Computational investigations of physicochemical, pharmacokinetic, toxicological properties and molecular docking of betulinic acid, a constituent of Corypha taliera (Roxb.) with Phospholipase A2 (PLA2)

The acid dissociation constant (pKa), distribution coefficient (logD), partition coefficient (logP) and aqueous solubility (logS) of betulinic acid (BA) over the pH range of 0.0 to 14.0at 298 K were calculated using MarvinSketch 15.6.29 (ChemAxon, Hungary) (http://​www.​chemaxon.​com). The consensus logP method was applied to calculate distribution coefficient of the molecule.

The rapid progress in computational methods such as Gaussian family of methods (G1, G2, G2MP2, G3) and complete basis set extrapolation (CBS) method enable researchers to perform highly sophisticated calculations of enthalpies and thermochemical properties with minor errors in comparison to experimental data [47‐51]. However, these methods are computationally very expensive. An alternative to these high cost calculations is the use of density functional theory (DFT) methods. Previous report revealed that [52] calculation of geometries (bond length, bond angle and dihedral angle) and thermochemical properties using DFT/B3LYP level of theory showed better agreement with the experiments. So, in the current investigation, the calculation of solvation free energy, dipole moment, polarizability, hyperpolarizability and global reactivity descriptor properties such as the chemical hardness, softness, chemical potential, electronegativity, electrophilicity index were conducted in gas phase and in different solvents namely water, dimethyl sulfoxide (DMSO), acetonitrile, n-octanol, chloroform and carbon tetrachloride with the B3LYP/6-31G(d) level of theory implemented in Gaussian 09 software package [53]. All calculations were conducted using the optimized geometry which was confirmed by the absence of imaginary frequency in the lowest energy state of the molecule. The Solvation Model on Density (SMD) [54] was used for all calculations involving the solvents. All calculations involving solvation were performed using the optimized solution-phase structures.

The pharmacokinetic and toxicological properties were calculated using online server PreADMET (https://​preadmet.​bmdrc.​kr/​). The pharmacokinetic properties such as human intestinal absorption (HIA), in vitro cellular permeability using Caco-2 cell model, skin permeability (P_Skin), plasma protein binding (PPB) and penetration of the blood-brain barrier (BBB), interaction with P-glycoprotein (Pgp) and metabolism (both phase I and phase II) were calculated and predicted. In addition, BA was virtually screened to evaluate toxicological properties such as mutagenicity, carcinogenicity and risk of inhibition of human ether-a-go-go-related (hERG) gene.

The calculation of pKa revealed that BA has two ionized species along with the unionized form (Fig. 5). It is clear from the figure that non-ionized form (Betulinic acid_1) predominates over the pH range of 0.0 to 4.6 whereas the betulinate form (Betulinic acid_3) dominates from pH 4.8 to 14.0. However, small amount of Betulinic acid_2 was present over the pH of 0.0 to 2.4.

The logD was calculated using consensus method implemented in MarvinSketch 15.6.29. The logD vs pH of BA is presented in Fig. 6. The logD value decreases with increasing pH of the solution suggesting that the prevalence of the unionized form of BA decreases and the ionized form (betulinate) increases with increasing pH of the solution. This result is in accordance with the calculation of pKa.

The partition coefficient (logP) of BA was also computed and presented in Table 1. The logP calculated in MarvinSketch 15.6.29 was 6.64 whereas the experimental logP of BA were reported as 6.61 [61] and 6.85 [62].

The aqueous solubility of BA in terms of logS is presented in Fig. 7. The figure indicates that the solubility increases with increasing pH of the solution. However, the intrinsic solubility of BA was found − 7.34 (0.000000046 mol/L) and the aqueous solubility at pH 7.4 was − 4.79 (0.000016 mol/L) indicating that BA is practically insoluble in water.

The solvation free energies of BA were calculated with the SMD model [54] The values are summarized in Table 2. The solvation energies of BA in water, DMSO, acetonitrile, n-octanol, chloroform and carbontetrachloride were − 41.74 kJ/mol, − 53.80 kJ/mol, − 66.27 kJ/mol, − 69.64 kJ/mol, − 65.96 kJ/mol and − 60.13 kJ/mol, respectively. It could therefore be concluded that solvation free energy increases with decreasing polarity of polar nonprotic solvent (DMSO and acetonitrile). The highest value of solvation energy was in non-polar protic solvent (n-octanol). The free energy decreases with decreasing polarity of non-polar aprotic solvent (chloroform and carbon tetrachloride). Thus, it could be stated that BA associates with non-polar portion (hydrocarbon chain of n-octanol) via hydrophobic interactions. Electrostatic interactions might arise due to the association of the carboxyl group of BA with the polar portion (hydroxyl group of n-octanol) (Table 5).

The dipole moment of BA is found to be higher in different solvents than that of the gas phase. Table 3 presents the dipole moments computed in the gas phase and different solvents (water, DMSO, acetonitrile, n-octanol, chloroform and carbon tetrachloride) at the B3LYP level of theory with 6-31G(d) basis set using SMD solvation model. The dipole moments were 2.98D, 5.13D, 4.50D, 4.53D, 4.54D, 3.96D and 3.46D in the gas phase, water, DMSO, acetonitrile, n-octanol, chloroform and carbon tetrachloride, respectively. Therefore, increasing dielectric constant of the solvent is accompanied by a gradual increase in the dipole moment. In other words, the dipole moment increases with the increasing polarity of the solvent (Fig. 8).

The polarizability(α) of BA was calculated using the following equation:

The quantities α_xx, α_yy and α_zz are known as principal values of polarizability tensor.

The calculated polarizability of BA in different solvents is presented in Table 4, which showed that polarizability (α _tot ) ranged from 351.90 to 448.67 a.u. The plot of polarizability vs. solvent is shown in Fig. 9. It is clear from the figure that the polarizability gradually increases when going from lower to higher dielectric constant. This suggests that the kinetic reactivity of BA increases with increasing polarity of the solvent [63].

The first order hyperpolarizability (β _tot ) is the measure of the nonlinear optical activity and can be calculated using the following equation:

Where,

The β _tot for different solvents was listed in Table 4, which displayed that the hyperpolarizability in different solvents ranged from 184.48 to 246.57 a.u. Moreover, the hyperpolarizability increases as the dielectric constant increases except for non-polar protic n-octanol where the hyperpolarizability is higher than that of acetonitrile and DMSO (Fig. 10).

The highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) energy gap represents the stability or reactivity of molecules. Higher energy gap indicates greater stability and vice versa. The values of HOMO-LUMO energy gap in various solvents are presented in Table 5 and their trend is shown in Fig. 11. From Fig. 11, it is clear that the HOMO-LUMO energy gap is the highest in DMSO and acetonitrile, indicating greater stability of BA in these solvents, however lowest HOMO-LUMO energy gap was found in water meaning the titled molecule is less stable and hence, more reactive in water which is in agreement with calculated polarizability, chemical hardness and softness. The Table 5 and Fig. 11 suggest that the molecule is stabilized with decreasing polarity of the solvent i.e. the molecule is less likely to be kinetically reactive.

The global chemical reactivity descriptors such as softness, hardness, chemical potential and electrophilicity index can be calculated from the HOMO-LUMO energy gap of a molecule [64‐68]. Using Koopman’s theorem for closed-shell molecules the hardness (η), chemical potential(μ), electronegativity (χ) and softness (S) are calculated according to the following equation:

Where

Ionization potential, I = -E_HOMO

Electron affinity, A = -E_LUMO

The electrophilicity index (ω) is calculated according to equation derived (by Parr et al., [68]) [58] as follows:

The molecular properties of BA in different solvents are presented in Table 6. No systematic trend was found in the case of chemical hardness & softness, chemical potential, electronegativity and electrophilicity index. However, the highest chemical softness, electronegativity and electrophilicity index was found in water (polar protic) followed by n-octanol (non-polar aprotic) and carbon tetrachloride (non-polar aprotic).

Pharmacokinetic studies such as absorption, distribution and metabolism of BA was done by using the web based application PreADMET (https://​preadmet.​bmdrc.​kr/). The calculated absorption, distribution and metabolism parameters are presented in Table 7.

The calculated human intestinal absorption HIA of BA (Table 7) was found to be 95.996% which suggests that BA is well absorbed through the intestinal cell [69]. BA was moderately permeable since values for absorption through Caco-2 cell (P_Caco-2) was 21.86 [70]. The skin permeability (P_Skin) is a vital parameter for assessment of drugs and chemical that might require transdermal administration [71]. BA was found to be impermeable through skin since a calculated P_Skin was − 2.11 encountered which proves BA is not permeable through the skin.

The distribution properties were assessed by measuring the brain to blood partitioning (C_brain/C_blood) and plasma protein binding (PPB). Generally, compounds with more than 90% of PPB are classified as strongly bound chemicals whereas less than 90% are weakly bound chemicals (https://​preadmet.​bmdrc.​kr/​adme-prediction/​). The calculated value of PPB for BA was 100.00% indicating BA is a strongly bound chemical and hence, the free form of BA will be less available in the systematic circulation. Based on C_brain/C_blood ratio all chemicals fall under three categories namely high absorption to CNS (C_brain/C_blood value more than 2.0), moderate absorption to CNS (C_brain/C_blood value within 2.0–0.1) and low absorption to CNS (C_brain/C_blood value less than 0.1) [72]. The ratio of C_brain/C_blood (8.20) suggests high absorption of BA to CNS indicating higher ability to cross blood brain barrier (BBB).

The exchangeable fraction (sum of unbound fraction and fraction dissociated from the protein) of drug determines the amount of drug that will penetrate the BBB [73]. Besides, Pardridge and colleagues have reported the permeation of protein-bound compounds through the BBB [74‐76]. Since BA is a strongly protein bound chemical (100%) so it could be assumed that BA will penetrate the BBB either due to the fraction dissociated from the plasma protein or as a protein-bound form or both.

P-glycoprotein (Pgp) is the product of the multi drug resistance (MDR) gene and an ATP dependent efflux transporter that affects the absorption, distribution and excretion of clinically important drugs [77]. Over-expression of this protein, which may result in MDR is a major cause of the failure of cancer chemotherapy, and decreased efficacy of antibiotics [78, 79]. The prediction of Pgp substrates, which facilitates early identification and elimination of drug candidates of low efficacy or high potential of MDR [80, 81]. Identifying molecules that interact with Pgp transporters is important for drug discovery, but it is commonly determined through laborious in vitro and in vivo studies [82]. Computational classification model can be used to screen molecules and predict the likeliness to be substrate for Pgp [82]. The in silico screening revealed that BA is a dual inhibitor and substrate for Pgp like quinidine [83]. This is because due to complex modulatory interactions with the Pgp which make BA to function as combination of substrate and inhibitor [83].

The computed metabolism of BA (Table 7) displayed that it is an inhibitor of CYP2C9 and a dual inhibitor and substrate for CYP3A4 due to complex modulation of CYP3A4 in phase I reaction. In phase II reaction BA is neither a substrate for UDP-glucuronosyltransferase (UGT) nor sulfotransferase (SULT).

The PreADMET server (https://​preadmet.​bmdrc.​kr/) was used to evaluate the carcinogenicity of BA. It was found that although BA is mutagenic but it demonstrated noncarcinogenicity in mice. On the other hand, BA is unlikely to be an inhibitor of human ether-a-go-go-related (hERG) gene (low risk). Inhibition of the hERG gene has been linked to long QT syndrome [84]. The results have been in summarized in Table 8.

YASARA Energy Minimization Server (http://​www.​yasara.​org/​minimizationserv​er.​htm) [85] was used to optimize the structure of PLA2. This server utilizes a new partly knowledge-based all atom force field derived from Amber, whose parameters have been optimized to attain the protein structure as close as to its native structure with maximum accuracy [85]. The simulation was performed in a water sphere containing ions.

The structure validation Z-scores and force field energies before and after the minimization are displayed in Table 9. The energies of the structure before and after optimization were − 68,549.6 and − 83,729.2 kJ/mol, respectively. This indicates that the protein structure was stabilized by an amount of − 15,179.6 kJ/mol. In addition, the structure validation Z-scores of before and after optimization was − 0.1 and 1.15 indicating good optimization of the target protein. The structure of PLA2 after energy minimization in a water sphere is presented in Fig. 11. PyMOL (Version 1.7.4.4, Schrödinger) was used to evaluate the root mean square deviation (RMSD) between the initial and final structure of PLA2 which was found to be 0.363.

Molecular docking of BA was carried out with human PLA2 using AutoDock Vina, to identify and understand the binding mode of BA and the intermolecular interactions with the target protein. All reported PLA2 inhibitors possess three key enzyme binding components such as Ca^+ 2 binding oxygen atom, a HIS47 binding H-bond donor and a hydrophobic component that bind the active site of the enzyme [55]. So, PLA2 inhibitors should form electrostatic interactions with Ca^+ 2 ion, H bond formation with the catalytic HIS47 and different types of hydrophobic interactions with hydrophobic residues which line the active site cavity of the enzyme. In addition, the inhibitors should interact and displace the HIS6; a unique residue to human PLA2 enzyme [55]. From our docking study, it was found that BA interacts with PLA2 at a binding affinity of − 41.00 kJ/mol whereas for BR4 it was − 33.89 kJ/mol. BA lies deep within the active site activity and makes numerous close contacts (< 5.4 Å) with the enzyme through hydrogen bond formation with GLY22 & GLY29 and hydrophobic interactions with the LEU2, PHE5, HIS6, ALA17, ALA18, TYR51 and the catalytic residue HIS47 (Figs. 2 and 3 and Table 10). However, no interactions of BA with calcium were observed as it was found in case of bovine PLA2 [5]. Therefore, BA has the ability to interact with the catalytic residue HIS47 and the base forming residue HIS6 of human PLA2.

On the other hand, the standard compound BR4 forms hydrogen bonding with LYS62, HIS47, GLY29, GLY31, CYS28, VAL30 and LYS62. Different types of hydrophobic interactions were also displayed by BR4 with GLY29-VAL30, LEU2 and ALA1. Moreover, BR4 exhibited attractive electrostatic interactions with Ca^+ 2 ions.

A comparison was made between the binding affinities and in vitro inhibitory activities (IC₅₀) of LY 311727, Manoalide, Indomethacin, Manoalogue, Aristolochic Acid and 1,1,1-Trifluoro-2-heptadecanone. The data has been presented in Table 11. Similar protocol described earlier was followed during performing docking of these compounds with PLA2 (1KQU). From the table it was clear that generally compounds with higher binding affinities also possess higher inhibitory activities. The binding affinities and reported PLA2 inhibitory concentration of BA are − 41.00 kJ/mol and ~ 2.5 μM for 30% & ~ 5 μM for 40% [5] inhibition which is consistent in accordance to our findings of binding affinity and in vitro inhibitory activity.