Abstract
A recently developed variational Bayesian analysis using pattern recognition and machine learning of single viral ribonucleoprotein (RNP) particle tracks in the cytoplasm of living cells provides a quantitative molecular explanation for active diffusion, a concept previously "explained" largely by hypothetical models based on indirect analyses such as continuum microrheology. Machine learning shows that vesicular stomatitis virus (VSV) RNP particles are temporarily confined to dynamic traps or pores made up of cytoskeletal elements. Active diffusion occurs when the particles escape from one trap to a nearby trap. In this paper, we demonstrate that actin filament disruption increased RNP mobility by increasing trap size. Inhibition of nonmuscle myosin II ATPase decreased mobility by decreasing trap size. Trap sizes were observed to fluctuate with time, dependent on nonmuscle myosin II activity. This model for active diffusion is likely to account for the dominant motion of other viral and cellular elements. IMPORTANCE RNA virus ribonucleoproteins (RNPs) are too large to freely diffuse in the host cytoplasm, yet their dominant motions consist of movements in random directions that resemble diffusion. We show that vesicular stomatitis virus (VSV) RNPs overcome limitations on diffusion in the host cytoplasm by hopping between traps formed in part by actin filaments and that these traps expand and contract by nonmuscle myosin II ATPase activity. ATP-dependent random motion of cellular particles has been termed "active diffusion." Thus, these mechanisms are applicable to active diffusion of other cellular and viral elements.