Computational protocol: A Universal Base in a Specific Role: Tuning up a Thrombin Aptamer with 5-Nitroindole

Similar protocols

Protocol publication

[…] TBA-N8 MD simulations were carried out using the Amber 10 program suite. The influence of the solvent was simulated with a TIP3P water model. The simulation was performed under periodical boundary conditions in a rectangular box. The parameters for the inter-atomic energy calculation were taken from the force field parmbsc0. Partial charges on NI atoms were calculated by first individually optimizing the NI residues (a methyl group was used instead of the sugar moiety) using DFT with hybrid exchange-correlation functional B3L LYP (Becke three-parameter (exchange), Lee, Yang and Parr (correlation)) and the aug-cc-pVTZ basis sets. Single point calculations were then performed at the MP2/aug-cc-pVTZ level. Finally, for the electron density distribution determined with DFT/B3LYP, partial atomic charges were obtained by charge fitting of the electrostatic potential at points selected according to the Merz-Singh-Kollman scheme. All quantum mechanic simulations were carried out using the Gaussian 09 program. A K+cation was placed at the centre of the aptamer between the G-tetrads to stabilize the structure. Adding Na+ ions neutralized the negative charges and 1947 water molecules were employed for solvation. The structures were minimized and an MD simulation performed as previously described for the TBA analog in complex with thrombin and the energies of the aptamers were estimated by using the MM-GBSA approach. The total mechanical energy of the molecule in gas phase (EMM) was calculated as the sum of the electrostatic energies (Eeq), van der Waals energies (EvdW), and the energies of internal strain (bonds, angles and dihedrals) by using a molecular-mechanics approach. The free energy of solvation was calculated as the sum of the polar and nonpolar contributions. The polar contribution (EGB) was computed using the Generalized Born (GB) method and the algorithm developed by Onufriev et al. for calculating the effective Born radii. The non-polar contribution to the solvation energy (Esurf), which includes solute-solvent van der Waals interactions and the free energy of cavity formation in solvent, was estimated from a solvent-accessible surface area (SASA).The docking procedure with a partially flexible aptamer was performed using Autodock 4.2. The quadruplex core of TBA/TBA-N8 was fixed, while in the loops heterocyclic bases could rotate around the glycosidic bonds. Naturally, this only allowed for a partial accounting of the actual conformational diversity in the loops. To consider at least several possible loop arrangements, we used several conformers of each aptamer taken from MD trajectories and differing in loop arrangements (not just 50 ns ones), in our docking experiments. The results shown in and refer to the conformers that demonstrated the best affinity for thrombin (the highest binding energy).The protein and the aptamers were prepared for docking by AutodockTools (ADT Version 1.5.4), and the partial atomic charges on the aptamer atoms that were calculated for MD simulations were preserved. The partial atomic charges on protein atoms were calculated using the Gasteiger-Hückel method.Two types of docking experiments were performed. The first type was ‘rigid’ docking with a fixed protein body. The second type was ‘half-rigid’ docking with flexible side chains of the key amino acid residues. The thrombin model used in the ‘half-rigid’ docking of the aptamers to exosite I and the proximal area was taken from PDB:1HAO (the amino acids with flexible side chains are His71, Arg73, Thr74, Arg75, Tyr 6, Glu77, Arg77A, Asn78, Glu80 and Tyr117). The model of the thrombin/heparin complex used for docking the aptamers to exosite II and the proximal area was taken from PDB:1XMN (the amino acids with flexible side chains are Arg93, Arg101, Arg126, Arg233 and Lys240). Three-dimensional grid maps (126 × 126 × 126 points) for each atom type were computed using AutoGrid4. In the case of ‘rigid’ docking to the whole protein, the grid centre was placed at the geometric centre of the protein, and the size of a lattice cell was 0.5 Å. In the case of docking to exosite I area, the grid centre was placed into the CA atom of Lys70 with a lattice cell of 0.375 Å. In the case of docking to the exosite II area, the lattice cell was also 0.375 Å, and the grid centre was placed into the CD1 atom of Ile103. Electrostatic and desolvation maps of the protein were calculated. The Lamarckian genetic algorithm (GA-LS), a hybrid of a genetic algorithm and a local search algorithm, was employed for identifying the most probable binding site. The number of GA-LS runs was 50, the maximum number of energy estimations was 2500000, the maximum number of generations was 27000, and the mutation and crossover rates were 0.02 and 0.8, respectively. Pseudo-Solis & Wets parameters were used for the local search, and the number of iterations was 300. Starting positions and aptamer conformations were random. The torsion angle rotation step was 50°. After docking, all the generated structures were clustered up with RMS tolerance of 2 Ǻ from the lowest energy. Binding free energies (ΔG) of the aptamer/protein complexes were calculated according to the formula: The electrostatic interaction free energy (ΔGel) was estimated using a distance-dependent dielectric function of Mehler and Solmajer. The van der Waals interaction free energy (ΔGvdW) was estimated using the L-J potential and atomic parameters from the AMBER Force Field. For scoring the H-bond energy ΔGH-bond, the 10/12 potential was used with a maximal well depth of 5 kcal/mol at 1.9 Å for hydrogen bonds with oxygen and nitrogen atoms, and then multiplied by the function estimation rate from the ideal H-bonding geometry. For estimating the desolvation energy ΔGdesolv, the atomic fragmental volume and solvation parameters derived from the method of Stouten et al. were used. Each term was multiplied by a semi-empirical weighting constant by fitting to the binding affinity data of a training set of complexes with known structures. The scoring function in the docking procedure also involved ΔGtors, the conformational entropy defined via the conformational mobility of the aptamer and calculated using a number of dihedral angles. In our case, ΔGtors could not be evaluated precisely (the weighting constant for ΔGtors might be incorrect because the regression multipliers for this parameter have been used for chemical structures, which are different from the aptamers we used). […]

Pipeline specifications

Software tools AMBER, AutoDock
Application Protein interaction analysis