Computer Simulation of Biomolecular Dynamics and Reactions

June 5-8, 2012 Location: Pittsburgh Supercomputing Center Pittsburgh, PA Organizer: Troy Wymore (PSC/NRBSC)

Introduction and Motivation A major challenge in computational biochemistry and molecular biology is elucidating how enzymes achieve their enormous catalytic power as well as discovering the governing principles by which enzymes evolve novel specific functions. Overcoming these challenges requires sophisticated computational approaches both as an aid and a guide to experimental research. The knowledge acquired from such detailed studies has led to i) the design of extremely tight binding transition state analogues that often serve as starting points in the design of new therapeutics and ii) the engineering of new enzymes that extend beyond natures repertoire. The latter may be harnessed for more energy efficient and environmentally friendly synthesis of useful chemicals. Finally, with todays modern computing platforms, QC/MM investigations of both native and several mutant enzymes within an enzyme family can more readily be performed enabling the construction of functional landscapes. These functional landscapes broaden our perspective of enzyme catalysis providing insight into mechanisms of molecular evolution and protein design principles. Arguably, the most compelling approach for the simulation of enzyme reactions is the use of hybrid quantum chemical (QC) and molecular mechanical (MM) potentials in combination with reaction-path finding and free-energy determination methodologies. QC/MM methods have become standard tools in computational science in the last decade or so and many simulation programs now implement them. Nevertheless, they are an active field of research with remaining questions concerning their accuracy and the best way in which they can be applied. For example, in cases where quantitatively accurate free energy profiles or surfaces are needed to distinguish between competing enzymatic mechanisms, how does an investigator balance the computational requirements of simulating sufficiently long enough to converge the free energy profile/surface while employing a QC method that captures the necessary physicochemical phenomena encompassing enzymatic reactions. Simply stated, applying QC/MM methods to the study of enzyme reactions is not as straightforward as performing QC calculations on small isolated systems or classic force field simulations of condensed phase biomolecular systems. The aim of this workshop is to introduce these critical questions and the enzyme simulation field in general using the pDynamo program (www.pdynamo.org) and ultimately to spur the participants to either employ these methods into their own research or possibly improve upon the way they currently employ hybrid QC/MM methods. In addition, the prediction and analysis of protein dynamics using normal mode analyses (NMA) and elastic network models (ENM) will be explored with the ProDy package (www.csb.pitt.edu/prody/). The pDynamo program has been developed in the laboratory of Dr. Martin Field since the beginning of the 1990's, is opensource software and is used for simulations of enzymes and other systems by laboratories worldwide. The program itself is under constant development with aims that include: (i) the addition of new features; (ii) to increase the ease of use and robustness of the simulation techniques; and (iii) to take advantages of advances in hardware, such as multicore processing and GPUs. Though pDynamo is a functional QC and MD simulation package, its strength lies in its focus on the hybrid QC/MM capabilities. The program can read and write files in formats produced by other popular MD simulation programs, including CHARMM, NAMD, AMBER, Desmond and GROMACS. Thus, simulation snapshots from these programs can be seamlessly integrated into pDynamo and investigated further with hybrid QC/MM methods. With these features, we anticipate several participants will be able to advance through the basic tutorial

material and be able to focus on the more critical questions in their own enzyme mechanism research. We anticipate that a new version will be released before the workshop that will include the current code optimizations that have already sped up typical QC/MM calculations on a single core (about 5 times faster) and is fine grain parallelized for use on multi-core platforms. In addition, new features including Particle Mesh Ewald for use with hybrid QC/MM potentials, new semiempirical molecular orbital methods and the ability to employ interpolated corrections between low and high level QC methods will be available. These developments facilitate more extensive and more accurate QC/MM calculations than was previously possible. Workshops on simulating enzyme reactions using Dynamo (www.pdynamo.org) have been held three times in the last few years. The Pittsburgh Supercomputing Center hosted the first and third workshops in 2007 and 2009, whereas the second was hosted by the Bioinformatics Institute in Singapore in 2008. Contents and Objectives The workshop will present computational approaches for the prediction and analysis of protein dynamics using ENM-based NMA, the construction of atomic enzyme models (assignment of protonation states, addition of solvent, force field parameterization of non-standard residues and ligands, and assignment of QC/MM boundaries) and the subsequent simulation of enzymatic reactions. Lectures will be held on the relevant theoretical background, on state-of-the-art methods and on current challenges in the enzyme simulation field. Example topics include the applicability of various quantum chemical (QC) methods for use in hybrid QC/MM simulations, parameterization of molecular mechanics (MM) force fields, hybrid QC/MM potential energy functions, simulation techniques for finding reaction paths, free-energy calculations and advanced topics, such as quantum dynamical effects observed in proton/hydride transfer reactions. The interaction of the participants with the organizers, other experienced users and developers promises to be a rich and rewarding experience for all involved. The tutorials will primarily utilize ProDy and the pDynamo package. In addition, i) construction of atomic enzyme models, energy minimizations, simulations and analysis of results will be demonstrated with the aid of a newly developed pDynamo extension to the popular molecular visualization program VMD and PyMol and ii) though pDynamo has some Density Functional Theory (DFT) methods available in the package, we will demonstrate how to employ the pDynamo interface to ORCA (www.thch.uni-bonn.de/tc/orca/), a freely available and parallel QC package, enabling an even more extensive range of hybrid QC/MM potentials to be utilized. Tutorials that cover a diverse range of enzyme reaction modeling issues including the role of solvent molecules, conformational transitions, proton transfers, specification of appropriate reaction coordinates and the use of various reaction path methods will be available giving participants the opportunity to follow those that most resemble their own research programs. Experience with pDynamo is helpful but not a prerequisite for attendance. The workshop is designed for graduate students, post-doctoral researchers and faculty in computational biophysics and physical organic chemistry who have experience with computational chemistry methods and seek to enhance their capabilities to include the modeling

of enzymatic reactions. Participants will be encouraged to discuss and work on bio-molecular systems from their own research during the workshop. Tentative Agenda Day 1: From Elastic Network Models to Force Field Parameterization 09:00-09:30 Introduction and Opening Remarks 09:30-10:30 Lecture: Biomolecular Dynamics Dr. Timothy Lezon 10:30-10:45 Coffee Break 10:45-1:00 Tutorial: Elastic Network Models using ProDy (Dr. Ahmet Bakan) 1:00-2:00 Lunch Break 2:00-3:00 Lecture: Parameterization of MM Force Fields and Semiempirical Molecular Orbital (SMO) Methods 3:00-3:15 Coffee Break 3:15-6:00 Tutorial: Developing MM Force Fields and SMO Corrections Day 2: Modeling Enzyme Structure with Hybrid Potentials 09:00-10:00 Lecture: Basic Functionality of pDynamo 10:00-10:30 Coffee Break 10:30-12:30 Tutorial: Constructing an enzyme model for MM energy calculations in pDynamo 12:30- 1:30 Lunch Break 1:30- 2:30 Lecture: Overview of Hybrid QC/MM Potential Methods 2:30- 2:45 Coffee Break 2:45- 6:00 Tutorial: QC/MM enzyme model construction and energy calculations Day 3: Modeling Enzyme Reactions with Hybrid Potentials 09:00-10:30 Lecture: Studying Biomolecular Reactions with Hybrid QC/MM Potentials 10:30-11:00 Coffee Break 11:00-12:30 Tutorial: Simulating Enzyme Reactions with pDynamo 12:30- 1:30 Lunch Break 1:30- 2:30 Lecture: Quantum Tunneling in Enzyme Catalyzed Reactions 2:30- 2:45 Coffee Break 2:45- 6:00 Tutorial: Simulating Enzyme Reactions with pDynamo (continued) Day 4: Analysis and Informal Discussions 09:00-10:00 Tutorial: Analysis of Hybrid Potential Simulations 10:00-10:30 Coffee Break and Evaluations 10:30-12:00 Open session with instructors 12:00- 1:00 Lunch 1:00- 4:00 Open session/discussion with instructors

Instructors Martin J. Field Timothy Lezon Ahmet Bakan Nikolay Simakov Troy Wymore Jose Fernado Bachega Basic Literature Some of the pre-requisite readings we have suggested in our past workshops include: Simulating Enzyme Reactions: Challenges and Perspectives. Martin J. Field. Journal of Computational Chemistry 2002. Vol. 23, Issue 1, pg. 48-58 Mechanisms and Free Energies of Enzymatic Reactions Jiali Gao, Shuhua Ma, Dan T. Major, Kwangho Nam, Jingzhi Pu, and Donald G. Truhlar Chemical Reviews, 2006. Vol. 106, pg. 31883209. QM/MM Methods for Biological Systems. Hans Senn and Walter Thiel. Topics in Current Chemistry, 2007. Vol. 268, pg. 173-290 More specific literature concerning the pDynamo program includes: The pDynamo Library for Molecular Simulations using Hybrid Quantum Mechanical and Molecular Mechanical Potentials, Martin J. Field J. Chem. Theo. Comp. 4, 1151-1161 (2008) A Practical Introduction to the Simulation of Molecular Systems, Martin J. Field, Cambridge University Press (2007) Institute of Structural Biology, Grenoble, France University of Pittsburgh Medical School University of Pittsburgh Medical School Pittsburgh Supercomputing Center Pittsburgh Supercomputing Center University of Sao Paulo