An analysis of the conformational paths of citrate synthase.

Previous simulation studies have provided reaction pathways leading from the closed to the open form of citrate synthase. We now undertake a detailed analysis of these pathways using a variety of different tools including backbone dihedral angles, P-Curves helicoidal parameters, inter-helix geometrical parameters, and accessibility calculations. The results point to a relatively small number of residues, mostly in loop regions, which are responsible for the majority of the conformational changes observed. An important role is attributed to transient changes in the backbone which facilitate movement along the reaction coordinate. Comparisons between the two pathways show that they share many common features despite the different algorithms used to generate them.

Theoretical determination of conformational paths in citrate synthase.

Two methods are developed for the theoretical determination of a conformational path between two well-documented forms, a closed form and the open form [Remington et al. (1982) J. Mol. Biol. 158, 111-152] of pig heart citrate synthase, a dimeric enzyme of 2 x 437 residues. The first method uses the minimization of the sum of the potential energies at a set of equidistant points, according to Elber and Karplus [(1987) Chem. Phys. Lett. 139, 375-380]. The initialization of the algorithm is modified to account for large-angle rotations of many groups by performing the interpolations in the space of internal polar coordinates of a set of generalized Jacobi vectors earlier introduced by Durup [(1991) J. Phys. Chem. 95, 1817-1829] and by carefully testing all choices of directions of rotation for determining the initialized midpoint between the known forms. The path includes intermediate points, created by successive splittings of each interval into two equal parts, with a partial energy minimization performed after each splitting. The minimization encounters the well-known local-minima problem, which here is handled by low-temperature molecular dynamics annealing. It is shown that the best ratio of potential energy decrease to rms deviation is achieved by running the dynamics at 50 K, as compared to 100 K and above. The main character of the path obtained is the occurrence of strong to-and-fro variations of some dihedral angles at specific stages along the path. The second method, which we name directed dynamics, uses only low-temperature molecular dynamics simulations by starting trajectories from each of the two known forms with initial velocities directed toward the other one. The procedure is iterated by restarting trajectory pairs after the points of closest approach of the preceding pair. The two half-paths thus built eventually meet after 70 iterations. This method provides a second path with strong similarities, as well as some differences, with respect to the path obtained by the first method.