QM/MM Studies of Enzymic Reactions and Conformational Studies of Peptides in Aqueous Solution - Methods Development and Applications
Parallel computing is becoming increasingly important in the study of systems of biomolecular interest. The replica-exchange molecular dynamics (REMD) method is a novel and powerful molecular simulation method for sampling configuration space that is well suited to parallel architectures. In the REMD simulation method, small amounts of data (simulation temperatures) are exchanged between randomly chosen pairs of processors. Such random walks in temperature space enforce random walks in potential energy space. In this project we are investigating the use of the REMD method for the study of activation and reaction free energies of enzyme catalysed reactions, and for the conformational analysis of peptide - prion protein (PrP) repeat - systems in aqueous solution. We are developing and testing methods that can utilize the sampling power of REMD together with hybrid quantum mechanical and molecular mechanical (QM/MM) methods to obtain free energy estimates in studies of enzymic reaction mechanisms. The QM/MM methodology, which forms the basis of our simulation studies on enzyme catalysis, is restricted to the less accurate and problematic semiempirical QM methods such as AM1 and PM3. Consequently, we are also making an effort to refine the underlying semiempirical-QM/MM methodology with respect to the force-field description of the hybrid QM/MM system in the REMD simulations. In order to obtain accurate free energies ab initio and/or density functional theory (DFT) calculations of energy derivatives with respect to the reaction coordinate are also required.
|
Principal Investigator Peter CumminsComputational Proteomics and Therapy Design, JCSMR Australian National University |
Project u51, d52 |
|
Co-Investigators Ivan RostovVladislav Vassiliev ANU Supercomputer Facility Australian National University Jill Gready Computational Proteomics and Therapy Design, JCSMR Australian National University |
RFCD Codes 250601, 250602, 270108 |
Significant Achievements, Anticipated Outcomes and Future Work
We have developed a modified semiempirical-QM/MM scheme that overcomes some of the geometrical instability problems that may arise from the use of semiempirical methods in molecular dynamics simulations [1], particularly at the higher temperatures that are required when using the REMD method. We are currently performing REMD simulations using up to 20 processors (i.e. simulation temperatures) to study effects of QM region size on the free energy of hydride-ion transfer in DHFR. The insights gained from these REMD simulations may lead to improved ways of computing free energies along reaction pathways using the more rigorous free energy perturbation (FEP) or thermodynamic integration (TI) methods, rather than the approximate linear response method. We are now in a position to compute free energy surfaces for identifying reactant, product and transition states. We will continue to apply these new methods to study unresolved issues regarding the mechanism of DHFR and in studies of the mechanisms in other enzymes.
We have performed REMD on a 10 residue marsupial PrP peptide using 24 processors with a temperature range of 290 K to 620 K and with a number of different solvent boundary conditions. The initial results correlate well with data obtained from fluorescence resonance energy transfer (FRET) experiments [2]. However, work is in progress to further refine the methods of analysis of the peptide structures obtained from REMD simulations, and to extend the REMD simulation studies to larger (up to 17 residue) PrP repeat peptides.
Computational Techniques Used
The semiempirical AM1 and PM3 QM methods coupled with the AMBER protein force field are used in the QM/MM calculations. AMBER protein force field parameters and the TIP3P water force field parameters are being used for the solvated PrP repeat peptides in the conformational studies. Standard molecular dynamics simulation algorithms are used to sample configuration space and generate the canonical ensembles. In order to enhance this sampling, the REMD method also involves a Monte Carlo simulation step in the temperature (i.e. processor) space. Linear response approximation, free energy perturbation (FEP) and thermodynamic integration (TI) techniques are used to compute free energies. All these techniques have been incorporated into our own locally developed QM/MM simulation package MOPS. Higher level ab initio or DFT ONIOM calculations of energy derivatives are performed using the GAUSSIAN 03 programs.
Publications, Awards and External Funding
External Funding and Awards
None.
Publications
[1] P. L. Cummins and J. E. Gready. Computational Methods for the Study of Enzymic Reaction Mechanisms II: An Overlapping
Mechanically Embedded Method for Hybrid Semiempirical-QM/MM Calculations. J. Mol. Struct. (THEOCHEM) 632, 2003, 247-257.
[2] M. Gustiananda, J. R. Liggins, P. L. Cummins and J. E. Gready. Average Distance between Tryptophan and the N-terminal
residue of Prion Protein Repeat Peptides probed using Fluorescence Resonance Energy Transfer and Molecular Dynamics
Simulation. Biophys. J., in press.