(627b) Multidimensional Free Energy Landscapes for the Binding of Functionalized Nanoparticles to Lipid Bilayers

Nanoparticles (NPs) functionalized with organic ligands offer a potential nanotechnology platform for drug/gene delivery and phototherapy. These applications require specific interactions with cell membranes to facilitate efficient cell uptake or membrane adsorption while minimizing cytotoxicity. These interactions can be tuned by varying the properties of the ligand coating. However, nanoparticle interactions with the cell membrane remain difficult to predict a priori, hindering nanoparticle design. To address this challenge, previous experiments have correlated NP binding to model lipid bilayers with the octanol/water partition coefficient of the grafted ligand, yet the mechanism for this binding process is still largely unexplored. Molecular dynamics (MD) simulations offer the capability to understand these processes, explore the impact of new design motifs, and predict binding behavior. A challenge for molecular simulations is modeling the free energy landscapes underlying the binding of large NPs. Past simulation studies of NP-bilayer binding have utilized the center-of-mass (COM) distance between the NP and the bilayer to calculate free energy differences for the binding of NPs with short (< 1nm) ligands. Recently, atomistic MD simulations of NPs grafted with longer functionalized ligands have shown hysteresis in the COM-based measurement and suggest that the number of contacts between the ligands and the bilayer must contribute to the pathway. Sampling contacts in simulations that only bias the COM distance is challenging, indicating that multiple collective variables must be simultaneously biased to obtain the correct reversible path and accurate binding predictions. In this work, we parameterize coarse-grained (CG) simulation models of functionalized NPs based on prior atomistic calculations. We utilize the CG models to determine the free energy for binding NPs to model lipid bilayers as a function of multiple collective variables using multi-dimensional free energy calculation techniques including 2D umbrella sampling and metadynamics. Using the string method, we then efficiently find minimum free energy pathways for binding and compare the binding properties of several functionalized NPs. These predictions offer new mechanistic insight into the binding pathway for these NPs, allowing us to better understand which NP features contribute to enhanced binding to guide further NP design.