Computationally Assisted Structural Elucidation of Cembranoids from the Soft Coral Sarcophyton tortuosum

A persistent study on soft coral Sarcophyton tortuosum resulted in the characterization of two new cembranolides, tortuolides A and B (1 and 2), and a new related diterpene, epi-sarcophytonolide Q. Their structures were determined not only by extensive spectroscopic analysis but also by DFT calculations of ECD and NMR data, the latter of which was combined with statistical analysis methods, e.g., DP4+ and J-DP4 approaches. Anti-inflammatory and cytotoxicity activities were evaluated in this study.


Introduction
Soft corals are known to produce a large variety of secondary metabolites [1]. In particular, soft coral of the genus Sarcophyton is a prolific source of promising bioactive cembranoids, some of which have exhibited antiviral [2], anti-inflammatory [3][4][5][6][7], and cytotoxic activities [4][5][6]8]. The flexibility of the macrocyclic ring in cembranoids makes the accurate determination of chemical structures particularly challenging. Despite the 2D NMR spectroscopic method being the most potent approach for structural elucidation, it suffers from inherently low accuracy for flexible structures, especially when there is no informative correlation in NMR spectrometry. Consequently, the computational approach and related statistical analysis methods, e.g., DP4+ and J-DP4 [9,10], have gradually became convenient and reliable alternatives.
Our pervious investigation on Sarcophyton tortuosum resulted in the isolation of several novel structures, including secotortuosenes A and B with a novel secoditerpenoidal skeleton, bistortuolide cyclobutane A with a novel cyclobutane biscembranoidal skeleton, and tortuosenes A and B with a rare tricyclic diterpenoidal skeleton [7,11]. As part of our continuing effort to explore bioactive marine natural products from soft corals [3][4][5][6][7][8], the chemical constituents of S. tortuosum collected at Lanyu Island were investigated in this study, and three new cembranoids, namely tortuolides A and B (1 and 2) and epi-sarcophytonolide Q (3), were characterized ( Figure 1). Structural elucidations were performed by a comprehensive 2D NMR spectroscopic analysis, as well as computational and statistical analysis methods. Their biological activities, including cytotoxicity and anti-inflammatory activities, were evaluated herein.

Results and Discussion
Defrosted specimens (1.3 kg) of the soft coral S. tortuosum was freeze-dried, minced, and extracted with EtOAc to yield a crude extract (10.2 g), which was repeatedly chromatographed on silica gel and subsequently purified by high-performance liquid chromatography (HPLC) to obtain compounds 1 (8.9 mg), 2 (2.7 mg), and 3 (2.1 mg) ( Figure 1).
Tortuolide A (1), obtained as a colorless oil, was found to have a molecular formula of C 23 H 32 O 7 based on the sodiated ion peak at m/z 443.2049 [M + Na] + (calcd for C 23 H 32 O 7 Na, 443.2040). The 1 H and 13 C NMR data ( Table 1)  Inspection of the NMR data revealed that the planar structures of 1 and emblide [11,12] were quite similar, with differences for the ∆ 1 double bond in emblide replaced by an epoxy ring in 1, as indicated by the heteronuclear multiple bond correlation (HMBC) correlations from H 3 -16, H 3 -17, H 2 -14, and H-2 to C-1, as well as the correlation spectroscopy (COSY) correlation between H-2 and H-3 ( Figure 2). The cis geometry of the epoxide was assigned by the nuclear Overhauser effect (NOE) correlation of H-15/H-3, whereas the E double bond was deduced based on the NOE correlation of OMe/H-3 ( Figure 3).  Figures S1-S8).
Evaluations for inhibitory effect toward the superoxide anion generation and elastase release in fMLF/CB-induced human neutrophils were performed on compounds 1-3. The result showed that compounds 1 and 2 exhibited weak inhibitory activity of 13.64 ± 2.27 and 14.15 ± 3.57%, respectively, on elastase release at a concentration of 10 µM. Compounds 1-3 were further screened for cytotoxicity toward murine leukemia (P388), human chronic myelogenous leukemia (K562), human colon carcinoma (HT-29), human lung adenocarcinoma (A-549), and lymphoblastic leukemia (Molt-4); unfortunately, the tested compounds were also found to be inactive against the above cell lines, with IC 50s over 40 µM.

Animal Material
The animal material, S. tortuosum, was manually collected by an underwater diver from the coral reef of Lanyu Island of Taiwan in August 2008. The specimen was identified by Prof. C.-F. Dai. A voucher specimen (specimen no. sheuCYJ-001) was deposited with the Department of Marine Biotechnology and Resources, National Sun Yat-sen University, Kaohsiung 804, Taiwan.

Computational Method
A conformational search was performed at the MMFF94 force field using GMMX package implemented in Gaussian 16 software [14]. The resulted conformers, within a 6 kcal/mol window, were subjected for further NMR and ECD calculations.
For iJ/dJ-DP4, the shielding tensors and Fermi contacts were calculated at the PCM/ B3LYP/6-31+G(d,p)//MMFF94 level. The resulting data were weighted based on Boltzmann population using energies calculated at the same level of theory. J-DP4 probabilities were generated using the Excel sheet provided by Zanardi et al [10]. For DP4+, the conformers were subjected to geometry optimization at B3LYP/6-31G(d) [9]. The NMR chemical shifts were computed at the PCM/mPW1PW91/6-31G+(d,p)//B3LYP/6-31G(d) level in chloroform with the Boltzmann population refined in the solvation model based on density (SMD) for CHCl 3 at a new level (M06-2X/6-31G+(d,p)) [15]. The DP4+ probability was determined using the Excel sheet provided in the literature [9].
The conformers resulting from MMFF94 calculations were subjected to geometry optimizations and frequency calculations at the M06-2X/def2svp level using IEFPCM in MeOH. The generated ECD spectra calculated at TDDFT/M06-2X/def2TZVP with IEF-PCM in MeOH were weighted based on the Boltzmann population using Gibbs free energy, obtained by the sum of single-point energy at M06-2X/def2TZVP and the thermal correction at M06-2X/def2svp. The calculated ECD spectra were generated using GaussView 6 by applying a Gaussian band shape with 0.20 and 0.28 eV width for 1 and 2, respectively. It should be noted that Grimme's dispersion (D3 version) was used for empirical dispersion correction in ECD calculation, and the g09defaults keywords were applied for NMR calculation.

Cytotoxicity Assay
The assay was implemented using the published Alamar Blue assay according to the published protocols [16,17]. Concisely, cancer cells, including P388, K562, HT-29, A549, and MOLT-4, were purchased from the American Type Culture Collection and individually seeded into a 96-well microtiter plate and incubated following the previously published procedure. The tested compounds were dissolved in DMSO and added to each well of cancer cells. After three days of culture, attached cells were treated with Alamar Blue for 4 h and subsequently measured at 595 nm using a microplate reader.

Anti-Inflammatory Assay
Freshly isolated human neutrophils from blood using dextran sedimentation were incubated according to the published procedure [18,19]. The incubated neutrophils (6 × 10 5 cells mL −1 ) were treated with compounds 1-3 in DMSO for 5 min. After the neutrophils were activated with fMLF (100 nM)/CB for 10 min, the amounts of superoxide generation and elastase release were measured at 550 nm and 405 nm, respectively, using a UV spectrometer apparatus.

Conclusions
Two cembranolides, namely tortuolides A and B (1 and 2), and a related cembranoid, namely epi-sarcophytonolide Q (3), were characterized from the persistent study of the soft coral Sarcophyton tortuosum. Compounds 1 and 2 are structurally related to emblide [12,20], featuring a C-8-C-20 α,β-unsaturated ε-lactone ring, and represented the first embliderelated cembranolide with a 1,2-epoxy functionality. The flexible nature of macrocyclic compounds, e.g. cembranoids, make the unambiguous assignment of chemical structures particularly challenging. In the present study, we showed the successful application of DFT calculations combined with statistical analysis methods, e.g. DP4+ and J-DP4, as well as the conventional NOESY approach.