Performances of DFT methods implemented in G09 for simulations of the dispersion-dominated CH-π in ligand–protein complex: A case study with glycerol-GDH

• Comparisons of DFT functionals, MP2, and CCSD(T) methods.
• Calculations of CH-π interactions in the transition state of glycerol-dehydratase complex.
• Use DFT methods in the dispersion dominated system.
• QM/MM simulations of large ligand–protein.

In this study, calculations of the dispersion-dominated CH-π interaction in the transition state of glycerol-dehydratase complex have been performed using 56 standalone functionals, other 325 separate pure generalized gradient approximation (GGA) density functional theory (DFT) functionals, MP2, and CCSD(T) methods. The calculation results showed that the ability of DFT methods to recover correlation energy does vary significantly from method to method. Compared with the interaction solvation corrected energy of −0.25 kcal/mol and −0.66 kcal/mol from CCSD(T)/aug-cc-pVDZ and CCSD(T)/CBS respectively, PW91B95, M06, M06L, and M06HF give close results of −0.20, −0.28, −0.36, and −0.45 kcal/mol respectively. However, the most popular B3LYP provides interaction energy of 2.18 kcal/mol, which is evidently deviated from the results calculated from other methods. Interaction energies calculated using the 156 separate GGA functionals are significantly improved by adding long range correction. All DFT SMD solvation energies range from 0.34 to 0.57 kcal/mol, which are comparable with the solvation energy of 0.46 kcal/mol from CCSD(T) calculation. The CH-π orientation obtained from QM/MM simulations shows a dependence on the DFT method applied in the quantum region. Among all the obtained CH-π geometries, those obtained from QM(wB97)/MM, QM(wB97XD)/MM, QM(M062X)/MM, QM(wB97X)/MM, QM(M06HF)/MM, and QM(M06)/MM methods represent the most accurate outcomes in comparison to that optimized with QM(MP2/6-31G*)/MM ONIOM simulation. Our results provide insights about how to efficiently use DFT methods in the dispersion dominated system and may aid in the development of new DFT functional to better address the correlated component of inter-molecular interaction.