(73d) Multiscale Modeling and Enhanced Sampling to Examine Self-Assembly in Peptide and Peptoid Systems

Molecular self-assembly driven by non-covalent interactions is a central phenomenon that regulates a multitude of biological functions, from the formation of cell membranes, to the folding of polypeptide chains into proteins. Synthetically, the principles of molecular self-assembly are applied to produce crystals, polymers, and scaffolds for implants. To achieve greater control over manufacturing self-assembled materials, it is imperative to study driving forces that lead to the association of these molecules. As molecular assemblies occur over a gamut of length scales, it is useful to have models at different levels of complexity. In this talk, I will discuss distinct computational approaches that were implemented to elucidate the molecular assembly in two different systems.

First, I studied two biomineralizing peptides R5 and LKα14, which interact with orthosilicic acid to precipitate silica in solution. Both peptides have different secondary structures – R5 is random, whereas LKα14 is helical. However, they both produce silica that is morphologically similar. This suggests that there are dominant interpeptide interactions that cause peptides to assemble in a way that enables uniform templating of minerals. Using enhanced sampling methods to capture aggregation behavior occurring over long times, I compared the dominant driving forces leading to the assembly of these peptides. This knowledge can potentially help manufacture hierarchical nanomaterials in a predictive way, under ambient conditions.

Next, I systematically developed a coarse-grained model for peptoid systems using a bottom-up approach. Peptoids, positional isomers of peptides, are robust, synthetic, bio-inspired polymers that fold and assemble into a variety of nanostructures. An atomistic peptoid forcefield was developed recently that was shown to accurately reproduce local structures from quantum mechanical calculations. I used this forcefield to explicitly parameterize my CG model. To ensure reasonable transferability, the model was further parameterized using different combinations of hydrophobic and hydrophilic sidechain chemistries, chain lengths and temperatures. The mapping is similar to the MARTINI forcefield, where 4 heavy backbone atoms are mapped to a single CG bead. Self-assembled CG peptoids were compared with experimentally observed peptoid networks, to elucidate functional properties of the self-assembled aggregates arising from underlying molecular structure.