IN SILICO STUDY OF SIRT1 ACTIVATORS USING A MOLECULAR DYNAMIC APPROACH

Objective: The importance of SIRT1 activator’s role in antidiabetic and anti-aging therapies is widely demonstrated. Drug discovery and development are time consuming. Drug design can be performed in silico using molecular dynamic approaches to accelerate and facilitate identification of the best compound candidates and their physicochemical characteristics and hit-to-lead selection. Methods: In silico study of SIRT1 activator for complexes using of Protein Data Bank (PDB) IDs 4ZZI, 4ZZJ 4ZZH, and 5BTR and 4TO ligand. Ligand– receptor interactions and bond energies were determined using molecular docking with the AutoDock4Zn program. Then, the complex with the best bond energy was identified using a simulation of the molecular dynamics (50 ns) using the Amber program, and values for root mean square deviation, root mean square fluctuation, and bond energy were determined using the Molecular Mechanic–Poisson Boltzmann (Generalized Born) surface area (MM-PB[GB]SA) calculation. Results: Interaction analysis between activator ligand (4TO) and the SIRT1 receptor (PDB IDs 4ZZJ and 5BTR) revealed the ligand’s selectivity for hydrophobic interaction at Leu206, Ile223, Ile227, and hydrophilic interaction at Asn226, Glu230. Hydrogen bond interactions between Glu230 and Arg234 (allosteric region) with Arg446, Val459, His473, and Asp475 (catalytic region) brought them close to the bounding substrate area. Bond energy values obtained using the MM-GB(PB)SA calculation showed 4TO interaction with 4ZZJ (MMGBSA ∆G, −31.4729–−26.6756; MMPBSA ∆G, −32.6292–−28.486]. The bond energy value of the 4TO interaction with 5BTR showed MMGBSA ∆G = −40.6255–−30.0653 and MMPBSA ∆G = −34.6713–−25.9951. Conclusions: These findings provide important information on the target interaction of the bonds to the more selective SIRT1 activator useful for drug discovery and development.


INTRODUCTION
SIRT1 enzyme activity has two primary roles as a target for therapy, namely SIRT1 inhibition in the catalytic region which is used in cancer treatment and SIRT1 activation in the allosteric region which is used in metabolic damage treatment, including diabetes [1,2] and anti-aging [3][4][5]. SIRT1 is a nicotinamide adenine dinucleotide (NAD+)-dependent deacetylase which catalyzes of deacetylating protein in lysine residues [6,7].
The structure of human sirtuin protein includes the SIRT1 isoform (UniProt accession code Q96EB6, available online: NCBI, Research Collaboratory for Structural Bioinformatics). The SIRT1 sequence comprises 747 amino acids; amino acids 1-180 are located at the N-terminal region, amino acids 195-240 in the allosteric region (active regulator of SIRT1), amino acids 244-512 in the catalytic region, and amino acids 513-747 in the C-terminal region [8].
The structure of SIRT1 PDB ID 4ZZJ (www.rcsb.org/structure/4ZZJ) shows interactions between 4TQ at the allosteric region of the HI of amino acids Thr219 and Ile223, hydrogen bonding at Asn226, and with CNA (carba-nicotinamide-adenine-dinucleotide) ligand at the catalytic region. The active conformation model of SIRT1 comprises a STAC complex with the Ac-p53 substrate, indicating the presence of amino acid hydrogen bond interactions at Glu230 and Arg446. The structure of the complex of SIRT1 with STAC compounds shows binding of STAC-1 to the STAC binding domain (SBD). Binding of STAC encourages SBD interaction with the central catalytic region to the substrate, increasing the activity of SIRT1 and stabilizing the deacetylase activity. Replacement of Glu (E) 230 with Lys residue disrupts enzymatic activation by reducing the electrostatic interactions between Glu230 in the SBD and Arg446 at the substrate-binding site [13]. The structure of SIRT1 PDB ID 5BTR (www.rcsb.org/structure/5btr) is the SIRT1 activator located in the allosteric region, a complex containing three resveratrol ligands (STL), and shows hydrogen bonding interactions at Gln222, Asn226, lys444, Asp298, Asp292, and Glu230 and HIs at Gly415, Ile223, Gln294, Ile223, and Arg446.
Interaction of 7-amino-4-methyl coumarin with 5BTR shows HIs at Phe412, Val445, His363, and Phe414 and hydrogen bond interactions at Arg446, Asn226, Gly415, Asp150, Glu416, and Val412 [12]. Therefore, the present study used an in silico study with molecular docking and dynamic simulations of the SIRT1 enzyme (PDB IDs 4ZZI, 4ZZJ, 4ZZH, and 5BTR) to determine the bonding interactions of SIRT1 required for ligand selectivity and selective identification of bond interactions and target bond interactions with biological function [14,15].

Materials
The following materials were used: IntelXeon(R) central processing unit E5620 at 2.40 GHz × 16, processor 2.6 GHz Intel Core i7, random access memory 32 GB 1600 MHz DDR3, and Graphics Intel Iris 1536 MB. The operating system used was Linux Ubuntu 12.04 LTS with an uninterrupted power supply.

Ligand preparation
The ligands used were the 4TO crystals of macromolecular SIRT1 (PDB IDs: 4ZZH). Minimization of ligand used the AMBER program, and 1NS compound (SIRT1 inhibitor) was used as a negative control.

Molecular docking validation
Autodock4Zn was used to perform molecular docking in this study [18,19]. Validation of the autodock4Zn program measured the root mean square deviation (RMSD) value analyzed using the visual molecular dynamics (VMD) program and the initial cocrystal structure of PDB against the conformational crystal positioning after molecular redocking was done.

Molecular docking
Molecular docking was performed using AutoDock4Zn, with ligandmacromolecular grid binding sites by determining npts and gridcenter.
Analysis of the docking results in the form of affinity and RMSD binding values (for docking performance validation) and visualization of docking results to analyze ligand interaction with macromolecules used LigandScout (Inte:Ligand, Austria) [20] and CHIMERA (UCSF, USA), respectively [21].

Molecular dynamics
The best compounds for molecular docking were selected for the molecular dynamics study, and the simulation was performed using Amber12 [17] for 50 ns.
General amber force field (GAFF) was used for ligand preparation [13]. Ligands and macromolecules topologies and coordinates were created in a vacuum and explicit waters environment. At this stage, the ligand structure was provided with an AM1-bond charge correction charge [14] using Antechamber software (UCSF, USA) accessed via PuTTY. The file output was obtained in the form of *.mol2, and the antechamber result created a *.frcmod file.
Preparation of peptide minimization (substrate) used parameters and coordinates of peptide-containing (NArg His Lys Lys Leu Met CPhe) macromolecular files [22]. Preparation for the formation of NAD + used NAD + as a cofactor with a positive charge. The creation of macromolecules with NAD + required NAD + .lib, and NAD + .frcmod parameter files were obtained from Ross Walker, and the coordinates were altered using the coordinates of the NAD + file associated with the macromolecules. Further complexes of ligand: Zn: NAD + : Macromolecules were used as topological and coordinate files with the addition of water molecules TIP3PBOX 12Å 3 , followed by minimization, heating, density, equilibration, and production.
Before continuing the dynamics simulation, verification of the system was balanced using the command "process_mdout. pl" to extract useful information from the output file: heat. out, density. out, and equil. out. A balanced system was seen from the temperature, density, and RMSD. After the balanced system was observed, production was continued for 50 ns. The ptraj program from AmberTools and VMD was used to perform trajectory analysis of the molecular dynamic simulation results [15]. The parameters analyzed were RMSD and root mean square fluctuation (RMSF), and the hydrogen bond conditions were analyzed using VMD program. Hydrogen bonds with an occupancy of >50% of the overall data from the hydrogen bond analysis were selected. The distance between the donor and acceptor of the hydrogen bond was set at 3.0Å, and the cutoff angle was set at 60°. The Molecular Mechanic-Poisson Boltzmann (Generalized Born) surface area (MM-PB[GB]SA) method was used to calculate the binding energy [23].

RESULTS AND DISCUSSION
The present study was performed by visualizing 4TO cocrystal interactions with SIRT1 macromolecules (SIRT1 PDB IDs 4ZZI, 4ZZJ, and 4ZZH) and one with three STLs (PDB ID 5BTR). Analysis of complex compound interactions (STACs) with SIRT1 activator receptors in the allosteric region was used to examine the bonding interactions that play an important role in ligand selectivity. Activator compounds act as sirtuin activators bound to the amino acid allosteric site at residues 183-243 and generally interact at the hydrophobic chains at Thr209, Pro211, Pro212, Leu215, Thr219, Ile223, and Ile227 at the shape of the (helix-turn-helix) amino acid and one hydrophilic interaction at Asn226 (Supplemental Fig. 1).
Visualization of macromolecular interactions with SIRT1 PDB IDs 4ZZI, 4ZZJ, 4ZZH, and 5BTR showed differences in the position of the allosteric region receptors. Residues at Glu230-Asn241 showed a flexible helix shape. As Fig. 1 shows, the superimposition of SIRT1 PDB ID 4ZZI demonstrates a more closed form between the allosteric and catalytic regions, containing 4TO ligand crystals in the allosteric region and 1NS ligand crystals in the catalytic region. SIRT1 PDB ID 4ZZJ contains CNA and 4TO ligand crystals in the allosteric region, and the allosteric area form was catalytically more open than that of 4ZZI. PDB 4ZZH showed an allosteric and catalytic form that was more open and contained only 4TO ligand crystals in the allosteric region. PDB 5BTR contained resveratrol compounds in the allosteric region adjacent to the catalytic region.

Validation of molecular docking
In the present study, AutoDock4Zn was used to visualize molecular docking. Validation of this Autodock program measured the RMSD value analyzed using the VMD program. The initial cocrystal PDB structure was analyzed against the conformational crystal positioning after molecular docking restarted. The RMSD values of 4TO as a crystal were 0.81Å, 0.79Å, and 0.89Å for 4ZZI, 4ZZJ, and 4ZZH, respectively.

Molecular docking of macromolecular 4ZZI
The macromolecule receptor of 4ZZI has ligand crystals in the catalytic region as an activator.   Fig. 2).

Molecular docking of macromolecular 4ZZJ
The AutodockZn program was used to analyze docking of the ligand crystal molecule and the active compound SIRT1 with 4ZZJ. The results obtained from the process showed a bond energy value of ∆G = −6.94 kcal/mol, a Ki value of 8.24 μM (predicted binding interaction), and the complex interaction of ligand and receptor bonds. Fig. 3 shows 4TO compound interaction. The complexes demonstrate four hydrophobic features at amino acids namely Leu206 (HI), Thr209, Leu215 (HI), Thr219 (HI), Ile223 (HI), and Ile227 (HI) and show acceptor hydrogen bonding interactions (HBA) at Asn226. Docking of the 1NS compound (SIRT1 inhibitor) as a negative control on the allosteric region showed interactions at Gln222 (HBD) and Thr219 (HBD) and no HIs (Supplemental Fig. 2).

Molecular docking of macromolecular 4ZZH
The AutodockZn program was used to generate molecular docking of the crystal ligand molecule and the active compound SIRT1 with 4ZZH. The results were obtained using a bond energy value of ∆G = −7.61 kcal/mol and Ki = 5.6 μM (predicted binding interactions).

Molecular docking of macromolecular 5BTR
Molecular docking 4TO with 5BTR using the AutodockZn program was obtained using a bond energy value of ∆G = −13.1 kcal/mol and Ki = 2.51 × 10 −4 μM (predicted binding interaction).
The 4TO ligand revealed bonds at the activating regions of essential amino acids Ile223 and Ile227. The complex showed hydrophobic features at amino acids namely Leu206 (HI), Thr219 (HI), Ile223 (HI), and Ile227 (HI) (Fig. 5). Docking of the 1NS compound (SIRT1 inhibitor) as a negative control on the allosteric region showed interactions at Leu206 (HBD) and no HIs (Supplemental Fig. 2).

Molecular dynamic simulation
Molecular dynamic simulation of 4TO compounds against 4ZZI, 4ZZJ, 4ZZH, and 5BTR receptors was performed for 50 ns using the Amber program. Analysis of the dynamics simulation result was carried out considering RMSD, RMSF, and hydrogen bonding conditions, and the binding energy was calculated using the MMGB/PBSA method.

RMSD
Conformational changes of the 4TO compound during the simulation were seen from the RMSD values. The RMSD curve for 50 ns showed a change in the stability of the 4TO complex dynamic simulation that corresponded to 4ZZI, 4ZZJ, 4ZZH, and 5BTR receptors. In Fig. 6, the 4TO:4ZZI, 4TO:4ZZJ, and 4TO:5BTR complexes showed an RMSD range of 2Å, whereas the 4TO:4ZZH complex had a RMSD range of 3Å.

RMSF
RMSF is the measure of the deviation between the atomic positions of each protein residue, i.e., the difference in fluctuations in the movement of each residue during the simulation is measured for 50 ns. Fig. 7 shows the RMSF value of the complex molecular dynamic simulation between the 4TO ligand (ligand crystal) and 4ZZI, 4ZZJ, 4ZZH, and 5BTR receptors. The 4ZZH receptor showed fluctuations of movement in the N-terminal domain of all RMSF compared with 4ZZI, 4ZZJ, and 5BTR receptors that did not show fluctuations, indicating that binding of the 4TO compound to the receptor is more stable.  Fig. 8 shows the RMSF of the backbone atom in a dynamic simulation system between 4TO ligands with 4ZZI, 4ZZJ, 4ZZH, and 5BTR receptors for 50 ns in allosteric regions. The 4TO ligand bond with the 4ZZI receptor and 5BTR on Leu206, Leu215, Thr219, Ile223, Asn226, Ile227, and Glu230 residues did not show fluctuations (low RMSF values), indicating that binding occurred in the residue. These data correspond to the bonding interaction in ligand crystals 4ZZI [7] and 5BTR [8].
The bond interaction between the 4TO ligand and the 4ZZJ receptor shows a low RMSF value at Leu206, Ile223, Asn226 Ile227, and Glu230, demonstrating strong binding to the residue. These data correspond to the bonding interaction of the 4ZZJ ligand crystal [7]. The binding interaction between the 4TO ligand and the 4ZZH receptor shows high RMSF values at Ile223, Asn226, Ile227, and Glu230, showing no strong binding of such residues, as in the crystal ligands of 4ZZH [7].  (Fig. 10). The results of this analysis indicate that the SIRT1 activator is showed by hydrogen bond interactions between Glu230 and Arg234 (allosteric region) with Arg446, Val459, His473, and Asp475 (catalytic region) which are close to the bound substrate region.

Free energy calculation (ΔG)
The result of the ΔG bond energy calculated from the molecular docking with AutoDock4Zn was then recalculated to determine the free bonding energy from the 4TO compound with SIRT1 PDB IDs 4ZZI, 4ZZJ, 4ZZH, and 5BTR using the MM-GB(PB)SA method, with simulation dynamics carried out for 50 ns. This was performed to ensure that the bond energy was more selective [23], and the energy value of the 4TO ligand bond with 4ZZJ receptor using the MMGBSA method was calculated, while MMPBSA measured bond energy on simulation dynamics for 50 ns.   Fig. 11 shows an example of the calculation. Fig. 12 shows the overall results.  the four different receptors (4ZZI, 4ZZJ, 4ZZH, and 5BTR) showed that the strength of the 4TO bond to 5BTR was stronger than that of the other receptors because the receptor form was more closed. A small difference in MM-GB(PB)SA range values was seen in the interaction between 4TO:Zn:4ZZI.

CONCLUSIONS
The present study describes an in silico study of SIRT1 bond interaction using a simulation approach of molecular dynamics over 50 ns using the Amber program. Analysis of activator ligand (4TO) binding to the SIRT1 receptor (PDB IDs 4ZZJ and 5BTR) showed selectivity of the ligand by marked hydrophobic bond features on Leu206, Ile223, Asn226, Ile227, and Glu230 of the 4ZZJ and 5BTR receptors.
Hydrogen bond interactions between Glu230 and Arg234 (allosteric regions) with Arg446, Val459, His473, and Asp475 (catalytic areas) ensured that they became close to the bounding substrate area. These results are important for drug discovery and development as they give insight into target interaction of the bonds to the more selective SIRT1 activator [24][25][26].