ABINITIO FORCE-FIELD FOR SIMULATIONS OF PROTEINS AND NUCLEIC-ACIDS

We present a new self-consistent set of ab initio analytical pair potential to predict specific nonbonded interactions of protein with nucleic acid, of protein with protein, and of nucleic acid with nucleic acid. The purpose of this study is to represent the interaction between biological molecules with an accuracy equivalent to the ab initio molecular orbital calculations, which are used as reference data to obtain the pair potentials. Atoms in nucleic acids and proteins are classified according to their chemical environments. An "effective charge," a modification of a charge obtained from the Mulliken population analysis, is introduced and used to represent the electrostatic energy. More than 30,000 SCF interaction energies have been calculated to provide the reference data for the fitting procedure that we have adopted in the parameterization of the potentials. The standard deviation is 1.61 kcal/mol for interaction energies spanning the range from about -220 kcal/mol to +20 kcal/mol. Molecular dynamics simulations, using the above new set of force field, have been performed successfully for the systems where adequate treatments of specific interactions are required: The stability of alpha-helix of C-peptide and the interaction of spermine with oligonucleotide are examined as preliminary examples.