Dihydroisocoumarins and Dihydroisoflavones from the Rhizomes of Dioscorea collettii with Cytotoxic Activity and Structural Revision of 2,2′-Oxybis(1,4-di-tert-butylbenzene)

The investigation of the constituents of the rhizomes of Dioscorea collettii afforded one new dihydroisocoumarin, named (−)-montroumarin (1a), along with five known compounds—montroumarin (1b), 1,1′-oxybis(2,4-di-tert-butylbenzene) (2), (3R)-3′-O-methylviolanone (3a), (3S)-3′-O-methylviolanone (3b), and (RS)-sativanone (4). Their structures were elucidated using extensive spectroscopic methods. To the best of our knowledge, compound 1a is a new enantiomer of compound 1b. The NMR data of compound 2 had been reported but its structure was erroneous. The structure of compound 2 was revised on the basis of a reinterpretation of its NMR data (1D and 2D) and the assignment of the 1H and 13C NMR data was given rightly for the first time. Compounds 3a–4, three dihydroisoflavones, were reported from the Dioscoreaceae family for the first time. The cytotoxic activities of all the compounds were tested against the NCI-H460 cell line. Two dihydroisocoumarins, compounds 1a and 1b, displayed moderate cytotoxic activities, while the other compounds showed no cytotoxicity.


Introduction
Dioscorea collettii HK. f. (Dioscoreaceae) is a perennial herbaceous plant of the Dioscorea genus that is widely distributed in Southwest China, Myanmar, and India. The rhizomes of D. collettii have been used clinically in combination with its sister species D. collettii HK. f. var. hypoglauca in the treatment of gouty arthritis, hyperuricemia, hyperlipidemia, inflammation, and tumors in China as a traditional medicine [1][2][3][4][5]. Previous studies have revealed that the major active constituents of D. collettii are steroidal saponins [6,7]. An unusual tricyclic diarylheptanoid derivative with thirty-four compounds has been reported on the basis of the previous work of our group [8,9]. Pharmacological studies have demonstrated that steroidal saponins of the Dioscorea genus have significant tumor-suppressive activities in human cancer cells [10,11]. In our previous study, tsaokoarylone, a diarylheptanoid obtained from D. collettii, also showed strong cytotoxic activity against the NCI-H460 cell line [8]. However, to date, the chemical constituents of D. collettii and whether its other types of compounds have significant cytotoxic activities have remained unclear.
As part of our continuous search for structurally unique and biologically valuable natural products from the Dioscorea genus [8,9,12,13], this study aimed to investigate D. collettii and whether its other types of compounds have significant cytotoxic activities have remained unclear.
As part of our continuous search for structurally unique and biologically valuable natural products from the Dioscorea genus [8,9,12,13], this study aimed to investigate the phytochemical constituents and cytotoxic activity of D. collettii. In the present study, one new dihydroisocoumarin, named (−)-montroumarin (1a), together with five known compounds were obtained from the rhizomes of D. collettii (Figure 1). Compound 1a is the enantiomer of the known compound montroumarin (1b). Compound 2 was first discovered as a natural product from the marine mollusk Onchidium struma [14]. The originally proposed structure of compound 2 (2′) (Figure 1) possessed a phenyl ether skeleton bearing four tert-butyl substituents located in the aromatic rings at the positions C-1, C-4, C-1′, and C-4′. However, during our structure elucidation of compound 2, a closer inspection of its NMR data indicated that the originally proposed structure of compound 2 might not be correct. Herein, details of the isolation and structure elucidation of these compounds, the structural revision of compound 2, and their cytotoxic activities against the NCI-H460 cell line are described.

Structural Elucidation of the Compounds
Compound 1 (1a/1b) was obtained as a colorless solid. Its molecular formula was elucidated to be C15H12O4 by HR-ESI-MS at m/z 279.0626 for [M + Na] + (calculated. for C15H12O4Na, 279.0633), with ten degrees of unsaturation. The 13 C NMR data (Table 1) in combination with analysis of the HSQC spectrum for 1 showed the presence of one carbonyl carbon, nine olefinic or aromatic carbons, one oxymethine, and one methylene. In its 1 H NMR spectrum, the five aromatic proton signals between δH 7.48 and 7.36 (δH 7.48 (2H, d, J = 7.5 Hz), 7.39 (2H, t, J = 7.5 Hz), and 7.36 (1H, t, J = 7.0 Hz)) indicated the presence of one mono-substituted benzene ring. Two meta-coupled aromatic signals at δH 6.26 (1H, br s) and 6.24 (1H, J = 1.8 Hz) suggested that 1 possessed one 1,2,3,5-tetrasubstituted aromatic ring. Additionally, one oxymethine proton at δH 5.56 (1H, dd, J = 12.0, 3.0 Hz) and one methylene group at δH 3.21 (1H, dd, J = 16.3, 12.2 Hz) and 3.06 (1H, dd, J = 16.4, 3.2 Hz) were observed in the 1 H NMR spectrum, perfectly matching the above 13 C NMR data. In addition to the two aromatic rings and one carbonyl, the remaining 1 degree of unsaturation suggested the existence of a one ring structure. All 1 H and 13 C chemical shifts for 1 were essentially identical to those observed for montroumarin [15]. Based on the interpretation of the HMBC and 1 H-1 H COSY spectra ( Figure 2) as well as a comparison with the literature data, the planar structure of 1 was characterized as 6, 8-dihydroxy-3-phenyl-3, 4-dihydroisocoumarin.

Structural Elucidation of the Compounds
Compound 1 (1a/1b) was obtained as a colorless solid. Its molecular formula was elucidated to be C 15 Na,279.0633), with ten degrees of unsaturation. The 13 C NMR data (Table 1) in combination with analysis of the HSQC spectrum for 1 showed the presence of one carbonyl carbon, nine olefinic or aromatic carbons, one oxymethine, and one methylene. In its 1 H NMR spectrum, the five aromatic proton signals between δ H 7.48 and 7.36 (δ H 7.48 (2H, d, J = 7.5 Hz), 7.39 (2H, t, J = 7.5 Hz), and 7.36 (1H, t, J = 7.0 Hz)) indicated the presence of one mono-substituted benzene ring. Two meta-coupled aromatic signals at δ H 6.26 (1H, br s) and 6.24 (1H, J = 1.8 Hz) suggested that 1 possessed one 1,2,3,5tetrasubstituted aromatic ring. Additionally, one oxymethine proton at δ H 5.56 (1H, dd, J = 12.0, 3.0 Hz) and one methylene group at δ H 3.21 (1H, dd, J = 16.3, 12.2 Hz) and 3.06 (1H, dd, J = 16.4, 3.2 Hz) were observed in the 1 H NMR spectrum, perfectly matching the above 13 C NMR data. In addition to the two aromatic rings and one carbonyl, the remaining 1 degree of unsaturation suggested the existence of a one ring structure. All 1 H and 13 C chemical shifts for 1 were essentially identical to those observed for montroumarin [15]. Based on the interpretation of the HMBC and 1 H-1 H COSY spectra ( Figure 2) as well as a comparison with the literature data, the planar structure of 1 was characterized as 6, 8-dihydroxy-3-phenyl-3, 4-dihydroisocoumarin.   Due to having a single C-3 stereogenic center, 1 might exist in two potential configurations, (3R)-1 and (3S)-1, and only the 3S-form (montroumarin) has been isolated previously [15]. Subsequent chiral HPLC separation on 1 afforded compounds 1a and 1b, a pair of enantiomers, in a ratio of 15:85 (Supplementary materials Figure S8), showing that 1 is a scalemic mixture. The opposite optical rotations and mirror-like electronic circular dichroism (ECD) spectra confirmed their enantiomeric relationship ( Figure 3). The absolute configuration of 1a and 1b was determined by comparing their ECD spectra and optical rotations with those of montroumarin [15] and other similar dihydroisocoumarin derivatives [16]. The absolute configuration of (+)-1 (1b) was established as S on the basis of the strong positive Cotton effect (CE) at 233 nm in its ECD spectrum (Figure 3), which is consistent with that of montroumarin. However, (−)-1 (1a) showed a strong negative CE at Due to having a single C-3 stereogenic center, 1 might exist in two potential configurations, (3R)-1 and (3S)-1, and only the 3S-form (montroumarin) has been isolated previously [15]. Subsequent chiral HPLC separation on 1 afforded compounds 1a and 1b, a pair of enantiomers, in a ratio of 15:85 (Supplementary materials Figure S8), showing that 1 is a scalemic mixture. The opposite optical rotations and mirror-like electronic circular dichroism (ECD) spectra confirmed their enantiomeric relationship ( Figure 3). The absolute configuration of 1a and 1b was determined by comparing their ECD spectra and optical rotations with those of montroumarin [15] and other similar dihydroisocoumarin derivatives [16]. The absolute configuration of (+)-1 (1b) was established as S on the basis of the strong positive Cotton effect (CE) at 233 nm in its ECD spectrum (Figure 3), which is consistent with that of montroumarin. However, (−)-1 (1a) showed a strong negative CE at 233 nm; thus, its absolute configuration was deduced to be R. A comparison of the optical rotations of 1a and 1b with that of montroumarin also supported the above conclusions. Thus, the structure of compound 1a was defined and named as (−)-montroumarin, and 1b was identified as the known compound montroumarin (see Figure 1).  Compound 2 was purified as a pale-yellow oil. The 1 H NMR spectrum ( Table 2) of 2 in CDCl3 showed signals attributable to an ABX-type aromatic ring at δH 7.54 (1H, d, J = 8.6 Hz, H-6),7.36 (1H, t, J = 2.5 Hz, H-3),and 7.13 (1H, dd, J = 8.6, 2.5 Hz, H-5), which indicated the presence of a 1,2,4-trisubstituted benzene ring [17,18]. In addition, two sets of non-equivalent signals each containing three methyl groups at δH 1.33 (9H, s, H-8/9/10) and 1.28 (9H, s, H-12/13/14) were also observed. The 13 C NMR and HSQC spectra displayed 12 carbon resonances, which were classified as eight olefinic or aromatic carbons (δC 147. 8 Figure S15) for compound 2 were highly consistent with the experimental data for 2, 2′oxybis (1, 4-di-tert-butylbenzene) (2′), suggesting that 2 was the same substance isolated by Sun B-N and coworkers [14]. The molecular formula of compound 2′ is C28H42O and the structure of 2′ was originally identified as a phenyl ether derivative bearing four tertbutyl substituents located in the aromatic rings at the positions C-1, C-4, C-1′, and C-4′ ( Figure 1). We subsequently performed 2D NMR experiments on 2. Careful examination of the HMBC spectrum ( Figure 2) of 2 revealed a strong HMBC correlation from H-5 (δH 7.13) to C-1 (δC 147.8), which is apparently inconsistent with structure 2′ (from H-2 to C-5 of 2′) but consistent with 2. This key HMBC correlation indicated that the structure elucidation of 2′ is clearly incorrect. The 13 C NMR data (δC 147.8, 147.8, 138.7, and 138.6) and the molecular formula of 2′ indicated that 2 was a dimer. In the HMBC spectrum, the correlation from the three equivalent methyl groups signals at δH 1.33 (9H, H-8/9/10) to the quaternary carbon signal at δC 35.0 (C-7) suggested a direct connection between C-8/9/10 (δC 30.4) and C-7, forming one tert-butyl substituent, which was further confirmed by the HMBC correlations from H-8 (δH 1.33, 9H) to C-9/10 (δC 30.4). This tert-butyl substituent attached to C-2 (δC 138.6) of the aromatic ring was clarified by the correlation from H-8/9/10 to C-2 in the HMBC spectrum. Similar, the HMBC correlations from H-12/13/14 (δH 1.28, 9H) to C-11 (δC 34.7) and C-4 (δC 147.2) suggested a connection between C-12/13/14 (δC 31.6) and C-11, forming another tert-butyl substituent attached to C-4 of the aromatic ring. The 1 H-1 H COSY correlation (Figure 2) of H-5/H-6 (δH 7.54) and the HMBC correlations of H-5/C-11 and H-6/C-4 elucidated the connectivity of C-4/C-5/C-6. The linkage between C-2 and C-3 was verified by the HMBC correlation from H-3 (δH 7.36) to C-7. The HMBC correlations from H-5 to C-3 (δC 124.6) and H-3 to C-5 (δC 124.1), together with the C-3 and C-5 located at the meta-position of the aromatic ring, indicated the connectivity of C-2/C-3/C-4/C-5/C-6. The established linkage of C-2/C-3/C-4/C-5/C-6 as well as the HMBC correlations from H-3 and H-5 to C-1 (δC 147.8) constructed the aromatic ring. Considering the downfield chemical shift of C-1, and the molecular formula required for 2, C-1 and C-1′ should be linked to the remaining one oxygen atom to form a dimer. The structure of 2, shown in Figure 1, was identified as 1,1′-oxybis(2,4-di-tert-butylbenzene).  Figure S15) for compound 2 were highly consistent with the experimental data for 2, 2 -oxybis (1, 4-di-tert-butylbenzene) (2 ), suggesting that 2 was the same substance isolated by Sun B-N and coworkers [14]. The molecular formula of compound 2 is C 28 H 42 O and the structure of 2 was originally identified as a phenyl ether derivative bearing four tert-butyl substituents located in the aromatic rings at the positions C-1, C-4, C-1 , and C-4 ( Figure 1). We subsequently performed 2D NMR experiments on 2. Careful examination of the HMBC spectrum ( Figure 2) of 2 revealed a strong HMBC correlation from H-5 (δ H 7.13) to C-1 (δ C 147.8), which is apparently inconsistent with structure 2 (from H-2 to C-5 of 2 ) but consistent with 2. This key HMBC correlation indicated that the structure elucidation of 2 is clearly incorrect. The 13 C NMR data (δ C 147.8, 147.8, 138.7, and 138.6) and the molecular formula of 2 indicated that 2 was a dimer. In the HMBC spectrum, the correlation from the three equivalent methyl groups signals at δ H 1.33 (9H, H-8/9/10) to the quaternary carbon signal at δ C 35.0 (C-7) suggested a direct connection between C-8/9/10 (δ C 30.4) and C-7, forming one tert-butyl substituent, which was further confirmed by the HMBC correlations from H-8 (δ H 1.33, 9H) to C-9/10 (δ C 30.4). This tert-butyl substituent attached to C-2 (δ C 138.6) of the aromatic ring was clarified by the correlation from H-8/9/10 to C-2 in the HMBC spectrum. Similar, the HMBC correlations from H-12/13/14 (δ H 1.28, 9H) to C-11 (δ C 34.7) and C-4 (δ C 147.2) suggested a connection between C-12/13/14 (δ C 31.6) and C-11, forming another tert-butyl substituent attached to C-4 of the aromatic ring. The 1 H-1 H COSY correlation (Figure 2) of H-5/H-6 (δ H 7.54) and the HMBC correlations of H-5/C-11 and H-6/C-4 elucidated the connectivity of C-4/C-5/C-6. The linkage between C-2 and C-3 was verified by the HMBC correlation from H-3 (δ H 7.36) to C-7. The HMBC correlations from H-5 to C-3 (δ C 124.6) and H-3 to C-5 (δ C 124.1), together with the C-3 and C-5 located at the meta-position of the aromatic ring, indicated the connectivity of C-2/C-3/C-4/C-5/C-6. The established linkage of C-2/C-3/C-4/C-5/C-6 as well as the HMBC correlations from H-3 and H-5 to C-1 (δ C 147.8) constructed the aromatic ring. Considering the downfield chemical shift of C-1, and the molecular formula required for 2, C-1 and C-1 should be linked to the remaining one oxygen atom to form a dimer. The structure of 2, shown in Figure 1, was identified as 1,1 -oxybis(2,4-di-tert-butylbenzene).
Compound 3 (3a/3b) was obtained as a colorless needle. The results of a comparison between the 1 H, 13 C NMR data of 3 and those reported by Guimarães et al. [24] suggested that the planar structure of 3 was 3 -O-methylviolanone. Compound 3 was optically inactive, suggesting that it was a racemic mixture. Both of the two configurations (3R)-3 and (3S)-3 have been reported previously [25,26], but the ECD spectra of (3R/3S)-3 and the specific rotation of (3R)-3 have remained undefined. Subsequent chiral resolution of 3 ( Figure S18) afforded a pair of enantiomers, 3a and 3b, and they displayed almost mirror-image ECD curves that showed opposite CEs at 193, 210, 230, 272, and 325 nm. The respective absolute configurations of (−)-3 and (+)-3 (3a and 3b) were defined as (3R) and (3S) via a comparison of their experimental and calculated ECD spectra (Figure 3).
Compound 4 was also obtained as a colorless needle. A comparison of the 1 H and 13 C NMR data of 4 with those of compound 3 showed that their structures are closely related, suggesting that 4 is also a dihydroisoflavone analogue. Through a comparison of its NMR data with those reported [27], the planar structure of 4 was identified as sativanone. Similarly to compound 3, 4 was also optically inactive, suggesting that it was a racemate. Both of the two configurations (3R)-4 and (3S)-4, including their ECD spectra and the specific rotations, have been reported [28,29]. Thus, compound 4 was identified as (RS)-sativanone.

Cytotoxic Activity against NCI-H460 Cell Line
All the isolated compounds 1-4 were further evaluated for their cytotoxicity against the human lung cancer NCI-H460 cell line ( Table 3). The results indicated that compounds 1a and 1b, two dihydroisocoumarins, possess moderate cytotoxic activities, with IC 50 values of 33.37 and 32.06 µM, respectively. Compounds 2-4 showed no activity, with an IC 50 > 50 µM.

Materials
The rhizomes of D. collettii were collected from Mount Emei, Leshan City, Sichuan province, China, and authenticated by Prof. Wenyuan Gao (School of Pharmaceutical Science and Technology, Tianjin University). A voucher specimen (ID: 245020328) was deposited in the School of Pharmaceutical Science and Technology, Tianjin University, Tianjin, China.

Extraction and Isolation
The air-dried rhizomes of D. collettii (16.2 kg) were extracted three times with 90% aqueous ethanol and three times with 60% aqueous ethanol under reflux (30 L, each for 2 h). After removal of the solvent under reduced pressure, the residue was combined and suspended in water to a final volume of 10 L, and then sequentially partitioned with petroleum ether (PE, 60-90 • C), ethyl acetate (EtOAc), and n-butyl alcohol (n-BuOH).

Cytotoxicity Assays
Compounds 1-4 were evaluated for their cytotoxic activities by the MTT method using NCI-H460 cell line. The NCI-H460 cells were seeded at a density of 1 × 104/well in a complete growth medium in 96-well plates. The cells were incubated with the test compounds for 24 h before the MTT assay. Then, a fresh solution of MTT (0.5 mg/mL) was added to each single well of the 96-well plate. The plate was incubated in a CO 2 incubator for another 4 h. Finally, the cells were dissolved with 100 µL of DMSO and then analyzed in a multiwall plate reader with a wavelength of 570 nm.

Conclusions
The chemical investigation of D. collettii led to the isolation of six compounds (1-4), including one new dihydroisocoumarin, (−)-montroumari (1a). The structure of 2,2oxybis(1,4-di-tert-butylbenzene) (2 ) was revised to be 1,1 -oxybis(2,4-di-tert-butylbenzene) (2), assisted by a careful re-examination of the structural elucidation process. All the compounds 1-4 were isolated from D. collettii for the first time. The isolation of compounds 3a, 3b, and 4 from the Dioscorea species has not been reported yet [30,31]. Notably, to the best of our knowledge, dihydroisoflavones have not been described from any other species in the Dioscoreaceae family. Our discovery of these dihydroisoflavones (3a, 3b, and 4) enriches the structural diversity of the Dioscoreaceae family. Dihydroisocoumarins (1a-1b) exhibited moderate cytotoxic activities against the NCI-H460 cell line, with IC 50 values ranging from 32.06 to 33.37 µM, whereas the other compounds, including the three dihydroisoflavones (3a-4), did not show any activities at the tested concentrations (IC 50 > 50 µM). Mounting evidence has revealed that steroidal saponins exert strong cytotoxic activities against human cancer cells, which means they could be considered as promising cytotoxic agents against human cancer cells. Dihydroisocoumarins and diarylheptanoids showed moderate cytotoxic activities against the NCI-H460 cell line. However, more selectivity studies are needed to determine whether these two types of compounds in D. collettii have cytotoxic activities against human cancer cells.