Exploring GB1 Dimerization Through Short Simulations and Markov State Models

Abstract

Protein-protein association, like dimerization, is essential for various biological functions. Here we explore dimerization of the GB1 protein, a well-known model system, through multiple short simulations and Markov State Model (MSM). To this end, Parallel Tempering Metadynamics (PTMetaD-WTE1) simulations were performed along two collective variables, the center of geometry distance (dcog) and inter-chain co-ordination number (ninter) to quickly explore the free energy landscape (FES). Next, several unbiased and independent short MD simulations were run by choosing initial configurations from the explored FES. MSMs were then developed at different timestep resolutions based on the 87.66 μs long aggregated short trajectories to identify the metastable states and transitions. Dimensionality reduction was performed using time-lagged independent component analysis (TICA2) to find the slow coordinates and density based clustering3 was used to identify the macrostates by dynamical coring4 and dynamical clustering. Preliminary analysis reveals that a 7-state MSM can capture the intermediate states onpathway to dimerization and highlights the role of input data resolution on the MSMs.