Identification of flavone and flavanone from ethyl acetate fraction of Caesalpinia sappan heartwood and tentative structure-activity relationship: An in vitro and in silico approach

Extraction and fractionation of C. sappan

Dry C. sappan heartwood simplicia collected from Malang, Indonesia, was purchased from Materia Medica, Batu, Malang, and authenticated as secang by Unit Pelaksana Teknis Materia Medica. The voucher sample was deposited in the Brawijaya University Herbarium of Microbiology Laboratory under code number 191113.SCG H.001. A 300 g of C. sappan refluxed with MeOH (volume of MeOH is 5 times dried weight) for 1 hour. The extraction was repeated four times until the bright color of the extract faded. The concentrated MeOH extract under vacuum gave crude extract as a dark red solid (10.2% yield). Crude MeOH extract was partitioned using ethyl acetate according to Ulfa’s et al. (2022) protocols as follows. The MeOH extract was resuspended in water with the ratio of extract and water 1:10 (w/v). The solution was then transferred in the separatory funnel and added with ethyl acetate in the ratio of 1:1 (v/v). The partition was carried out four times until the color of the solution faded. The ethyl acetate fraction was collected from the upper layer, separated, and evaporated using a vacuum. From 30.61 g of MeOH extract, the obtained ethyl acetate fraction was 22.64 g (74%).

Ten grams of the ethyl acetate fraction was purified using a silica gel column and eluted with a gradient solvent (CHCl₃: EtOAc, 6:4 to 0:1), giving ethyl acetate fraction (fraction A) (Ningsih et al., 2020). Fraction A was purified further to get A1–A11 using stepwise gradient elution (CHCl₃: EtOAc = 6:4 to 0:1). Subfraction A3 was purified using the gradient solvent EtOAc: n-hexane 2:8 to 1:0 (v/v), which gave A3.3 as 2’,5,6-trimethoxyflavone (TM) and A3.4 as cirsimaritin (CR). Subfraction A4 was purified using EtOAc:n-hexane 7:3 to 1:0 (v/v)to get A4.1–A4.10. Subfraction A4.2 consists of pinostrobin (PN). Further purification of subfraction A4.2 using CHCl₃: EtOAc in the ratio 8:2 to 0:1 gave subfractions A4.2.4 and A4.2.5 as NR and alpinone (AL), respectively (Fig. 1). The subfraction A5–A11 was not chosen to be purified because their intense red color indicated a high amount of brazilin or brazilein, which was reported as a major compound in C. sappan (Ngamwonglumlert et al., 2020).

In vitro antioxidant activity using DPPH

The antioxidant activity test method followed the standard protocol with some modifications (Udenigwe et al., 2009). All fractions and partitions were dissolved in methanol as a stock of 200 ppm. About 100 μl of each fraction/partition was added to the upper row of the 96-well costar microplate well, diluted stepwise until row 10, and followed by the addition of 100 μl DPPH (0.1 mM) in each well. DPPH solution and methanol were added in rows 11 and 12, respectively. The mixture was incubated for 30 minutes at room temperature. The absorbance of each fraction was analyzed using a microplate reader at 517 nm without shaking. Ascorbic acid (AA) was used as a positive control. DPPH radical scavenging ability was measured using the following equation:

$\%\mathrm{Inhibiton}=\left(\frac{{\mathit{A}}_{\mathit{c}}-{\mathit{A}}_{\mathit{s}}}{{\mathit{A}}_{\mathit{c}}}\right)\times 100\%$

where A_c is the absorbance of the control containing DPPH solution and A_s is the absorbance of the sample. The IC₅₀ value was calculated from y = ax ± b obtained from plotting a graph between %inhibitons (y-axis) and concentrations (x-axis).

Figure 1. Purification of ethyl acetate fraction of C. sappan. [Click here to view]

Chemical constituent analysis using LC-QToF-MS/MS

The identification methods of the molecules were performed by LC-QToF-MS/MS, which was conducted in the Forensic Laboratory of Indonesia National Police. The analysis used a C₁₈ column (1.8 μm, 2.1 × 100 mm) with an injection volume of 5 μl. The combination of water + 5 mM ammonium formic and acetonitrile + 0.05% formic acid was used. The conditions of the column were as follows: column temperature = 50°C, flow rate = 0.2 ml/minute, and 23 minutes of running. Electrospray ionization was used to elucidate the compounds, with a positive mode of detection, mass analysis range = 50–1,200 m/z, source temperature = 100°C, desolvation temperature = 350°C, collision energy = 4 volts, and ramp collision energy = 25–60 volts. The mzCloud Library was used to analyze the proposed compounds from C. sappan extract. From this analysis, subfraction A3.3 was recorded as 2’,5,6-TM and subfraction A3.4 as CR; subfraction A4.2 consisted of PN; subfraction A4.2.4 was identified as NR and subfraction A4.2.5 as AL (Fig. 1).

Molecular docking and molecular dynamics (MDs) studies

All ligands were collected from PubChem in 3D .sdf format and converted to .pdb using Discovery Studio 2019. CP450 (PDB id: 1OG5), LO (1N8Q), NO (2CDU), and XO (3NRZ) macromolecules were retrieved from www.rcsb.org and optimized using Schrodinger Maestro 11.8. Autodocktools 1.5.6 was used for docking all ligands with their receptors. Water residue in the receptors was removed, and Kollman charge was added to the macromolecules. Docking was repeated several times to get root mean square deviation (RMSD) <2 Å (Rizvi et al., 2013). Discovery Studio was used to visualize 2D and 3D interactions between ligands and receptors. The same procedure was also applied for native ligands/controls.

The ligand topology for MDs was prepared using PRODRG and MD and conducted on https://simlab.uams.edu/ with adjusted parameters such as a temperature of 300 K, 20 ns simulation time, 0.15 M NaCl, GROMOS96 43a1 for forcefield, and pressure at 1 atm (Bekker et al., 1993; Schüttelkopf and van Aalten, 2004).

The bond dissociation energy (BDE) and bond dissociation free energy (BDFE) of each –OH from flavone and flavanone derivatives were calculated using density functional theory (DFT) on www.bde.ml.nrel.gov (St. John et al., 2020).

Statistical analysis

The data presented in the figures are representative of three independent analyses that gave similar results. The values are presented as the means ± SD. The differences between data were analyzed using Student’s t-test. The statistical significance was set at p < 0.05 and p < 0.01.

Extraction and fractionation of C. sappan

Dried simplisia of C. sappan extracted using MeOH gave 10.2% of crude extract. Fractionation of the crude extract using ethyl acetate (hydrophobic solvent) afforded 74.0% of fraction A. Based on the radical scavenging activity test using the DPPH method, the constituents in fraction A supposed to have potency as antioxidants, shown by the IC₅₀ = 0.36 ± 0.03 μg/ml (very strong category) (Riza Marjoni and Zulfisa, 2017). Purification of fraction A carried out by SiO₂ open column chromatography gave subfractions A1–A11. Among them, subfractions A3 and A4 are our subjects of interest based on their antioxidant activities. Subfractions A5–A11 collected as a powder with an intense red color suggested the major constituents should be brazilin or brazilein (Ngamwonglumlert et al., 2020).

The purification of subfractions A3 and A4 is depicted in Figure 1. Five compounds consisting of subfractions A3.3, A3.4, A4.2, A4.2.4, and A4.2.5 are determined as classes of flavone and flavanone, that is, 2′,5,6-TM, CR, PN, NR-5-methyl ether, and AL determined by LC-MS/QToF (Table 1). The structures are depicted in Figure 2. All the detected compounds have similar parent fragments to those in previous studies. The 2′,5,6-TM has a molecular formula C₁₈H₁₆O₅ with m/z 312.1074 (M+H)⁺ as reported by Wollenweber et al. (1990). CR (C₁₇H₁₄O₆) was found with m/z 314.0880 (M+H)⁺, as published by Suleimenov et al. (2008). PN (C₁₆H₁₄O₄) was found with m/z 270.0972 (M+H)⁺, similar to the results in the report by Hua et al. (2011). NR-5-methyl ether with the molecular formula (C₁₆H₁₄O₅) with m/z 287.0913 (M+H)⁺ has the same fragment as that published in Hammami et al. (2004). AL (C₁₆H₁₄O₅) was established to have m/z 287.0930 (M+H)⁺ in agreement with Singh et al. (2019).

Table 1. LC-MS/QToF analysis of flavone and flavanone class compounds. [Click here to view]

Figure 2. Structure of the identified compounds. [Click here to view]

Figure 3. Antioxidant activity of subfraction based on the IC50 analysis using the DPPH method. The data represent the mean ± SD (n = 3). **p < 0.01 and *p < 0.05. [Click here to view]

In vitro antioxidant activity of the isolated compounds

The antioxidant activity of subfractions A3.3, A3.4, A4.2, A4.2.4, and A4.2.5 was performed using the DPPH method and compared with AA as a positive control (Fig. 3). Subfractions A4.2 (IC₅₀ = 0.15 ± 0.02 μg/ml), A4.2.4 (IC₅₀ = 0.10 ± 0.03 μg/ml), and A4.2.5 (IC₅₀ = 0.21 ± 0.01 μg/ml) have comparable IC₅₀ values to that of AA (IC₅₀ = 0.29 ± 0.02 μg/ml). Thus, it is suggested that the major constituent in these subfractions should be responsible for stabilizing free radicals. Subfractions A3.3 and A3.4 showed a higher IC₅₀ than that of AA, but in general, they possess a very strong antioxidant property (Riza Marjoni and Zulfisa, 2017). From the LC-MS analysis, PN (28.30%) is the major compound from A4.2, contributing to its antioxidant activity because it has a hydroxyl group. PN was analyzed for its antioxidant activity, with IC₅₀ > 500 μg/ml against DPPH (Christov et al., 2005). NR is a potent antioxidant agent due to its ability to scavenge hydroxyl, hydrogen peroxide, superoxide, and nitric oxide radicals (Rashmi et al., 2018). The NR 5-methyl is suspected of having similar antioxidant activities to the reported data.

Figure 4. Effect of the presence of π-bond at C2=C3 on flavone and flavanone. The data represent the mean ± SD (n = 3). **p < 0.01 and *p < 0.05. [Click here to view]

The variation of the substituent in the core structure of flavone (A3) and flavanone (A4) gives different activities for radical trapping. Comparing the kind and the position of substituent give useful insight for SAR analysis. Three cases were evaluated based on structural differences, as follows: (1) the presence of π-bond at C2=C3 in flavone and flavanone; (2) the presence of OH in flavone; and (3) the presence of OH in aromatic ring (A and B) in flavanone.

From Figure 4, PN in A4.2 exerts better scavenging ability than A3.4 with this only one hydroxyl at rge C5 position attached to PN. In this case, it seems that there is no contribution to antioxidant activity from the extended conjugation system in the ring B and C by the presence of a double bond at the C2=C3 position on the CR (Wang et al., 2018). Even though the IC₅₀ value of PN is lower than that of CR, CR still has a low IC₅₀ value, which might be caused by the methoxyl substituent at C7 and C6 (Sarian et al., 2017).

Comparing the structures of TM and CR in the subfractions A3.3 and A3.4 showed that TM did not have any hydroxyl in the ring (Fig. 5). In contrast, CR has two hydroxyls attached in C5 (ring A) and C4’ (ring B), which are supposed to contribute to their antioxidant activities by the HAT mechanism (Liang and Kitts, 2014). Two hydroxyl groups at ring A and ring B of flavonoids have been reported to exert higher antioxidant activity when compared to one hydroxyl group in either ring (Wang et al., 2018). On the other hand, the O-methylation at the C5 position in TM weakens its scavenging ability (Sarian et al., 2017). For this reason, TM has a lower IC₅₀ than that of CR.

The PN, NR, and AL structures are the major flavonoid compounds in A4 and their subfractions. Among those compounds, NR ranks one as an antioxidant because two hydroxyl groups at ring A (C-7) and B (C-4’) contribute the most to its activity (Fig. 6). However, AL also has two hydroxyls at C-3 and C-5 and supposedly has comparable antioxidant activity with NR. In fact, the BDE calculation of the hydroxyl in AL is 93.2–108.8 kcal/mol higher than that in NR, suggesting that the hydrogen transfer from AL to the DPPH radical is more difficult (Vo et al., 2019). The order of easiness of hydrogen transfer is NR>PN>AL, and this result is supported by in vitro (Fig. 6) and DFT calculation (Table 2). Further detailed investigation of in vitro analysis and apoptosis using cell lines is in progress.

Figure 5. The effect of the presence of OH in flavone. The data represent the mean ± SD (n = 3). **p < 0.01 and *p < 0.05. [Click here to view]

Figure 6. The effect of the presence of OH in aromatic ring (A and B) in flavanone. The data represent the mean ± SD (n = 3). **p < 0.01 and *p < 0.05. [Click here to view]

Table 2. BDE and BDFE of flavone and flavanone from C. sappan. [Click here to view]

In silico study of the isolated compounds

The known enzyme that generates ROS is CP450, 5-LO, NO, and XO (Dharmaraja, 2017). Prior to molecular docking, a validation method is needed to see whether the applied method reproduces valid data compared with the published one. The RMSD values between the calculated and experimental X-ray crystallographic data must be less than 2.0 Å to be considered a success (Hevener et al., 2009). The similarity of conformation between the calculated (blue) and from this experiment (yellow) was tested using molecular docking protocol (Fig. 7). Co-crystal ligands were chosen in this validation due to their inhibition of the macromolecules’ active sites. If the result shows that the RMSD of all receptors is less than 2.0 Å, then the method is validated.

The SARs of TM, CR, PN, NR, and AL are analyzed using those enzymes. First, the analysis suggested that all the identified compounds were able to inhibit the active site of the receptors and have similar or better binding free energy compared to the cocrystal ligands, except for LO (Fig. 8).

The study about the S-warfarin complex with CYP450 reveals a new active site, including the interaction with Arg97, Gly98, Ile99, Phe100, Leu102, Ala103, Val113, Phe114, Asn217, Thr364, Ser365, Leu366, Pro367, and Phe476 residues (Williams et al., 2003). Figure 9 shows that all flavonoid compounds have better binding free energy than 5-fluorouracil (FLU). CR has the best binding energy because it interacts with CP450 residues via a hydrogen bond, but most interactions are not similar to those of S-warfarin. The hydroxyl group in the C7 and C4′ of NR helps to inhibit the active site by interacting with key residues, such as Val113 and ASN217. For the flavone ring, –OCH₃ in the R² position enables it to interact with Phe476, while –OH from CR at R⁴ interacts with Gly98. The hydroxyl group from PN (R¹) interacts with Ala103. From the MD simulation, the most stable ligand is AL with RMSD around 2 Å (Fig. 10).

Figure 7. Molecular docking protocol validation for the receptors: CP450, LO, NO, and XO. [Click here to view]

Quercetin is a known potent inhibitor of the LO enzyme because of its competitive inhibition. The crucial residues that quercetin interacts with are His523 and Gln514 via a hydrogen bond (Borbulevych et al., 2004). The only drug commercially available that can inhibit LO is zileuton (ZIL) (Rossi et al., 2010). In this study, ZIL can interact with His523 and Gln514 by pi-pi T-shaped and hydrogen bonds, respectively. The identified flavonoids from C. sappan only inhibit either one of them, so they are not a good candidate as a LO inhibitor. CR and AL are excluded from Figures 10 and 11 because of their low binding energy. But the interesting part is the hydroxyl group at the C3 position of AL (ΔG = −1.7 kcal/mol) is negligible in the active site of LO because, without the hydroxyl group at the C3 position, the binding free energy of PN (ΔG = −4.05 kcal/mol) is better. Both –OCH₃ from TM (R¹) and NR (R²) interact with His523 via pi-sigma and are stable at 2.4 Å from the MD result (Fig. 10).

Adenosine 5-diphosphate (ADP) is a ligand that occupies the binding site of NO. The residues that interact with the ADP molecule are Ile160, Tyr159, Tyr188, Asp179, and His181. ADP acts as an H₂O₂ blocker, so the hydrogen peroxide will stay inside the complex NO and not dissociate from it (Lountos et al., 2006). All the compounds inhibit the same active site and interact similarly with amino acid residues as ADP (Fig. 12). Dextromethorphan (DEX) is a known NO inhibitor with ΔG = −8.88 kcal/mol⁻¹, but it only interacts with Ile160 and Tyr188 (Chen et al., 2018). The similarity of TM and CR to ADP is better than DEX because the methoxy group interacts with the residues in the active site. The double bond C2=C3 in flavone supports the interaction with Asp179. AL has the most hydrogen bond interaction due to hydroxyl groups, which increases the stability of the complex, but it does not help interact with the important residues (Majewski et al., 2019). However, the –OCH₃ from TM interacts with Tyr188, which might contribute to the stability of TM inside the active site of NO with RMSD around 2 Å (Fig. 10).

Figure 8. Binding energy values generated by molecular docking between ligands and receptors. Control molecules for C450, LO, NO, and XO were 5-FLU, ZIL, DEX, and FEB, respectively. [Click here to view

Figure 9. Interactions of the flavonoid and control (FLU) compounds against the CP450 receptor. [Click here to view]

XO generates ROS in our body while converting hypoxanthine to xanthine by the hydroxylation reaction (Kalra et al., 2007). Glu 802, Glu1261, and Arg880 are critical residues for substrate hydroxylation. A known XO inhibitor like febuxostat (FEB) has been proven to interact with Glu802, Arg880, Thr1010, and Asn768 (Šmelcerovi? et al., 2017). In this experiment, FEB only interacts with Arg880, Thr1010, and Asn768 because of the different molecular docking programs used in the literature (Liu et al., 2022). All the flavonoid compounds do not interact with Glu802, but most interact with Glu1261, except folerogenin (Fig. 13). TM shows the best binding free energy (ΔG = −8.37 kcal/mol⁻¹) because it interacts with two essential residues. The methoxy group of TM at the C-6 position is an important functional group because it interacts with Arg880. To interact with Arg880, the flavonoid compounds must have two methoxy groups at the C-6 and C-5 positions, like TM. From the MD result, TM is relatively stable below 2 Å until 5 ns, but it changes its position after that (Fig. 10).

From Table 3, the antioxidant activities of flavonoids can be enhanced by altering the position of –OH and –OCH₃ groups. In the flavanone structure, the hydroxyl group is crucial for contacting key residues in CP and XO. In this experiment, two –OH groups of NR interact with important residues in those macromolecules, unlike AL. This result correlates with the data obtained from the DPPH method. When comparing in vitro and in silico data, two hydroxyl groups of NR outweigh the ability to scavenge DPPH radicals in the O-methylation groups in TM (Moalin et al., 2011). However, according to MD and molecular docking results, the methoxy groups of TM have better interactions by inhibiting crucial residues than the –OH from NR.

Figure 10. MDs simulation of RMSD ligand movement. [Click here to view]

Figure 11. Interactions of the flavonoid and control (ZIL) compounds against the LO receptor. [Click here to view]

Figure 12. Interactions of the flavonoid and control (DEX) compounds against the NO receptor. [Click here to view]

Figure 13. Interactions of the flavonoid and control (FEB) compounds against the XO receptor. [Click here to view]

Table 3. Interactions between functional groups in flavon and flavanone against prooxidant enzymes. [Click here to view]