Grid-steered molecular dynamics
Grid-steered molecular dynamics (G-SMD) is a flexible method for applying forces that our group has implemented in NAMD. Much as in experiment, simulation studies often involve perturbing the system in some way and monitoring the result. As simulations have become bigger, longer, and more complex, the need for more sophisticated forcing techniques has increased. In the G-SMD method, an arbitrary external potential field, defined on a grid, is applied to desired target atoms with arbitrary coupling, making it a very flexible tool. Because it is coded natively in NAMD, the performance impact of G-SMD is minimal. G-SMD provides tremendous flexibility in the forces that can be applied. It was originally designed to apply an enhanced electrostatic field to DNA threaded through the membrane protein α-hemolysin in order to realistically increase the DNA translocation speed. It is the basis for a method of combining crystallographic structures and cryo-EM maps to obtain an all-atom model developed by Trabuco et al. and called Molecular Dynamics Flexible Fitting (MDFF). A variation of this approach was also recently used by our group to study the mechanical properties of a complete microtubule.
The transport of biomolecules across cell boundaries is central to cellular function. While structures of many membrane channels are known, the permeation mechanism is known only for a select few. Molecular dynamics (MD) is a computational method that can provide an accurate description of permeation events at the atomic level, which is required for understanding the transport mechanism. However, due to the relatively short time scales accessible to this method, it is of limited utility. Here, we present a method for all-atom simulation of electric field-driven transport of large solutes through membrane channels, which in tens of nanoseconds can provide a realistic account of a permeation event that would require a millisecond simulation using conventional MD. In this method, the average distribution of the electrostatic potential in a membrane channel under a transmembrane bias of interest is determined first from an all-atom MD simulation. This electrostatic potential, defined on a grid, is subsequently applied to a charged solute to steer its permeation through the membrane channel. We apply this method to investigate permeation of DNA strands, DNA hairpins, and alpha-helical peptides through alpha-hemolysin. To test the accuracy of the method, we computed the relative permeation rates of DNA strands having different sequences and global orientations. The results of the G-SMD simulations were found to be in good agreement in experiment.
Microtubules are ubiquitous biological filaments found in all eukaryotic cells. They are the largest type of cellular filament, and are essential in processes ranging from mitosis and meiosis to flaggelar motility. Due to their structural importance, the mechanical properties of microtubules have been extensively studied. However, because of the small size of microtubules and their high rigidity, experimental studies have determined the Young's modulus only indirectly. Molecular dynamics (MD) simulations allows the elastic properties of a biopolymer to be determined computationally. However, while the atomic structures of the building blocks of a microtubule (α- and β-tubulin) have been solved, the only published structures of a complete microtubule are cryo-electron microscopy maps far from atomic resolution. Using the cryo-EM map as a guide, we have produced the first all-atom structure of a complete microtubule. With this model, we applied tension, compression, and shear to determine the elastic moduli of an effectively infinite microtubule, yielding results in agreement with previous estimates. This work is one of the first to combine cryo-EM and crystallographic structures for subsequent all-atom MD simulation. The successful performance of such a model opens the door to the simulation of many other systems whose constituent units are known in atomic detail but whose complete structure is known only at lower resolution.