Chromone-Derived Polyketides from the Deep-Sea Fungus Diaporthe phaseolorum FS431

Five new chromone-derived polyketides phaseolorins A-F (1–5), together with nine known compounds, were isolated from the deep-sea derived fungus Diaporthe phaseolorum FS431. The structures of new compounds were determined by analysis of their NMR and high-resolution electrospray ionization mass spectroscopy (HRESIMS) spectroscopic data. The absolute configurations were confirmed by chemical transformations, extensively experimental electron capture detection (ECD) calculations, or X-ray crystallography. Among them, compound 2 represented the first example for a new family of chromone derivative possessing an unprecedented recombined five-member γ-lactone ring. Moreover, the new compounds (1–5) were evaluated for in vitro cytotoxic activities against a panel of human cancer cell lines.


Introduction
Marine-derived fungi are emerging as a promising source for discovering novel natural products with significant biological and pharmacological properties. Deep-sea fungi have attracted even more attention in recent years because they are naturally forced to metabolize more potent bioactive compounds to acclimatize the extreme and variable environments of the deep sea, which has resulted in the generation of numerous chemically diverse and structurally unique secondary metabolites. Therefore, deep-sea fungi have been proverbially respected to be one of the most potential resources for the discovery of biologically meaningful natural products [1,2].

Structure Elucidation
Compound 1 was obtained as a yellow powder. Its molecular formula was established as C15H16O7 on the basis of its high-resolution electrospray ionization mass spectroscopy (HRESIMS) with the protonated molecular peak discovered at m/z 309.0971 ([M + H] + , calcd 309.0969), requiring eight unsaturation degrees. The IR spectrum displayed the classical absorption bands at 3362, 1771, and 1614 cm −1 , which were characteristic for the hydroxy, γ-lactone carbonyl, and conjugated ketone carbonyl functionalities, respectively. The 1 H NMR spectrum of 1 exhibited a series of proton resonances responsive for two hydroxymethyls at δH 3.78 (m, H-14) and 3.59 (m, H-13) and three aromatic ones of a 1,2,3-trisubstituted benzene at δH 6.43 (d, J = 8. 3 Hz, H-6 and H-8) and 7.38 (t, J =

Structure Elucidation
Compound 1 was obtained as a yellow powder. Its molecular formula was established as C 15 H 16 O 7 on the basis of its high-resolution electrospray ionization mass spectroscopy (HRESIMS) with the protonated molecular peak discovered at m/z 309.0971 ([M + H] + , calcd 309.0969), requiring eight unsaturation degrees. The IR spectrum displayed the classical absorption bands at 3362, 1771, and 1614 cm −1 , which were characteristic for the hydroxy, γ-lactone carbonyl, and conjugated ketone . The 13 C NMR spectrum showed two typical carbon resonances at δ C 179.0 and 198.1, which were unambiguously supported the presence of a γ-lactone and ketone carbonyl functional groups, respectively.
With a careful inspection and analyses of the 13 C NMR data (Table 1) in conjunction with HSQC spectrum, it could readily reveal the existence of 15 carbon resonances including four methylenes, five methines, and six nonprotonated carbons. Notably, an elaborative comparison of its 1D (Table 1) and 2D NMR spectra data of 1 with those of the known compound mangorvamide H [23] disclosed that compound 1 should also share the similar chromone-derived skeleton. The major differences between 1 and mangorvamide H were attributed to the replacements of the C-13 and C-14 methyls in the known compound with hydroxymethyls in 1, and the assumption could be rationally verified by the 13 C NMR chemical shifts (δ C 64.6 (C-13), 62.9 (C-14)) and the critical HMBC correlations from H 2 -14 (δ H 3.78 (m)) to C-2 and C-3 as well as H 2 -13 (δ H 3.59 (m)) to C-9, C-10, and C-11 ( Figure 2 With a careful inspection and analyses of the 13 C NMR data (Table 1) in conjunction with HSQC spectrum, it could readily reveal the existence of 15 carbon resonances including four methylenes, five methines, and six nonprotonated carbons. Notably, an elaborative comparison of its 1D (Table 1) and 2D NMR spectra data of 1 with those of the known compound mangorvamide H [23] disclosed that compound 1 should also share the similar chromone-derived skeleton. The major differences between 1 and mangorvamide H were attributed to the replacements of the C-13 and C-14 methyls in the known compound with hydroxymethyls in 1, and the assumption could be rationally verified by the 13 C NMR chemical shifts (δC 64.6 (C-13), 62.9 (C-14)) and the critical HMBC correlations from H2-14 (δH 3.78 (m)) to C-2 and C-3 as well as H2-13 (δH 3.59 (m)) to C-9, C-10, and C-11 ( Figure 2). The relative configuration of compound 1 was established with the aid of the single-crystal Xray diffraction experiment, which was successfully carried out using CuKα radiation ( Figure 3). Although it unfortunately failed to clarify its absolute configuration, it unambiguously revealed the planar and relative configuration of compound 1 as depicted in Figure 1. The absolute configuration of 1 was finally determined by the calculated CD spectrum method, as expected, it provided a satisfying agreement between the calculated CD spectrum for the 2S,9S,10R isomer and the experimental one ( Figure 4). Therefore, the structure elucidation of compound 1 was completely finished, and its absolute structure was deduced to be 2S,9S,10R and trivially named as phaseolorin A.  The relative configuration of compound 1 was established with the aid of the single-crystal X-ray diffraction experiment, which was successfully carried out using CuKα radiation ( Figure 3). Although it unfortunately failed to clarify its absolute configuration, it unambiguously revealed the planar and relative configuration of compound 1 as depicted in Figure 1. The absolute configuration of 1 was finally determined by the calculated CD spectrum method, as expected, it provided a satisfying agreement between the calculated CD spectrum for the 2S,9S,10R isomer and the experimental one ( Figure 4). Therefore, the structure elucidation of compound 1 was completely finished, and its absolute structure was deduced to be 2S,9S,10R and trivially named as phaseolorin A. . The 13 C NMR spectrum showed two typical carbon resonances at δC 179.0 and 198.1, which were unambiguously supported the presence of a γ-lactone and ketone carbonyl functional groups, respectively. With a careful inspection and analyses of the 13 C NMR data (Table 1) in conjunction with HSQC spectrum, it could readily reveal the existence of 15 carbon resonances including four methylenes, five methines, and six nonprotonated carbons. Notably, an elaborative comparison of its 1D (Table 1) and 2D NMR spectra data of 1 with those of the known compound mangorvamide H [23] disclosed that compound 1 should also share the similar chromone-derived skeleton. The major differences between 1 and mangorvamide H were attributed to the replacements of the C-13 and C-14 methyls in the known compound with hydroxymethyls in 1, and the assumption could be rationally verified by the 13 C NMR chemical shifts (δC 64.6 (C-13), 62.9 (C-14)) and the critical HMBC correlations from H2-14 (δH 3.78 (m)) to C-2 and C-3 as well as H2-13 (δH 3.59 (m)) to C-9, C-10, and C-11 ( Figure 2). The relative configuration of compound 1 was established with the aid of the single-crystal Xray diffraction experiment, which was successfully carried out using CuKα radiation ( Figure 3). Although it unfortunately failed to clarify its absolute configuration, it unambiguously revealed the planar and relative configuration of compound 1 as depicted in Figure 1. The absolute configuration of 1 was finally determined by the calculated CD spectrum method, as expected, it provided a satisfying agreement between the calculated CD spectrum for the 2S,9S,10R isomer and the experimental one ( Figure 4). Therefore, the structure elucidation of compound 1 was completely finished, and its absolute structure was deduced to be 2S,9S,10R and trivially named as phaseolorin A.   Compound 2 was also purified as yellow powder. It has the same molecular formula C15H16O7 as that of 1 based on the positive HRESIMS (m/z 331.0794 [M + Na] + ; calcd for C15H16O7Na, 331.0788). Analysis of its 1 H and 13 C NMR data revealed similar structure features to those of 1, which in turn gave rise to the conclusive information for the structural isomeric relationship between 1 and 2. The further careful interpretation of the HMBC correlations could reconfirm this deduction. The major differences between 1 and 2 were the noticeable chemical shifts of C-9 (δC 84.4 in 1; 73.1 in 2) and C-13 (δC 64.6 in 1; 71.5 in 2) together with the critical HMBC correlation from H2-13 to C-12 (δC 177.1) in 2, indicating that they might be a pair of regioisomers with the formation of a new five-member γlactone ring in compound 2. Notably, compound 2 was discovered as the first natural product possessing such a five-member γ-lactone ring from nature. Furthermore, the experimental electron capture detection (ECD) spectrum of 2 ( Figure S18) was very similar to that of compound 1, which showed good accordance with the theoretical one with obvious Cotton effects found at 353 nm. The aforementioned informative results strongly suggested that the absolute stereogenic centers of compound 2 might be the same as those of compound of 1 Compound 2 was also purified as yellow powder. It has the same molecular formula C 15 H 16 O 7 as that of 1 based on the positive HRESIMS (m/z 331.0794 [M + Na] + ; calcd for C 15 H 16 O 7 Na, 331.0788). Analysis of its 1 H and 13 C NMR data revealed similar structure features to those of 1, which in turn gave rise to the conclusive information for the structural isomeric relationship between 1 and 2.
The further careful interpretation of the HMBC correlations could reconfirm this deduction. The major differences between 1 and 2 were the noticeable chemical shifts of C-9 (δ C 84.4 in 1; 73.1 in 2) and C-13 (δ C 64.6 in 1; 71.5 in 2) together with the critical HMBC correlation from H 2 -13 to C-12 (δ C 177.1) in 2, indicating that they might be a pair of regioisomers with the formation of a new five-member γ-lactone ring in compound 2. Notably, compound 2 was discovered as the first natural product possessing such a five-member γ-lactone ring from nature. Furthermore, the experimental electron capture detection (ECD) spectrum of 2 ( Figure S18) was very similar to that of compound 1, which showed good accordance with the theoretical one with obvious Cotton effects found at 353 nm. The aforementioned informative results strongly suggested that the absolute stereogenic centers of compound 2 might be the same as those of compound of 1 and be rationally clarified as 2S,9S,10R. Moreover, the compounds 1 and 2 should share the same biosynthetic pathway, which could further strengthen this conclusion. In order to achieve the direct evidence to unambiguously determine the absolute configuration of compound 2, the further chemical transformations between compounds 1 and 2 using the various chemical reactions were performed.
After an extensive screen of chemical transformations in different base or acid conditions, the compound 2 was disclosed to be readily transformed to compound 1 under the trifluoroacetic acid (TFA) condition with the CH 2 Cl 2 /THF (3:1) as the combined solvent (60% yield, 24 h) or under the neat state in air condition at room temperature (80% yield, 30 d); whereas the compound 2 could be correspondingly transformed into compound 1 under a methanol solution of 0.25 M NaOH as the strong based condition (15% yield, 6 h) ( Figure 5). The characteristic for the mutual transformation between compounds 1 and 2 thus successfully established that the absolute configuration of compound 2 was the same with that of compound 1 and assigned as 2S,9S,10R. Therefore, the absolute structure of compound 2 was completely established, and it represented a new family of chromone derivatives and was given the trial name as phaseolorin B.
Mar. Drugs 2019, 17, x FOR PEER REVIEW 5 of 12 and be rationally clarified as 2S,9S,10R. Moreover, the compounds 1 and 2 should share the same biosynthetic pathway, which could further strengthen this conclusion. In order to achieve the direct evidence to unambiguously determine the absolute configuration of compound 2, the further chemical transformations between compounds 1 and 2 using the various chemical reactions were performed.
After an extensive screen of chemical transformations in different base or acid conditions, the compound 2 was disclosed to be readily transformed to compound 1 under the trifluoroacetic acid (TFA) condition with the CH2Cl2/THF (3:1) as the combined solvent (60% yield, 24 h) or under the neat state in air condition at room temperature (80% yield, 30 d); whereas the compound 2 could be correspondingly transformed into compound 1 under a methanol solution of 0.25 M NaOH as the strong based condition (15% yield, 6 h) ( Figure 5). The characteristic for the mutual transformation between compounds 1 and 2 thus successfully established that the absolute configuration of compound 2 was the same with that of compound 1 and assigned as 2S,9S,10R. Therefore, the absolute structure of compound 2 was completely established, and it represented a new family of chromone derivatives and was given the trial name as phaseolorin B.  Compound 3 was isolated as yellow oil. The molecular formula was absolutely assigned as C17H18O8, attributable to the positive ion mode HRESIMS (m/z 351.1072 [M + H] + (calcd for C17H19O8, 351.1074), corresponding to nine degrees of unsaturation. The 1 H and 13 C NMR spectra of 3 closely resembled to those of 1 and 2, collectively pointing to the reality of a chromone-derived derivative for compound 3 as those for compounds 1 and 2. The major differences between the compounds 3 and 1 were attributed to the existence of two additional carbons (δC 171.0, 20.7) in 3 (Table 2), which were characteristically responsive for an acetyl group. As referring to the 2D NMR spectra, the key HMBC correlations from H3-16 to C-15 and H2-14 to C-15 could be obviously distinguished, indicating the acetoxy ought to be located at C-14. The absolute configuration of 3 was theoretically deduced to be the same as that of 1 with the aid of the calculated CD spectrum method, which expectedly launched a calculated CD spectrum of the 2S,9S,10R isomer perfectly matched with the experimental one ( Figure 6). Therefore, the configuration of 3 was conclusively assigned as shown in Figure 1 and given the tentative name as phaseolorin C.  Compound 3 was isolated as yellow oil. The molecular formula was absolutely assigned as C 17  .1074), corresponding to nine degrees of unsaturation. The 1 H and 13 C NMR spectra of 3 closely resembled to those of 1 and 2, collectively pointing to the reality of a chromone-derived derivative for compound 3 as those for compounds 1 and 2. The major differences between the compounds 3 and 1 were attributed to the existence of two additional carbons (δ C 171.0, 20.7) in 3 (Table 2), which were characteristically responsive for an acetyl group. As referring to the 2D NMR spectra, the key HMBC correlations from H 3 -16 to C-15 and H 2 -14 to C-15 could be obviously distinguished, indicating the acetoxy ought to be located at C-14. The absolute configuration of 3 was theoretically deduced to be the same as that of 1 with the aid of the calculated CD spectrum method, which expectedly launched a calculated CD spectrum of the 2S,9S,10R isomer perfectly matched with the experimental one ( Figure 6). Therefore, the configuration of 3 was conclusively assigned as shown in Figure 1 and given the tentative name as phaseolorin C. and be rationally clarified as 2S,9S,10R. Moreover, the compounds 1 and 2 should share the same biosynthetic pathway, which could further strengthen this conclusion. In order to achieve the direct evidence to unambiguously determine the absolute configuration of compound 2, the further chemical transformations between compounds 1 and 2 using the various chemical reactions were performed.
After an extensive screen of chemical transformations in different base or acid conditions, the compound 2 was disclosed to be readily transformed to compound 1 under the trifluoroacetic acid (TFA) condition with the CH2Cl2/THF (3:1) as the combined solvent (60% yield, 24 h) or under the neat state in air condition at room temperature (80% yield, 30 d); whereas the compound 2 could be correspondingly transformed into compound 1 under a methanol solution of 0.25 M NaOH as the strong based condition (15% yield, 6 h) ( Figure 5). The characteristic for the mutual transformation between compounds 1 and 2 thus successfully established that the absolute configuration of compound 2 was the same with that of compound 1 and assigned as 2S,9S,10R. Therefore, the absolute structure of compound 2 was completely established, and it represented a new family of chromone derivatives and was given the trial name as phaseolorin B.  Compound 3 was isolated as yellow oil. The molecular formula was absolutely assigned as C17H18O8, attributable to the positive ion mode HRESIMS (m/z 351.1072 [M + H] + (calcd for C17H19O8, 351.1074), corresponding to nine degrees of unsaturation. The 1 H and 13 C NMR spectra of 3 closely resembled to those of 1 and 2, collectively pointing to the reality of a chromone-derived derivative for compound 3 as those for compounds 1 and 2. The major differences between the compounds 3 and 1 were attributed to the existence of two additional carbons (δC 171.0, 20.7) in 3 (Table 2), which were characteristically responsive for an acetyl group. As referring to the 2D NMR spectra, the key HMBC correlations from H3-16 to C-15 and H2-14 to C-15 could be obviously distinguished, indicating the acetoxy ought to be located at C-14. The absolute configuration of 3 was theoretically deduced to be the same as that of 1 with the aid of the calculated CD spectrum method, which expectedly launched a calculated CD spectrum of the 2S,9S,10R isomer perfectly matched with the experimental one ( Figure 6). Therefore, the configuration of 3 was conclusively assigned as shown in Figure 1 and given the tentative name as phaseolorin C.     (Table 3) coupling with the HSQC spectrum of 1 further successfully clarified the existence of 15 carbon resonances corresponding to one methyl, two methylenes, six methines, and six quaternary carbons including a carbonyl one. Furthermore, the 1 H− 1 H correlated spectroscopy (COSY) and HSQC spectra unambiguously disclosed the presence of two spin coupling systems as depicted with bold lines in Figure 7    After a careful inspection of the NMR spectra of 4 with those of the known chromone mangrovamide J [23], it could be readily disclosed that they showed very close similarity in all the NMR spectra with most profiles. The major differences between them were the methyl moiety at C-5a in the known compound mangrovamide J was replaced by a hydroxymethyl one in 4 and the absence of a double bond at C-6 and C-7 positions in 4, which were substantiated by its chemical shifts (δH 4.38; δC 68.5) in conjunction with the HMBC correlations from H2-10 (δH 4.38, (t, J = 2.9 Hz)) to C-5, C-5a, and C-8a as well as H3-11 to C-5, C-6, and C-7. Therefore, the planar structure of 4 was elucidated as shown in Figure 1.
With the aim to grow its single crystals, many tentative efforts by using different solvents or solvent combinations were conducted. Fortunately, a single crystal of 4 was successfully obtained from the methanol solution after repeated attempts. The following single-crystal X-ray diffraction experiment was carried out using CuKα radiation with a Flack parameter of 0.00(4) (Figure 8), which unambiguously verified the absolute configuration of compound 4. Thus, the structure elucidation After a careful inspection of the NMR spectra of 4 with those of the known chromone mangrovamide J [23], it could be readily disclosed that they showed very close similarity in all the NMR spectra with most profiles. The major differences between them were the methyl moiety at C-5a in the known compound mangrovamide J was replaced by a hydroxymethyl one in 4 and the absence of a double bond at C-6 and C-7 positions in 4, which were substantiated by its chemical shifts (δ H 4.38; δ C 68.5) in conjunction with the HMBC correlations from H 2 -10 (δ H 4.38, (t, J = 2.9 Hz)) to C-5, C-5a, and C-8a as well as H 3 -11 to C-5, C-6, and C-7. Therefore, the planar structure of 4 was elucidated as shown in Figure 1.
With the aim to grow its single crystals, many tentative efforts by using different solvents or solvent combinations were conducted. Fortunately, a single crystal of 4 was successfully obtained from the methanol solution after repeated attempts. The following single-crystal X-ray diffraction experiment was carried out using CuKα radiation with a Flack parameter of 0.00(4) (Figure 8), which unambiguously verified the absolute configuration of compound 4. Thus, the structure elucidation of compound 4 was completely finished, and its absolute structure was finally deduced to be 5S,5a S,8S,8aR and trivially named as phaseolorin D.
Mar. Drugs 2019, 17, x FOR PEER REVIEW 7 of 12 of compound 4 was completely finished, and its absolute structure was finally deduced to be 5S,5a S,8S,8aR and trivially named as phaseolorin D. Compound 5 was obtained as yellow oil. Its molecular formula was established as C15H16O7 on the basis of the protonated molecule peak at m/z 331.0788 [M + Na] + in its HRESIMS, requiring eight degrees of unsaturation. The 1D NMR data (Table 3) of 5 were almost in accordance with those of 4, except for the possible replacement of hydroxyl group at C-5 position in 4 by an additional carbonyl functionality (δC 207.1) in 5, which could be further strengthened by the determinative 1 H-1 H COSY cross-peaks of H-6/H-7/H-8 and the predominant HMBC correlations from H3-11 to C-5, C-6, and C-7, as well as H2-7 to C-5. Moreover, this deduction could be also evidenced by the extensive 2D NMR analysis as depicted in Figure 2. Interestingly, compound 5 showed an ECD spectrum almost consistent with that of 4 (see supporting information), which strongly illustrated a mode of sharing a very similar absolute configuration by the consideration of the same biogenesis. Therefore, the configuration of 5 was conclusively assigned as shown in Figure 1 and given the trial name as phaseolorin E.  Compound 5 was obtained as yellow oil. Its molecular formula was established as C 15 H 16 O 7 on the basis of the protonated molecule peak at m/z 331.0788 [M + Na] + in its HRESIMS, requiring eight degrees of unsaturation. The 1D NMR data (Table 3) of 5 were almost in accordance with those of 4, except for the possible replacement of hydroxyl group at C-5 position in 4 by an additional carbonyl functionality (δ C 207.1) in 5, which could be further strengthened by the determinative 1 H-1 H COSY cross-peaks of H-6/H-7/H-8 and the predominant HMBC correlations from H 3 -11 to C-5, C-6, and C-7, as well as H 2 -7 to C-5. Moreover, this deduction could be also evidenced by the extensive 2D NMR analysis as depicted in Figure 2. Interestingly, compound 5 showed an ECD spectrum almost consistent with that of 4 (see supporting information), which strongly illustrated a mode of sharing a very similar absolute configuration by the consideration of the same biogenesis. Therefore, the configuration of 5 was conclusively assigned as shown in Figure 1 and given the trial name as phaseolorin E.

Biological Activity
Compounds 1-6 were evaluated for cytotoxicities against three human cancer cell lines: HepG-2 (liver cancer), MCF-7 (breast cancer), and SF-268 (human glioblastoma carcinoma). However, all of them were found to be devoid of significant cytotoxicity activity even at a concentration of 100 µM.

Fungal Material
The fungus strain FS431 was isolated from a marine sediment sample collected from the Indian Ocean (depth 3605m, 7

Fermentation and Extraction
Diaporthe phaseolorum FS431 was cultured for 5 days at 28 • C in a potato dextrose agar (PDA) culture plate. The mycelial plugs were transferred to ten 500-mL Erlenmeyer flasks each containing 250 mL potato dextrose broth (20% potato, 2% glucose, 0.3% KH 2 PO 4 , 0.15% MgSO 4 •7H 2 O, and 250 mL water with 1.5% sea salt), and then incubated on a rotary shaker at 120 r/m and 28 • C for 4 days as seed cultures. After that, each of the seed cultures (10 mL) was transferred into autoclaved 1000-mL Erlenmeyer flasks with 500 mL potato dextrose broth. Then, the strain was incubated on a rotary shaker for 7 days at 28 • C and 120 r/m. The culture (120 L) was centrifuged to give the broth and mycelia. The broth was exhaustively extracted with EtOAc for four times, and then the EtOAc layers were combined and evaporated under reduced pressure at a temperature not exceeding 40 • C to yield a dark brown gum (53 g).

X-ray Crytallographic Data of Compounds 1 and 4
The single-crystal X-ray diffraction data for compounds (1 and 4) were collected at 100K on Agilent Xcalibur Nova single-crystal diffractometer using CuKα radiation. Their crystal structures were refined by the program with full-matrix least-squares calculation. Hydrogen atoms bonded to carbons were located by the geometrically ideal positions by the "ride on" method. Hydrogen atoms bonded to oxygen were placed on the difference Fourier method and were included in the calculation of structure factors with isotropic temperature factors. Crystallographic data for the reported structures have been deposited in the Cambridge Crystallographic Data Centre. (Deposition number: CCDC 1890682 for 1, CCDC 1890683 for 4). Copies of the data can be obtained, free of charge, on application to the CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax, +44-(0)-1223-336033; e-mail, deposit@ccdc.cam.ac.uk).

The Chemical Transformation Between Compounds 1 and 2
Sodium hydroxide (10 mg, 0.25 mmol) was slowly dissolved in anhydrous methanol (1 mL) at room temperature. To this clear solution, compound 1 (3.0 mg) was added carefully and the mixture stirred for 6 h under air condition. The crude mixture was then quenched with 2 N HCl (2 mL) and extracted with ethyl acetate (3 × 5 mL), washed with brine, and concentrated in vacuum. The crude product was purified by flash chromatography (silica gel, CHCl 3 /MeOH, 20:1→10:1) to provide 2 (0.45 mg, 15% yield) as a slight yellow powder.
A flame-dried 5 mL flask containing a mixture solvent of THF (0.25 mL) and CH 2 Cl 2 (0.75 mL) was charged with compound 2 (1.5 mg) and TFA (20 µL), and the reaction mixture stirred vigorously under air condition at room temperature for 24 h. The mixture was quenched by addition of 2 mL water and extracted with ethyl acetate (5 × 3 mL). The combined organic phases were washed with brine (4 mL), dried over Na 2 SO 4 , and filtered. Removal of solvent by rotary evaporation and purification by flash column chromatography (silica gel, CHCl 3 /MeOH, 20:1→10:1) afforded 1 (0.90 mg, 60% yield).
The compound 2 (2.5 mg) was put at a 5 mL round flask and kept in neat state under air condition at the room temperature for~30 d. Then, the resulting mixture was purified by 5 cm long flash chromatography (silica gel, CHCl 3 /MeOH, 20:1→10:1) afforded the desired product 1 (2.0 mg) with 80% yield.

Cytotoxic Activity Assay
The cytotoxic activities of compounds (1-6) against HepG-2, MCF-7 and SF-268 cell lines were evaluated by using the Sulforhodamine B (SRB) method [24] with cisplatin as the positive control.