Chemistry and Selective Tumor Cell Growth Inhibitory Activity of Polyketides from the South China Sea Sponge Plakortis sp.

Simplextone E (1), a new metabolite of polyketide origin, was isolated with eight known analogues (2–9) from the South China Sea sponge Plakortis sp. The relative configuration of the new compound was elucidated by a detailed analysis of the spectroscopic data and quantum mechanical calculation of NMR chemical shifts, aided by the newly reported DP4+ approach. Its absolute configuration was determined by the TDDFT/ECD calculation. Simplextone E (1) is proven to be one of the isomers of simplextone D. The absolute configuration at C-8 in alkyl chain of plakortone Q (2) was also assigned based on the NMR calculation. In the preliminary in vitro bioassay, compounds 6 and 7 showed a selective growth inhibitory activity against HCT-116 human colon cancer cells with IC50 values of 8.3 ± 2.4 and 8.4 ± 2.3 μM, corresponding to that of the positive control, adriamycin (IC50 4.1 μM). The two compounds also showed selective activities towards MCF-7 human breast cancer and K562 human erythroleukemia cells while compound 3 only displayed weak activity against K562 cells.

Mar. Drugs 2017, 15, 129 2 of 13 polyketides leads to a challenge in the absolute configuration identification.The absolute configurations were determined at an early stage mainly by chemical methods, such as chemical degradation [30] and total synthesis [29,31,32].In recent years, the calculating of NMR parameters provided a strong and convenient approach for the assignment of relative or absolute configuration of these highly flexible systems [33][34][35].
Simplextone E (1) was isolated as a colorless oil.The molecular formula of 1 was determined as C 17 H 30 O 4 from the HRESIMS data (316.2485[M + NH 4 ] + ) and it required three degrees of unsaturation.The IR spectrum of 1 indicated the presence of hydroxy (3438 cm −1 ) and ester carbonyl (1749 cm −1 ) groups.The 13 C NMR and DEPT data (Table 1) displayed seventeen carbon signals, including one sp 2 carbon atom (one ester carbonyl group) at a lower field and sixteen sp 3 carbon atoms at a higher field (four methyl, six methylenes, three methines, one oxygenated methine, two oxygenated tertiary carbons), accounting for one double bond equivalents.The remaining double bond equivalents were due to the presence of two rings in the molecule.2).The HMBC correlations from Me-17 to C-7, C-8 and C-9 revealed the linkage of moieties b and c by C-8-C-9 bond.The HMBC correlations from H-5, H-7, and Me-16 to C-6 and from H-5 and H-15 to C-7 give a connection of fragments b, c, e by the oxygenated tertiary carbon C-6, leading to the formation of a cyclopentane ring.The HMBC correlations from H-2 and Me-14 to C-4, and from H-5 to C-3, C-4, and C-13 led the connection of a, d, and the cyclopentane ring by another oxygenated tertiary carbon C-4.The HMBC correlation from H-2 to C-1 indicated the ester carbonyl group was connected to fragment a.The remaining one double bond equivalent in the molecule allowed for the linkage from the ester carbon C-1 to the tertiary carbon C-4 by an oxygen bridge to form a γ-butyrolactone ring.The planar structure of simplextone E (1) was then established.The assigned planar structure of 1 was identical to that of simplextone D [12], which was previously The relative configuration of 1 was determined by 1D and 2D NOE experiments (Figure 3).The distinct NOESY cross peak between H-3 and H-5 indicated the same orientation of these protons.This observation was in agreement with the NOE between H-3 and H-2α and H-5 in the NOE difference spectrum (Figure S9).The NOE correlation Between H-2β and Me-14 in the NOESY spectrum supported the above conclusion.On the other hand, the NOE correlations of Me-16 with H-5 and H-9, and of H-9 with Me-17 indicated that these groups were oriented on the same side of the cyclopentane ring (Figure 3).Thus, the relative configurations of γ-butyrolactone and cyclopentane ring were determined, respectively.However, the relative configuration of the two conjoined bicyclic systems could not be identified unambiguously by NMR data because of the free rotating nature of C-4/C-5 single bond.Hence, quantum mechanical calculation of 13 C NMR chemical shifts and DP4+ method were applied to determine the relative configuration of 1.There were two possible candidate configurations of 1 (1a and 1b in Figure 4).An extensive conformational search at the empirical level was carried out for 1a-1b (see Computational Details, Experimental Section).Then, the obtained conformers were submitted to a geometry and energy optimization phase at the density functional level (DFT), and afterwards 13 C and 1 H NMR chemical shifts were predicted for 1a-1b at the DFT.Mean absolute error (MAE) values were used to compare calculated and experimental 13 C and 1 H NMR chemical shifts (see Table 2 and Figure 5).Although mean absolute errors (MAE) values of 1 H NMR and 13 C NMR were not in accordance, 13 C NMR MAE values clearly pointed out 1a as the most probable isomer (MAE values: 1.83 for 1a, 2.14 for 1b) (Figure 5).To unambiguously assign the relative configuration of 1, we then computed the DP4+ probabilities [42] (Table 2) for the two possible considered stereoisomers (1a-1b).Specifically, the best DP4+ performance in predicting the correct stereochemistry of organic compounds has been associated to the combination of both 1 H and 13 C chemical shift data.In fact, when 1 H/ 13 C DP4+ probabilities point to two different stereoisomers causing the uncertainty in assigning the relative configuration, the combination of both has been shown to determine a clear enhancement of DP4+ performance, indicating that data from the two The relative configuration of 1 was determined by 1D and 2D NOE experiments (Figure 3).The distinct NOESY cross peak between H-3 and H-5 indicated the same orientation of these protons.This observation was in agreement with the NOE between H-3 and H-2α and H-5 in the NOE difference spectrum (Figure S9).The NOE correlation Between H-2β and Me-14 in the NOESY spectrum supported the above conclusion.On the other hand, the NOE correlations of Me-16 with H-5 and H-9, and of H-9 with Me-17 indicated that these groups were oriented on the same side of the cyclopentane ring (Figure 3).Thus, the relative configurations of γ-butyrolactone and cyclopentane ring were determined, respectively.The relative configuration of 1 was determined by 1D and 2D NOE experiments (Figure 3).The distinct NOESY cross peak between H-3 and H-5 indicated the same orientation of these protons.This observation was in agreement with the NOE between H-3 and H-2α and H-5 in the NOE difference spectrum (Figure S9).The NOE correlation Between H-2β and Me-14 in the NOESY spectrum supported the above conclusion.On the other hand, the NOE correlations of Me-16 with H-5 and H-9, and of H-9 with Me-17 indicated that these groups were oriented on the same side of the cyclopentane ring (Figure 3).Thus, the relative configurations of γ-butyrolactone and cyclopentane ring were determined, respectively.However, the relative configuration of the two conjoined bicyclic systems could not be identified unambiguously by NMR data because of the free rotating nature of C-4/C-5 single bond.Hence, quantum mechanical calculation of 13 C NMR chemical shifts and DP4+ method were applied to determine the relative configuration of 1.There were two possible candidate configurations of 1 (1a and 1b in Figure 4).An extensive conformational search at the empirical level was carried out for 1a-1b (see Computational Details, Experimental Section).Then, the obtained conformers were submitted to a geometry and energy optimization phase at the density functional level (DFT), and afterwards 13 C and 1 H NMR chemical shifts were predicted for 1a-1b at the DFT.Mean absolute error (MAE) values were used to compare calculated and experimental 13 C and 1 H NMR chemical shifts (see Table 2 and Figure 5).Although mean absolute errors (MAE) values of 1 H NMR and 13 C NMR were not in accordance, 13 C NMR MAE values clearly pointed out 1a as the most probable isomer (MAE values: 1.83 for 1a, 2.14 for 1b) (Figure 5).To unambiguously assign the relative configuration of 1, we then computed the DP4+ probabilities [42] (Table 2) for the two possible considered stereoisomers (1a-1b).Specifically, the best DP4+ performance in predicting the correct stereochemistry of organic compounds has been associated to the combination of both 1 H and 13 C chemical shift data.In fact, when 1 H/ 13 C DP4+ probabilities point to two different stereoisomers causing the uncertainty in assigning the relative configuration, the combination of both has been shown to determine a clear enhancement of DP4+ performance, indicating that data from the two However, the relative configuration of the two conjoined bicyclic systems could not be identified unambiguously by NMR data because of the free rotating nature of C-4/C-5 single bond.Hence, quantum mechanical calculation of 13 C NMR chemical shifts and DP4+ method were applied to determine the relative configuration of 1.There were two possible candidate configurations of 1 (1a and 1b in Figure 4).An extensive conformational search at the empirical level was carried out for 1a-1b (see Computational Details, Experimental Section).Then, the obtained conformers were submitted to a geometry and energy optimization phase at the density functional level (DFT), and afterwards 13 C and 1 H NMR chemical shifts were predicted for 1a-1b at the DFT.Mean absolute error (MAE) values were used to compare calculated and experimental 13 C and 1 H NMR chemical shifts (see Table 2 and Figure 5).Although mean absolute errors (MAE) values of 1 H NMR and 13 C NMR were not in accordance, 13 C NMR MAE values clearly pointed out 1a as the most probable isomer (MAE values: 1.83 for 1a, 2.14 for 1b) (Figure 5).To unambiguously assign the relative configuration of 1, we then computed the DP4+ probabilities [42] (Table 2) for the two possible considered stereoisomers (1a-1b).Specifically, the best DP4+ performance in predicting the correct stereochemistry of organic compounds has been associated to the combination of both 1 H and 13 C chemical shift data.In fact, when 1 H/ 13 C DP4+ probabilities point to two different stereoisomers causing the uncertainty in assigning the relative configuration, the combination of both has been shown to determine a clear enhancement of DP4+ performance, indicating that data from the two nuclei should be accounted for, when available.In our case, all data DP4+ corroborated 1a as the most probable relative stereoisomer, and accordingly, the relative configuration of 1 was established as 3S*, 4R*, 5S*, 6R*, 8R*, 9S*.nuclei should be accounted for, when available.In our case, all data DP4+ corroborated 1a as the most probable relative stereoisomer, and accordingly, the relative configuration of 1 was established as 3S*, 4R*, 5S*, 6R*, 8R*, 9S*.Once the relative configuration of 1 was proposed, we then determined the absolute configuration applying to the TDDFT/ECD calculation [37,43].The conformers of 1a produced in the above reported step of NMR QM calculations were then submitted to a further optimization step at the DFT, reproducing experimental solvent effects (CH3CN) and using the integral equation formalism version of the polarizable continuum model (IEFPCM).We built the final ECD spectrum for the identified diastereoisomer, considering the influence of each conformer on the total  nuclei should be accounted for, when available.In our case, all data DP4+ corroborated 1a as the most probable relative stereoisomer, and accordingly, the relative configuration of 1 was established as 3S*, 4R*, 5S*, 6R*, 8R*, 9S*.Once the relative configuration of 1 was proposed, we then determined the absolute configuration applying to the TDDFT/ECD calculation [37,43].The conformers of 1a produced in the above reported step of NMR QM calculations were then submitted to a further optimization step at the DFT, reproducing experimental solvent effects (CH3CN) and using the integral equation formalism version of the polarizable continuum model (IEFPCM).We built the final ECD spectrum Once the relative configuration of 1 was proposed, we then determined the absolute configuration applying to the TDDFT/ECD calculation [37,43].The conformers of 1a produced in the above reported step of NMR QM calculations were then submitted to a further optimization step at the DFT, reproducing experimental solvent effects (CH 3 CN) and using the integral equation formalism version of the polarizable continuum model (IEFPCM).We built the final ECD spectrum for the identified diastereoisomer, considering the influence of each conformer on the total Boltzmann distribution and taking into account the relative energies.The experimental ECD curve aligned well with the calculated ECD curve for 3S, 4R, 5S, 6R, 8R, 9S isomer (Figure 6), and this allowed us to assign the absolute configuration for compound 1.
Mar. Drugs 2017, 15, 129 6 of 13 aligned well with the calculated ECD curve for 3S, 4R, 5S, 6R, 8R, 9S isomer (Figure 6), and this allowed us to assign the absolute configuration for compound 1.Compound 2 was isolated as a colorless oil.The HRESIMS and NMR spectroscopic data of 2 reveal the identical structure of plakortone Q, recently obtained from the Chinese sponge Plakortis simplex [13].However, the absolute configuration of C-8 in the alkyl chain was not determined in the literature.We then applied the above reported method to assign the absolute configuration of C-8.There were two possible absolute configurations of C-8, 8R and 8S.Thus, there were two candidate stereostructures of 2 (2-8R and 2-8S in Figure 4).Therefore, the configuration of C-8 of 2 was assigned as S by using the quantum mechanical calculation of 1 H and 13 C NMR chemical shifts ( 13 C MAE values: 1.52 for 2-8R, 1.44 for 2-8S; 1 H MAE values: 0.14 for 2-8R, 0.13 for 2-8S, Figure 7), this was also confirmed by the analysis of all data DP4+ probabilities as previously mentioned in 1 (Table 3).Finally, the TDDFT/ECD calculations were also performed starting from conformers of 2-8S produced in the QM step of computation of the NMR chemical shift data, and they were submitted to another round of geometry optimization at the DFT in CH3CN IEFPCM.The obtained curves for the two possible enantiomeric species superimposed with the experimental one (Figure 8) disclosed 3S, 4S, 5S, 6S, 8S as the absolute configuration of 2, thus conforming the stereochemistry pattern at the bicyclic portion as previously reported [13].Compound 2 was isolated as a colorless oil.The HRESIMS and NMR spectroscopic data of 2 reveal the identical structure of plakortone Q, recently obtained from the Chinese sponge Plakortis simplex [13].However, the absolute configuration of C-8 in the alkyl chain was not determined in the literature.We then applied the above reported method to assign the absolute configuration of C-8.There were two possible absolute configurations of C-8, 8R and 8S.Thus, there were two candidate stereostructures of 2 (2-8R and 2-8S in Figure 4).Therefore, the configuration of C-8 of 2 was assigned as S by using the quantum mechanical calculation of 1 H and 13 C NMR chemical shifts ( 13 C MAE values: 1.52 for 2-8R, 1.44 for 2-8S; 1 H MAE values: 0.14 for 2-8R, 0.13 for 2-8S, Figure 7), this was also confirmed by the analysis of all data DP4+ probabilities as previously mentioned in 1 (Table 3).Finally, the TDDFT/ECD calculations were also performed starting from conformers of 2-8S produced in the QM step of computation of the NMR chemical shift data, and they were submitted to another round of geometry optimization at the DFT in CH 3 CN IEFPCM.The obtained curves for the two possible enantiomeric species superimposed with the experimental one (Figure 8) disclosed 3S, 4S, 5S, 6S, 8S as the absolute configuration of 2, thus conforming the stereochemistry pattern at the bicyclic portion as previously reported [13].3).Finally, the TDDFT/ECD calculations were also performed starting from conformers of 2-8S produced in the QM step of computation of the NMR chemical shift data, and they were submitted to another round of geometry optimization at the DFT in CH3CN IEFPCM.The obtained curves for the two possible enantiomeric species superimposed with the experimental one (Figure 8) disclosed 3S, 4S, 5S, 6S, 8S as the absolute configuration of 2, thus conforming the stereochemistry pattern at the bicyclic portion as previously reported [13].3. Figure 7. 13 C (red bars) and 1 H (green bars) mean absolute errors (MAE) histograms related to compounds 2-8R and 2-8S, as indicated in Table 3.

General Experimental Procedures
Column chromatography (CC) was performed on silica gel (200-300; 400-500 mesh, Yantai, China), RP silica gel (43-60 µm, Merck, Darmstadt, Germany) and Sephadex LH-20 (GE Healthcare Bio-Sciences AB, SE-751 84 Uppsala, Sweden).TLC was performed on precoated silica gel plates (HSGF-254, Yantai, China) and RP silica gel (RP-18 F254, Macherey-Nagel, Düren, Germany).Anisaldehyde-sulphuric acid reagent was used for detecting spots on TLC.An Agilent 1100 system (refractive index detector, YMC Pack ODS-A column (10 × 250 mm, 5 µm)) was put to use to carry on HPLC purification.The NMR data were measured on Bruker DRX 400 at 300 K. Parts per million (δ) was used to report chemical shifts applying residual CDCl 3 signal as an internal standard ( 1 H NMR: δ H 7.26 ppm, 13 C NMR: δ C 77.00 ppm, Hz for coupling constants (J)).Autopol IV polarimeter was used to record the optical rotations in CH 2 Cl 2 at the sodium D line (590 nm).Nexus 470 FT-IR spectrophotometer (Nicolet, Ramsey, MN, USA) was applied to measure the infrared spectra of compounds in thin polymer films and peaks are reported in cm −1 .Varian Cary 100 UV-Vis spectrophotometer was adopted to record UV absorption of compounds and wavelengths are reported in nm.JASCO J-810 circular dichroism spectropolarimeter was used to measure circular dichroism spectra.A Q-TOF micro mass spectrometer was utilized to measure the HRESIMS, and the reference compound was sodium iodide dissolved in isopropyl alcohol (2 mg/mL).

Animal Material
The marine sponge Plakortis sp.(1.1 kg, dry weight) was collected near Yongxing Island in the South China Sea in November 2011 and identified by Yalan Zhou.A voucher specimen (LG-10) was deposited in the Second Military Medical University, Shanghai, China.

Extraction and Isolation
The frozen specimen (1.1 kg, dry weight) was cut into small pieces and extracted ultrasonically with acetone (3 L × 3) and MeOH (3 L × 2), respectively.The organic extracts were concentrated under a vacuum to give a residue, which was partitioned between H 2 O and diethyl ether to afford 13.8 g of an Et 2 O extract.The Et 2 O extract was subjected to column chromatography on silica gel to give seven fractions, using petroleum ether and EtOAc (from 100:1 to 0:100) as eluent.Fraction 1 was chromatographed on a silica gel column eluting with a petroleum ether/acetone solvent gradient system (from 100:1 to 0:1), to give 10 subfractions (Fr.1-1 to Fr.1-10).Fr.1-8 (1.08 g) was

Computational Details
The chemical structures of compounds 1a, 1b, 2-8R, 2-8S were built by Maestro 10.2 [44], and optimized by MacroModel 10.2 [45] with the OPLS force field [46] and the Polak-Ribier conjugate gradient algorithm (PRCG, maximum derivative less than 0.001 kcal/mol).Conformational search rounds for the above mentioned compounds were performed by MacroModel 10.2 [44,45] at the empirical molecular mechanics (MM) level, with Monte Carlo Multiple Minimum (MCMM) method and Low Mode Conformational Search (LMCS) method.Also, molecular dynamic simulations were performed at 450, 600, 700, 750 K, with a time step of 2.0 fs, an equilibration time of 0.1 ns, and a simulation time of 10 ns.All the produced conformers were then analyzed, and non-redundant conformers were selected by using the "Redundant Conformer Elimination" module of Macromodel 10.2 [44].Then, the obtained conformers were optimized at quantum mechanical (QM) level by using the MPW1PW91 functional and the 6-31G(d) basis set.The selected conformers accounted for the subsequent computation of the 13 C and 1 H NMR chemical shifts, using the MPW1PW91 functional and the 6-31G(d,p) basis set.The final 13 C NMR and 1 H NMR spectra were built considering the influence of each conformer on the total Boltzmann distribution and taking into account the relative energies.All the 13 C and 1 H NMR calculated chemical shifts were scaled to tetramethylsilane (TMS).
Experimental and calculated 13 C and 1 H NMR chemical shifts were compared in detail, computing the ∆δ parameter (see Tables S1-S4, Supplementary Materials): where, δ exp (ppm) and δ calc (ppm) are the 13 C/ 1 H experimental and calculated chemical shifts, respectively.
The mean absolute errors (MAEs) for all the possible diastereoisomers (see Tables 2 and 3 and Tables S1-S4, Supplementary Materials) were computed: MAE = ∑ (∆δ)/n defined as the summation through n of the absolute error values (difference of the absolute values between corresponding experimental and 13 C-1 H chemical shifts), normalized to the number of the chemical shifts considered.
Furthermore, DP4+ probabilities related to all the considered stereoisomers of 1a, 1b, 2-8R, 2-8S were computed considering both 13 C and 1 H chemical shifts, and they were compared with the related experimental data, using the available DP4+ Toolbox (Excel file).
Once obtained, the relative configurations of the investigated compounds and the prediction of CD spectra were performed using all the conformers previously obtained from DFT calculations, by submitting them to another round of geometry and energy optimization at DFT, in acetonitrile IEFPCM.Afterwards, QM calculations were performed at TDDFT (NStates = 40) MPW1PW91/6-31g(d,p) level, in acetonitrile IEFPCM to reproduce the experimental solvent environment.The final CD spectra for both the enantiomers related to the determined relative stereoisomers were built considering the influence of each conformer on the total Boltzmann distribution while taking into account the relative energies, and were graphically plotted using SpecDis software [47].In order to simulate the experimental CD curve, a Gaussian band-shape function was applied with the exponential half-width (σ/γ) of 0.20 eV.All QM calculations were performed using Gaussian 09 software package [48].

Cytotoxicity Assay
HCT116, MCF-7 and SMMC-7721 cell lines were grown in DMEM medium supplemented with 10% FBS (fetal bovine serum), K562 cell line was cultured in RPMI 1640 and supplemented with 10% FBS.All cell lines were incubated at 37 • C with 5% CO 2 in air atmosphere.The density of tumor cells plated in a 96-well plate was 1 × 10 4 cells/well.The cells were treated with the isolated compounds in six concentration gradients (threefold dilution starting from 100 µM dissolved in 1% DMSO in the final cell medium), and the cells treated with medium containing 1% DMSO served as a control.Adriamycin was used as positive control.Triplicate wells with untreated cells were set up to serve as a vehicle control.After 48 h, the relative cell viability was assayed by CellTiter-Blue ® Cell viability assay kit, and the fluorescence was recorded in a microplate reader (Synergy 2, BioTek Instruments, Inc., Winooski, VT, USA) at 530/590 nm.The IC 50 was determined by fitting the relative cell viability curve by a dose-response model in the Prism program from GraphPad Software [49].The in vitro cytotoxicity assay was performed three times with triplicates in each experiment.p values were calculated using the Student's t test, and a probability of 0.05 or less was considered statistically significant.IC 50 results are given as the mean of three independent experiments in triplicate in each experiment.

Conclusions
A new polyketide, simplextone E (1), together with eight known analogues (2-9), were isolated from the South China Sea sponge Plakortis sp.Polyketides of this family contain flexible moieties that may rotate freely around a single bond.Determination of the absolute configuration of these segments is one of the most challenging parts in structural elucidation of natural products.Recently, quantum mechanical calculation has provided novel and reliable approaches to resolve this tricky issue.Calculation for 13 C and 1 H NMR chemical shifts gives relative configuration and the TDDFT/ECD calculation leads to the assignment of the absolute configuration.The research gives an example for structure determination of flexible molecules by combination of spectroscopic analysis and quantum mechanical calculation.In the biotest in vitro, compounds 6 and 7 with peroxide moiety showed a potent growth inhibitory effect against HCT-116 cells with similar IC 50 values as that of the positive control adriamycin, while polyketides with lactone fragment showed weak cytotoxic activity.Compounds 6 and 7 also showed selective activity towards MCF-7 and K562 cells, while compound 3 displayed weak activity against K562 cells.All the tested compounds were inactive towards SMMC-7721 cell lines.The tumor cell growth inhibitory activity of compounds 6 and 7 may be attributed to the peroxide moiety in the structure, as reported previously [50].

Figure 9 .
Figure 9. Histograms of relative cell viability percentage for compounds 6 and 7 against HCT-116 cell lines.

Figure 9 .
Figure 9. Histograms of relative cell viability percentage for compounds 6 and 7 against HCT-116 cell lines.

Table 2 .
The calculation results of compound 1, with mean absolute errors (MAE) values and DP4+ probabilities.MAE Values (

Table 2 .
The calculation results of compound 1, with mean absolute errors (MAE) values and DP4+ probabilities.

Table 2 .
The calculation results of compound 1, with mean absolute errors (MAE) values and DP4+ probabilities.MAE Values (

Table 3 .
The calculation results of compound 2, with MAE values and DP4+ probabilities.

Table 3 .
The calculation results of compound 2, with MAE values and DP4+ probabilities.