ABSTRACTPrediction of ligand-induced protein conformational transitions is a challenging task due to a large and rugged conformational space, and limited knowledge of probable direction(s) of structure change. These transitions can involve a large scale, global (at the level of entire protein molecule) structural change and occur on a timescale of milliseconds to seconds, rendering application of conventional molecular dynamics simulations prohibitive even for small proteins. We have developed a computational protocol to efficiently and accurately predict these ligand-induced structure transitions solely from the knowledge of protein apo structure and ligand binding site. Our method involves a series of small scale conformational change steps, where at each step linear response theory is used to predict the direction of small scale global response to ligand binding in the protein conformational space (dLRT) followed by construction of a linear combination of slow (low frequency) normal modes (calculated for the structure from the previous step) that best overlaps with dLRT. Protein structure is evolved along this direction using molecular dynamics with excited normal modes (MDeNM) wherein excitation energy along each normal mode is determined by excitation temperature, mode frequency, and its overlap with dLRT. We show that excitation temperature (ΔT) is a very important parameter that allows limiting the extent of structural change in any one step and develop a protocol for automated determination of its optimal value at each step. We have tested our protocol for three protein–ligand systems, namely, adenylate Kinase – di(adenosine-5’)pentaphosphate, ribose binding protein – β-D-ribopyranose, and DNA β-glucosyltransferase – uridine-5’-diphosphate, that incorporate important differences in type and range of structural changes upon ligand binding. We obtain very accurate prediction for not only the structure of final protein–ligand complex (holo-structure) having a large scale conformational change, but also for biologically relevant intermediates between the apo and the holo structures. Moreover, most relevant set of normal modes for conformational change at each step are an output from our method, which can be used as collective variables for determination of free energy barriers and transition timescales along the identified pathway.
A Genetic Algorithm with Zooming for the Determination of the Optimal Open Pit Mines Layout
