New Diterpenes with a Fused 6-5-6-6 Ring System Isolated from the Marine Sponge-Derived Fungus Trichoderma harzianum

New diterpenes, namely, trichodermanins F–H, with a fused 6-5-6-6 ring system were isolated from the fungus Trichoderma harzianum OUPS-111D-4 separated from a marine sponge Halichondria okadai. These chemical structures were elucidated by 1D and 2D NMR as well as high-resolution fast atom bombardment mass spectrometry (HRFABMS) spectral analyses. We established their absolute stereostructures by application of the modified Mosher’s method or circular dichroism (CD) spectroscopy. In addition, their cytotoxicities were assessed using several cancer cell lines, with 1 and 2 exhibiting modest activities.


Introduction
A number of secondary metabolites derived from marine microorganisms have diverse structures and exhibit unexpected biological activities [1][2][3][4]. This motivated us to target marine microorganisms as a source of seed compounds for antitumor chemotherapy agents, and we have reported many cytotoxic metabolites to date [5][6][7][8]. As part of this ongoing study, we examined the metabolites of the fungus Trichoderma harzianum separated from a piece of the marine sponge Halichondria okadai. We have already reported the isolation, structural determination, and cytotoxicity of the decalin derivatives tandyukisins A-F derived from this fungus [9][10][11] as well as those of trichodermanins C, D, and E (1), classified as diterpenes with a rare fused 6-5-6-6 ring system [12]. In the continuing search for cytotoxic metabolites from this fungal strain, we isolated three new compounds, namely, trichodermanins F-H (2)(3)(4). Except for our study, there are only two other reports of natural products consisting of this ring system, i.e., trichodermanin A [13] and wickerols A and B [14]. We have previously reported the first determination of the absolute configurations of trichodermanins C and D by application of the modified Mosher's method [15]. In the present study, we established the absolute configuration of trichodermanin E (1), which was confirmed by circular dichroism (CD) spectroscopy. In addition, we also examined the absolute stereostructures of the newly isolated trichodermanins F-H (2)(3)(4). We herein describe the evaluation of the cytotoxic activities of 1-4, although they exhibited zero or weak activities.

Results and Discussion
T. harzianum, a fungus from H. okadai, was cultured at 27 • C for six weeks in a medium (70 L) containing 1% glucose, 1% malt extract, and 0.05% peptone in artificial seawater adjusted to pH 7.5. After incubation, the EtOAc extract of the culture filtrate was purified via cytotoxic assay-directed fractionation using a stepwise combination of silica gel column and Sephadex LH-20 chromatography, followed by reverse-phase HPLC, affording trichodermanins E (1) (0.93 mg, 0.011%), F (2) (3.7 mg, 0.043%), G (3) (4.6 mg, 0.053%), and H (4) (1.3 mg, 0.015%) The molecular formula of trichodermanin E (1) was established as C20H34O3 based on the molecular ion m/z 345.2397 [M + Na] + in high-resolution fast atom bombardment mass spectrometry (HRFABMS) ( Figure S7). Absorption in the IR spectrum at 3383 cm −1 indicated the presence of hydroxy groups. The relative stereostructure of 1 has previously been reported using distortionless enhancement by polarization transfer (DEPT), heteronuclear multiple quantum coherence spectroscopy (HMQC), and NOESY experiments with the 1 H and 13 C NMR spectra of 1 ( Figure 1, and Figures S1-S6, Table 1, and Table S1) [12]. However, the assignment of its absolute configuration was not performed. In the present study, the absolute stereostructure of 1 was elucidated by acylation with p-bromobenzoyl chloride to apply the dibenzoate rule [16]. In the CD spectrum of the dibenzoate 1a, the negative Cotton effect (Δε258-15) demonstrated that the screw sense of two benzoate chromophores at C-1 and C-2 was counterclockwise. Therefore, the absolute configurations for C-1 and C-2 were revealed as R and R, respectively ( Figure 2) [16].     Figure S7). Absorption in the IR spectrum at 3383 cm −1 indicated the presence of hydroxy groups. The relative stereostructure of 1 has previously been reported using distortionless enhancement by polarization transfer (DEPT), heteronuclear multiple quantum coherence spectroscopy (HMQC), and NOESY experiments with the 1 H and 13 C NMR spectra of 1 ( Figure 1, and Figures S1-S6, Table 1, and Table S1) [12]. However, the assignment of its absolute configuration was not performed. In the present study, the absolute stereostructure of 1 was elucidated by acylation with p-bromobenzoyl chloride to apply the dibenzoate rule [16]. In the CD spectrum of the dibenzoate 1a, the negative Cotton effect (Δε258-15) demonstrated that the screw sense of two benzoate chromophores at C-1 and C-2 was counterclockwise. Therefore, the absolute configurations for C-1 and C-2 were revealed as R and R, respectively ( Figure 2) [16].    , which contained one less oxygen atom than that of 1 ( Figure S14). Consideration of the 1 H and 13 C NMR spectra of 2 (Table 1 and Table S2, and Figures S8, S9 and S12), including 2D NMR spectra, suggested that compound 2 had one secondary methyl (C-17); four tertiary methyls (C-16, C-18, C-19, and C-20); six sp 3 -hybridized methylenes (C-1, C-7, C-9, C-10, C-13, and C-14); five sp 3 methines (C-2, C-3, C-6, C-11, and C-12), including one oxygen-bearing sp 3 methine (C-2); and four quaternary sp 3 carbons (C-4, C-5, C-8, and C-15), including one oxygen-bearing quaternary sp 3 carbon (C-15). 1 H-1 H correlation spectroscopy (COSY) revealed three partial structures ( Figure 3 and Figure S10). In the HMBC spectrum ( Figure 3 and Figure S13), the correlation from 15-methyl to C-11, C14, and C-15; from 8-methyl to C-7, C-8, C-9, and C-12; from germinal dimethyl (H-18 and H-19) to C-4, C-5, and C-6; from H-12 to C-3, C-4, and C-5; and from H-13 to C-12 showed three cyclohexane rings and a cyclopentane comprising a fused 6-5-6-6 ring system to give the planar structure of 2, as shown in Figure 3. For the stereochemistry of 2, the relative configuration and conformation were investigated by NOESY experiments (Table S2 and Figure 4 and Figure S11). In the NOESY analysis of 2, correlations from H-2α to H-19; from H-3 to H-11, H-14β, and H-20; and from H-1β to H-20 indicated that cyclohexane ring A existed in the boat conformation with 3-CH3 in the α-orientation. In cyclohexane ring B, the above correlations and the NOESY cross peaks (H-7α/H-12 and H-18) demonstrated that ring B existed in the chair conformation and that 5-CH3 (C-18), H-7α, and H-12 were oriented in coaxial arrangements. The NOESY correlations between H-3, H-11, and H-14β indicated that cyclohexane ring C also existed in the chair conformation. In addition, the correlations (H-18/H-13α, H-11/H-20, H-14α/H-16, and H-12/H-16) revealed that H-11, H-13α, H-14β, and 15-CH3 (C-16) were oriented in coaxial arrangements and that the ring juncture between cyclohexane rings B and C was trans ( Figure 4 and Figure S11, and Table S2). A difference in the structural features of 2 and 1 was the absence of a secondary hydroxyl group at C-1. Therefore, we applied the modified Mosher's method [15] to determine their absolute stereostructures. The 1 H chemical shift differences between the (S)-and (R)-MTPA esters 2a and 2b showed positive on the right side and negative on the left side of the MTPA plane; therefore, we revealed an S configuration at C-2 ( Figure 5 and Table S5). For the stereochemistry of 2, the relative configuration and conformation were investigated by NOESY experiments (Table S2 and Figure 4 and Figure S11). In the NOESY analysis of 2, correlations from H-2α to H-19; from H-3 to H-11, H-14β, and H-20; and from H-1β to H-20 indicated that cyclohexane ring A existed in the boat conformation with 3-CH 3 in the α-orientation. In cyclohexane ring B, the above correlations and the NOESY cross peaks (H-7α/H-12 and H-18) demonstrated that ring B existed in the chair conformation and that 5-CH 3 (C-18), H-7α, and H-12 were oriented in coaxial arrangements. The NOESY correlations between H-3, H-11, and H-14β indicated that cyclohexane ring C also existed in the chair conformation. In addition, the correlations (H-18/H-13α, H-11/H-20, H-14α/H-16, and H-12/H-16) revealed that H-11, H-13α, H-14β, and 15-CH 3 (C-16) were oriented in coaxial arrangements and that the ring juncture between cyclohexane rings B and C was trans ( Figure 4 and Figure S11, and Table S2). For the stereochemistry of 2, the relative configuration and conformation were investigated by NOESY experiments (Table S2 and Figure 4 and Figure S11). In the NOESY analysis of 2, correlations from H-2α to H-19; from H-3 to H-11, H-14β, and H-20; and from H-1β to H-20 indicated that cyclohexane ring A existed in the boat conformation with 3-CH3 in the α-orientation. In cyclohexane ring B, the above correlations and the NOESY cross peaks (H-7α/H-12 and H-18) demonstrated that ring B existed in the chair conformation and that 5-CH3 (C-18), H-7α, and H-12 were oriented in coaxial arrangements. The NOESY correlations between H-3, H-11, and H-14β indicated that cyclohexane ring C also existed in the chair conformation. In addition, the correlations (H-18/H-13α, H-11/H-20, H-14α/H-16, and H-12/H-16) revealed that H-11, H-13α, H-14β, and 15-CH3 (C-16) were oriented in coaxial arrangements and that the ring juncture between cyclohexane rings B and C was trans ( Figure 4 and Figure S11, and Table S2). A difference in the structural features of 2 and 1 was the absence of a secondary hydroxyl group at C-1. Therefore, we applied the modified Mosher's method [15] to determine their absolute stereostructures. The 1 H chemical shift differences between the (S)-and (R)-MTPA esters 2a and 2b showed positive on the right side and negative on the left side of the MTPA plane; therefore, we revealed an S configuration at C-2 ( Figure 5 and Table S5). A difference in the structural features of 2 and 1 was the absence of a secondary hydroxyl group at C-1. Therefore, we applied the modified Mosher's method [15] to determine their absolute stereostructures. The 1 H chemical shift differences between the (S)-and (R)-MTPA esters 2a and 2b showed positive on the right side and negative on the left side of the MTPA plane; therefore, we revealed an S configuration at C-2 ( Figure 5 and Table S5). Trichodermanin G (3) was assigned the formula C20H34O3 by HRFABMS m/z 345.2395 [M + Na] + (calcd for C20H34O3Na: 345.2396), which was the same as that of 1 ( Figure S21). The NMR spectral features (Table 1 and Table S3, and Figures S15,16, and S19) resembled those of 1, except for the proton signals for H-2 (δH 1.45, m and δH 2.66, ddd), H-3 (δH 2.10, m), and H-10 (δH 4.38, ddd) and the carbon signals of C-1 (δC 72.6), C-2 (δC 41.8), C-3 (δC 26.7), C-9 (δC 52.2), C-10 (δC 72.9), and C-11 (δC 55.0) in 3. The 1 H-1 H COSY correlations (H-1/H-6, H-1/H-2, and H-2/H-3), and HMBC correlation from H-17 to C-2 indicated that the hydroxy methine at C-2 in 1 were replaced by a methylene in 3 (Table 1 and Table S3, and Figures S17 and S20). In addition, the 1 H-1 H COSY correlations (H-9/H-10, H-10/H-11, and H-11/H-12) and HMBC correlations (H-20/C-9, H-12/C-10, and H-16/C-11) demonstrated that the methylene at C-10 in 1 were replaced by a hydroxy methine in 3 (Table 1 and  Table S3, and Figures S17 and S20). The HMBC correlations for the 6-5-6-6 skeleton were the same as those of 1 and 3 (Table S3 and Figure S20). The above evidence confirmed the planar structure of 3.
The study for the stereochemistry of 3 is described later (together with that of 4).
The relative stereochemical configurations and conformations of 3 and 4 were investigated by NOESY experiments (Tables S3 and S4, Figure 6, and Figures S18 and S25).  Figure 6, and Figures S18 and S25). The absolute configurations of 3 and 4 were not positively elucidated because of the small amount of material available. However, it is unreasonable to assume that only 3 and 4 derived from the same fungus are enantiomers of other trichodermanins. Therefore, we suggest the absolute stereostructures of 3 and 4 as shown in Figure 1 based on assumed biosynthetic relatedness. These stereostructural hypotheses will be confirmed using the modified Mosher's method [15], the quantum chemical equivalent circulating density (ECD) calculation method [17], or synthetic research in the future.
Mar. Drugs 2019, 17, x FOR PEER REVIEW 6 of 10 4 as shown in Figure 1 based on assumed biosynthetic relatedness. These stereostructural hypotheses will be confirmed using the modified Mosher's method [15], the quantum chemical equivalent circulating density (ECD) calculation method [17], or synthetic research in the future. As a primary screen for antitumor activity, the cancer cell growth inhibitory properties of trichodermanins E-H (1-4) were examined using murine P388 leukemia, human HL-60 leukemia, and murine L1210 leukemia cell lines. The results are shown in Table 2. Both 1 and 2 exhibited modest activities against the cell lines, although they were inferior to that of 5-fluorouracil (Table 2). However, 3 and 4 did not inhibit cell growth at all (Table 2). In the study of the structure-activity relationship inferred from the activities of 1-4 and previous trichodermanins E and D, the binding positions (C-1, C-2, C-6, C-10, and C-17) of the hydroxy groups did not contribute to the enhancement of activity. The number had some influence, i.e., the activities of diol derivatives as 2 and trichodermanins C were more potent than those of the triol derivatives as 1, 3, 4, and trichodermanin D. In particular, trichodermanins C with a carbonyl group at C-1 showed strong activity comparable to 5-fluorouracil. We speculate that the significant cell growth inhibition of trichodermanins C was caused by the change of conformation of A ring due to the carbonyl group at C-1. A comparison of cytotoxicity with normal cell and an elucidation of the detailed mechanism of activity will continue to be examined after supply of these fungal metabolites.

General Experimental Procedures
These are the same procedures as those in recent reports [7,8,9,10,11]. NMR spectra were recorded on an Agilent-NMR-vnmrs (Agilent Technologies, Santa Clara, CA, USA) 600 with tetramethylsilane (TMS) as an internal reference. FABMS was recorded using a JEOL JMS-7000 mass spectrometer (JEOL, Tokyo, Japan). IR spectra were recorded on a JASCO FT/IR-680 Plus (Tokyo, Japan). Optical rotations were measured using a JASCO DIP-1000 digital polarimeter (Tokyo, Japan). As a primary screen for antitumor activity, the cancer cell growth inhibitory properties of trichodermanins E-H (1-4) were examined using murine P388 leukemia, human HL-60 leukemia, and murine L1210 leukemia cell lines. The results are shown in Table 2. Both 1 and 2 exhibited modest activities against the cell lines, although they were inferior to that of 5-fluorouracil (Table 2). However, 3 and 4 did not inhibit cell growth at all (Table 2). In the study of the structure-activity relationship inferred from the activities of 1-4 and previous trichodermanins E and D, the binding positions (C-1, C-2, C-6, C-10, and C-17) of the hydroxy groups did not contribute to the enhancement of activity. The number had some influence, i.e., the activities of diol derivatives as 2 and trichodermanins C were more potent than those of the triol derivatives as 1, 3, 4, and trichodermanin D. In particular, trichodermanins C with a carbonyl group at C-1 showed strong activity comparable to 5-fluorouracil. We speculate that the significant cell growth inhibition of trichodermanins C was caused by the change of conformation of A ring due to the carbonyl group at C-1. A comparison of cytotoxicity with normal cell and an elucidation of the detailed mechanism of activity will continue to be examined after supply of these fungal metabolites.

Fungal Material
This study is a follow-up report for this fungal strain. As shown in the previous reports [9], the fungus Trichoderma harzianum was isolated from a piece of inner tissue of the marine sponge Halichondria okadai collected at Osaka Bay, Japan, in October 2008. The fungal strain was identified by Techno Suruga Laboratory Co., Ltd. The sponge was wiped with EtOH and a cutting applied to the surface of nutrient agar layered in a Petri dish. Serial transfers of one of the resulting colonies provided a pure strain of T. harzianum.

Culturing and Isolation of Metabolites
The EtOAc extract (8.6 g) of the culture filtrate, which was obtained as described in previous reports [9][10][11][12], was chromatographed on a silica gel column with a CHCl 3 /MeOH gradient as the eluent to afford Fr.  Table 1 and Table S1 and the previous report [12].

Formation of Dibenzoate of 1
To a solution of 1 (1.8 mg, 5.6 µmol) in abs. pyridine (0.3 mL), p-bromobenzoyl chloride (3.0 mg, 13.6 µmol) was added, and the reaction mixture was stirred at room temperature for 15 h. Water (1.0 mL) was added to the reaction mixture and extracted using CH 2 Cl 2 . The organic layer was evaporated under reduced pressure, and the residue was purified by HPLC using 100% MeOH as the eluent to afford dibenzoate 1a ( Figure S36). 1 H NMR data and spectrum including 2D NMR spectra are listed in Table S5, and Figures S33-S35 of Supplementary Materials.

Assay for Cytotoxicity
Cytotoxic activities of 1-4 were examined by the same procedure to date [5][6][7][8][9][10][11][12]18], i.e., the 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) method. P388, HL-60, and L1210 cells were cultured in RPMI 1640 Medium (10% fetal calf serum) at 37 • C in 5% CO 2 . The test materials were dissolved in dimethyl sulfoxide (DMSO) to give a concentration of 10 mM, and the solution was diluted with the Essential Medium to yield concentrations of 200, 20, and 2 µM, respectively. Each solution was combined with each cell suspension (1 × 10 5 cells/mL) in the medium, respectively. After incubating at 37 • C for 72 h in 5% CO 2 , grown cells were labeled with 5 mg/mL MTT in phosphate-buffered saline (PBS), and the absorbance of formazan dissolved in 20% sodium dodecyl sulfate (SDS) in 0.1 N HCl was measured at 540 nm with a microplate reader (MTP-310, CORONA electric). Each absorbance value was expressed as percentage relative to that of the control cell suspension that was prepared without the test substance using the same procedure as that described above. All assays were performed three times, semilogarithmic plots were constructed from the averaged data, and the effective dose of the substance required to inhibit cell growth by 50% (IC 50 ) was determined.

The Origin of the Cell Lines
P388 cell line was obtained from Dr. Numata. HL-60 cell line was obtained from Dr. Kawai. L1210 cell line was from Dr. Endo.

Conclusions
In this study, three new terpenes with a rare fused 6-5-6-6 ring system, namely, trichodermanins E-H (1-4), were isolated from the fungus T. harzianum separated from the marine sponge H. okadai. Spectral analyses and chemical transformations were utilized to elucidate the absolute stereostructures of 1 and 2. The relative stereostructures of 3 and 4 were assumed to be those of the same class of metabolites. Therefore, we deduced that their absolute configurations were also the same as those of other trichodermanins. In cytotoxic assays performed using three cancer cell lines, 1 and 2 exhibited moderate activity against the cell lines, although they were inferior to that of 5-fluorouracil.
Author Contributions: T.Y., A.F., and T.K conceived and designed the experiments; T.Y. and A.F. performed the experiments; T.Y. analyzed the data; and T.Y. wrote the paper.
Funding: This research received no external funding.