Unbiased Clustering of Molecular Dynamics for Spatially Resolved Analysis of Chemically Heterogeneous Surfaces

A technique is described for resolving and interpreting molecular interactions with a chemically heterogeneous surface. Using total internal reflection fluorescence microscopy, dynamic single molecule trajectories were accumulated simultaneously for fluorescently labeled fatty acid (interacting primarily via hydrophobic interactions) and dextran (interacting via hydrogen-bonding interactions) probe molecules at the interface between an aqueous solvent and a photopatterned solid surface with distinct regions of amine-terminated and poly(ethylene glycol) self-assembled monolayers. Using dynamic properties of the probe molecules (adsorption rate, surface diffusion coefficient, residence time), an unsupervised Gaussian mixture model algorithm was used to identify areas of the surface that were chemically related to each other, and the dynamic behaviors of the probe molecules were studied statistically on these distinct regions. The dynamic data were compared to data from homogeneous surfaces of known chemistry to provide a chemical identification of each location on the surface. Spatial maps were also constructed, allowing for spatial visualization of surface chemistry on a hydrophilic substrate. This work enables the direct study of interactions between single-molecule probes and distinct surface chemistries, even in the presence of spatial heterogeneity, without human bias, assumptions about surface structure, or model-dependent analysis.