Samuel Flores

The prediction of conformational change in proteins has a long history, with many contributions and outstanding problems. Our goal is to predict the dynamics of the very slowest degrees of freedom of the system, those which cannot be computed for all atoms by frontal solution of Newton’s equations of motion, by Monte Carlo, or by other techniques without large expenditure of computer time. To simplify we focused on the largest class of motion: domain hinge bending.

The first step in this process is the determination of the hinge location. Many techniques exist to compute flexibility, relying on normal mode expansions, crystal lattice vibrations, and Nuclear Magnetic Resonance data. However none focus specifically on the problem of determining the location of points of flexibility separating highly interconnected and stable regions of the protein, the structural domains. We first describe some structural, physicochemical, and biological characteristics of hinges. We then present novel algorithms that analyze the stability of various regions of the protein, their normal mode motional correlations, and the topology of the connections between atoms, seeking to identify the boundaries of these domains and hence the probable hinge points. We present the development, testing, integration, and results of these algorithms.

Hinge points having been determined, it is possible to reduce the dimensionality of the problem, since the motions within each domain can be assumed to be highly correlated. This assumption allows us to displace one of the domains without recomputing its internal coordinates. Some improbable interactions result but we show how these can be annealed by computing the short-time trajectory of motion. The relative orientation of domains is selected from many possibilities based on the free energy of interaction between the protein and its ligand, stability, compactness, and conservation of domain structure.