Development of a molecular dynamics based method to accelerate sampling of large domain motions in proteins: Applications to adenylate kinase

Enzymes can undergo large conformational rearrangements when they bind their substrates. To study these rearrangements with the Molecular Dynamics (MD) computational method requires new methods to improve the ability of exploring conformational space. We study large scale motions of Adenylate Kinase (AK), an enzyme that has domains that close over its substrates to form a catalytically active complex.; Carrying out Molecular Dynamics on proteins requires an efficient method to handle the long range-part of the MD force field. We devise a parallel version of a Particle Mesh Ewald method and implement it in the CUKMODY MD code that permits accurate and efficient simulation of large systems such as solvated proteins.; Adenylate kinase from E. coli (AKE), composed of AMP binding (AMP-bd), Lid and Core domains, is simulated in its ligand free form with MD at 303 and 500 K, in order to study its domain motions and to compare the results with those of energy transfer experiments that monitor the distance between domains. The simulation trajectories are analyzed using a Principal Component Analysis (PCA) method that extracts large scale domain fluctuations. We find that AKE has motions that do indicate a closing motion is present, but the limited simulation time of conventional MD does not permit sampling of more closed conformations.; Increasing the sampling efficiency of configurations by MD is accomplished using a variant of the Replica Exchange Method (REM). When a reaction coordinate is known, for example, a distance between protein domains that serves as a measure of how open or closed the protein is, the different systems in the REM correspond to different restraint potentials that keep this interdomain distance around desired values along the reaction coordinate. This Distance REM (DREM) is implemented in CUKMODY by a combination of Message Passage Interface (MPI) and pThread methods. DREM also has the advantage that the different systems can be run in parallel on different nodes of a cluster computer.; DREM is applied to AKE by changing the distance between its AMP-bd and Core domains. The DREM greatly accelerates the rate and extent of configurational sampling, and leads to equilibrium sampling as measured by monitoring collective modes obtained from a PCA. The potential of mean force along the reaction coordinate reveals a rather flat region for distances from the open to a relatively closed AKE conformation. The potential of mean force for smaller distances has a distinct minimum that is quite close to that found in the closed form x-ray structure. In concert with a decrease in the reaction coordinate distance (Amp-binding to Core distance), the Lid to Core distance of AKE also decreases. Therefore, apo-AKE can fluctuate from its open form to conformations that are quite similar to its closed form x-ray structure, even in the absence of its substrates. We wrote ANALYZER to carry out a set of analyses for MD trajectories.