Computational protocol: Prediction of Reduction Potentials of Copper Proteins with Continuum Electrostatics and Density Functional Theory

Similar protocols

Protocol publication

[…] The 4AZU pdb file was used as a model of the azurin structure and the coordinates of each of the four chains were extracted. Mutations were created by replacing specific amino acid sidechains using the SCWRL 4.0 algorithm to find the optimum position of the side chain. The PDB2PQR software package was used to add hydrogen atoms to the structure, refine hydrogen bonds and assign partial charges and atomic radii to each atom using the CHARMM27 forcefield. Protonation states of titratable residues at pH 7.0 were determined by calculating their pK a values using the in‐house finite difference/Debye–Hückel (FD/DH) method. All histidine amino acid residues were singly protonated at Nδ (His35, His46, His83 and His117). Glu and Asp residues were in the anionic form and Arg/Lys in the protonated form. The only exception to this rule was the Lys121 residue in the M121K mutant, which was left unprotonated. Partial charges for the copper ion and its direct ligands, namely His46, Cys112 and His117 in both oxidation states were determined from small gas phase DFT clusters with Gaussian09. Geometries were optimized with the B3LYP method in combination with a 6‐31G** basis set on all atoms except copper for which we utilized the effective core potential LANL2DZ basis set (basis set BS1). To improve the energetics and partial charges, single point calculations were performed using a 6‐311++G(2df,p) basis set (BS2) on all atoms and partial charges were calculated using natural bond orbital (NBO) analysis. The atomic radius of copper was set to 1.71 Å in both oxidation states in accordance with previously reported observations. Relative reduction potentials were estimated by comparing the difference in total computed electrostatic energy of mutants in each oxidation state, with that of wild type. This was done for each of the four chains (asymmetric units) of the protein and the results were averaged. Electrostatic energies were calculated using a finite‐difference Poisson–Boltzmann method. The linearized finite difference Poisson–Boltzmann equation was solved using adaptive Poisson–Boltzmann solver (APBS). Calculations were performed on a 57.8 Å3 cubic grid with 0.2 Å grid spacing. Charges were mapped onto the grid using cubic B‐spline discretization. The dielectric constant was 4 for regions of the protein and 78 for regions of the solvent. The protein region was defined by a molecular surface determined by a solvent probe with radius 1.4 Å. The value for the dielectric constant was smoothed at the protein‐solvent boundary using 9‐point harmonic averaging. The ionic strength was 0.15 M. [...] Active site models of azurin were constructed in GaussView 5, based on chain A of the 4AZU pdb crystal structure coordinates. The model included the copper ion with its three main ligands (His46, Cys112 and His117), the two axial ligands (Gly45 and Met121), two hydrogen bonds between backbone nitrogen atoms and the Cys112 sulfur as well as the two hydrogen bonds that form an interaction between Asn47 and Thr113. Following a procedure reported previously, the model for the F114N mutant also included the additional environment surrounding Asn114 (namely Pro115, Gly116) to aid with its orientation of the polar sidechain. A WT model including this additional environment was also constructed to calculate relative reduction potential of the F114N mutant.Geometries of both oxidation states were partially optimized using the B3LYP/BS1 method., To mimic the rigidity of the protein backbone and imposition of the axial ligand some atoms were fixed in the models (see Supporting Information for details). As shown before, differences between zero‐point energies, thermal energies, and entropies of models only make minor contributions to the relative reduction potential, due to the similarity of each of the models. Single point energies were calculated using the basis set 6‐311++G(2df,p) for all atoms: Basis set BS2. Optimizations and single point calculations were performed in the gas phase as well in implicit solvent using the integral equation formalism polarization continuum model (IEFPCM) method. Chlorobenzene was chosen to mimic “protein‐like” conditions (ϵ=5.6968) and calculations were also performed in water (ϵ=78.3553). In general, the value of the dielectric constant appeared to be less critical for the results as long as a value larger than one is used. As shown previously, for these types of calculations, zero‐point and entropic corrections cancel out, and hence have not been considered. […]

Pipeline specifications

Software tools SCWRL, PDB2PQR, GaussView
Applications Drug design, Protein structure analysis
Chemicals Copper