261222 Modeling of Globular Protein Crystal Growth Via Kinetic Monte Carlo Simulations
The production of highly-ordered, high quality protein crystals through batch crystallization processes is vital in devising proteins for therapeutic purposes. Despite extensive experimental and theoretical work on understanding protein structure and function, there is lack of a systematic framework that relies on fundamental understanding of the nucleation and growth mechanisms of protein crystals at the microscopic level and utilizes such information to model and operate protein batch crystallization processes at the macroscopic level.
Our work focuses on investigating the growth of protein crystals via molecular dynamics and kinetic Monte Carlo simulations and using such knowledge to model and control batch protein crystallizers through population, mass and energy balances. The modeling and simulation work is tested against available experimental data.
The present work focuses on modeling and simulation of globular protein crystal growth. The model protein used for this work is tetragonal hen egg white lysozyme. Since crystal growth is a non-equilibrium process, kinetic Monte Carlo methods are developed. As is common practice in simulations of crystal growth, the solid-on-solid model is assumed. In this approximation, particles are deposited on the growing crystal lattice without voids or overhands, resulting in a highly compact crystal. In this work we consider three types of microscopic events: 1) molecular attachment, 2) detachment, and 3) migration events on the (101) and (110) faces of model protein. The implementation of the kinetic Monte Carlo methodology requires knowledge of the binding energies, the impingement rate, and the crystallization driving force. Previous work assigned a range of values to these parameters until satisfactory agreement between the calculated and the experimental growth rates was achieved.
Furthermore, the surface kinetics of the model protein is described for each microscopic event. For molecular attachment it is assumed that each lattice site is available for attachment, and thus the attachment rate is equal over the lattice. Conversely, for molecular detachment and migration events, the rates are dependent on the local environment. Each local environment comprises up to four nearest neighbors. This allows us to classify our local environment into five classes (zero to four nearest neighbors) in order to increase the computational efficiency when calculating the rates associated with executing a kinetic Monte Carlo event. The specific details of our simulation model and extensive simulation results covering the various aspects discussed above will be presented.