Files
Abstract
The inclusion of zero-mass point charges (commonly called lone-pairs or extended points) around electronegative atoms, such as oxygen, within force fields is known to improve hydrogen bonding directionality. In parallel, inclusion of lone-pairs in the TIP5P water modelincreased its ability to reproduce both gas-phase and condensed- phase properties over its non-LP predecessor, TIP3P.In general, force field parameters relevant for modeling solvent are developed independently of those for biomolecules, generally focusing on the ability of the parameters to reproduce bulk properties of the solvent. Such an approach is not generally extendible to solutes, which themselves never exist in the pure liquid phase. Therefore, a different ansatz has to be implemented to derive the most optimal lone-pair description for each type of solute, which is not always consistent with the solvent model. Currently, the partial charges for most biomolecular parameter sets are computed via fitting of the classical molecular electrostatic potential to the quantum molecular electrostatic potential. Application of this methodology to optimize lone-pair description is therefore consistent with current approaches of modeling electrostatics and is straightforward to implement. Here, we present an atom-type specific lone-pair model, which leads to the most optimal lone-pair placement for each atom-type, and notably, results in reproduction of the lone-pair description in TIP5P. Carbohydrates are rich in hydroxyl groups and development of a lone-pair inclusive carbohydrate parameter set for use with a lone-pair containing water model, such as TIP5P, ensures compatibility between these two models. Implementation of this lone-pair model into the GLYCAM series of parameter sets improves the geometry of a series of hydrogen bonded clusters over the nonlone-pair containing parameter set and is shown to reproduce crystalline and aqueous-phase properties for mono- and disaccharides. In addition, this methodology can easily be implemented to derive a general biomolecular parameter set and a preliminary investigation of this application to the protein segment, in conjunction with the carbohydrate segment, is examined.