Can the results of quantum refinement be improved with a continuum-solvation model?

Quantum refinement has repeatedly been shown to be a powerful approach to interpret and improve macromolecular crystal structures, allowing for the discrimination between different interpretations of the structure, regarding the protonation states or the nature of bound ligands, for example. In this method, the empirical restraints, used to supplement the crystallographic raw data in standard crystallographic refinement, are replaced by more accurate quantum mechanical (QM) calculations for a small, but interesting, part of the structure. Previous studies have shown that the results of quantum refinement can be improved if the charge of the QM system is reduced by adding neutralizing groups. However, this significantly increases the computation time for the refinement. In this study, we show that a similar improvement can be obtained if the original highly charged QM system is instead immersed in a continuum solvent in the QM calculations. The best results are typically obtained with a high dielectric constant (ɛ). The continuum solvent improves real-space Z values, electron-density difference maps and strain energies, and it normally does not affect the discriminatory power of the calculations between different chemical interpretations of the structure. However, for structures with a low charge in the QM system or with a low crystallographic resolution (>2 A), no improvement of the structures is seen.