Tho Huu Le‡
abc,
Diem Ngoc Thi Lu‡ab,
Hai Xuan Nguyenabc,
Phu Hoang Dang
abc,
Truong Nhat Van Doabc,
Nguyen Thien Han Lebd,
Thang Quoc Truongab,
Minh Hien Nguyen
*bd and
Mai Thanh Thi Nguyen*abc
aFaculty of Chemistry, University of Science, Vietnam National University, Ho Chi Minh City, Vietnam. E-mail: nttmai@hcmus.edu.vn
bVietnam National University, Ho Chi Minh City, Vietnam. E-mail: nmhien@uhsvnu.edu.vn
cResearch Lab for Drug Discovery and Development, University of Science, Ho Chi Minh City, Vietnam
dUniversity of Health Sciences, Vietnam National University, Ho Chi Minh City, Vietnam
First published on 26th June 2025
The phytochemical investigation of the EtOAc-soluble extract of the rhizomes of Curcuma zedoaria (Berg.) Roscoe led to the isolation of five labdane-type diterpenes, including a previously undescribed norditerpene, zedolabdin A (CZ1), and four known compounds (CZ2–CZ5). The structures of these compounds were elucidated using NMR, HR-ESI-MS, and IR spectroscopy, supported by comparisons with literature data. The anti-α-glucosidase evaluation revealed that all compounds exhibited potent inhibitory activity, with zerumin (CZ3) and coronarin C (CZ4) displaying the most potent inhibition, achieving IC50 values of 6.2 μM and 3.0 μM, respectively, significantly lower than the positive control, acarbose (IC50 = 190.6 μM). Molecular docking and dynamics studies identified coronarin C (CZ4) and zedolabdin A (CZ1) as the most promising candidates for α-glucosidase inhibition, exhibiting strong interactions and structural stability. In silico ADMET and toxicity predictions indicated that CZ1 and CZ4 had favorable safety and pharmacokinetic profiles, whereas CZ2 and CZ3 posed higher toxicity risks, with CZ3 also showing potential CYP3A4 inhibition. These findings suggest that CZ1 and CZ4 hold significant potential as novel α-glucosidase inhibitors (AGIs), supporting their further development as safe and effective antidiabetic agents. Moreover, the structural features of CZ1, particularly its hydrogen bonding and hydrophobic interactions, contribute to its enhanced binding affinity and stability within the enzyme's active site. Similarly, CZ4's favorable interactions and pharmacokinetic properties reinforce its potential as a promising AGI candidate, warranting further optimization for drug development.
α-Glucosidases are generally categorized into two main groups, GH-family 13 and 31, based on their sequence similarity. The enzyme's catalytic GH-31 domain (residues 334–779) is conserved, while a variable loop from the N-terminal domain (residues 271–288) plays a role in shaping the substrate binding site.5 Previous studies have focused on the inhibition of α-glucosidase through the docking of diterpenoids using a homology model constructed from the α-glucosidase sequence. This analysis revealed that the 18,19-γ-lactone forms a water-mediated hydrogen bond with H245, while the diterpene structure interacts with a hydrophobic surface within a five-residue binding pocket.6 Despite the availability of several AGIs, many are associated with side effects, highlighting the need for new, safer compounds with enhanced efficacy.
Although in vitro assays to evaluate AGIs are well-established and crucial in drug development, the gap between in vitro findings and the development of drug candidates remains significant. From preliminary studies to a compound's commercial and clinical use, it often spans over a decade and incurs costs exceeding $1 billion.7 Moreover, statistics reveal that a significant proportion of drug candidates fail during development, primarily due to toxicity concerns and, most critically, a lack of clinical efficacy.8,9 Current computational tools extend beyond activity prediction to include the assessment of absorption, distribution, metabolism, excretion, and toxicity (ADMET) profiles of compounds. These predictions are instrumental in identifying compounds with favorable pharmacokinetic and safety profiles, reducing the risk of late-stage failures. By integrating virtual screening, a process that uses computer simulations to identify potential drug candidates, with ADMET predictions, researchers can prioritize compounds for subsequent in vitro and in vivo studies, streamlining the selection process and enhancing the efficiency of drug discovery pipelines.10,11 Furthermore, computational tools enable to predict natural compound interactions with drug targets, providing valuable insights into how specific functional groups or structural motifs contribute to biological activity.12,13
Curcuma zedoaria (Berg.) Roscoe, commonly known as “Nghệ đen” in Vietnam, is a member of the Zingiberaceae family and has a long-standing history of medicinal use. The rhizomes of C. zedoaria are traditionally used in folk remedies to treat gastrointestinal issues, including bloating, indigestion, and gastritis.14,15 This plant has gained recognition for its bioactive compounds, particularly sesquiterpenoids and diterpenoids.16–19 Among the diterpenoids, labdane-type compounds are noteworthy for their broad pharmacological activities, including anti-inflammatory,20 antimicrobial,21 and cytotoxic effects.22 However, their potential as AGIs has not been fully explored. Given the therapeutic importance of AGIs in managing postprandial glucose levels, investigating labdane-type diterpenoids from C. zedoaria could lead to the identification of novel bioactive compounds for the development of new antidiabetic therapies.
In this study, five labdane-type diterpenoids, including a new compound, were isolated from the rhizomes of C. zedoaria. These compounds were evaluated for α-glucosidase inhibitory activity, and all isolates exhibited potent inhibition. The promising results suggest that these compounds have potential as AGIs, offering new possibilities for diabetes management. Although further studies, including molecular docking, and in silico toxicity predictions, are necessary to understand their mechanisms and safety profiles fully, this research serves as a crucial step towards the development of labdane-type diterpenoids as antidiabetic agents. By isolating and characterizing bioactive diterpenoids, this study bridges the gap between virtual screening predictions and experimental validation, providing a comprehensive approach to exploring the therapeutic potential of natural products in diabetes management.
Compound CZ1 was obtained as a colorless gel. The HR-ESI-MS spectrum revealed m/z 349.23775 for the protonated molecular ion [M + H]+ (calcd for 349.23734), indicating its molecular formula C21H32O4. The IR spectrum exhibited absorption for a double bond (3084 and 1715 cm−1), carboxylic acid (2924 and 1698 cm−1), and ester (1735 cm−1). The 1H-NMR spectrum of CZ1 revealed three olefinic protons [δH 7.12 (1H, t, J = 6.5 Hz, H12), 4.83 (1H, brs, H17b) and 4.38 (1H, brs, H17a)], one oxymethylene group [δH 4.86 (2H, s, H14)], two methine groups [δH 1.87 (1H, d, J = 10.4 Hz, H9), 1.12 (1H, dd, J = 12.6 and 3.0 Hz, H5)], six methylene groups [δH 1.07–2.55] along with three methyl groups [δH 0.88 (3H, s, H18), 0.82 (3H, s, H19) and 0.73 (3H, s, H20)] and one acetyl group [δH 2.06 (3H, s, 14–OCOCH3)]. The 13C NMR spectrum exhibited resonances of 21 carbons, including one carboxylic carbon [δC 170.7 (C16)], one carbonyl carbon of an acetoxyl group [δC 170.8 (14–OCOCH3)], four olefinic carbons [δC 153.7 (C12), 148.1 (C8), 125.6 (C13), 107.7 (C17)]. In the high field exhibited signals of one oxymethylene carbon [δC 58.0 (C14)], two quaternary sp3 carbons [δC 39.5 (C10) and 33.7 (C4)], two methines [δC 56.5 (C9) and 55.3 (C5)], six methylenes [δH 19.2–41.9], three methyl carbons [δC 33.7 (C18), 21.9 (C19), and 14.6 (C20)] together with an acetyl group [δC 20.7 (14–OCOCH3)] (Table 1).
Position | δH mult. (J in Hz) | δC |
---|---|---|
1 | 1.71 m | 39.1 CH2 |
1.07 ddd (12.9, 12.8, 3.9) | ||
2 | 1.58 ddddd (13.8, 13.4, 12.8, 4.1, 4.0) | 19.2 CH2 |
1.51 m | ||
3 | 1.41 br d (13.3) | 41.9 CH2 |
1.18 ddd (13.4, 13.3, 4.2) | ||
4 | — | 33.5 qC |
5 | 1.12 dd (12.6, 3.0) | 55.3 CH |
6 | 1.74 m | 24.2 CH2 |
1.34 dddd (12.9, 4.4, 4.4, 3.0) | ||
7 | 2.00 ddd (12.9, 5.1, 4.4) | 37.8 CH2 |
2.40 m | ||
8 | — | 148.1 qC |
9 | 1.87 d (10.4) | 56.5 CH |
10 | — | 39.5 qC |
11 | 2.55 ddd (13.5, 10.4, 6.5) | 24.0 CH2 |
2.40 m | ||
12 | 7.12 t (6.5) | 153.7 CH |
13 | — | 125.6 qC |
14 | — | 170.7 qC |
15 | — | — |
16 | 4.86 s | 58.0 CH2 |
17 | 4.38 brs | 107.7 CH2 |
4.83 brs | ||
18 | 0.88 s | 33.5 CH3 |
19 | 0.82 s | 21.6 CH3 |
20 | 0.73 s | 14.3 CH3 |
14–OCOCH3 | 2.06 (3H, s) | 170.8 qC, 20.7 CH3 |
Combining the 1D-2D NMR data, compound CZ1 exhibits the structure of a norditerpenoid labdane-type skeleton. This structure was clarified by HMBC and COSY correlations (Fig. 2). The acid carboxylic group attached at the C13 was confirmed through HMBC correlations of protons H14/H12 with C16 and the chemical shift at [δC 170.7 (C16)]. The acetoxyl group was identified at the C14 through the HMBC correlations from oxymethylene protons H14 to the carbonyl carbon of the acetoxyl group, together with the chemical shift at [δC 170.8 (14–OCOCH3)]. The presence of double bonds at C12C13 was deduced from HMBC correlations of H12 to C9/C16, as well as H9/H14 to C12, supported by the chemical shift of olefinic carbon C12 (δC 153.7 ppm) and H11/H16 with C13, alongside the chemical shift at C13 (δC 125.6 ppm). Similarly, the HMBC interactions from H17 to C7/C9, and H9/H7 to C17, with the carbon signals at C17 (δC 107.7 ppm), together with the correlations between H6/H11 and C8, supported by the chemical shift of the C8 (δC 148.1 ppm) were confirmed the double bonds at C8 and C17 (Fig. 2). The 1D and 2D NMR spectra of compound CZ1 exhibited a pattern comparable to that of compound CZ2, differing only in the substitution at carbon C16, where a hydroxyl group in CZ2 is replaced by an acetoxyl group in CZ1.
![]() | ||
Fig. 2 The HMBC (solid arrows), COSY (bond lines), and NOESY (dashed arrows) correlations for zedolabdin A (CZ1). |
The relative stereochemical structure of CZ1 was elucidated by analyzing the NOESY spectrum. The NOESY correlations from H7 (δH 2.40)/H6 (δH 1.34), H6 (δH 1.34)/H318, and H5/H9 indicated protons H5, H9, H6 (δH 1.34), H7 (δH 2.40), and methyl groups (C(18)H3) located on the same plane of the decalin moiety. Meanwhile, the NOESY interactions of H1 (1.71)/H11, H2 (1.58)/H320, H320/H319, H319/H3 (1.41) and H320/H6 (1.74) assigned that protons H1 (1.71), H2 (1.58), H3 (1.41), H6 (1.74), H11 together with two methyl groups (C(19)H3 and C(20)H3) oriented on the same plane of the decalin moiety. In addition, the NOESY correlations observed between protons H11 and H214, indicated an E-configuration for the double bond C12 and C13 (Fig. 2). The NOESY experiment confirmed its relative stereochemistry of the skeleton. Based on a comparison of experimental and calculated electronic circular dichroism (ECD) data, the absolute configuration of compound CZ1 was determined. The process began with initial conformational searches utilizing the MMFF force field. The resulting structures were subsequently re-optimized at the B3LYP/6-31G* level using Spartan'14. All conformers contributed over 95% Boltzmann distribution were selected for subsequent DFT (Density Functional Theory) calculations in Gaussian 16. TDDFT (Time-Dependent Density Functional Theory) calculations were then carried out at the B3LYP/6-31G*/CAM-B3LYP/aug-cc-pVDZ level to generate Boltzmann-weighted ECD spectra. The calculated ECD spectrum for the (5S,9S,10S)-CZ1 showed a close match with the experimental spectrum (Fig. 3). All the aforementioned evidence facilitated the elucidation of the structure of CZ1, which is named zedolabdin A.
![]() | ||
Fig. 3 Calculated ECD spectra of (5S,9S,10S)– and (5R,9R,10R)-CZ1 and the experimental ECD spectrum of CZ1 in methanol. |
Five isolated compounds were examined to assess their inhibitory effect on α-glucosidase at concentrations ranging from 1 to 100 μM. The concentration of these compounds for the 50% inhibition of α-glucosidase (IC50) is displayed in Table 2. This study used acarbose, a pharmacological α-glucosidase inhibitor in managing hyperglycaemia, as a positive control (PC).27
Compounds | Inhibition rate (%) | IC50 (μM) | |||
---|---|---|---|---|---|
100 μM | 50 μM | 25 μM | 10 μM | ||
CZ1 | 98.8 ± 0.4 | 96.2 ± 0.4 | 81.4 ± 0.5 | 29.5 ± 1.8 | 14.1 |
CZ2 | 99.1 ± 2.0 | 72.4 ± 2.0 | 46.3 ± 3.1 | 24.6 ± 4.1 | 25.5 |
CZ5 | 99.8 ± 0.2 | 75.8 ± 4.4 | 16.6 ± 3.8 | 3.5 ± 3.6 | 37.6 |
10 μM | 5 μM | 2.5 μM | 1.0 μM | ||
CZ3 | 73.4 ± 2.3 | 35.2 ± 1.9 | 20.8 ± 3.6 | 6.9 ± 4.0 | 6.2 |
CZ4 | 95.4 ± 1.5 | 81.0 ± 1.7 | 34.0 ± 0.9 | 13.7 ± 5.4 | 3.0 |
250 μM | 100 μM | 50 μM | 25 μM | ||
PC | 66.0 ± 1.2 | 32.1 ± 1.7 | 19.7 ± 1.3 | 8.6 ± 1.1 | 190.6 |
The results indicated that all five compounds (CZ1–CZ5) exhibited remarkable α-glucosidase inhibitory activity, with IC50 values ranging from 3.0 to 37.6 μM, surpassing the potency of the positive control. Among them, zerumin (CZ3) and coronarin C (CZ4) demonstrated exceptional inhibitory effects, with IC50 values of 6.2 μM and 3.0 μM, respectively. These findings suggest that the lactone ring closure at the C16 position significantly enhances the structural potency of these compounds, consistent with previous reports in the literature.28,29 Moreover, acetylation at the C14 position was observed to double the enzyme inhibitory activity. This was evident from the bioactivity results of compound CZ1 (IC50 = 14.1 μM) and compound CZ2 (IC50 = 25.5 μM), further emphasizing the critical influence of specific structural modifications on α-glucosidase inhibitory potential.
Although regarding in vitro enzyme assay, the Saccharomyces cerevisiae α-glucosidase enzyme (EC 3.2.1.20) was used due to its commercial availability, its 3D crystallographic structure is not available in public databases, the in silico study was processed on S. cerevisiae isomaltase structure (PDB: 3A4A) with high sequence similarity (84%) compared to S. cerevisiae maltase structure.30 Moreover, both α-glucosidase and isomaltase hydrolyze sucrose and the synthetic substrate p-nitrophenyl-α-D-glucopyranoside.31 These reasons suggest that isomaltase is a potential alternative model for computational analysis, allowing for the primary investigation of enzyme–ligand interactions.
The binding interactions of five compounds from C. zedoaria with α-glucosidase are summarized in Table 3 and Fig. 4. Three compounds, zerumin (CZ3), coronarin C (CZ4), and zedolabdin A (CZ1), demonstrated the highest binding scores, with binding energies of −8.8, −8.7, and −8.4 kcal mol−1, respectively. In contrast, (E)-14-hydroxy-15-norlabda-8(17),12-dien-16-oic acid (CZ2) and (E)-14,15,16-trinorlabda-8(17),11-dien-13-oic acid (CZ5) showed lower binding energies of −7.6 and −7.8 kcal mol−1. These computational results align well with the in vitro assay findings (Table 2), where compounds CZ3 and CZ4 exhibited the most potent α-glucosidase inhibitory activity, followed by CZ1, CZ2, and CZ5. Notably, all five compounds share a decaline moiety, which plays a crucial role in binding affinity by forming π-alkyl interaction with the Tyr158 residue of α-glucosidase and blocking the entrance active site of α-glucosidase. This aligned with previous studies that identified these compounds as competitive inhibitors binding near the active site of α-glucosidase.32,33 Additionally, Arg315 showed a role in stabilizing the enzyme–compound complex through various interactions, further enhancing binding affinity. On the other hand, the lactone moiety of compounds CZ3 and CZ4 interacted with Glu277 and Gln279 (Fig. 4E and G), effectively blocking the enzyme's exit site. This obstruction inhibited the enzyme product's release, impeding the catalytic activity of α-glucosidase. The difference in the position of hydroxyl substituents on the furanone moiety significantly influenced the binding affinity with the enzyme. This observation is consistent with the binding energy scores and in vitro findings, where compound CZ4 demonstrated the lowest IC50 value, followed by compound CZ3, indicating their superior inhibitory activity. To sum up, this finding suggests that compounds CZ3 and CZ4 may inhibit α-glucosidase through a competitive or uncompetitive mechanism, potentially interfering with substrate entrance or releasing exit. Within the scope of the current study, molecular docking provided preliminary insights into the potential mode of inhibition, suggesting either competitive or uncompetitive binding. However, future studies should incorporate enzyme kinetic assays to experimentally validate computational predictions and gain a more comprehensive understanding of the inhibitor's mechanism of action.
Cp | Binding energy (kcal mol−1) | Residue interacts |
---|---|---|
CZ1 | −8.4 | Hbond: Gln279 (2.6 Å) |
π-alkyl/alkyl: Tyr158 (3.6 Å), Phe314 (3.4 Å), Arg315 (3.5 Å) | ||
Unfavorable bond: Arg442 (4.4 Å) | ||
CZ2 | −7.6 | Hbond: Tyr158 (3.7 Å), His280 (2.4 Å) |
Alkyl: Arg315 (3.9 Å), Lys156 (3.4 Å) | ||
π-alkyl: Tyr158 (4.0 Å) | ||
CZ3 | −8.8 | H bond: Tyr158 (2.4 Å), Gln279 (2.4 Å) |
π-alkyl/alkyl: Tyr158 (4 Å), Arg315 (3.8 Å) | ||
CZ4 | −8.7 | H bond: Glu277 (2.0 Å), Gln279 (2.5 Å), Arg442 (2.2 Å) |
π-alkyl: Tyr158 (3.8 Å) | ||
CZ5 | −7.8 | H bond: Tyr158 (1.9 Å) |
Alkyl: Arg315 (3.7 Å) |
![]() | ||
Fig. 4 3D docking poses and 2D interactions diagrams of α-glucosidase binding compounds: CZ1 (A, B); CZ2 (C, D); CZ3 (E, F); CZ4 (G, H) and CZ5 (I, J). |
A molecular dynamics simulation assessed the stability, flexibility, and interactions between α-glucosidase and the five compounds during 100 ns, as shown in Fig. 5. The initial 50 ns allowed the complexes to achieve a stable stage. Among the six simulations, the backbone root-mean-square deviation (RMSD) values for the complexes of α-glucosidase with zerumin (CZ3) and coronarin C (CZ4) were the lowest, stabilizing around 3 Å, which was lower than the unbound α-glucosidase and the complexes with compounds CZ2 and CZ5 (Fig. 5A). Conversely, the RMSD values for the complexes with compounds CZ2 and CZ5 were higher, indicating greater flexibility and less stability. Overall, the RMSD results revealed that α-glucosidase–compound complexes had lower RMSD values compared to the unbound enzyme, aligning with decreased radius of gyration values (Fig. 5B). These findings suggest that the complexes adopt more rigid and stable conformations than the unbound α-glucosidase. Moreover, the MD simulation further revealed that CZ4 formed the highest number of hydrogen bonds among all five compounds, highlighting its robust binding interaction and stability within the complex.
These molecular dynamics results are consistent with molecular docking and in vitro assay results, which demonstrated that coronarin C (CZ4) exhibits promising α-glucosidase inhibitory potential. The MD simulation further revealed the flexibility across all amino acids via root-mean-square fluctuation (RMSF) (Fig. 5D). Residues located in the B domain loop (amino acids 130–160), the active site lid (amino acids 230–236), and the A domain side (amino acids 310–316) exhibited greater fluctuations, corresponding to the substrate-binding pocket formed at the interface of domains A and B of the enzyme.34 Notably, the α-glucosidase–CZ4 complex exhibited lower fluctuations compared to α-glucosidase and other enzyme-compound complexes (Fig. 5D, brown line). The reduced RMSF values indicate that CZ4 stabilized the enzyme structure by limiting residue fluctuations through strong interactions. This observation aligns with the molecular docking results (Fig. 4G and H), confirming the stability of the α-glucosidase–CZ4 complex.
The comprehensive results of the in silico ADMET prediction of five compounds from SwissADME web tool were demonstrated in Table 4. Importantly, all five compounds adhere to Lipinski's Rule of Five, suggesting they are likely to be bioavailable orally. The findings also indicate a high probability of gastrointestinal absorption for these compounds, contributing to their favourable bioavailability.
Compounds | CZ1 | CZ2 | CZ3 | CZ4 | CZ5 |
---|---|---|---|---|---|
Molecular weight | 348.48 | 306.44 | 316.43 | 318.45 | 262.39 |
No. H-bond acceptor | 4 | 3 | 3 | 3 | 2 |
No. H-bond donor | 1 | 2 | 1 | 1 | 1 |
No. Rotatable bonds | 6 | 4 | 2 | 3 | 2 |
TPSA (Å2) | 63.6 | 57.53 | 46.53 | 46.53 | 37.3 |
Log P | 3.11 | 2.72 | 3.39 | 3.24 | 2.85 |
Log S | −4.63 | −4.14 | −4.69 | −4.81 | −4.37 |
Gastrointestinal absorption | High | High | High | High | High |
Log kp (cm s−1) | −5.02 | −5.17 | −4.82 | −4.64 | −4.49 |
CYP1A2 inhibitor | No | No | No | No | No |
CYP2C19 inhibitor | Yes | Yes | Yes | Yes | Yes |
CYP2C9 inhibitor | Yes | Yes | Yes | Yes | Yes |
CYP2D6 inhibitor | No | No | No | No | No |
CYP3A4 inhibitor | No | No | Yes | No | No |
Lipinski violations | 0 violation | 0 violation | 0 violation | 0 violation | 0 violation |
Cytochrome P450 (CYP) is a class of enzymes predominantly found in the liver and intestines, responsible for metabolizing most drugs through oxidation processes. Compounds that inhibit these CYP enzymes can lead to drug interactions, which may reduce drug efficacy or even cause toxicity. Noteworthy, all five compounds likely inhibit CYP2C9 and CYP2C19. All five compounds tested do not inhibit CYP1A2 and CYP2D6, two critical metabolizing enzymes in the liver. However, compounds CZ1 and CZ3 have a potential probability of inhibiting CYP3A4, while compounds CZ2, CZ4, and CZ5 do not inhibit this enzyme. Therefore, it is essential to consider clinically significant pharmacokinetic interactions for compounds CZ1 and CZ3 when they are coadministered with other drugs, as mechanism-based inhibition of CYP3A4 can lead to pharmacokinetic-pharmacodynamic drug–drug interactions. In future studies, it will be important to assess the in vivo pharmacokinetic profiles of all compounds in the presence of CYP3A4 substrates to evaluate interaction risks and ensure safe therapeutic use.
The computational tool Deep-PK was used to assess the predicted toxicity profiles of the compounds. Compounds CZ1, CZ2, CZ4, and CZ5 exhibited mild toxicity, with LD50 values in a rat model of 1667.93 and 1889.56, 1633.26, and 1068.93 mg kg−1, respectively. These values suggest that these compounds have a relatively low risk of acute toxicity. In contrast, compounds CZ3 displayed significant toxicity, with LD50 values falling below 1000 mg kg−1, highlighting the need for caution in their potential applications (Table 5). Moreover, the prediction model identified a specific concern regarding compound CZ3, as it demonstrated a potential to bind to the glucocorticoid receptor. This interaction could contribute to adverse clinical outcomes by disrupting normal glucocorticoid signaling pathways. Since the prediction models for each endpoint and values are established based on different model type, training dataset, there may have discrepant results.35 Thus, although CZ3 is predicted to be safe in terms of liver toxicity and non-carcinogenicity, its LD50 value of 302.2 mg kg−1 indicates a relatively high level of acute toxicity. Despite this, none of the compounds showed significant interactions with key receptors such as the androgen receptor, estrogen receptor, thyroid receptor, or hERG channels, often associated with hormonal imbalances and cardiac arrhythmias. Additionally, the compounds were predicted to be safety concerns to carcinogenicity, liver injury, or micronucleus formation. The in silico toxicity prediction results above were among the most commonly investigated in toxicology.36 These in silico predictions suggest that while compounds CZ1, CZ2, CZ4, and CZ5 may hold promise for further research, the potential risks associated with compound CZ3 require careful consideration and further investigation. Although in silico techniques cannot fully replace in vitro and in vivo methods, they provide valuable insights into mechanism-based toxicity for subsequent in vitro and in vivo experimental validation in establishing toxicity profiles. Numerous computational models based on chemical analogs have been designed for toxicity assessment.36 The in silico approach in toxicity studies allows researchers to identify potential adverse effects early in the development process, ultimately reducing time and resources.
Compounds | CZ1 | CZ2 | CZ3 | CZ4 | CZ5 |
---|---|---|---|---|---|
Acute LD50 (mg kg−1) | 1667.93 | 1889.56 | 302.20 | 1633.26 | 1068.93 |
Carcinogenesis | Safe | Safe | Safe | Safe | Safe |
Liver injury | Safe | Safe | Safe | Safe | Safe |
Micronucleus | Safe | Safe | Safe | Safe | Safe |
hERG inhibitors | Safe | Safe | Safe | Safe | Safe |
Androgen receptor | Safe | Safe | Safe | Safe | Safe |
Estrogen receptor | Safe | Safe | Safe | Safe | Safe |
Glucocorticoid receptor | Safe | Safe | Toxic | Safe | Safe |
Thyroid receptor | Safe | Safe | Safe | Safe | Safe |
Fraction M (6.8 g) was chromatographed on silica gel to give 10 subfractions (subfractions M1–M10). Subfraction M4 (1.6 g) was loaded to silica gel column, eluting with a gradient of n-hexane–EtOAc (v/v, 100:
0 → 0
:
100) to afford four subfractions, including M4.1 (70 mg), M4.2 (238 mg), M4.3 (1.0 g) and M4.4 (11 mg). Subfraction M4.1 (70 mg) was further chromatographed on silica gel via n-hexane–chloroform gradient mixtures (v/v, 100
:
0 → 20
:
80) to yield the compound CZ5 (5.9 mg). Subfraction M4.2 (238 mg) was fractionated by a silica gel column chromatography employing n-hexane–EtOAc mixtures in ascending order of polarity (v/v, 100
:
0 → 40
:
60), producing three subfractions of M4.2.1 (54 mg), M4.2.2 (72 mg), and M4.2.3 (14 mg). 72 mg of subfraction M4.2.2 was further chromatographed on silica gel with a gradient of a n-hexane–EtOAc system (v/v, 100
:
0 → 50
:
50) followed by purified over preparative TLC with n-hexane–EtOAc–i-PrOH mixtures (v/v/v, 80
:
18
:
2) to furnish the compound CZ3 (8.6 mg). Silica gel column chromatography was utilized for further isolation from fraction M5 (1.7 g) using a gradient n-hexane–EtOAc mixtures (v/v, 100
:
0 → 0
:
100) as eluents, this process resulted in four subfractions, consisting of M5.1 (86 mg), M5.2 (437 mg), M5.3 (266 mg), and M5.4 (457 mg). Subfraction M5.3 (266 mg) was subjected to silica gel column chromatography and eluted with n-hexane–EtOAc gradient mixtures (v/v, 100
:
0 → 30
:
70) to afford three fractions: M5.3.1 (17 mg), M5.3.2 (92 mg), and M5.3.3 (94 mg). Subfraction M5.3.3 (94 mg) was submitted to reversed-phase silica gel column chromatography, eluting with an acetone–H2O gradient solvent system (v/v, 10
:
90 → 50
:
50). Thereafter, this fraction was purified by normal-phase preparative TLC with a CH2Cl2–CHCl3–EtOAc–i-PrOH (50
:
30
:
18
:
2) mixture to yield the compound CZ4 (3.5 mg). Fraction M7 (830 mg) was fractionated over normal-phase silica gel column chromatography, eluted with a gradient solvent of n-hexane–EtOAc (v/v, 100
:
0 → 20
:
80) to divide four subfractions: M7.1 (374 mg), M7.2 (76 mg), M7.3 (191 mg), and M7.4 (71 mg). Subfraction M7.1 (374 mg) was passed over a silica gel column eluted with n-hexane–EtOAc mixtures (v/v, 100
:
0 → 20
:
80) to yield three subfractions: M7.1.1 (184 mg), M7.1.2 (56 mg), and M7.1.3 (95 mg). The compound CZ2 (21.7 mg) was obtained through column chromatography on silica gel of subfraction M7.1.1 (184 mg), and consecutively eluted using an isocratic solvent system of n-hexane–EtOAc (v/v, 80
:
20). Finally, the compound CZ1 (5.0 mg) was isolated from M7.4 (71 mg) by chromatography on reverse-phase silica gel, using gradient acetone–H2O mixtures (v/v, 10
:
90 → 40
:
60) as the eluent.
Molecular docking was performed to obtain the binding energies and the interactions between the α-glucosidase and compounds CZ1–CZ5. The 3D crystal structure of the α-glucosidase was obtained from the RCSB Protein Data Bank database (PDB ID: 3A4A) with non-standard residues removed and SMILES formats of natural compounds were converted to 3D structures using the OpenBabel tool. All protein and compound structures were prepared prior to docking by removing all non-standard residues, adding missing hydrogen atoms and assigning charges using the Dock Prep tool of UCSF Chimera program version 1.17.3. Next, the docking process was conducted using AutoDock Vina 1.1.2. Finally, Biovia Discovery Studio 22.1 software was used to visualize docking poses and the interaction types between α-glucosidase and ligands.
Footnotes |
† Electronic supplementary information (ESI) available: Spectral data of five compounds (CZ1–CZ5). See DOI: https://doi.org/10.1039/d5ra03418c |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2025 |