMR data were the chemical shifts around proline and isopentenyl moieties ( Figure 3). Compared to the minor set, obvious higher chemical shifts for C-7, C-8, C-9, C-11, C-22, and C-24, and lower chemical shifts for C-6 and C-21 in the major set were observed ( Figure 3). Based on the above deviation, a distinction in configuration of nitrogen atom N-10 between 1a and 1b was considered, and two candidate structures isomers 1 and 2 were proposed ( Figure 4). The C-N bond in the amino carbonyl cannot freely rotate due to the delocalization of the nitrogen atom's lone electron pairs, thus resulting in the different orientation of the methyl ester group in isomers 1/2, which explained well the differences in chemical shifts of proline and isopentenyl moieties. Besides, in the major set, the influence of π-systems of the aromatic rings led to obvious lower chemical shifts of H-6 and H-7α/β, whereas H-8α/β and H-9α/β remained unaffected ( Figure 3). In the minor set, on the other hand, the chemical shifts for H-8α/β and H-9α/β were pushed to lower values by the aromatic rings, while H-6 and H-7α/β were unaffected ( Figure 3). Thus, the major NMR data were assigned to isomer 1 (1a) and the minor NMR data were assigned to isomer 2 (1b).    To further confirm the assignment, a comparison of the observed NMR data with those of computed values for two possible isomers using DFT-NMR calculations with DP4+ probability analysis (see excel files in the supplementary material) was carried out [12]. As a result, the experimental NMR data of the major 1 H and 13 C NMR resonances corresponded to the computed NMR data for isomer 1 (100% probability, Table S5 in the supplementary material), while the calculated chemical shifts for isomer 2 were consistent with the minor 1 H and 13 C NMR resonances (100% probability, Table S6 in the supplementary material).
To determine the configuration of proline, a chiral HPLC analysis of the mixture's acidic hydrolysate was carried out. The result of the HPLC analysis showed that the retention time of the acidic hydrolysate of mixture 1 was identical with that of L -Pro ( Figure S1 in the supplementary material), indicating an L -proline, accordingly the 6S configuration of 1a/1b. Thus, the structures of 1a and 1b were assigned as shown in Figure 1 and named secofumitremorgin A and secofumitremorgin B, respectively.
Compound 6 was obtained as a white solid. Its molecular formula was deduced as C 24 Figure S17 in the supplementary material). The 1 H-and 13 C-NMR spectra ( Table 2, and Figures S11 and S12 in the supplementary material) displayed 1 methyl, 2 sp 3 hybrid methylenes with 1 oxygenated, 8 aromatic methines and 3 sp 3 hybrid methines connected to heteroatoms, and 10 quaternary carbons with 3 amino carbonyls. Detailed analysis of the 1D and 2D NMR data (Figures S13-S15 in the supplementary material) revealed that 6 showed close similarity to fumiquinazoline C (5) [9], except that the doublet methyl CH 3 -29 (δ H 1.07/δ C 18.8) of 5 was replaced by an oxygenated methylene (δ H 3.41 H a , 3.21 H b /δ C 61.8) in 6. This was further supported by 1 H-1 H COSY correlations for the spin system of 19-NH/H-20/H-21/21-OH, to propose the structure of 6 ( Figure 2). The relative configuration of 6 was deduced from analysis of NOESY data ( Figure 5, and Figure S16 in the supplementary material). The NOESY correlations from H 3 -16 to H-14 indicated the cofacial orientation of them. Moreover, a NOE cross-peak from H-18 to H-20 suggested the same spatial orientation of them, while NOESY correlation from H-29 to H-15 placed these groups on the opposite face. The above observation also revealed the β-orientation of the ether bridge between C-3 and C-17 [13]. Thus, the relative configuration of 6 was identical to that of 5. The absolute configuration of 6 was subsequently assigned as 3R,14R,17S,18S,20S, based on the same sign of the optical rotation to that of 5 [13]. To further verify the absolute configuration of 6, time-dependent, density functional (TDDFT)-ECD calculations at the BH&HLYP/TZVP level were performed. The calculated ECD spectrum for the (3R,14R,17S,18S,20S)-6 matched well with that of the experimental curve ( Figure S18 in the supplementary material), allowing the establishment of the absolute configuration of 6 as 3R,14R,17S,18S,20S ( Figure 6). The trivial name 29hydroxyfumiquinazoline C was assigned to 6. Compound 7 was isolated as a yellowish solid. The molecular formula was determined as C 23 8 , 446.1809), which was 14 amu more than that of the previously-reported pseurotin A (9) [11], and accounted for 11 degrees of unsaturation ( Figure S25 in the supplementary material). The 1 H-, 13 C-, and HSQC NMR spectra ( Table 2, and Figures S19, S20, S22 and S23 in the supplementary material) revealed signals for three methyls (including one methoxyl), two aliphatic methylenes, ten methines (including three oxygenated and seven sp 2 hybridized), two oxygenated, sp 3 hybridized quaternary carbons, three sp 2 hybridized quaternary carbons, two keto carbonyls (δ C 196.7 and δ C 196.3), and one amino carbonyl (δ C 166.5). Detailed comparison of its NMR data with those of pseurotin A (9) showed close similarity in the planar structure, except for the additional appearance of an aliphatic methylene C-15 (δ H 1.30/δ C 22.2). Moreover, the significantly higher chemical shift for C-14 (δ C 29.2) was also observed in 7. The above observation indicated that 7 was the 15-methylation derivative of 9. 1  To clarify the relative configurations of the chiral centers C-10 and C-11, J-based configuration analysis [14] was performed using 3 J H-H , 3 J H-C , and 2 J H-C coupling constants obtained from the 1 H NMR and J-HMBC spectra ( Figure S24 in the supplementary material). The medium 3 J H10-H11 (5.6 Hz), the small 3 J H10-C12 (1.5 Hz), the small 3 J H11-C3 (2.0 Hz), and the medium 2 J H10-C11 (4.6 Hz) matched with a pair of gauche/anti equilibrating rotamers in Figure 7, indicating the syn/syn orientation of H-10 and H-11. The absolute configurations of C-5 and C-8 were assigned based on electronic circular dichroism (ECD) spectrum of 7 ( Figure S26 in the supplementary material), according to the previously reported rule [15]. As described in the previous reference, a negative Cotton effect at around 280 nm in the ECD spectrum ( Figure S26 in the supplementary material) revealed S configuration of C-8, while the negative Cotton effect at around 230 nm and positive Cotton effect at 250 nm suggested the 5S configuration. The cis configurations of 8-OCH 3 and 9-OH were deduced from the large coupling constant (J = 9.0 Hz) between H-9 and 9-OH [16,17]. Therefore, the absolute configuration of C-9 was assigned as 9R. Subsequently, NMR calculations with DP4+ probability analysis (see the excel files in the supplementary material) were carried out to correlate the stereochemical relationship between C-5/C-8/C-9 and C-10/C-11. The experimental 1 H and 13 C NMR data of 7 were compared with the calculated 1 H and 13 C NMR data of 7a and 7b (two possible isomers of 7, Figure 8) and matched well with those calculated for the isomer 7a (5S,8S,9R,10R,11S) with a DP4+ probability of 100% (  Figure S34 in the supplementary material). The 1D NMR spectra (Table 3, and Figures S28 and S29 in the supplementary material) suggested 10 was a pentacyclic triterpenoid containing six singlet methyls, eleven methylenes with two oxygenated, seven methines with two oxygenated, and six quaternary carbons with one oxygenated. Combined with 2D NMR ( Figures S30-S32 in the supplementary material), 10 was suggested to be similar to the previously reported hopane-22,30-diol [19]. However, the signals for two methylenes at C-3 (δ C 42.2) and C-12 (δ C 24.2), and one methyl at C-23 (δ C 33.4) present in the NMR spectra of hopane-22,30-diol were not detected in those of 10, while resonances for two oxygenated methines at C-3 (δ H 3.16/δ C 78.6) and C-12 (δ H 3.67/δ C 67.9), and one oxygenated methylene at C-23 (δ H 3.25, 3.80/δ C 62.8) were observed in the NMR of 10. These data indicated that C-3, C-12, and C-23 in hopane-22,30-diol were all substituted by hydroxy in 10. This deduction was supported by key 1 H-1 H COSY correlations from 3-OH to H-3, from 12-OH to H-12, and from 23-OH to H-23 ( Figure 2). The relative configuration of 10 was partially assigned by NOESY spectrum ( Figure 5, and Figure S33 in the supplementary material). NOE cross-peaks from H-5 to H-9 and from H-9 to H-27 indicated the cofacial orientation of these groups. Further, NOE correlations from H-23 to H-25, from H-13 to H-26, and from H-11α to H-25 and H-26 placed them on the same side. However, relative configurations of the rest of the chiral centers could not be assigned through analysis of NMR data.
After many attempts, the single crystal of 10 suitable for X-ray diffraction was obtained by slowly crystallizing in solvent MeOH at −4 • C (Figure 9), which not only confirmed the planar structure but also determined its relative configurations. Since the Flack parameter [−0.04(5)] was negative, calculation of optical rotation (OR) was performed at three different levels including BH&HLYP/TZVP, CAM-B3LYP/TZVP, and PBE0/TZVP. The calculated OR values (Table S1 in Figure S40 in the supplementary material. The 1D NMR (Table 3, and Figures S35 and S36 in the supplementary material) and HSQC spectra ( Figure S38 in the supplementary material) displayed signals for a triplet methyl, 11 aliphatic methylenes, 3 oxygenated methines, 3 olefinic methines, 3 quaternary carbons with 2 carbonyls, and an amino group. The lipid side chain was deduced from the highly overlapped peak at δ H 1.22-1.27, with an integral of 20 protons in the 1 H-NMR spectrum. These NMR features were similar to those of sphingofungin H, which was isolated from Aspergillus penicilliodes Speg [20], except that the signals of an oxygenated methine C-3 (δ H 4.57/δ C 70.1) and a nitrogen-bearing methine C-2 (δ H 4.70/δ C 55.5) in sphingofungin H were replaced by an olefinic methine C-3 (δ H 7.32/δ C 127.4) and an olefinic quaternary carbon C-2 (δ C 126.7) in 11. These observations indicated that 11 was a dehydrated derivative of sphingofungin H at C-3 and C-2 with an additional double bond. This deduction was supported by key 1 H-1 H COSY correlation from H-3 to H-4 ( Figure S37 in the supplementary material) and HMBC correlation from H-3 to C-1 ( Figure S39 in the supplementary material).  The absolute configurations of C-4 and C-5 in 11 were established as 4S,5S based on the ECD spectrum ( Figure S41 in the supplementary material), which showed a negative Cotton effect at~210 nm and a positive Cotton effect at~240 nm, which is similar to those of the previously reported acetyl derivative of malondungin [21]. The trivial name sphingofungin I was assigned to 11.

General Experimental Procedures
Detailed information for apparatus, reagents, solvents, and materials are same as described in our previous publication [3].

Fungal Material
The fungus Aspergillus fumigatus SD-406 was isolated from the deep-sea sediment of the East China Sea (121 • 20.2 E, 26 • 45.5 N), collected in September 2017. The fungal strain was identified as Aspergillus fumigatus according to the ITS (internal transcript spacer) region sequence, which is the same (100%) as that of Aspergillus fumigatus (accession No. MT635279). The sequence data of SD-406 have been deposited in GenBank with the accession No. OL662987. The strain is preserved at the Key Laboratory of Experimental Marine Biology, Institute of Oceanology, Chinese Academy of Sciences (IOCAS).

Fermentation, Extraction and Isolation
The fungal strain Aspergillus fumigatus SD-406 was cultivated on potato dextrose agar medium at 28 • C for 7 days, which was then transferred into 100 × 1 L Erlenmeyer flasks with rice solid medium (each flask containing 70 g rice, peptone from animal 0.3 g, yeast extract 0.5 g, corn steep liquor 0.2 g, monosodium glutamate 0.1 g and naturally sourced seawater) and incubated at room temperature for 30 days. Then, the solid fermented substrate was extracted three times with EtOAc and the combined extracts were concentrated under reduced pressure to give a dark brown crude extract (76 g).

Acidic Hydrolysis of Compound 1
Compound 1 (1 mg) was dissolved in 10 mL of 6 N HCl and heated in a sealed tube at 110 • C for 24 h. The solutions were then evaporated to dryness under reduced pressure. Each sample, including the standard amino acids L-Pro and D-Pro, were dissolved in 1 mL of eluting solvent (2 mM CuSO 4 ·5H 2 O in 100 mL of H 2 O). Chiral HPLC analysis, both alone and by co-injection with standards, was carried out using a Phenomenex-Chirex-3126 column (150 mm × 4.60 mm, 5 µm; flow rate 1.0 mL/min at 25 • C; detection at 254 nm).

X-ray Crystallographic Analysis of Compound 10
Crystallographic data were collected on an Agilent Xcalibur Eos Gemini CCD plate diffractometer utilizing graphite-monochromatic Cu-Kα radiation (λ = 1.54178 Å) at 293 (2) K [22]. The data were corrected for absorption using the program SADABS [23]. The structures were solved by direct methods with the SHELXTL software package [24]. All nonhydrogen atoms were refined anisotropically. The H atoms connected to C atoms were calculated theoretically, and those to O atoms were assigned by difference Fourier maps [25]. The structures were optimized by full-matrix, least-squares techniques.
Crystal data for compound 10:

ECD Calculation of Compound 6 and OR Calculation of Compound 10
Conformational searches were carried out via molecular mechanics with the MM+ method in HyperChem 8.0 software, and the geometries were optimized at the gas-phase B3LYP/6-31G(d) level in Gaussian09 software (Version D.01; Gaussian, Inc.: Wallingford, CT, USA) [26] to afford the energy-minimized conformers. Then, the optimized conformers were subjected to the calculations of ECD spectra using the TD-DFT at BH&HLYP/TZVP levels, and solvent effects of the MeOH solution were evaluated at the same DFT level using the SCRF/PCM method.
Optical rotations of the optimized conformers were calculated using the TDDFT method at BH&HLYP/TZVP, CAM-B3LYP/TZVP and PBE0/TZVP levels in methanol (λ = 589.4 nm). The calculated optical rotations were later obtained according to the Boltzmann weighting of each conformer.

Computational NMR Chemical Shift and DP4+ Analyses
All the theoretical calculations were conducted in Gaussian09 program package [26]. Conformational searches for possible isomers based on molecular mechanics were conducted with the MM+ method in HyperChem 8.0 software. The corresponding stable conformers, whose Boltzmann distributions were higher than 2%, were further optimized at the B3LYP/6-31G(d) PCM level in DMSO (Tables S2-S4, and Figures S9, S10 and S27 in the supplementary material). Then, all optimized conformers were subjected to the DFT method at the mPW1PW91/6-31+G(d) PCM level in DMSO to acquire calculated shielding tensors. Then, the calculated shielding tensors were averaged based on Boltzmann distribution theory. Finally, the DP4+ analysis of the calculated shielding tensors and experimental chemical shifts was applied, using the Excel template provided by the original authors [12].

Bioassay
The antimicrobial activities against human pathogenic bacteria (Pseudomonas aeruginosa), aquatic pathogens (Edwardsiella tarda and Vibrio alginolyticus), and plant pathogens (Fusarium oxysporum and Fusarim graminearum Schw.) were determined by a serial dilution technique, using 96-well microtiter plates with minor modifications as per our previous report [27]. The human, aquatic, and plant pathogenic strains were offered by the Institute of Oceanology, Chinese Academy of Sciences. Chloramphenicol was used as a positive control for the bacteria, and amphotericin B was used as a positive control for the fungi.

Conclusions
In summary, 11 compounds, including 6 new compounds (1a, 1b, 6, 7, 10 and 11), were obtained from the deep-sea, sediment-derived fungus Aspergillus fumigatus SD-406. Among them, secofumitremorgins A/B featured an unusual seco-diketopiperazine scaffold and the formation of a pyridine moiety. The stereoconfigurations of isolated compounds were determined by chiral HPLC analysis of the acidic hydrolysate, X-ray crystallographic analysis, J-based configuration analysis, and quantum chemical calculations of ECD, OR, and NMR (with DP4+ probability analysis). Compounds 1, 4, 6, 7, 10 and 11 showed inhibitory activities against pathogenic bacteria and plant-pathogenic fungus, with MIC values ranging from 4 to 64 µg/mL, possessing the potential to be developed as antibiotic lead compounds.