This award to Stanford University facilitates scientific research using the large new computational resource named Blue Waters scheduled to be deployed at the University of Illinois at Urbana-Champaign.
The PI will investigate membrane fusion, the process by which neuronal exocytosis and infection by enveloped viruses occur. This mechanism has been notoriously difficult to characterize at a molecular level. Part of the problem is that the underlying reaction that fusion proteins catalyze is not fully understood. The development of robust predictive models for the mechanism of lipid membrane fusion and its catalysis by viral fusion proteins will greatly aid in the understanding of the underlying physical process and how to effectively target it with antiviral agents. Simulations on Blue Waters will be performed using the Gromacs software package.
This project will contribute substantially to the development of peta-scale molecular dynamics simulation software, the integration of tightly-coupled supercomputers with distributed computing, and our understanding of membrane fusion. Molecular dynamics is poised on the edge of a seminal expansion in its capability to make computational predictions of chemical kinetics and mechanism for biomolecules. This project aims to realize the vision of making such predictions for complex systems of physiological relevance. Such entails not merely scaling to systems of larger size but developing analytic methodology for complex reactions. The collaborative team includes developers of both the GROMACS simulation software and the Folding@Home simulation and analysis platform. Simulation and analysis methodology developed as part of this project will be shared with the scientific community in the form of improvements to the publicly available GROMACS software and the open-source software packages on the SimTK website.
Summary. We have both completed and ongoing research projects that utilize the extensive, massively parallel computing infrastructure available on the Blue Waters supercomputing facility. Our workflow involves running massively parallel biomolecular simulations, followed by application of a statistical model that aggregates the simulations to give a reduced set of stable, connected receptor states. We can analyze the connectivity of states in this model to find the most probable pathways between two receptor end states. We have completed work on G-protein coupled receptors (GPCRs), which are membrane proteins that are the targets of approximately 40% of all commercially available drugs (Cherezov, et al. Science, 2007). We focus on the specific GPCR β2 adrenergic receptor and use our approach to provide the first atomistic description of this receptor’s ligand-modulated activation pathways (Kohloff, et al, Nature Chem. in press). On Blue Waters, we targeted states along these pathways with massively parallel, retrospective small molecule virtual screens, resulting in over 2.9 million calculations. By clustering top-scoring molecules based on 3D shape and chemical similarity, we demonstrate that docking to intermediate pathway states allow discovery of unique ligand chemotypes that would have been omitted without knowledge of the full activation pathways (Figure 1). These results show that our model of biomolecular simulations can improve our understanding of drug efficacy at GPCRs receptors and can be incorporated into an effective structure-based drug design approach. Second, we have ongoing biomolecular simulations that will study the folding and binding of the intrinsically disordered protein p53, which is a cancer tumor suppressor. Activation of p53 prevents tumorigenesis and maintains normal cell growth. We study the binding of a key portion of p53 to two proteins, s100b and a member of the sirtuin family, Sir2Tm, that antagonize the p53 activation pathway. This study gives biophysical insight into fly-casting mechanisms for disordered proteins (A. Shoemaker, et al, Proc. Nat. Ac. Sci., 2000), as p53 binds in two distinct structural conformations with the different binding partners. In the most probable pathways connecting the unfolded, unbound state and the folded, bound state, we find a fly-casting mechanism for both proteins, illustrated in Figures 2 and 3. An initial capture process involves broad, mostly polar, protein contacts that allows flexibility for p53 rearrangement toward the bound state. For s100b, we see that p53 can rearrange into a stable misfolded off-pathway intermediate that may contribute the bottleneck rate for finding the native state. In both proteins, disordered regions on the p53 binding partner cooperatively shuttle the p53 peptide to the native, bound state. Understanding p53 interactions with these proteins can inform development of potential inhibitors for cancer therapy, to restore p53 activation and normal cell growth cycles in tumors (G. Botta, et al, Curr. Med. Chem. 2012). We suggest that the green interfaces highlighted in Figures 2 and 3 could be targeted for small molecule or peptide inhibitors to prevent p53 capture. Intellectual Merit. Our work helps shed light on key dynamical processes in biological molecules, especially proteins, the building block of much of biology. In particular, by understanding the basic biophysics of how these molecules behave, we can better understand their function in a cellular context. Broader Impacts. More broadly, the methods we have pioneered can very efficiently use many (thousands to tends of thousands) processors and GPUs to simulate the long timescale dynamics of proteins. This method would empower many other groups to fully use the dramatic power available through Blue Waters for many other simulations of the dynamics of proteins on long timescales.
