Novel Simulation Methods Targeting the Protein Crystal Folding Problem

The protein folding problem has been a recognized experimental, theoretical and simulation challenge for more than 50 years. The ability to computationally predict the structure, thermodynamics and kinetics of proteins helps open the door to understanding biological function and the role of genetic missense variations, and provides insight into designing proteins with novel functions. The related protein crystal folding problem is also of great importance, not only for research purposes (i.e., to predict amenable crystallization conditions), but also for human health due to the prevalence of protein aggregation diseases such as those resulting from prions.

Despite significant progress in our understanding of the protein folding problem, theories and computational tools targeted at the protein crystal folding problem remain limited. This is due to the latter presenting a variety of challenges that go beyond protein folding. First, the time scales for nucleation and protein crystal growth are necessarily longer than for protein folding alone. Second, even minor force field imperfections (i.e., the potential energy functions used to describe intra- and intermolecular energetics) result in the prediction of the wrong space group, unit cell parameters and/or molecular conformation(s).

To address the challenges of the protein crystal folding problem, novel physics-based simulation methods will be presented that focus on accelerating the phase transition between the solvated and crystalline states. Our approach combines novel electrostatics algorithms for the polarizable atomic multipole AMOEBA force field (e.g., a generalized Kirkwood implicit solvent for the solvated state and space group symmetry constraints for the crystalline state) with advanced alchemical thermodynamic sampling schemes to search for the most stable space groups, unit cell parameters and molecular conformations. The talk will conclude with ongoing work to predict NNQQ tetra-peptide polymorphs, which is a fibril-forming segment of the yeast prion protein Sup35. This represents one the largest ab initio crystal structure predictions attempted to date.