1. Introduction
The
Trigonostemon genus (Euphorbiaceae family) includes over 50 species growing throughout the dense forests and riverside thickets of tropical and subtropical Asia, among which nine species are endemic to south China [
1].
Trigonostemon plants have been utilized as traditional remedies for treating asthma, diarrhea, skin diseases, poisonous snake bites, and food poisoning over a considerable period of time [
1], [
2]. A large array of chemical constituents in this genus have been demonstrated to show diverse therapeutic properties, including antiviral [
3], [
4], [
5], [
6], [
7], anti-tumor [
8], [
9], [
10], [
11], [
12], antimicrobial [
13], [
14], [
15], [
16], and anti-inflammatory [
17], [
18], [
19], [
20], [
21], [
22] activities. Among the
Trigonostemon metabolites, daphnane diterpenoids—especially highly modified daphnane diterpenoid orthoesters—are recognized as the major components, with over 110 diverse structures discovered in previous phytochemical investigations [
2], [
20], [
22], [
23]. Norditerpenoid dimers and monomers represent another characteristic metabolite category. Since the isolation of the first phenanthrenone norditerpenoid, trigonostemone, from
Trigonostemon reidioides in 1990, over 60 norditerpenoid dimers and/or monomers from
Trigonostemon plants have been reported [
2], [
13], [
24], [
25], [
26].
Trigonostemon fragilis (
T. fragilis; formerly used name:
T. lutescens) [
27] is a shrub growing mainly in Guangxi Zhuang Autonomous Region [
18]. Previously, a few coumarins [
28], [
29], lignans [
30], [
31], sterols [
31], [
32], ellagitannins [
18], alkaloids [
33], [
34], triterpenoids [
32], sesquiterpenoids [
34], [
35], and diterpenoids [
32], [
36], [
37] were identified from
T. fragilis. In our continuous search for bioactive ingredients from
Trigonostemon species [
8], [
9], [
10], [
11], [
16], [
38], [
39], four undescribed norditerpenoid heterodimers, trigofragiloids A-C (denoted as compounds 1-3 in this paper) and (+)- and (−)-trigofragiloid D (compound 4), and three undescribed phenanthrenone norditerpenoids, trigofragiloids E-G (compounds 5-7), alongside four known compounds, (+)- and (−)-8, (+)- and (−)-9, 10, and 11, were isolated and characterized from
T. fragilis (
Fig. 1 [
1], [
2], [
3], [
4], [
5], [
6], [
7], [
8], [
9], [
10], [
11]). Among them, compounds 1 and 2 possess unprecedented heterodimeric frameworks of a tetra-norditerpenoid and an ennea-norditerpenoid, respectively. Compound 3 is a rare phenylpropanoid-norditerpenoid adduct featuring an unprecedented dimerization pattern. Compounds (+)- and (−)-4, a pair of enantiomers, represent the first example of S-shaped 1,4-dioxane-fused norditerpenoid heterodimers. In addition, two previously reported 1,4-dioxane-fused norditerpenoid dimers, actephilol A and epiactephilol A [
40], which were reported previously as two C-shaped C-2 epimers, were structurally revised as S-shaped and C-shaped geometrical isomers and were further identified as two pairs of enantiomers, (+)- and (−)-8 and (+)- and (−)-9, respectively. In the adenosine triphosphate-citrate lyase (ACLY) inhibition assay, compound 7 exhibited highly potent activity comparable to that of the positive control, BMS-303141. The interaction mode of compound 7 with ACLY was further analyzed by means of a molecular docking experiment. Herein, the isolation, structure elucidation, biosynthetic origin, and bioassays of these norditerpenoids are reported.
2. Materials and methods
2.1. General experimental procedures
Detailed information on the general experimental procedures has been incorporated into Section S1 in Appendix A.
2.2. Plant material
Details on the collection and authentication of plant samples are provided in Section S2 in Appendix A.
2.3. Extraction and isolation
Twig and leaf material of T. fragilis (8 kg) was powdered and extracted in 95% ethanol (EtOH, 25 L) at ambient temperature three times. It was then filtered and evaporated to provide a crude residue (220 g), which was dissolved in water (H2O) and then underwent liquid-liquid partitioning with ethyl acetate (EtOAc) to yield a medium-polarity organic phase (80 g). The organic phase then underwent macroporous resin column chromatography (CC) with 25% and 90% EtOH-H2O mixtures as the mobile phase, yielding two parts: D1 and D2. Eight fractions, D1M1-D1M8, were obtained by partitioning fraction D1 (70 g) using middle chromatogram isolated (MCI) gel CC eluted with MeOH-H2O (from 30% to 95%).
Fraction D1M2 (2.8 g) was separated over a silica gel CC eluting with petroleum ether-acetone (from 90% to 50%) to produce eight subfractions, D1M2P1-D1M2P8. Of these, D1M2P5 (128 mg) was purified via semipreparative high-performance liquid chromatography (HPLC; XBridge C18; Waters, USA; 3.0 mL∙min−1) with 37% CH3CN-H2O as the mobile phase. In this way, compound 5 (25 mg) was obtained.
Fraction D1M3 (7.0 g) was loaded onto silica gel CC eluting with petroleum ether-acetone (from 90% to 50%) to produce nine subfractions, D1M3P1-D1M3P9. Subfraction D1M3P6 (260 mg) was separated using silica gel CC with CH2Cl2-MeOH (v/v, 200:1 to 10:1) as the mobile phase; it was then purified by means of semipreparative reversed-phase HPLC (Triart octadecylsilyl (ODS)-A; YMC, Japan; 3.0 mL∙min−1) eluting with 55% CH3CN-H2O to yield compound 1 (1.0 mg). The recrystallization of D1M3P7 (1.5 g) in petroleum ether-acetone yielded compound 7 (65 mg). Fraction D1M4 (2.5 g) was separated over silica gel CC eluting with gradient CH2Cl2-MeOH (v/v, 200:1 to 10:1), which was followed by recrystallization, yielding compound 6 (106 mg). Nine subfractions, D1M5P1-D1M5P9, were obtained by separating fraction D1M5 (7.7 g) through silica gel column eluting with petroleum ether-acetone (from 95% to 50%). Subfraction D1M5P7 (700 mg) was fractionated by means of Sephadex LH-20 CC with 50% CHCl3-MeOH as the eluents, yielding two subfractions: D1M5P7S1 and D1M5P7S2. Fraction D1M5P7S2 (120 mg) was separated over silica gel CC (CH2Cl2-MeOH; v/v, 300:1 to 10:1) to give D1M5P7S2C1-D1M5P7S2C5. Fraction D1M5P7S2C3 (8 mg) was then purified by means of semipreparative HPLC (XBridge C18; 3.0 mL∙min−1) eluting with 60% CH3CN-H2O to obtain compound 2 (2.0 mg). Fraction D2 (10 g) was separated into five subfractions, D2P1-D2P5, through silica gel CC using petroleum ether-acetone (from 95% to 50%) as the eluents. Subfraction D2P3 (700 mg) was divided into four subfractions, D2P3S1-D2P3S4, through Sephadex LH-20 CC using 50% CHCl3-MeOH as the eluting solvents. Subfraction D2P3S1 (21 mg) was then purified with silica gel CC, eluting with CHCl3-MeOH (v/v, 300:1 to 50:1) to obtain compound 3 (0.9 mg). Subfraction D2P3S4 (125 mg) was further purified in the same way and yielded trigonochinene E (compound 11; 120 mg). Four subfractions, D2P4S1-D2P4S4, were obtained from the separation of D2P4 (450 mg) over Sephadex LH-20 CC eluting with 50% CHCl3-MeOH. Further purification of D2P4S4 (300 mg) was conducted through silica gel CC eluting with petroleum ether-EtOAc (from 85% to 60%), yielding compound 4 (70 mg). Further separation of subfraction D2P5 (1.17 g) was conducted using a gel column on Sephadex LH-20 CC with 50% CHCl3-MeOH elution, yielding six subfractions: D2P5S1-D2P5S6. The reversed-phase HPLC preparation (XBridge C18; 3.0 mL∙min−1) of D2P5S4 (84 mg) with 70% CH3CN-H2O yielded actephilol A (compound 8; 3.0 mg) and epiactephilol A (compound 9; 4.0 mg). Fraction D2P5S6 (thrigonosomone E) was identified as compound 10 (50 mg).
The chiral separation of compound 4 was performed to yield a pair of enantiomers (−)- and (+)-4 (each 35 mg) using chiral HPLC semi-preparation (Lux cellulose-1; Phenomenex, USA; 3.0 mL∙min−1) eluted with 20% isopropanol-n-hexane. The chiral analysis of compound 8 was implemented on an HPLC-circular dichroism (CD) apparatus equipped with a Chiralpak AD-H (Daicel, Japan) column (1.0 mL∙min−1) eluted with 12% isopropanol-n-hexane, while that of compound 9 was conducted using a Chiralpak IC-H (Daicel) column (1.0 mL∙min−1) eluted with 5% isopropanol-n-hexane. Subsequent chiral separations of compounds 8 and 9 by means of semipreparative HPLC under the same conditions yielded (−)- and (+)-8 (each 1.1 mg) and (−)- and (+)-9 (each 1.3 mg), respectively. The compounds were characterized as follows:
(1)
Trigofragiloid A (compound 1). Brownish-red amorphous powder;
$[\alpha ]_{\text{D}}^{20}$ 0 (
c 0.08, MeOH); ultraviolet (UV) (MeOH)
λmax (log
ε) 211 (4.64), 245 (4.54), and 280 (4.24) nm; infrared (IR) (KBr)
νmax 3443, 2924, 2849, 1587, 1384, 1151, and 1042 cm
−1;
1H (600 MHz, CD
3OD) and
13C (125 MHz, CD
3OD) nuclear magnetic resonance (NMR) data (
Table 1); (−)-low-resolution mass spectrometry (LRMS) (electrospray ionization (ESI))
m/z 455.1 [M − H]
−; (−)-high-resolution mass spectrometry (HRMS) (ESI)
m/z 455.1494 [M − H]
− (calculated for C
28H
23O
6, 455.1495). (
α: observed rotation in degrees; D: wavelength of sodium light (589 nm);
c: concentration of sample;
λmax: maximum absorption wavelength;
ε: molar absorption coefficient; KBr: film on potassium bromide pellet;
νmax: frequencies of absorption-band maxima;
m/
z: mass to charge.)
(2)
Trigofragiloid B (compound 2). Brownish-red amorphous powder;
$[\alpha ]_{\text{D}}^{20}$ +2 (
c 0.10, MeOH); UV (MeOH)
λmax (log
ε) 212 (4.42), 244 (4.33), and 277 (4.00) nm; IR (KBr)
νmax 3442, 2920, 2852, 1575, 1404, and 1119 cm
−1;
1H (500 MHz, CDCl
3) and
13C (125 MHz, CDCl
3) NMR data (
Table 1); (+)-LRMS (ESI)
m/z 471.3 [M + H]
+; (+)-HRMS (ESI)
m/z 471.1819 [M + H]
+ (calculated for C
29H
27O
6, 471.1808).
(3)
Trigofragiloid C (compound 3). Yellow amorphous powder; UV (MeOH)
λmax (log
ε) 202 (4.10) and 296 (3.03) nm; IR (KBr)
νmax 3440, 2921, 2848, 1587, 1452, 1254, and 1143 cm
−1;
1H (600 MHz, CD
3OD) and
13C (125 MHz, CD
3OD) NMR data (
Table 2); (−)-LRMS (ESI)
m/z 417.0 [M − H]
−; (−)-HRMS (ESI)
m/z 835.3488 [2M − H]
− (calculated for C
52H
51O
10, 835.3482).
(4)
Trigofragiloid D (compound 4). Brownish-red crystals; melting point (mp) 254-256 °C; UV (MeOH)
λmax (log
ε) 226 (4.33), 250 (4.40), 264 (4.23), 289 (4.13), and 316 (3.79) nm; IR (KBr)
νmax 3444, 2939, 2850, 1732, 1634, 1608, 1464, 1250, 1149, 1064, 1031, and 847 cm
−1;
1H (500 MHz, CDCl
3) and
13C (125 MHz, CDCl
3) NMR data (
Table 3); (−)-LRMS (ESI)
m/z 628.9 [M + 2H
2O − H]
−; (−)-HRMS (ESI)
m/z 1187.4448 [2M − H]
− (calculated for C
72H
67O
16, 1187.4429).
(5) (-)-Trigofragiloid D (compound 4). $[\alpha ]_{\text{D}}^{20}$ −205 (c 0.20, MeOH); electronic circular dichroism (ECD) (MeOH) λ (Δε) 225 (−6.4), 248 (−30.0), 265 (+2.6), and 322 (+1.9) nm.
(6) (+)-Trigofragiloid D (compound 4). $[\alpha ]_{\text{D}}^{20}$ +205 (c 0.20, MeOH); ECD (MeOH) λ (Δε) 224 (+7.6), 248 (+30.7), 264 (−2.4), 278 (+3.3), and 322 (−2.0) nm.
(7)
Trigofragiloid E (compound 5). Yellow crystals; mp 197–198 °C;
$[\alpha ]_{\text{D}}^{20}$ +13 (
c 0.10, MeOH); UV (MeOH)
λmax (log
ε) 208 (4.05), 243 (4.10), and 320 (3.79) nm; ECD (MeOH)
λ (Δ
ε) 226 (+1.7), 255 (+7.8), and 328 (−2.4) nm; IR (KBr)
νmax 3442, 2926, 1661, 1591, 1320, and 754 cm
−1;
1H [600 MHz, (CD
3)
2CO] and
13C [150 MHz, (CD
3)
2CO] NMR data (
Table 4); (−)-LRMS (ESI)
m/z 273.2 [M − H]
−; (+)-HRMS (ESI)
m/z 845.3529 [3M + Na]
+ (calculated for C
48H
54O
12Na, 845.3513).
(8)
Trigofragiloid F (compound 6). Brownish-red amorphous powder; UV (MeOH)
λmax (log
ε) 253 (4.32) and 211 (4.29) nm; IR (KBr)
νmax 3300, 2926, 1655, 1580, 1331, 1159, and 1103 cm
−1;
1H (500 MHz, C
5D
5N) and
13C (125 MHz, C
5D
5N) NMR data (
Table 4); (−)-LRMS (ESI)
m/z 297.1 [M − H]
−; (−)-HRMS (ESI)
m/z 297.0764 [M − H]
− (calculated for C
17H
13O
5, 297.0763).
(9)
Trigofragiloid G (compound 7). Brownish-red amorphous powder; UV (MeOH)
λmax (log
ε) 254 (4.19) and 211 (4.18) nm; IR (KBr)
νmax 3449, 3332, 1650, 1575, 1333, 1277, 1163, and 750 cm
−1;
1H (500 MHz, C
5D
5N) and
13C (125 MHz, C
5D
5N) NMR data (
Table 4); (−)-LRMS (ESI)
m/z 283.1 [M − H]
−; (−)-HRMS (ESI)
m/z 283.0608 [M − H]
− (calculated for C
16H
12O
5, 283.0606).
(10) Actephilol A (compound 8). Brownish-red amorphous powder; UV (MeOH) λmax (logε) 229 (4.44), 251 (4.51), 262 (4.39), 287 (4.26), and 315 (3.93) nm; IR (KBr) νmax 3401, 2924, 2853, 1728, 1633, 1608, 1432, 1252, 1148, 1065, 1003, and 758 cm−1; 1H (800 MHz, CD3OD) and 13C (200 MHz, CD3OD) NMR data (Table S1 in Appendix A); (−)-LRMS (ESI) m/z 579.3 [M − H]−; (−)-HRMS (ESI) m/z 579.2024 [M − H]− (calculated for C35H31O8, 579.2024).
(11) (−)-Actephilol A (compound 8). $[\alpha ]_{\text{D}}^{18}$ −151 (c 0.11, MeOH); ECD (n-hexane/isopropanol = 88/12) λ (Δε) 227 (−7.9), 250 (−23.4), and 266 (+5.8) nm.
(12) (+)-Actephilol A (compound 8). $[\alpha ]_{\text{D}}^{18}$ +149 (c 0.11, MeOH); ECD (n-hexane/isopropanol = 88/12) λ (Δε) 230 (+6.5), 249 (+23), and 266 (−3.8) nm.
(13) Epiactephilol A (compound 9). Brownish-red amorphous powder; UV (MeOH) λmax (logε) 228 (4.57), 253 (4.62), 263 (4.52), 287 (4.42), and 315 (4.08) nm; IR (KBr) νmax 3424, 2923, 1728, 1633, 1609, 1585, 1433, 1274, 1148, and 1067 cm−1; 1H (600 MHz, CD3OD) and 13C (150 MHz, CD3OD) NMR data, (Table S1); (−)-LRMS (ESI) m/z 579.2 [M − H]−; (−)-HRMS (ESI) m/z 579.2021 [M − H]− (calculated for C35H31O8, 579.2024).
(14) (-)-Epiactephilol A (compound 9). $[\alpha ]_{\text{D}}^{18}$ −328 (c 0.13, MeOH); ECD (n-hexane/isopropanol = 95/5) λ (Δε) 248 (+14.1), 264 (−17.3), and 289 (−6.0) nm.
(15) (+)-Epiactephilol A (compound 9). $[\alpha ]_{\text{D}}^{18}$ +325 (c 0.13, MeOH); ECD (n-hexane/isopropanol = 95/5) λ (Δε) 247 (−14.0), 262 (+15.5), and 287 (+6.2) nm.
2.4. Crystal structure analysis
Suitable crystals of compounds 4 and 5 were acquired in MeOH via slow evaporation at room temperature. Crystallographic experiments for compounds 4 and 5, following the previously reported protocol [
41], [
42], [
43], were conducted on a Bruker APEX-II charge coupled device (CCD) diffractometer using Cu Kα radiation (
λ = 1.54178 Å = 1.54178 × 10
−10 m) and operating in the
ϕ-
ω scan mode. Crystal structure determination and refinement procedures were conducted with the ShelXT and ShelXL program using Olex2 [
44], [
45], [
46]. The crystallographic information pertaining to compounds 4 and 5 was submitted to the Cambridge Crystallographic Data Centre (CCDC) and assigned deposition numbers CCDC 2218152 (compound 4) and CCDC 2218153 (compound 5). Complimentary copies of the aforementioned data are readily accessible on the CCDC website
†. Details of the X-ray crystallographic data of compounds 4 and 5 are provided in Tables S10 and S11 in Appendix A.
2.5. Computational details
The potential energy surface scan and ECD calculations were carried out using established methods, as previously reported [
47], [
48]. Detailed procedures are included in Section S3 in Appendix A.
2.6. ACLY inhibition assay
The activity of ACLY depends on the adenosine triphosphate (ATP)-consuming and adenosine diphosphate (ADP)-generating procedures. The ADP-Glo assay (Promega, USA) is a luminescent quantification assay that correlates ADP concentrations with ACLY inhibitory activities. According to the ADP-Glo assay protocols, the testing compounds were dissolved in dimethyl sulfoxide (DMSO) in fourfold serial dilutions with a starting concentration of 400 μmol∙L−1 for a total of seven gradients. Once the reaction between the ACLY buffer (20 nmol∙L−1) and substrate mixture (5 μmol∙L−1 ATP, 30 μmol∙L−1 citrate, and 50 μmol∙L−1 coenzyme A (CoA)) was complete (37 °C, 30 min), ADP reagent (2.5 μL) was added to stop the reaction (90 min, room temperature). Then, the ADP detection reagent (5 μL) was added to catalyze the conversion of ADP to ATP, which allowed the quantification of freshly produced ATP through a luciferase/luciferin reaction. The luminescence value, which reflects the amount of ADP generated and is indicative of the ACLY activity, was recorded using an EnVision plate reader (PerkinElmer, USA). BMS-303141 was used as the positive control.
2.7. Molecular docking study
A molecular docking simulation of compound 7 with the ACLY binding site was conducted following previously reported methods [
49], [
50], [
51]. Pymol
‡ was used to visualize the docking outputs. Details of the procedures are included in Section S4 in Appendix A.
3. Results and discussion
Compound 1, a brownish-red powder, was assigned the molecular formula C
28H
24O
6 incorporating 17 double-bond equivalents (DBEs), based on an analysis of the
13C NMR data and the (−)-high-resolution electrospray ionization mass spectrometry (HRESIMS) ion peak at
m/z 455.1494 [M − H]
− (calculated 455.1495). Scrutiny of its one-dimensional (1D) NMR data (
Table 1), in combination with the heteronuclear single quantum coherence (HSQC) data, unveiled a highly substituted aromatic skeleton characterized by ten aromatic and/or olefinic double bonds and two carbonyl groups (
13C NMR chemical shift value (
δC) 210.0, 186.3), which accounted for 12 out of 17 DBEs. The remaining five DBEs thus indicated a pentacyclic architecture for compound 1. Analysis of the heteronuclear multiple bond correlation (HMBC) (
Fig. 2) and rotating-frame Overhauser effect spectroscopy (ROESY) (
Fig. 3) spectra enabled the assembly of two substructures, units A and B. The structure of unit A was identical with the partial structures of the reported homodimer neoboutomannin [
52] and heterodimers trigohowilols C and D [
15], as supported by the HMBC signals of H
318(19)/C3, C4, and C5; H6/C4, C8, and C10; H11/C8, C10, C12, and C13; H14/C7, C9, and C12; and H
317/C12, C13, and C14 (red arrows in
Fig. 2). The unit B of compound 1 was elucidated as a penta-substituted naphthalene, 5-methoxy-3-methyl-naphthalene-2,7-diol, by the HMBC signals from H6′ to C5′ (
δC 154.6), C7′, C8′, and C10′; CH
3O-7′ to C7′; H11′ to C8′, C10′, and C13′; H14′ to C7′, C9′, and C12′ (
δC 157.4); and H
317′ to C12′, C13′, and C14′ (blue arrows in
Fig. 2). Finally, the linkage of these two units via a carbon-carbon bond between the two non-proton-bearing carbons, C1 (
δC 141.9) and C10′ (
δC 103.3), was assigned to complete the structure of compound 1. Although the HMBC correlations were unavailable for the connection, this assignment was corroborated based on the crucial ROESY cross-peak of H11′/H
318. From a biosynthetic perspective, unit B of compound 1 probably originated from the degradation of a diterpenoid (
Fig. 4). As far as we know, compound 1 represents the first heterodimer of a tetra-norditerpenoid and an ennea-norditerpenoid.
Compound 2 was assigned the molecular formula C
29H
26O
6, as deduced by an analysis of the (+)-HRESIMS and
13C NMR data. Comparison of its 1D NMR data (
Table 1) with those of compound 1 demonstrated the occurrence of an additional methoxy moiety (
1H NMR chemical shift value (
δH) 2.87;
δC 54.7) in compound 2. An analysis of the HMBC (Fig. S10 in Appendix A) and nuclear Overhauser effect spectroscopy (NOESY) spectra (
Fig. 3) confirmed that compound 2 is a 12-
O-methylated derivative of compound 1. Intriguingly, the additional CH
3O-12 was spatially close to the through-space shielding zone of the naphthalene ring; thus, the CH
3O-12 protons were dramatically shielded (Δ
δH ≈ −1.0 parts per million (ppm)) [
53].
Similar to the other planar biaryl norditerpenoid dimers neoboutomannin [
52] and trigohowilols C-G [
15], both compounds 1 and 2 exhibited no photoactivity due to their close-to-zero specific rotation values. However, it appears that there would be potential axial chirality about the C1-C10′ bond for compounds 1 and 2, as evidenced by the different chemical shifts of CH
318 and CH
319 in these compounds (
Table 1). To estimate the rotation energy (Δ
Erot) barrier, the relaxed potential energy surface scanning (PES) method on the C3-C1-C10′-C5′ dihedral angle at the B3LYP/6-31G* level was performed [
47]. The result revealed that the Δ
Erot for compounds 1 and 2 was 28.7 and 28.4 kcal·mol
−1 (1 kcal = 4.1868 × 10
3 J) (Fig. S11 in Appendix A), respectively, which suggested that these compounds are class 2 atropisomers with elimination half-life (
t1/2) values in the range of hours to days, according to LaPlante’s atropisomer classification method [
54], [
55].
Compound 3 was assigned the molecular formula C
26H
26O
5 incorporating 14 DBEs, as supported by the
13C NMR and (−)-HRESIMS data analysis. Its
1H NMR spectrum (
Table 2) showed resonances for six aromatic protons, three aromatic methyl groups (
δH 2.25, 2.39, and 2.40), and three methoxy groups. Analysis of the
13C NMR data (
Table 2), taken together with the distortionless enhancement by polarization transfer (DEPT) and HSQC data, revealed the presence of ten aromatic double bonds, accounting for 10 out of 14 DBEs; the remaining four DBEs indicated that compound 3 is tetracyclic. As shown in
Fig. 2, the HMBC signals of H
31′/C1, C2′, and C3′; H
318/C3′, C4, and C5; H1/C3′, C5, and C9; H6/C4, C7, C8, and C10; H11/C8, C10, and C12; H14/C7, C9, and C12; H
317/C12, C13, and C14; and CH
3O-7/C7 delineated a hexa-substituted phenanthrene adorned with three methyl groups, a hydroxy group, and a methoxy group. Next, the HMBC signals of CH
3O-6′(8′)/C6′(8′) and H5′(9′)/C3′; and of C4′, C6′(8′), and C7′ revealed a symmetrical 3,5-dimethoxy-4-hydroxy phenyl group at C3′. In light of our present knowledge, compound 3 possesses an unprecedented carbon skeleton; it could be biosynthetically derived from a norditerpenoid and phenylpropanoid via an intermolecular Diels-Alder (DA) reaction (
Fig. 4).
Compound 4, which took the form of brownish-red crystals, originally existed as a racemate and had the molecular formula C
36H
34O
8 with 20 DBEs, based on an analysis of its
13C NMR and (−)-HRESIMS data. Analysis of the 1D NMR data (
Table 3), combined with the HSQC data, revealed a highly substituted aromatic skeleton with seven uncoupled aromatic protons (
δH 6.76, 6.76, 7.19, 7.50, 7.87, 7.91, and 7.98, each singlet (s);
δC 95.8, 99.8, 105.0, 106.7, 106.8, 123.4, and 124.4), three aromatic methyl groups (
δH 2.32, 2.38, and 2.42, each 3H, s), three methoxyl groups (
δH 3.58, 3.83, and 4.01, each 3H, s), and a
gem-dimethyl group (
δH 1.49 and 1.69, each 3H, s). Further scrutiny of the HMBC (
Fig. 2) and ROESY correlations (
Fig. 3) established two independent substructures, units A and B, which featured a highly aromatized tri- and tetra-norditerpenoid, respectively. With two oxygen atoms and one DBE remaining, it was obvious that units A and B were conjugated by a 1,4-dioxane ring. However, the regio-chemistry of the 1,4-dioxane moiety (C1-O-C2′/C2-O-C3′ or C1-O-C3′/C2-O-C2′), as well as the relative configurations of the C1 and C2 stereocenters, were uncertain due to the lack of reliable two-dimensional (2D) NMR correlations.
Fortunately, after several recrystallization attempts, high-quality crystals were successfully obtained. Further X-ray diffraction study of a suitable single crystal (
Fig. 5) not only confirmed the regio-chemistry of the 1,4-dioxane moiety as C1–O–C3′/C2–O–C2′ but also established the relative
cis-configuration between H1 and CH
3O-2. The centrosymmetric space group
P$\bar{1}$ [
56] indicated that compound 4 was racemic, which was congruent with the close-to-zero optical rotation value. A pair of enantiomers, (−)- and (+)-4, were acquired by leveraging chiral HPLC separation, with a ratio of about 1:1. The specific rotation values were
$[\alpha ]_{\text{D}}^{20}$ −205 and
$[\alpha ]_{\text{D}}^{20}$ +205 for the two enantiomers (−)- and (+)-(4), respectively. Subsequently, a pair of mirror-image curves for the two enantiomers were obtained in the ECD experiment (
Fig. 6). Eventually, the absolute configurations of (−)- and (+)-(4) were assigned as 1
S, 2
S, and 1
R, 2
R, respectively, based on a comparison of experimental and calculated ECD data (
Fig. 6) by using time-dependent density functional theory (TDDFT) [
57], [
58] at the PCM/
ωB97XD/6-311G**//B3LYP/6-31G* (PCM: polarizable continuum model;
ωB97XD: a functional of the
ωB97 family; B3LYP: Becke’s 3 parameters and Lee-Yang-Parr; 6-31G* and 6-311G**: two basis sets) level in methanol. In light of current knowledge, although there are eight members in the C-shaped 1,4-dioxane-fused norditerpenoid heterodimer family [
24], [
40], [
59], [
60], compound 4 represents the first S-shaped 1,4-dioxane-fused norditerpenoid dimer hitherto.
The structure elucidation of compound 4 inspired us to reinvestigate the co-occurring known analogues, compounds 8 and 9 [
40], whose original stereochemical and geometrical elucidation was superficial and questionable. Their NOESY spectra were recollected, and both exhibited H1/CH
3O-2 correlations (
Fig. 7), indicating that the 1,4-dioxane moieties are both
cis-fused. This result suggested that compounds 8 and 9 might be geometrical isomers rather than C2 epimers. In the NOESY spectra, weak correlations of H11/H
318′ in the former and CH
3O-2/H
318′ in the latter were observed, indicating that compounds 8 and 9 are S-shaped and C-shaped, respectively. The assignments were further verified based on a comparison of the chemical shifts of H
318′ of compounds 8 (
δH 2.37 in compound 8 vs
δH 2.42 in compound 4) and 9 (
δH 2.67 in compound 9 vs
δH 2.74 in fimbricalyx C) with those of the S-shaped 1,4-dioxane-fused dimer compound 4 and the C-shaped analogue fimbricalyx C [
59], respectively (
Fig. 7). The observed differences presumably occur because the CH
318′ group in the S-shaped 1,4-dioxane-fused dimer is situated close to the shielding zone of the naphthyl ring in unit A, while its counterpart in the C-shaped 1,4-dioxane-fused dimer is adjacent to the deshielding zone of the C-3 ketone group. Therefore, the structures of compounds 8 and 9 were revised as shown in
Fig. 7. Chiral HPLC-CD analysis revealed that both compounds existed in racemic forms with two pairs of enantiomers featuring mirror-image ECD curves, respectively (
Fig. 8). Chiral separation followed by polarimeter measurement determined the specific rotation values of compounds (+)- and (−)-8 (+149 and −151, respectively) and (+)- and (−)-9 (+325 and −328, respectively). Further TDDFT-ECD calculations (
Fig. 8) confirmed the absolute configurations of the optically pure enantiomers, (+)-8 (1
R, 2
R), (−)-8 (1
S, 2
S), (+)-9 (1
R, 2
R), and (−)-9 (1
S, 2
S), as depicted in
Fig. 7.
In the original report, the specific rotation values of compounds 8 and 9 were +11 and −81 [
40], respectively. As compared with our measured data, the smaller absolute values of the specific rotation indicated that neither compound was enantiomerically pure. It should be noted that both the structures of compounds 8 and 9 incorporate a methyl ketal, which might be formed from the corresponding hemiketal via a reaction with the solvent MeOH in the purification process. However, an early study on their analogue, fimbricalyx C, demonstrated that 1,4-dioxane-fused norditerpenoid dimers containing such methyl ketals are not artifacts [
59]. An HPLC-mass spectrometry (MS) analysis of the EtOH extract of
T. fragilis led to the same conclusion, which revealed that compounds 4, 8, and 9 are all true natural products (Fig. S12 in Appendix A).
Compound 5 had the molecular formula C
16H
18O
4 with eight DBEs, as deduced from its (+)-HRESIMS and
13C NMR data analysis. A comparison of the NMR data (
Table 4) with those of compound 1 disclosed that it possessed similar B/C rings with the unit A partial structure of compound 1. The main difference was the existence of a sp
3 oxygenated quaternary carbon at C10 (
δC 76.9) in compound 5 instead of a sp
2 one in compound 1, which was verified by the HMBC signals from H6 and H11 to C10 (
Fig. 2). A CH
21-CH(OH)
3 spin-spin coupling system was distinguished by the pairwise
J coupling constants (
δH1α 2.48, doublet of doublet (dd),
J = 13.9, 1.1 Hz;
δH1β 2.13, dd,
J = 13.9, 5.4 Hz;
δH3 3.92, dd,
J = 5.4, 1.1 Hz). The HMBC correlations from H
21 to C4, C5, and C10; H3 to C5 and C10; and H
319 to C3, C4, C5, and C18 constructed a five-membered A-ring incorporating the above coupling system. The NOESY signals (
Fig. 3) supported the 2D structure of compound 5 but were unable to determine the relative configuration. Gratifyingly, high-quality crystals were obtained, and the X-ray diffraction study successfully confirmed the absolute configuration (3
R, 10
S) of compound 5 (flack parameter = 0.01 (7)) (
Fig. 5).
Compound 6 was obtained as a brownish-red powder. Its chemical formula was established to be C
17H
14O
5 by means of a
13C NMR and (−)-HRESIMS data analysis. Its 1D NMR data, which featured strongly downfield shifted signals (
Table 4), were extremely similar to those of fimbricalyx B [
59], a highly oxygenated phenanthrenone norditerpenoid. The only difference was the occurrence of one methoxy group in compound 6, as opposed to two in the latter. The HMBC signals (
Fig. 2) from CH
3O-6 (
δH 3.77) to C6 (
δC 152.9) located the methoxy group at C6, and the signals from H11, H14, and H
317 to C12 (
δC 164.7) revealed the existence of a phenolic hydroxyl group at C12 replacing CH
3O-12 in fimbricalyx B. The structure of compound 6 was therefore established.
The molecular formula, C
16H
12O
5, for compound 7 was determined by means of a
13C NMR and (−)-HRESIMS data analysis, which showed 14 mass units less than compound 6. Compound 7's 1D NMR data (
Table 4) displayed remarkable resemblances to those of compound 6, except for the lack of signals for MeO-6, suggesting that compound 7 was a 6-
O-demethylated derivative of compound 6. Further interpretation of the HMBC correlations (Fig. S13 in Appendix A) verified the above assignment.
Aside from compounds 1-9, two known norditerpenoids, thrigonosomone E (compound 10) [
61] and trigonochinene E (compound 11) [
16], were also obtained and characterized by comparing their spectroscopic and MS data with those reported.
Considering the similar structural features of compounds 1, 3, 4, 8, and 9, their hypothetic biosynthetic pathways were thus proposed (
Fig. 4), and the co-isolated known tetra-norditerpenoid compound 10 was considered to be the common biosynthetic precursor. Two norditerpenoid intermediates, i and ii, biogenetically derived from compounds 10 and 5 via oxidation procedures, respectively, would undergo a Michael addition to produce the dimeric intermediate iii, which would then be converted into compound 1 via oxidative degradation. Compound 10 would also be oxidized to the lactonized product iv, which would undergo a DA cycloaddition with a phenylpropanoid, v, to generate a norditerpenoid-phenylpropanoid adduct, vi. Subsequent tandem oxidation/decarboxylation/aromatization transformation of vi would produce the highly aromatized compound 3. In another way, the oxidation of compound 10 would yield the
o-quinone intermediate vii, which could undergo an intermolecular DA cycloaddition with the intermediate viii in two possible regioselective ways to yield the S-shaped (compounds 4 and 8) and C-shaped (compound 9) 1,4-dioxane-fused norditerpenoid heterodimers, respectively.
ACLY is the cytosolic enzyme that converts citrate and CoA to acetyl-CoA, which is an essential precursor in both the glycolytic and lipidic metabolism [
62]. Inhibition of aberrant ACLY expression and activity has been demonstrated to be an effective therapeutic strategy for treating diabetes, fatty liver, and obesity [
63]. Compounds 4-11 were evaluated for ACLY inhibitory activities. Compound 7 exhibited remarkable inhibition (half-maximal inhibition concentration (IC
50) = (0.46 ± 0.11) μmol∙L
−1), with a potency similar to that of the positive control BMS-303141 (IC
50 = (0.60 ± 0.72) μmol∙L
−1). None of the other compounds was active. In particular, a comparison of compound 7 and the inactive compound 6 indicated that the methylation of hydroxyl-6 (HO-6) would dramatically reduce the activity of compound 7. As far as we know, compound 7 is the first naturally occurring norditerpenoid with ACLY inhibitory activity.
In order to better understand the interaction between compound 7 and ACLY, a molecular docking study was performed. The results showed that the interacting pattern of compound 7 in the binding pocket was similar to those of herbacetin [
49] and NDI-091143 [
64]. As depicted in
Fig. 9, compound 7 was located in the hydrophobic cavity of the citrate domain. The aromatic C ring formed an edge-on π-π stacking interaction with phenylalanine (PHE) 354, and the HO-3 and HO-12 hydroxy groups had strong hydrogen-bonding interactions with methionine (MET) 278 and glycine (GLY) 380, respectively.
4. Conclusions
In conclusion, seven new highly aromatic norditerpenoid heterodimers and monomers, compounds 1-7, along with four biosynthetically related known compounds (compounds 8-11), were discovered and characterized from T. fragilis. It is noteworthy that four norditerpenoid heterodimers, compounds 1-4, are strikingly different from the previously reported ones, and all of them feature unprecedented dimerization patterns. Quantum chemical research on the atropisomerism of compounds 1 and 2 provides a valuable template for future studies on similar heterodimers. In addition, two co-occurring norditerpenoid dimers, compounds 8 and 9, were reinvestigated and structurally revised as two pairs of enantiomers, (+)- and (−)-8 and (+)- and (−)-9, respectively. Last but foremost, a highly potent ACLY inhibitor (compound 7) was discovered, and the interaction mode of compound 7 with ACLY was illustrated through a molecular docking study. The good efficacy of compound 7 indicates that the hit compound could be a promising lead structure for the treatment of metabolic diseases.
Acknowledgments
The financial support from the National Natural Science Foundation of China (22237007 and 22177122), the Biological Resources Program of Chinese Academy of Sciences (CAS) (KFJ-BRP-008-001), and the Youth Innovation Promotion Association of Chinese Academy of Sciences (2022282) is gratefully acknowledged. We thank Prof. Shi-Man Huang, Department of Biology, Hainan University, China, for the identification of the plant material.
Compliance with ethics guidelines
Jun-Su Zhou, Long Cheng, Yuan Gao, Zhan-Peng Ge, Bin Zhou, Jing-Ya Li, Jin-Xin Zhao, and Jian-Min Yue declare that they have no conflict of interest or financial conflicts to disclose.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2023.09.015.