2.1. Experimental Results
An HPLC method was applied to purified compounds (
1–
3) from
C. longa extract using Agilent 1260 HPLC system and RP-C18 column (
Figure 1).
Chemical structures of curcumin (
1), demethoxycurcumin (
2), and bisdemethoxycurcumin (
3) were characterized by analyses of NMR spectral data and comparing with literature (
Figure 2).
Compound
1 was obtained as a yellow crystal. The
1H NMR spectrum (
Table 1) showed two sets of ABX aromatic protons at [
δH 7.28 (2H, d,
J = 1.8 Hz, H-6/H-6′), 6.87 (2H, d,
J = 7.8 Hz, H-9/H-9′), and 7.14 (2H, dd,
J = 1.8, 7.8 Hz, H-10/H-10′], which characterized for the 1,3,4-trisubstituted benzene ring, and two methoxy group [
δH 3.89 (6H, s, 7-OCH
3/7′-OCH
3]. Two pairs of
trans-olefinic protons [
δH 6.68 (2H, d,
J = 16.2 Hz, H-3/H-3′) and 7.58 (2H, d,
J = 16.2 Hz, H-4/H-4′], and a methylene group at
δH 5.98 (2H, s, H-1) were also observed in the
1H NMR (
Figure 2 and
Table 1). Consistent with the above
1H NMR analysis, the
13C NMR and DEPT spectra displayed signals of two methoxy groups at
δC 56.2 (7-OCH
3/7′-OCH
3), four olefinic carbons [
δC 122.1 (C-3/C-3′) and 141.4 (C-4/C-4′), two conjugated ketones at
δC 184.4 (C-2/C-2′), a methylene carbon at
δC 101.6 (C-1), and 12 carbons ranging from
δC 111.6 to 150.1 ppm corresponding to two aromatic rings (
Figure 2 and
Table 1). The arrangement of proton-carbon groups were characterized by the aid of HMQC, revealing the structural skeleton of a curcuminoid [
13]. The HMBC showed correlations between H-6 and H-10 to C-4 and C-8; H-6′ and H-10′ to C-4′ and C-8′; H-4 to C-2, C-6, and C-10; H-4′ to C-2′, C-6′, and C-10′; H-3 to C-1 and C-5; H-3′ to C-1 and C-5′; as well as H-1 to C-2, C-3, C-2′, and C-3′ (
Figure 3). The two methoxy groups were found attaching to C-7 and C-7′ due to the correlations of the methoxy protons with carbon C-7 and C-7′ in the HMBC spectrum (
Figure 3). The ESI-MS spectrum of
1 displayed an ion peak at
m/
z 369.1 [M+H]
+, revealing a molecular formula of C
21H
20O
6 for compound
1. Thus, compound
1 was characterized as 1,7-bis(4-hydroxy-3-methoxyphenyl)-1,6-heptadiene-3,5-dione [
28,
29] with a trivial name curcumin.
Compound
2 appeared similar morphology with
1 as yellow crystal. The molecular formula of compound
2 was deduced as C
20H
18O
5 due to the presence of an ion peak at
m/
z 339.2 [M+H]
+ on its ESI-MS. 1D-NMR spectra of compound
2 were quite similar to those of
1, showing an A
2B
2 aromatic spin system [
δH 7.52 (2H, dd,
J = 1.8, 6.6 Hz, H-6′/H-10′), 6.87 (2H, dd,
J = 1.8, 6.6 Hz, H-7′/H-9′)] for one benzene ring and one ABX aromatic spin for one another (
Figure 2 and
Table 1). A difference with
1 that compound
2 possessed only one methoxy group at C-7′ [
δH 3.89 (3H, s) and
δC 56.3]. Thus, compound
2 was determined as 1-(4-hydroxy-3-methoxyphenyl)-7-(4-hydroxyphenyl)hepta-1,6-diene-3,5-dione with a trivial name demethoxycurcumin [
30].
Compound
3 also appeared as a yellow crystal with a molecular formula of C
19H
16O
4 deduced by an ion peak at
m/
z 309.1 [M+H]
+ in its ESI-MS. There was no methoxy group found on its 1D NMR. While two A
2B
2 spin system of two benzene rings [
δH 7.53 (4H, dd,
J = 1.8, 7.8 Hz, H-6/H-10/H-6′/H-10′), 6.88 (4H, dd,
J = 1.8, 7.8 Hz, H-7/H-9/H-7′/H-9′)] were observed. Detailed analyses of the
1H and
13C NMR data of compound
3 and compare with those of compounds
1 and
2 lead to the structural identification of compound
3 to be 1,7-bis(4-hydroxyphenyl)-1,6-heptadiene-3,5-dione (bisdemethoxycurcumin) [
30].
Inhibitory effects of curcumins
1‒
3 on protein tyrosine phosphatase 1B (PTP1B) enzyme activity were tested in vitro using a triterpene (ursolic acid) and a quinone emodin as positive control [
31]. As shown in
Table 2, all compounds potential inhibited this enzyme action with IC
50 values of 37.8 ± 1.4 (curcumin
1), 45.3 ± 0.7 (demethoxycurcumin
2), and 72.6 ± 1.1 µM (bisdemethoxycurcumin
3). Emodin possessed potential inhibition with an IC
50 value of 7.6 ± 0.3 µM. Among these curcumins, curcumin (
1) bearing two methoxy moieties at C-7 and C-7′ (
Figure 2), displayed the most potential inhibition against PTP1B (IC
50 value of 37.8 ± 1.4 µM), demethoxycurcumin (
2) bearing one methoxy group at C-7 (loss of one methoxygroup at C-7′) exhibited less inhibitory activity (IC
50 value of 45.3 ± 0.7 µM) than
1. Bisdemethoxycurcumin (
3) without methoxy moiety in its structure showed moderate inhibition (IC
50 value of 72.6 ± 1.1 µM). This experimental result suggests that substitution of methoxy moiety at C-7 and/or C-7′ in curcuminoid compounds may play a key role for enhancing their inhibitory effect on PTP1B enzyme activity.
Li et al. reported that curcumin (
1) possessed inhibition on PTP1B action and affected on the liver of rats with fructose-fed. At 60 mg/kg, curcumin showed similar effect with pioglitazone on liver PTP1B expression, but showed less effect against PTP1B activity than pioglitazone at a concentration of 10 mg/kg [
16]. Recently, Kostrzewa et al. demonstrated curcumin showing antidiabetic and anticancer potential via inhibition of PTP1B [
17]. Curcumin also reduced the enzymatic activity of PTP1B phosphatase and MCF7 cell viability by 50% inhibition at a concentration of 100 μM. In our in vitro study, curcumin (
1) showed more potent inhibition with 50% inhibition at a concentration of 37.8 µM on PTP1B action. This assay has reported for the first time the inhibitory activities of demethoxycurcumin (
2) and bisdemethoxycurcumin (
3) against PTP1B.
Regarding to α-glucosidase inhibitory experimental assay, acarbose and curcumins (
1‒
3) were tested using α-glucosidase enzyme from rat intestinal tract. In this assay, acarbose displayed a similar inhibitory IC
50 value (147.2 ± 1.0 μM) toward α-glucosidase [
22]. Furthermore, curcumin (
1) potentially inhibited the action of α-glucosidase with an IC
50 value of 78.2 ± 0.2 μM, almost two times stronger than acarbose. Demethoxycurcumin (
2) showed an IC
50 value of 82.4 ± 0.8 μM and bisdemethoxycurcumin (
3) showed less activity than
1 and
2, with an IC
50 value of 90.6 ± 1.0 μM. Similar aspect with PTP1B enzyme, the inhibitory effect of curcumins
1‒
3 against α-glucosidase was also decreased by the demethoxylation in the curcuminoid structure. This leads to suggest that methylation or methoxyl substitution in the design and synthesis of curcumin analogs may be a key for the search and development of α-glucosidase inhibitory agents.
A published result reported that curcumin (
1) exhibited a mild inhibition effect on α-glucosidase through a competitive inhibition mechanism with its IC
50 value of 20.54 ± 1.02 mg/mL [
32]. Du et al. studied the inhibition of natural curcumins and the synthetic curcumin analogs on α-glucosidase enzyme. However, only natural bisdemethoxycurcumin (
3) showed inhibitory activity with an IC
50 value of 23.0 μM against α-glucosidase [
33], while the inhibitory effects of curcumin (
1) and demethoxycurcumin (
2) were ambiguously mentioned. Another report demonstrated the inhibition of curcumin (
1) and other curcuminoids on enzymatic α-glucosidase activity. But, their inhibitory effects were moderate at 50 μg/mL and at mM level [
34,
35]. In our experimental assay, α-glucosidase enzyme was obtained from rat intestine and the positive control acarbose showed a stable IC
50 value [
22]. Consequently, the result obtained in this study was accurate and reachable. And that the inhibitory activity of demethoxycurcumin (
2) on α-glucosidase enzyme has first reported.
2.2. Computational Studies
In this study, DFT and MD were the two utilized techniqs used to investigate the interaction between curcumins (1–3) and the targeted protein (3W37, 3AJ7 and PTP1B).
Figure 4 shows the geometrically optimized structures of compounds 1–3 and
Figure 5 shows the HOMO and LUMO frontier molecular orbitals,
Table 3 listed the related quantum parameters for these compounds. These compound structures are identical, compounds 1, 2, and 3 contains -OH and C=O functional groups, the C=O bonds are ~1.21Å, whereas C-OH are 1.35 Å. These were reported to be highly conducive to polarity and solubility of the host compound [
36,
37], thus implying promising inhibitability towards protein molecules based on polar interactions with highly polarized amino acids. Moreover, these compounds are suitable for intermolecular inhibition suggested by bonding analysis on frontier molecular orbitals. As shown in
Figure 5, the large lobes of the HOMO and LUMO are both mainly located on the aromatic ring, with O (C=O bonds), O (-OH group), C (C=C-C bonds). This is indicated that to uphold electron-transferring interactability, the molecules are able to initiate intermolecular inhibition from certain approaching manners. In addition, the parameter values which calculated for these curcumins are not significantly differences. Particularly, the energy gaps (DEGAP) of curcumins (1–3) are -5.962, -5.724 and -5.717 eV, respectively. These obtained values are considered to be low values which may lead to chemical reactivity and inhibitory stability [
38,
39].
The reason is thought to be that electrons of inhibitory molecules are easily activated and transferred to their surface, ready for intermolecular activities. Furthermore, electronegativity (c), or the chemical potential (m) in a negative value, could be considered as a reliable inhibition indicator since it presents an electron-attracting tendency. In principle, a higher electronegativity implies a stronger attraction of electrons towards the host molecule. Thus, these curcumins (1–3) seem promising for docking study.
As the X-ray structure of PTP1B was not well characterized, we built a three-dimensional structure of PTP1B by homology modeling on the Swiss-Model webserver. It is commonly assumed that a good quality model is expected to have a score over 90% in the most favored regions [
35]. Obtained data from Ramachandran plot showed that 98.27% amino acid residues of the PTP1B model located in the most favored regions (
Figure 6A,B). In addition, the modeled structure of PTP1B was superimposed with three previously published crystal structure of PTP1B (
Figure 6C). The sequence alignment between these models exhibit significant high identity (at least 96.98%), thus, this model could be considered as a liable model for further docking studies.
Amongst various docking softwares, AutoDock4 (AD4) is a non-commercial package that is recognized and used during the last ten years with about 6000 publicaitons. This is an useful tool to rapidly predict the binding affinity of ligands towards a specific protein/ targeted enzyme [
41,
42], thus, we chose AD4 to perform docking simulation. According to the ranking criteria of Autodock4, the negative value of docking energy express the binding affinity of the compound towards targeted receptor, this means that the value becomes more negative showing better binding affinity [
41,
43].
The binding site of targeted proteins are investigated using 3DLigandSite webserver (
https://www.wass-michaelislab.org/3dlig) in which they are marked as site 1 (yellow), site 2 (cyan), site 3 (magenta), site 4 (blue) (
Figure 7). Particularly, the key residues of each site are presented in
Table 4. Emodin, ursolic acid and acarbose were selected as reference ligands.
Obtained data showed that all the binding sites are constituted from a larger number of different amino acids, thus, they are assumed as highly conductive to peripheral interactions. Considering α-glucosidase protein (PDB ID: 3W37), a total of 21 amino acids were detected in site 2 binding region which is significantly higher than the other sites. The docking score also proved this is the most suitable site for curcumin compounds given the lowest value of binding free energy (varying from -6.99 to -12.36 kcal/mol) (
Table 6). In terms of 3AJ7, site 1 and 3 also expected to be favorable for inhibitors to form interaction based on their dominant in the numbers of constitute residues. Docking analysis has pointed out that site 3 provide the highest binding affinity towards studied ligands, especially curcumin compounds (varying from -9.99 to -10.65 kcal/mol). Regarding protein PTP1B, all the detected binding sites showed no noticeable differences in the number of in-pose amino acids. In this case, docking score revealed that site 3 is the most preferential region for inhibitors to bind with (varying from -8.09 to -12.75 kcal/mol).
Table 5.
Docking score of studied compounds on 3W37, 3AJ7 and PTP1B proteins.
Table 5.
Docking score of studied compounds on 3W37, 3AJ7 and PTP1B proteins.
Compounds |
3W37 |
3AJ7 |
PTP1B |
Site 1 |
Site 2 |
Site 3 |
Site 4 |
Site 1 |
Site 2 |
Site 3 |
Site 4 |
Site 1 |
Site 2 |
Site 3 |
Site 4 |
1 |
-4.49 |
-12.36 |
-8.38 |
-7.42 |
-9.98 |
-9.09 |
-10.65 |
-8.51 |
-7.47 |
-8.06 |
-9.57 |
-8.12 |
2 |
-5.29 |
-12.16 |
-7.50 |
-7.42 |
-9.80 |
-9.06 |
-9.99 |
-9.58 |
-6.85 |
-7.87 |
-8.93 |
-7.49 |
3 |
-6.43 |
-11.14 |
-7.63 |
-6.42 |
-9.04 |
-7.76 |
-10.60 |
-9.90 |
-6.47 |
-8.27 |
-8.09 |
-7.52 |
Emodin |
-4.65 |
-6.99 |
-6.25 |
-5.83 |
-7.53 |
-7.16 |
-7.90 |
-7.74 |
-6.03 |
-7.07 |
-11.31 |
-5.51 |
Ursolic acid |
-5.27 |
-7.36 |
-3.70 |
-6.20 |
-8.81 |
-8.01 |
-7.68 |
-8.80 |
-7.43 |
-7.28 |
-12.37 |
-5.98 |
Acarbose |
-6.61 |
-12.22 |
-6.06 |
-12.78 |
-13.15 |
-12.46 |
-12.72 |
-13.24 |
-12.22 |
-8.60 |
-12.75 |
-10.59 |
The detailed docking simulation interaction of curcuminoid compounds (
1–
3) are summarized in Table 7. Overall, all curcumins (
1–
3) exhibit the most inhibitory effect toward α-glucosiase (PDB ID: 3W37) than the others. The most negative values of binding free energies were recorded for curcumin (
1), demethoxycurcumin (
2) and bisdemethoxycurcumin (
3) as -12.36, -12.16 and -11.14 kcal/mol, respectively. The reference ligand, acarbose, was recorded to exhibit high affinity toward this protein with dock score of -12.22 kcal/mol. This ranking were highly correlate with experimental data since the IC
50 of tested compounds were: compound
1 (78.2 ± 0.2 µM), compound
2 (82.4 ± 0.8 µM), compound
3 (90.6 ± 1.0 µM), acarbose (147.2 ± 1.0 µM). As reported in previous study [
22], Arg676 was assumed to display important role in the functional of α-glucosidase (3W37). Interaction formed with this residue seem to induce sevear conformational changes on the enzyme PTP1B, thus, causing loss of its normal functionality. Obtained data proved that, all three studied ligands form H-bond with Arg676, particularly, the number of hydrogen bonds formed between 3W37 and curcumin compounds (
1–
3) were 3, 5 and 4, respectively. Additionaly, their interaction were further strengthen through Van der Waals bonding (
Figure 8). In terms of oligo-1,6-glucosidase protein (PDB ID: 3AJ7), although the experiments are not included in this study, docking simulation were conducted to gain better insight regarding mechanism of action of studied compounds towards this protein since it is a well-known drug target for diabetes treatment. Initially, docking conformation analysis revealed that residue Asn317 is assumed to play an important role in the fuction of this protein. This hypotheses would need further experiment to validate the inhibitory mechanism of action. Regarding PTP1B protein, the dock score ranking of studied ligands (ursolic acid > emodin >
1 >
2 >
3) showed high correlation with experiment-based assay result on PTP1B inhibition, in which the IC
50 value are: ursolic acid (4.3 ± 0.4 µM), emodin (7.6 ± 0.1 µM), compound
1 (37.8 ± 1.4 µM), compound
2 (45.3 ± 0.7 µM) and compound
3 (72.6 ± 1.1 µM). Dock pose analysis showed that amino acid Leu204 participated in forming hydrogen bonds with all the studied curcumin compounds within the site 3 binding region. Therefore, this residue might play a key role in the development of potential compounds to inhibit the function of this protein.