Many globular soluble proteins consist of relatively rigid domains connected by flexible loops or linkers. During functionally important processes, they often undergo large-scale conformational changes known as domain motions, which play a critical role in regulating biological activity. Understanding the interplay between these large conformational changes and their underlying mechanisms remains a major challenge. Advances in experimental and computational techniques have enabled high-resolution structural analyses of proteins with multiple chains and domains, as well as long-timescale molecular dynamics (MD) simulations that capture their extensive structural dynamics. However, extracting essential domain motions from the highly complex and large-scale MD trajectories of such molecules remains extremely difficult. In this study, we propose a reduced-particle representation of proteins, in which a small set of particles effectively captures hinge and twist motions characteristic of domain motions. Each domain is represented by its centroid, hinge particles are placed at inter-domain boundaries, and an additional “rotational point” is introduced to capture the rotational motion of each domain. Since these rotational points cannot be determined analytically even when the domains and hinge positions are determined, we optimized their positions using simulated annealing to minimize the difference between principal component (PC) vectors obtained from Cα-level resolution and those from the reduced-particle representation. We applied this method to multidomain soluble proteins, including the S-protein and demonstrated that it allows intuitive characterization of large-scale conformational changes.
Kobayashi et al. (Sun,) studied this question.