What determines sub-diffusive behavior in crowded protein solutions?

The aqueous environment inside cells is densely packed. A typical cell has a macromolecular concentration in the range 90-450 g/L, with 5%-40% of its volume being occupied by macromolecules, resulting in what is known as macromolecular crowding. The space available for the free diffusion of metabolites and other macromolecules is thus greatly reduced, leading to so-called excluded volume effects. The slow diffusion of macromolecules under crowded conditions has been explained using transient complex formation. However, sub-diffusion noted in earlier works is not well characterized, particularly the role played by transient complex formation and excluded volume effects. We have used Brownian dynamics simulations to characterize the diffusion of chymotrypsin inhibitor 2 in protein solutions of bovine serum albumin and lysozyme at concentrations ranging from 50 to 300 g/L. The predicted changes in diffusion coefficient as a function of crowder concentration are consistent with NMR experiments. The sub-diffusive behavior observed in the sub-microsecond timescale can be explained in terms of a so-called cage effect, arising from rattling motion in a local molecular cage as a consequence of excluded volume effects. By selectively manipulating the nature of interactions between protein molecules, we determined that excluded volume effects induce sub-diffusive dynamics at sub-microsecond timescales. These findings may help to explain the diffusion-mediated effects of protein crowding on cellular processes.

Definition of the Minimal Contents for the Molecular Simulation of the Yeast Cytoplasm.

The cytoplasm is a densely packed environment filled with macromolecules with hindered diffusion. Molecular simulation of the diffusion of biomolecules under such macromolecular crowding conditions requires the definition of a simulation cell with a cytoplasmic-like composition. This has been previously done for prokaryote cells (E. coli) but not for eukaryote cells such as yeast as a model organism. Yeast proteomics datasets vary widely in terms of cell growth conditions, the technique used to determine protein composition, the reported relative abundance of proteins, and the units in which abundances are reported. We determined that the gene ontology profiles of the most abundant proteins across these datasets are similar, but their abundances vary greatly. To overcome this problem, we chose five mass spectrometry proteomics datasets that fulfilled the following criteria: high internal consistency, consistency with published experimental data, and freedom from GFP-tagging artifacts. Using these datasets, the contents of a simulation cell containing a single 80S ribosome were defined, such that the macromolecular density and the mass ratio of ribosomal-to-cytoplasmic proteins were consistent with experiment and chosen datasets. Finally, multiple tRNAs were added, consistent with their experimentally-determined number in the yeast cell. The resulting composition can be readily used in molecular simulations representative of yeast cytoplasmic macromolecular crowding conditions to characterize a variety of phenomena, such as protein diffusion, protein-protein interactions and biological processes such as protein translation.