Machine learning-informed predictions of nanoparticle mobility and fate in the mucus barrier.

Nanomaterial diffusion through mucus is important to basic and applied areas of research such as drug delivery. However, it is often challenging to interpret nanoparticle dynamics within the mucus gel due to its heterogeneous microstructure and biochemistry. In this study, we measured the diffusion of polyethylene glycolylated nanoparticles (NPs) in human airway mucus ex vivo using multiple particle tracking and utilized machine learning to classify diffusive vs sub-diffusive NP movement. Using mathematic models that account for the mode of NP diffusion, we calculate the percentage of NPs that would cross the mucus barrier over time in airway mucus with varied total solids concentration. From this analysis, we predict rapidly diffusing NPs will cross the mucus barrier in a physiological timespan. Although less efficient, sub-diffusive "hopping" motion, a characteristic of a continuous time random walk, may also enable NPs to cross the mucus barrier. However, NPs exhibiting fractional Brownian sub-diffusion would be rapidly removed from the airways via mucociliary clearance. In samples with increased solids concentration (>5% w/v), we predict up to threefold reductions in the number of nanoparticles capable of crossing the mucus barrier. We also apply this approach to explore diffusion and to predict the fate of influenza A virus within human mucus. We predict only a small fraction of influenza virions will cross the mucus barrier presumably due to physical obstruction and adhesive interactions with mucin-associated glycans. These results provide new tools to evaluate the extent of synthetic and viral nanoparticle penetration through mucus in the lung and other tissues.