Cyclodepsipeptides and Sesquiterpenes from Marine-Derived Fungus Trichothecium roseum and Their Biological Functions

On the basis of the ‘one strain, many compounds’ (OSMAC) strategy, chemical investigation of the marine-derived fungus Trichothecium roseum resulted in the isolation of trichomide cyclodepsipeptides (compounds 1–4) from PDB medium, and destruxin cyclodepsipeptides (compounds 5–7) and cyclonerodiol sesquiterpenes (compounds 8–10) from rice medium. The structures and absolute configurations of novel (compounds 1, 8, and 9) and known compounds were elucidated by extensive spectroscopic analyses, X-ray crystallographic analysis, and ECD calculations. All isolated compounds were evaluated for cytotoxic, nematicidal, and antifungal activities, as well as brine shrimp lethality. The novel compound 1 exhibited significant cytotoxic activities against the human cancer cell lines MCF-7, SW480, and HL-60, with IC50 values of 0.079, 0.107, and 0.149 μM, respectively. In addition, it also showed significant brine shrimp lethality, with an LD50 value of 0.48 μM, and moderate nematicidal activity against Heterodera avenae, with an LC50 value of 94.9 μg/mL. This study constitutes the first report on the cytotoxic and nematicidal potential of trichomide cyclodepsipeptides.


Introduction
In the search for novel bioactive metabolites from various natural resources, marine-derived fungi have gained increasing attention due to their potential capacity to produce structurally unique and biologically active metabolites [1,2]. However, metabolites isolated from marine-derived fungi generally perform far below their biosynthetic potential, indicating the existence of silent biosynthetic pathways [3,4]. The culture strategy of 'one strain, many compounds' (OSMAC) is able to significantly increase the chemical diversity of fungi [5][6][7]. One successful application was the isolation of a series of cytotoxic cytochalasins from a marine-derived fungus Spicaria elegans KLA03, via the modification of the culture media and the addition of tryptophan [8][9][10].

Structural Elucidation
The molecular formula of trichomide D (compound 1), C 31 H 52 N 5 O 8 Cl, was obtained using HRESIMS ( Figure S2 in the Supplementary Materials). The one-dimensional NMR data (Table 1)   Further detailed analyses of one-dimensional and two-dimensional NMR data ( Figures S3-S7) resulted in the identification of one δ-chloro-α,γ-dihydroxypentanoic acid (δ-Cl-DHPA 1 ) and five amino acids, including β-Me-proline (β-Me-Pro 2 ), isoleucine (Ile 3 ), N-Me-isoleucine (N-Me-Ile 4 ), N-Me-alanine (N-Me-Ala 5 ), and β-alanine (β-Ala 6 ) residues ( Figure 2). Based on the characteristic 13 C NMR signal of δ C 48.3, the chlorine atom should be connected to the δ-CH 2 group in the DHPA 1 residue. The above-mentioned residues were preliminarily connected by the observed HMBC cross-peaks from α-CH or N-Me signals to the carbonyl carbons, indicating the presence of a cyclic hexadepsipeptide with the sequence of cyclo-(δ-Cl-DHPA 1 -β-MePro 2 -Ile 3 -NMeIle 4 -NMeAla 5 -β-Ala 6 ). The structure of compound 1 was unambiguously confirmed by single-crystal X-ray diffraction using Cu Kα radiation, which showed R configurations of both α-CH and γ-OH in the δ-Cl-DHPA 1 residue ( Figure 3). The amino acid units in compound 1 were all assigned as L-configured ( Figure 3).      Cyclonerodiol C (compound 8) was confirmed to have the molecular formula C 17 H 30 O 4 by its HRESIMS data ( Figure S8), implying three degrees of unsaturation. The one-dimensional NMR (Figures S9 and S10) and HSQC spectra ( Figure S12) showed the characteristic signals of one carbonyl (CO, δ C 171.5) and two olefinic carbons (CH, δ C 129.6/δ H 5.47 and C, δ C 129.9) (Table 2), which were responsible for two degrees of unsaturation. Therefore, a remaining degree of unsaturation due to one carbon ring can be deduced. The consecutive COSY cross-peaks ( Figure S11) from H-1 to H-2, 6, 5, and 4, as well as the HMBC peaks ( Figure S13) from H-13 to C-2, 3, and 4, could be identified as one cyclopentane residue ( Figure 2). Detailed analyses of the one-and two-dimensional NMR data (Figures S9-S13) suggested that the structure of compound 8 was similar to that of cyclonerodiol (compound 11), except that the signals of the methyl group (δ C 26.1/δ H 1.67) in cyclonerodiol (compound 11) were absent from the 1 H and 13 C NMR spectra of compound 8 [23]. Instead, additional oxymethylene (CH 2 , δ C 69.9/δ H 4.45) and acetoxyl groups (CH 3 , δ C 19.4/δ H 2.08 and CO, δ C 171.5) were observed in the spectrum of compound 8 ( Table 2). The key HMBC correlation between H-12 and the carbonyl carbon further confirmed the linkage between C-12 and the acetoxyl group ( Figure 2). The NMR data of compound 8 (Figures S14-S17) were also measured in DMSO-d6, showing two OH signals of δ H 3.81 and 3.89 (Table S1). The connections between 3-OH (δ H 3.81) and C-3, as well as between 7-OH (δ H 3.89) and C-7, were confirmed by the HMBC cross-peaks from 3-OH to C-3, 4, and 13, as well as from 7-OH to C-7, 8, and 14, respectively ( Figure 2). The molecular formula of compound 9, C 15 H 28 O 3 , was determined via HRESIMS ( Figure S19). The one-dimensional NMR (Figures S20 and S21) and HSQC ( Figure S23) data exhibited marked similarities to those of cyclonerodiol (compound 11) [23], except for the chemical shifts of the two olefinic carbons (CH-9, δ C 122.1/δ H 5.70 and CH-10, δ C 142.3/δ H 5.70), and the presence of another oxygenated quaternary carbon (C-11, δ C 70.8) in compound 9 ( Table 2). CH-9 in the double bond was connected to CH 2 -8, based on the COSY correlation ( Figure S22) between H-8 and H-9, while CH-10 was linked to the quaternary C-11 by the key HMBC cross-peaks ( Figure S24) from H-12 and 15 to C-10 and C-11. Thus, the planar structure of compound 9 was confirmed, and it was named cyclonerodiol D.
The relative configuration of the cyclopentane residue in compound 8 was deduced by the NOESY experiment ( Figure 4 and Figure S18). The key NOE correlation between H 3 -13 and H-2 suggested an α orientation of these protons, while the cross-peaks from H 3 -1 to H-6 and 3-OH indicated a β orientation. The E-geometry for the double bond in compound 8 was also confirmed by the NOE correlation between H-10 and H 2 -12. The relative configuration of C-7 in compound 8 was initially assigned via the NOE cross-peaks from H 3 -14 to H-2, from 7-OH to H 3 -1, and from H 2 -8 to H 2 -5 ( Figure 4, Figure S18), which suggested that the rotation of the single bond C6−C7 was restricted by the surrounding groups [24]. Although the specific rotation of compound 8 [23], evidence for the absolute configuration of compound 8 was still weak. However, the biosynthetic pathway of cyclonerodiol (compound 11) in T. roseum had been already confirmed [25][26][27], suggesting that the cyclonerodiol sesquiterpene compound 8 could also been biosynthesized via the same pathway in T. roseum. Therefore, the plausible biosynthetic pathway of compound 8 was deduced from cyclonerodiol (compound 11), further indicating the same absolute configurations between compounds 8 and 11 ( Figure 5). In addition, comparison of the NMR data between compound 8 and cyclonerodiol (compound 11) showed marked similarities. Based on the above-mentioned analyses, the absolute configuration of compound 8 is likely the same as cyclonerodiol (compound 11). On the basis of the same analyses of NOE correlations and the plausible biosynthetic pathway of compound 9 ( Figure 5), its absolute configuration is likely also the same as that of compound 8 and cyclonerodiol (compound 11). Although the cotton effects (CEs) from 200 nm to 240 nm in the measured CD spectrum of compound 9 were relatively low, its CEs (negative CE near 210 nm, and positive CE near 230 nm) were very similar to those of the calculated CD spectrum. The result of this ECD (Electronic Circular Dichroism) calculation, more or less confirms the absolute configuration of compound 9 as 2S, 3R, 6R, 7R, and 9E ( Figure S25). Detailed methods of the ECD calculation are also described in the Supplementary Materials.
The one-dimensional NMR (Figures S20 and S21) and HSQC ( Figure S23) data exhibited marked similarities to those of cyclonerodiol (compound 11) [23], except for the chemical shifts of the two olefinic carbons (CH-9, δC 122.1/δH 5.70 and CH-10, δC 142.3/δH 5.70), and the presence of another oxygenated quaternary carbon (C-11, δC 70.8) in compound 9 ( Table 2). CH-9 in the double bond was connected to CH2-8, based on the COSY correlation ( Figure S22) between H-8 and H-9, while CH-10 was linked to the quaternary C-11 by the key HMBC cross-peaks ( Figure S24) from H-12 and 15 to C-10 and C-11. Thus, the planar structure of compound 9 was confirmed, and it was named cyclonerodiol D.
The relative configuration of the cyclopentane residue in compound 8 was deduced by the NOESY experiment (Figures 4 and S18). The key NOE correlation between H3-13 and H-2 suggested an α orientation of these protons, while the cross-peaks from H3-1 to H-6 and 3-OH indicated a β orientation. The E-geometry for the double bond in compound 8 was also confirmed by the NOE correlation between H-10 and H2-12. The relative configuration of C-7 in compound 8 was initially assigned via the NOE cross-peaks from H3-14 to H-2, from 7-OH to H3-1, and from H2-8 to H2-5 ( Figure 4, Figure S18), which suggested that the rotation of the single bond C6−C7 was restricted by the surrounding groups [24]. Although the specific rotation of compound 8 was similar to that of cyclonerodiol (compound 11) (  [23], evidence for the absolute configuration of compound 8 was still weak. However, the biosynthetic pathway of cyclonerodiol (compound 11) in T. roseum had been already confirmed [25][26][27], suggesting that the cyclonerodiol sesquiterpene compound 8 could also been biosynthesized via the same pathway in T. roseum. Therefore, the plausible biosynthetic pathway of compound 8 was deduced from cyclonerodiol (compound 11), further indicating the same absolute configurations between compounds 8 and 11 ( Figure 5). In addition, comparison of the NMR data between compound 8 and cyclonerodiol (compound 11) showed marked similarities. Based on the above-mentioned analyses, the absolute configuration of compound 8 is likely the same as cyclonerodiol (compound 11). On the basis of the same analyses of NOE correlations and the plausible biosynthetic pathway of compound 9 ( Figure 5), its absolute configuration is likely also the same as that of compound 8 and cyclonerodiol (compound 11). Although the cotton effects (CEs) from 200 nm to 240 nm in the measured CD spectrum of compound 9 were relatively low, its CEs (negative CE near 210 nm, and positive CE near 230 nm) were very similar to those of the calculated CD spectrum. The result of this ECD (Electronic Circular Dichroism) calculation, more or less confirms the absolute configuration of compound 9 as 2S, 3R, 6R, 7R, and 9E ( Figure S25). Detailed methods of the ECD calculation are also described in the supplementary materials.

Biological Evaluation
The novel compounds (compounds 1, 8, and 9) were evaluated for their cytotoxic activities against five human cancer cell lines (MCF-7, SW480, HL-60, A-549, and SMMC-7721) [32][33][34]. Compound 1 showed significant cytotoxicity against MCF-7, SW480, and HL-60, with IC50 values of 0.079, 0.107, and 0.149 μM, respectively-better than the positive control of cisplatin (Table 3). However, compounds 8 and 9 were inactive in the cytotoxic assay (IC50 > 40 μM). This is the first report on the cytotoxic activity of trichomide cyclodepsipeptides. Although the cytotoxic mechanisms of trichomides have not yet been revealed, the cytotoxic mechanisms of the structurally similar destruxins have been reported to be associated with the inhibition of the phosphoinositide-3kinase (PI3K)/Akt pathway, and the disturbance of the intracellular redox balance. Therefore, trichomide cyclodepsipeptides might show the similar cytotoxic mechanisms to destruxins [13,35]. Brine shrimp (Artemia salina), an aquatic species characterized by high sensibility to toxins, can be used as a model organism for quick preliminary insecticidal screening [36,37]. Therefore, in order to identify the leading insecticidal compounds, all of the isolated compounds (compounds 1-10) were evaluated for lethal activity against brine shrimp, and furthermore, for nematicidal activity against Heterodera avenae [38] (Table 4). In the brine shrimp assay, the cyclodepsipeptide compounds 1, 2, and 4-6 exhibited significant lethal activity, with LD50 values of 0.48, 0.74, 3.22, 2.47, and 2.81 μM, respectively. The nematicidal assay showed that compounds 1 and 2 exhibited moderate activity, with LC50 values of 94.9 and 143.6 μg/mL, respectively. Compounds 1 and 2 exhibited obviously better insecticidal and nematicidal activity against brine shrimp and H. avenae than the other cyclodepsipeptides, which was probably due to the structural diversity of the DHPA 1 residues and the presence of N-Me-Ile 4 residue in the trichomide cyclodepsipeptides.  Besides the above-mentioned three novel compounds (compounds 1, 8, and 9), three known trichomide cyclodepsipeptides were isolated from the PDB medium: destruxin A5 (compound 2) [11], trichomide A (compound 3) [11], and homodestruxin B (compound 4) [11]. Three typical destruxin cyclodepsipeptides, destruxin chlorohydrin (compound 5) [28], roseotoxin B (compound 6) [29], and C (compound 7) [30], and one cyclonerodiol sesquiterpene, ascotrichic acid (compound 10) [31] were isolated from the rice medium. The structures of these compounds were determined by detailed analyses of their spectroscopic data and comparisons with previously published reports.

Biological Evaluation
The novel compounds (compounds 1, 8, and 9) were evaluated for their cytotoxic activities against five human cancer cell lines (MCF-7, SW480, HL-60, A-549, and SMMC-7721) [32][33][34]. Compound 1 showed significant cytotoxicity against MCF-7, SW480, and HL-60, with IC 50 values of 0.079, 0.107, and 0.149 µM, respectively-better than the positive control of cisplatin (Table 3). However, compounds 8 and 9 were inactive in the cytotoxic assay (IC 50 > 40 µM). This is the first report on the cytotoxic activity of trichomide cyclodepsipeptides. Although the cytotoxic mechanisms of trichomides have not yet been revealed, the cytotoxic mechanisms of the structurally similar destruxins have been reported to be associated with the inhibition of the phosphoinositide-3-kinase (PI3K)/Akt pathway, and the disturbance of the intracellular redox balance. Therefore, trichomide cyclodepsipeptides might show the similar cytotoxic mechanisms to destruxins [13,35]. Brine shrimp (Artemia salina), an aquatic species characterized by high sensibility to toxins, can be used as a model organism for quick preliminary insecticidal screening [36,37]. Therefore, in order to identify the leading insecticidal compounds, all of the isolated compounds (compounds 1-10) were evaluated for lethal activity against brine shrimp, and furthermore, for nematicidal activity against Heterodera avenae [38] (Table 4). In the brine shrimp assay, the cyclodepsipeptide compounds 1, 2, and 4-6 exhibited significant lethal activity, with LD 50 values of 0.48, 0.74, 3.22, 2.47, and 2.81 µM, respectively. The nematicidal assay showed that compounds 1 and 2 exhibited moderate activity, with LC 50 values of 94.9 and 143.6 µg/mL, respectively. Compounds 1 and 2 exhibited obviously better insecticidal and nematicidal activity against brine shrimp and H. avenae than the other cyclodepsipeptides, which was probably due to the structural diversity of the DHPA 1 residues and the presence of N-Me-Ile 4 residue in the trichomide cyclodepsipeptides. However, only compound 9 showed moderate antifungal activity against Valsa mali, with an MIC value of 64 µg/mL. The trichomides (compounds 1 and 4) and sesquiterpene (compound 8) exhibited weak bioactivities against V. mali and Rhizoctonia cerealis, with MIC values from 128 to 256 µg/mL (Table S2). None of the isolated compounds exhibited activity against Fusarium. oxysporum f. sp. vasinfectum.

Fungal Material
The fungal strain T. roseum was isolated from marine driftwood collected from the intertidal zone of Lingshan Island, Qingdao, China in November 2013. The fungus was identified on the basis of morphological characteristics and molecular analyses of ITS [20]. The strain was preserved in the Natural Products Laboratory, College of Chemistry and Pharmacy, Qingdao Agricultural University.

Fermentation and HPLC Analyses
Fresh mycelia of the fungus were statically fermented at 28 • C for 30 days on the liquid PDB and solid rice media. The liquid culture was conducted in 40 × 1 L conical flasks containing 300 mL of PDB medium (1000 mL natural seawater, 20 g glucose, and 200 mL potato juice, pH 6.5−7.0), while the solid one was kept in 40 × 1 L flasks containing rice (100 g/flask), peptone (0.6 g/flask), and natural seawater (100 mL/flask).
Ethyl acetate (EtOAc) extracts of the two fermentations were analyzed using HPLC with a MeOH-H 2 O eluting gradient and a detection wavelength of 225 nm ( Figure S1). Details are described in the Supplementary Materials.

Crystal Structure Determination
A colorless single crystal of compound 1 was obtained by the slow evaporation of a methanol solution (containing trace water), thus, its crystal structure contained two molecules of H 2 O. H 2 O molecules can form intermolecular hydrogen bonds with the cyclodepsipeptide compound 1, which was helpful to the crystallization of the compound. All crystallographic data were collected at 150.01 K on a Bruker Smart-1000 CCD diffractometer (Bruker-AXS, Saarbrucken, Germany) equipped with graphite-monochromatic Cu-Kα radiation (λ = 1.54178 Å). The adsorption data were obtained using the program SADABS [39]. The structures were elucidated by direct methods, using the SHELXTL software package [40]. All non-hydrogen atoms were refined with anisotropic displacement parameters. The hydrogen atoms were located via geometrical calculations, and their positions and thermal parameters were fixed during structure refinement. The structures were refined using full-matrix least-squares techniques [41]. Crystallographic data of compound 1 was deposited in the Cambridge Crystallographic Data Centre as CCDC 1858313.
Crystal data for compound 1: C 31 (13) in the final refinement for all 6124 reflections with 443 Friedel pairs. There were also crystallographic disorders of δ-CH 2 −Cl/δ'-CH 2 −Cl in the residue of δ-Cl-DHPA 1 , probably due to the flexibility of the δ-Cl-DHPA 1 residue.

Cytotoxicity against Human Cancer Cell Lines
The in vitro cytotoxic effects of the novel compounds 1, 8, and 9 were evaluated on five human cancer cell lines using the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) method [32][33][34]. The human cancer cell lines were as follows: HL-60, human myeloid leukemia; A-549, lung cancer; MCF-7, breast cancer; SW-480, human colon cancer; and SMMC-7721, liver cancer. All cells were cultured in RPMI-1640 medium containing 10% fetal bovine serum (FBS), and kept in a humidified atmosphere containing 5% CO 2 ayt 37 • C. The novel compounds and the cisplatin positive control were dissolved and diluted in DMSO to obtain sample solvents with a series of different concentrations. In brief, the cell suspensions (200 µL, 5 × 10 4 cells/mL) were seeded into 96-well culture plates and kept at 37 • C for 12 h, then the sample solvents (20 µL) were added into each well and further cultured at 37 • C for 24 h. Subsequently, MTT (100 µg) was added into each well and incubated at 37 • C for 4 h. After removal of the 100 µL culture medium, the cells were lysed with 20% SDS-50% DMF (100 µL). The remaining lysates were subjected to measurements of the optical density at 595 nm with a 96-well microtiter plate reader. Reed and Muench's method was used to calculate IC 50 values [32].

Brine Shrimp Lethality and Nematicidal Activity
Brine shrimp (Artemia salina) toxicity was evaluated as previously reported [36,37]. The plant-parasitic nematode H. avenae was selected for nematicidal bioassay using 24-well plates. Second stage juveniles (J2s) of H. avenae were collected to prepare the nematode suspension based on the protocol reported previously [34]. The isolated compounds (1-10) and abamectin positive control were dissolved and diluted in DMSO to obtain sample solvents with a series of different concentrations. The sample solvents (5 µL) were added to separate wells with the nematode suspension (495 µL), containing about 100 J2s, while the same amount of DMSO (5 µL) was added for the negative control. The plates were maintained at 25 • C for 48 h, and then observed using a stereomicroscope to evaluate the nematode mortalities. Nematodes were defined to be dead if their bodies became straight and did not react to mechanical touches [38]. The experiment was repeated three times under the same conditions.

Conclusions
The chemical investigation of marine-derived fungus T. roseum resulted in isolation of trichomide cyclodepsipeptides (compounds 1-4) from the liquid PDB medium, and, based on the OSMAC culture strategy, isolation of destruxin cyclodepsipeptides (compounds 5-7) and cyclonerodiol sesquiterpenes (compounds 8-10) from solid rice medium. The absolute configuration of novel compound 1 was determined by single crystal X-ray diffraction analysis, while the configurations of compounds 8 and 9 were determined by NOESY experiments, comparisons of specific rotations, ECD calculation, and plausible biosynthetic pathways. The novel compound 1 exhibited significant cytotoxic activities against human cancer cell lines MCF-7, SW480, and HL-60, with IC 50 values of 0.079, 0.107, and 0.149 µM, respectively. In addition, it also showed significant brine shrimp lethality with an LD 50 value of 0.48 µM, and moderate nematicidal activity against H. avenae with an LC 50 value of 94.9 µg/mL. Its cytotoxic, brine shrimp lethality and nematicidal activities suggest potential applications in the areas of medicine and agriculture. This is also the first time the cytotoxic and nematicidal potential of trichomide cyclodepsipeptides has been reported.