Molecular docking based screening of triterpenoids as potential G-quadruplex stabilizing ligands with anti-cancer activity

Triterpenoids isolated from Ganoderma lucidum (GLTs) exhibit a broad spectrum of anti-cancer properties, including anti-proliferative, anti-metastatic and anti-angiogenic activities. Current research studies revealed the role by GLTs in inducing apoptosis and suppression of telomerase activity of cancer cells with much lower toxicity to healthy cells. Compounds selectively binding and stabilizing G-quadruplex structures could inhibit the telomerase or downregulate the oncogenes and may act as anti-cancer agents. Targeting human telomeric G-quadruplex DNA could be one of the mechanisms by which these GLTs exert anti-cancer activity. In this study, 208 GLTs were screened for ligands with high binding affinity and selectively to stabilize the pG4DNA by using the docking tool AutoDock4. The results showed that ganoderic acid A and ganoderic acid Df exhibit high binding affinity and selectively bind to the lateral groove of pG4DNA. Based on our findings, we suggest that the triterpenoid represents a new class of G-quadruplex groove binding ligands and thus act as potential anti-cancer agents.


Background:
Ganoderma lucidum (GL) possesses various pharmacological properties, which are also documented in the ancient reports where GL is praised for its effects on the promotion of health and longevity. It has been known to have numerous pharmacological effects including anti-aging, anti-cancer, anti-diabetic, anti-HIV-1, anti-inflammatory, anti-hypertensive, anti-oxidative, immunemodulating, and radical-scavenging effects [1]. Recently, more than 400 bioactive compounds have been isolated and identified from GL [2]. The main bioactive natural components from GL are triterpenoids, polysaccharides, nucleotides, fatty acids, glycoproteins, sterols, steroids, proteins and trace elements [3]. GL is one attractive source of anti-cancer products, which have been used for centuries as an herbal medicine for the prevention and treatment of cancer and improving immune function. The triterpenoids, structurally highly oxidized lanostanes, have been isolated and characterized with ganoderic acids (GAs) such as ganoderic acid A (GA A), GA D, GA Df, GA T [3 -4]. Accumulated data have shown that GAs exhibits a broad spectrum of anticancer properties, including anti-proliferative, anti-metastatic and anti-angiogenic activities [5]. In 2008, Yuen et al. reported that GL extract (GLE) is a potential source of chemopreventive agents for human bladder cancer [6]. Cancer cell growth inhibition induced by GLE is mediated via apoptosis associated with suppression of telomerase activity and oxidative DNA damage [6]. Telomeric DNA contains repetitive DNA sequence (TTAGGG) n forming G-quadruplex (G4) structures; this structure inhibits telomerase activity that is required to maintain telomeres [7]. Therefore, the compounds those selectively binds to and stabilize G4 complex structures could inhibit the telomerase and suppress the gene transcription of oncogenes, which will result in senescence and apoptosis of cancer cells [8].
Several research groups have focused on the structure-based design approaches to develop the potential anti-cancer ligands with the capability to stabilize G4 [9]. Most G4 ligands, such as BRACO19, PIPER, quercetin, RHPS4, telomestatin and TmTyP4, are planar molecules, which comprise a planar p-rich rings structure, allowing them to intercalate into G4 and form a stable conformation [10][11][12][13][14][15]. Recently, non-planar G4 ligands such as berberine, peimine, peiminine and steroid FG, could stabilize G4 complex through interaction in the lateral groove [16][17][18][19]. GLTs are non-planar molecules, structurally similar to steroid FG, which may stabilize G4 [17,18]. GA A was the first GA isolated from GL in 1982 and it shows no cytotoxicity to normal and cancer cells [20,21]. GA A displays anti-cancer effects, such as anti-invasion, inhibition of NF-kB AP1/uPA, anti-proliferation, inhibition of JAK-STAT3, inhibition of farnesyl protein transferase [22][23][24][25]. In this study, GA A was selected to elucidate the capability of GA A to stabilize G4. The parallel stranded DNA quadruplex d-(TTAGGGT) 4 (PDB code 1NP9) were selected for exploiting the interaction of GA A and G4 structure [26]. The aims of the present work were to search for novel GLT ligands with high binding affinity and selectivity for the pG4DNA, which may lead to the discovery of novel natural molecules as lead, compounds having potential anti-cancer activity.

Methodology: Preparation of ligand:
The two-dimensional (2D) structures of a total of 208 triterpenoids isolated from Ganoderma lucidum were downloaded from the SCiFinder database. The 3D structures were created with SYBYL 8.0 of NECTEC server. Energy minimization was performed to find the optimum structure with lowest energy. Energy minimization of each structure was achieved by using standard Tripos force field (Powell method and 0.0001 kcal/mole energy gradient convergence criteria). Electrostatic charge was assigned by Gasteiger-Huckel, and iterations number was set to 3,000 rounds.

Molecular docking:
The structure of the parallel G-quadruplex d-(TTAGGGT)4 containing the human telomeric repeat was retrieved from the protein data bank (PDB code 1NP9); http://www.rcsb.org/pdb) (Gavathiotis and Searle 2003). Dockings of GLTs to unbound pG4DNA were carried out using AutoDock 4 with AutoDockTools 1.5.6rc3 (ADT) as described by Li J et al. [18]. G4 structures were prepared for docking using Sybyl 8 software (Certara Inc. Princeton, USA) and ADT. The Gasteiger atomic charges and Kollman united atom partial charges were assigned for the GLTs and G4, respectively. Grid maps were set at four grooves and two terminal places for G4 and the grid boxes were centered at G4. The size of grid box was 60 x 60 x 60 in three dimensions. The grid was set to be sufficiently large to cover significant portions of the active sites. Lamarckian genetic search algorithm was employed. Maximum number of energy evaluation was 2, 500, 000 per run and population size was set at 150. All other parameters were set to default values. Two hundred fifty independent docking runs were performed. Results were divided into groups using the clustering module in ADT according to the 2.0 root-mean-square deviation (RMSD) criteria. Besides RMSD clustering, the binding free energies were evaluated for the binding conformations of ligand by using ADT, and the low-energy conformations were selected from the largest cluster

Molecular dynamics:
Molecular dynamic (MD) simulations for the GLTs-G4 systems were carried out with the sander module of AMBER 12.0 program package as described by Li J et al. [18]. The PARM 99 parameters and General Amber Force Field (GAFF) parameters were set for G4 and GLTs, respectively. Partial-atomic charges for the triterpenoids were derived using Gaussian 03 with the HF/6-31G (d) basis set followed by RESP calculation. Internal K + ion was added into the channels of the cavities between consecutive guanine tetrads. Then, the TIP3P water model was chosen and extended to a distance of 10 Å from any solute atom. Counter K+ ion neutralized the systems.
Firstly, a 1000-step minimization was carried out with the solute molecules fixed, and the equilibration was continued by 20 ps of PME dynamics with the same restriction. Subsequently, five rounds of 1,000-step minimization followed this equilibration with solute restraints from 20 to 0 kcal/mol Å -2 reduced by 5 kcal kcal/mol Å -2 in the course of each round. Then, the system was heated from 0 to 300 K with the rate of 50 K for every 5 ps of MD run, and another 100 ps MD simulation was continued to equilibrate the system. After the minimization and equilibration, MD simulations were run under NPT condition at 300 K. During the MD simulations, SHAKE was used for constraining hydrogen atoms and a 9 cutoff was applied to non-bonded interaction. Simulation time step was set at 2 fs and the translational center of mass motion was removed every 10 ps [18].

Results and Discussion: Molecular docking and MD stimulations:
GA A, which exhibits anti-cancer effects and shows no cytotoxicity to normal and cancer cells [20,21], was selected to evaluate the ability to stabilize pG4DNA. The docking result of GA A in pG4DNA binding site is shown in Figure 1. The lowest energy docked conformation of the most populated cluster (the largest cluster) was selected and taken into account for study the binding against pG4DNA. The estimated inhibition constant (Ki) and estimated free binding energy (G) of GA A which stabilized the pG4DNA are shown in Table 1. The lowest binding free energy conformation of GA A binding in pG4DNA was selected for further MD stimulation.  a ∆VDW is the change in van der Waals energy in the gas phase upon complex formation. b ∆SUR is the change in energy due to the change in surface area upon complex formation. c ∆GBELE is the change in GB reaction field energy + gas phase electrostatic energy upon complex formation. d ∆TOT = ∆VDW + ∆SUR + ∆GBELE is the change in potential energy in water upon complex formation. . Therefore, in silico screening of GLTs was performed to search for the GLTs with high binding affinity and selectivity for the pG4DNA.

Virtual screening of GLTs as potential G4 stabilizing ligands :
In this study, 208 triterpenoids isolated from GL were screened for the ligand with high binding affinity and selectivity for the pG4DNA. GLTs were docked to the pG4DNA. The structures, estimated Ki and estimated G of GLTs, which stabilized the pG4DNA, are shown in Table 1. The results indicated that 131 GLTs interact with the pG4DNA with high affinity (Ki < 1 mM). GA Df was the most potent GAs to stabilize the pG4DNA with Ki = 13.97 nM.

MD stimulations:
MD stimulations were performed on GA A and GA Df with pG4DNA to explore the binding poses in depth. Molecular Mechanics and Generalized Born Surface Area (MM/GBSA) were determined for the best ranking conformation molecule on the solvation forces involved in the stabilization of GA-pG4DNA complex. The estimated rG, estimated Ki and target residues involved in the hydrogen bonding of the best-docked poses are given in Table 2. The pG4DNA (1NP9: containing the human telomeric repeat) consists of four equivalent grooves [26]. The results showed that GA A interacts with pG4DNA in the groove through hydrogen bond and van der Waals interactions. One hydrogen bond was formed by side chain carbonyl group of GA A and guanine base position 11 of lateral groove of pG4DNA (DG11) with hydrogen bond length of 2.21 Å. Two methyl groups (C18 and C19) were pointed into the groove and bound with guanine bases by hydrophobic and van der Waals interactions. The distance between methyl group (C18) and carbon atom of DG11 was 3.7 Å, and the distance between methyl group (C19) and nitrogen atom of DA10 was 3.6 Å (Figure 3a and Table 2). GA Df stabilized pG4DNA with 3 van der Waals interactions and 2 hydrogen bonds with pG4DNA at DG 11 and adenine base position 3 (DA3) with hydrogen bond length of 2.12 Å and 2.74 Å, respectively (Figure 3b and Table 2). Hydroxyl group of ring B of GA Df formed hydrogen bond with DG 11 and hydroxyl group of ring C formed H-bond with DA 3. Two methyl groups (C18 and C19) were pointed into the groove and bound with guanine base by hydrophobic and van der Waals interactions. The distance between methyl group (C18) and nitrogen atom of DG11 was 3.7 Å and the distance of this methyl group and nitrogen atom of DA3 was 3.9 Å. The distance between methyl group (C19) and carbon atom of DA10 was 3.8 Å. The result also provided new insight into the design of G4 groove-targeted agents.
Furthermore, Table 2 shows the MM/GBSA binding energy calculation of GA A and GA Df to pG4DNA. GA A displayed a lower total binding energy (-23.46 ± 1.70 kcal/mol) than GA Df (-13.32 ± 2.21 kcal/mol). However, in docking experiment, GA Df was about 30 times more active than GA A ( Table 1). As for the results from MD, the interaction of GAs with pG4DNA was in solution which mimicked the physiological condition, the total binding energy of GA A was approximately 2 times better than GA Df. Further studies are required for clarify these results.
The obtained results are in agreement with the published nonplanar G4 ligands that GLTs stabilized G4 through the groove binding [16][17][18][19]. The GLTs interacted with the pG4DNA and enhanced G4 stabilization through hydrogen bonds and van der Waals interactions. At physiological condition GA A, noncytotoxic GLTS [20, 21], might be potential lead compounds for the development of new telomerase inhibitors. Thus, GA A may serve as the starting point for the design of a new class of highly selective groove binding of pG4DNA with anti-cancer effect.

Conclusion:
In conclusion, the first virtual screening of GLTs as potential G4 stabilizing ligands was presented. Binding poses and binding energies for GLT-pG4DNA complexes were calculated using molecular docking and molecular dynamics. The results indicated that GLTs significantly stabilized the pG4DNA through interaction with the lateral groove of G4 by hydrogen bonds and van der Waals forces. GA A and GA Df exhibit high binding affinity and selectivity for lateral groove of pG4DNA with theoretical binding efficiency in nanomolar range. The triterpenoid represents a new class of highly selective groove-binding molecules. Thus, GLTs exert their novel anti-cancer mechanism by stabilizing the pG4DNA through the groove binding.