Molecular packing is studied over a dataset of crystal structures representing 500 diverse globular proteins in their native state ranging in size from 34 to 839 residues. Different size cavities are ubiquitously present due to the presence of void between atoms not accessible to solvent. Using a new algorithm we recently developed, void within a protein is classified as a cavity when large enough to hold a spherical shaped probe of radius, R, otherwise a microvoid. Although microvoid cannot fit an object (e.g. molecule or ion) that is the size of the probe or larger, total microvoid volume is a major contribution to protein volume. As microvoid volume increases the free volume of solution decreases. Importantly, the cavity and microvoid classification depends on probe radius. As probe size decreases, less microvoid forms in favor of more cavities, and vice versa. Microvoid is partitioned into distinct clusters, where a cluster is defined by a contiguous region of space that allows a minimum size object of width, w, to transverse. Cavity and microvoid pathways are visualized and their cluster statistics collected. The linear length scale of a protein, L, is also defined. Universal scaling is demonstrated through data collapse in microvoid cluster statistics as a function of probe size, R, spatial resolution, w, and protein size, L. Characteristics for microvoid, cavity, and solvent accessible boundary volumes are quantified. As probe size is varied from large to small, many disconnected cavities merge to form a percolating path. For fixed probe size, microvoid, cavity and solvent accessible boundary volume properties reflect conformational fluctuations. These results suggest interconversion between microvoid and cavity pathways regulate the dynamics of solvent and substrate penetration that are important to protein function, as well as crowding and pressure effects.
Read full abstract