Conformational States of Amino Acids in Proteins and Peptides by Means of Computational Chemistry Methods

The application of the physical principles governing biomolecules in order to adopt their unique 3D structure implies a need for reasonably accurate methods able to map conformational hyperspace at the Gibbs free-energy level. In the case of proteins, the question is rather complex and includes the issue of at which level the local conformational preferences of amino acids in short sequential context are propagated into the final protein structure. The role of amino acid side chains is not quite clear in the structural context because the backbone preferences define the number of possibilities available for the side chains to interact. This could crucially constrain the spatial possibility of finding the near-energy optimum geometry for each interacting side chain. It is well-known that the precise balance between propensities for helical and extended structures is an essential property of any force field intended for simultaneous description of folding alpha-helical, beta-sheet, or mixed protein architectures, as well as their unfolded states. As was documented in the not so distant past, the simulations of short alanine-based peptides have revealed overrated helical tendencies of current force fields. In this paper we discussed the issue from three different perspectives. First, how to improve empirical force field parameters by corrections obtained from high level ab initio calculations. Second, how these corrections can be implemented into a force field in a relatively simple way, and third, how the sampling of conformational space by metadynamics reflects the corrections towards a better agreement with experimental results.