(13h) Simulation of Peg/Poly Aspartic Acid Micelles Provides Insight for Design of Drug-Delivery Systems

We use an on-the-fly biasing method to explore the conformational profile of poly(ethylene glycol)/poly(aspartic acid) block copolymer molecules in lipozomal micelles. The configuration of assembled lipozome micelles is of importance when designing micelle-based small drug molecule delivery systems. However, these configurations are difficult to measure experimentally, and the timescales required for micelle assembly in-silico can be prohibitive as well. In this work, we use a custom plugin written in our group, HOOMD-TF, that allows GPU-accelerated machine learning via Google's TensorFlow library in tandem with the HOOMD-blue molecular dynamics simulation engine to simulate a coarse-grained poly(ethylene glycol) and poly(aspartic acid) block copolymer micelle in water. The polymer concentration is 1.25mg/mL, with a size of 117 ethylene glycol and 48 aspartic acid residues, and is simulated in the NVT ensemble at room temperature. We use HOOMD-TF to employ an on-the-fly biasing method to match the system configuration to experimental X-ray scattering data and investigate the conformational space of these micelles. We perform cluster analysis of the polymer backbone angles to find the most common configurations, and sample the free energy surface along angular space and radius of gyration. We also measure the observed micelle internal and external diameters and compare with experimental observations. This information can be used to help select suitable small drug molecules for encapsulation, and to design future polymer micelle drug-delivery systems.