Combined Autodock and comparative molecular field analysis study on predicting 5-lipoxygenase inhibitory activity of flavonoids isolated from Spatholobus suberectus Dunn

2.1.1 Ligand preparation:

The two-dimensional structures of the S. suberectus flavonoids (Figure 1) were constructed by ChemDraw Ultra and then converted into three-dimensional structures by Chem3D Ultra in which energy minimization was performed using the MOPAC program [16] and the AM1 semiempirical Hamiltonian. Before the docking trials, nonpolar hydrogen atoms were merged and Gasteiger charges were added. The rotatable bonds in the ligands were detected by Autodocktools 1.5.4 [17] and were all released with the active torsions that move the fewest atoms.

2.1.2 Protein preparation:

The three-dimensional structure of 5-LOX (pdb code: 3O8Y) was obtained from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (New York, NY, USA). 5-LOX has two same chains: A-chain and B-chain [10]. This structure is used for docking by deleting B-chain and retaining A-chain. Before the docking trial, nonpolar hydrogen atoms were merged, nonintegral charges on 5-LOX were corrected, and Gasteiger charges were also added by Autodocktools 1.5.4.

2.1.3 Docking parameter setting:

The number of points in the x-, y-, and z-dimensions was 100, 100, 100, respectively. The grid spacing was 0.375 Å, and the coordinates of the grid center were 8.725, 22.431, and −1.338. A Lamarckian genetic algorithm was used to search for the optimal binding mode within the 5-LOX active site. The number of genetic algorithm runs and the maximum number of energy evaluations were 100 and 2.5 million, respectively. The translation step was 0.2, and the quaternion step and the torsion step were both set to 5.0 during the search. All other parameters not mentioned were set to default.

3.1.1 Results with established inhibitors of 5-LOX

Before predicting the inhibitory activity of flavonoids from S. suberectus against 5-LOX, as shown in Table 2, seven established inhibitors of 5-LOX were selected to evaluate the performance of Autodock 4.2 used in this study. There was a good relationship between the binding free energies and the pIC₅₀ (negative decadic logarithm of IC₅₀) values. The value of the coefficient of determination (R²) is 0.838; SD and P are 0.216 and 0.00355, respectively.

Table 2

Seven established inhibitors of 5-LOX.

Compound	Structure	IC50, μM	References
Caffeic acid		46	[24]
Baicalein		1.2	[24]
Catechol		62	[24]
Quercetin		2.1	[24]
Esculetin		4.5	[24]
Luteolin		0.8	[25]
Epicatechin		22	[26]

3.1.2 Prediction of the inhibitory activities of the S. suberectus flavonoids

The more negative the binding free energy of a ligand, the stronger is its inhibitory activity. As can be seen in Table 3, the S. suberectus flavonoids are predicted to inhibit 5-LOX to different extents, with 3′,4′,7-trihydroxyflavone, plathymenin, butin, catechin, liquiritigenin, eriodictyol, dihydroquercetin, dihydrokaempferol, and gallocatechin having stronger inhibitory activity.

3.2.1 Results for the inhibitors in the training set

Before generating the CoMFA model, 21 inhibitors of 5-LOX were chosen as the training set (see Table 1). The experimental IC₅₀ values, in molar concentration, were converted to the pIC₅₀ (negative decadic logarithm of IC₅₀) values. The optimum number of components generated is 6 by one cross-validated analysis (i.e., the leave-one-out), and the R² is 0.710. The value of R² is 0.997 with no validated partial least square (PLS) analysis with the optimum number set to 6; the SE of estimate and the F value (n₁=6, n₂=14) are 0.0960 and 692.059, respectively, and the probability (R²=0) is 0.000. The contribution of steric and electrostatic factors to the generated CoMFA model is 0.478 and 0.522, respectively.

As shown in Figure 3, in the CoMFA model generated of the electrostatic map, more positive charges near C3 and C6 and more negative charges near C2 may increase binding. In the steric map, more bulky groups near C2 and C7 and less bulky groups near C3, C5, and C6 may increase binding.

Figure 3:

7-{[5-fluoro-3-(1,1,1,3,3,3-hexafluoro-2-hydroxypropan-2-yl)phenyl]sulfanyl}-4-(furan-3-yl)-2H-chromen-2-one, the most potent inhibitor of 5-LOX in the training set, within the contour of the CoMFA model.

Based on the steric contours, the green areas mean that more bulky groups near these areas are favorable to increasing binding and the yellow areas mean that less bulky groups near these areas are favorable to increasing binding. Based on the electrostatic contours, the blue areas mean that more positive charges near these areas are favorable to increasing binding and the red areas mean that more negative charges near these areas are favorable to increasing binding.

The contour of the CoMFA model of the flavonoid inhibitor luteolin is shown in Figure 4. In the electrostatic map, more positive charges near C3, C6, C2′, and C5′ and more negative charges near C2 and C4′ may increase binding, whereas, in the steric map, more bulky groups near C2, C5′, and C7 and less bulky groups near C3, C5, C6, C4′, and C5′ may increase binding.

Figure 4:

Luteolin, a flavonoid inhibitor of 5-LOX, within the contour of the CoMFA model.

Based on the steric contours, the green areas mean that more bulky groups near these areas are favorable to increasing binding and the yellow areas mean that less bulky groups near these areas are favorable to increasing binding. Based on the electrostatic contours, the blue areas mean that more positive charges near these areas are favorable to increasing binding and the red areas mean that more negative charges near these areas are favorable to increasing binding.

3.2.2 Prediction of the inhibitory activities of the compounds from S. suberectus

Before predicting the inhibitory activity of compounds from S. suberectus, the performance of the model generated by CoMFA must be evaluated. To confirm the performance of the model, the inhibitory activities of five inhibitors in the test set whose IC₅₀ values had been reported were predicted by the model (see Table 4). The predicted pIC₅₀ values of the five inhibitors in the test set using the generated CoMFA model are close to those obtained experimentally. The value of the R² is 0.875; SD and P are 0.236 and 0.0195, respectively.

Table 4

Five inhibitors in the test set of CoMFA.

Compound	Structure	IC50 (μM)	References
7-{[3-Fluoro-5-(3-hydroxypentan-3-yl)phenoxy]methyl}-4-(furan-3-yl)-2H-chromen-2-one		1.5×10−2	[22]
7-({3-Fluoro-5-[1-hydroxy-1-(thiazol-2-yl)propyl]phenoxy}methyl)-4-(4-fluorophenyl)-2H-chromen-2-one		0.175	[22]
7-{[3-Fluoro-5-(1,1,1,3,3,3-hexafluoro-2-hydroxypropan-2-yl)phenoxy]methyl}-4-(4-fluorophenyl)-2H-chromen-2-one		5.5×10−2	[22]
7-({3-Fluoro-5-[1-hydroxy-1-(6,8-dioxa-bicyclo[3.2.1]octan-1-yl)methyl]phenoxy}methyl)-4-(furan-3-yl)-2H-chromen-2-one		0.2	[22]
7-({3-Fluoro-5-[1-hydroxy-1-(6,8-dioxa-bicyclo[3.2.1]octan-1-yl)methyl]phenoxy}methyl)-4-phenyl-2H-chromen-2-one		0.3	[22]

The larger the value of the pIC₅₀ value obtained, the stronger is the inhibitory activity of the corresponding compound. As shown in Table 3, 3′,4′,7-trihydroxyflavone, liquiritigenin, catechin, gallocatechin, butin, and plathymenin are the strong inhibitors of 5-LOX.

3.2.3 Molecular binding mode

The molecular binding mode of the 5-LOX ligands is illustrated in Figure 5. The conformation of the ligands was obtained based on the lowest binding free energy scored by Autodock 4.2.

Figure 5:

Interactions between the ligands (A) baicalein, (B₁ and B₂) eriodictyol, and (C) (2S)-7-hydroxy-6-methoxy-flavanone and 5-LOX.

The thin green solid lines represent the hydrogen bonds. The thin red solid lines represent the ligands. The amino acid residues in the active site are shown in bold lines. As there are six hydrogen bonds between eriodictyol and 5-LOX, it is difficult to distinguish the hydrogen bonds in one presentation, so two presentations (B₁ and B₂) are provided for eriodictyol.

Three compounds from S. suberectus [i.e., baicalein and eriodictyol in the conformation of their lowest calculated binding free energy, and (2S)-7-hydroxy-6-methoxy-flavanone, with the highest binding free energy] were selected as examples to define the molecular binding mode and the likely inhibitory mechanism. As shown in Figure 5A, there are four hydrogen bonds in the complex between baicalein and 5-LOX. One hydrogen bond is formed between the oxygen atom of the 5-hydroxy group in ring A and a hydrogen atom of the imino group in the chain of ARG401 (distance 1.868 Å). The other two hydrogen bonds are located between the hydrogen atoms of the 6- and 7-hydroxy groups in ring A and a carboxyl oxygen atom in the chain of ASP170, whose distances are 1.845 and 1.947 Å, respectively. The fourth hydrogen bond is formed between the oxygen atom of the 7-hydroxy group in ring A and the hydrogen atom on the nitrogen of the indole ring of TRP102 (distance 1.837 Å). Figure 5B₁ and B₂ shows the complex formed between eriodictyol and 5-LOX. There are six hydrogen bonds: ASP170, 5-OH of ring A (1.906 Å); TRP102, 5-OH of ring A (2.178 Å); GLU622, 7-OH of ring A (1.797 Å); ALA405, 3′-OH of ring B (1.998 Å); ILE167, 3′-OH of ring B (1.893 Å); and VAL400, 4′-OH of ring B (1.964 Å). On the contrary, there is only a single hydrogen bond in the complex of (2S)-7-hydroxy-6-methoxy-flavanone and 5-LOX (ASP170, 7-OH of ring A; 1.839 Å). Therefore, the hydrogen bonds formed could stabilize hydrophobic interactions.