| Popis: |
The huge computational expense of atomistic simulations of protein-protein interactions and the lack of mechanistic detail in heuristic methods motivate the development of colloid-like models for proteins. Patchy particle "minimal" models of anisotropically shaped and functionalized colloids have been highly successful at predicting colloidal assembly. In this dissertation, we focus on the effect of shape and examine the role of entropic forces in various colloidal and biological systems. In the first project, we isolate the effect of shape complementarity in protein dimerization and study the reversible binding processes of fifty protein dimer pairs using hard particle Monte Carlo simulations with depletion interactions. We employ depletants using a generic, implicit depletion model to amplify the magnitude of the entropic forces arising from lock-and-key binding. We find that for a subset (13%) of dimers, protein shape is sufficient to predict native complexes as equilibrium assemblies. We analyze the importance of competing binding configurations and show how they affect assembly. Finally, we implement a machine learning classifier with support vector machine to identify the cases where shape alone can predict the native protein interface. The classifier achieves a precision of 89% and a recall of 77%. In the second project, we explore the limit of shape-based models for higher-order protein assembly, and investigate the assembly pathway of insulin as an example. We report the native insulin assembly pathway can be reproduced in a hierarchical manner with a highly simplified shape-based Monte Carlo simulation model despite the absence of enthalpic interactions. We demonstrate that insulin shape is important for explaining the formation of different insulin associated structures including dimer, hexamer and crystal. We further discuss the observed shape complementarity of insulin interfaces, and show that the formation of different interfaces can be tuned by the strength of the shape affinity. Finally, we demonstrate that the coordination of ion particles provides additional thermodynamic stabilization to the hexameric form of insulin. Lastly, we study how shape anisotropy and particle size affect the structural color behavior of discoidal particles. Using hard particle Monte Carlo simulations, we demonstrate that engineering the size and aspect ratio of the discoids can change the degree of ordering in their assembled structures, thereby leading to tunable structural color responses. The simulated structures and reflection spectra agree well with experimental observations, suggesting the power of shape-based models in capturing the assembly structures and properties. Results from this dissertation give insight into the role of shape in protein assembly, which has not been closely investigated, and provide new shape-based minimal models for understanding colloidal and protein assemblies. |
