A simplified representation of anisotropic charge distributions within proteins

Effective coarse-grained representations of protein-protein interaction potentials are vital in the modeling of large scale systems. We develop a method to fit an arbitrary number of effective charges to approximate the electrostatic potential of a protein at a given pH in an ionic solution. We find that the effective charges can reproduce an input potential calculated from a high resolution Poisson-Boltzmann calculation. Since the effective charges used in this model are not constrained to the locations of the original charged groups, the extra degrees of freedom allows us to reproduce the field anisotropy with fewer charges. The fitting procedure uses a number of approximations in the charge magnitudes, initial conditions, and multipoles to speed convergence. The most significant gains are found by fitting the multipole moments of the effective charge potential to the moments of the original field. We show that the Yukawa potential is not only sufficient as a pairwise summation in reproducing the potential, but comes naturally from the linearized expansion of the Poisson-Boltzmann equation. We compute interaction energies and find excellent agreement to the original potential. From the effective charge model we compute the electrostatic contribution to the second virial coefficient.