Investigation of Membrane Protein Dynamics using Correlative Single-Particle Tracking and Super-resolution Microscopy Combined with Bayesian Inference of Diusion in Arbitrary Landscapes

Many experiments have shown that the diffusive motion of lipids and membrane proteins are slower on the cell surface than those in artificial lipid bilayers or blebs. One hypothesis that may partially explain this mystery is the effect of the cytoskeleton structures on the protein dynamics. To test this hypothesis, we designed a high-speed single particle tracking microscope and use a hybrid tracking and super-resolution approach on the same cell. We labeled the high-affinity FceRI receptor as a transmembrane protein and GPI-anchored proteins as an example of outer leaflet protein in Rat Basophilic Leukemia (RBL) cells and tracked these membrane proteins at up to 500 frames per second. The cells were fixed immediately after tracking and further labeled for super-resolution imaging of actin filaments. To achieve a reliable correlation between the results from live-cell imaging and super-resolution of the same cells, we evaluated several common and custom fixation protocols with respect to changes in cell morphology, maintenance of thin actin filaments and the speed of fixing proteins, selecting Glutaraldehyde as the best choice. Bright field images allow re-alignment of cell with about ~ 10 nm precision. This sequential approach allowed use of far-red dyes for tracking and super-resolution, ameliorating chromatic aberrations. Our studies provide evidence of an influence of actin on the motion of the transmembrane protein, but not on the GPI-anchored outer leaflet protein. The dynamics of membrane proteins can be characterized by the diffusion constant. An accurate estimate of the diffusion constant from single particle tracking requires proper treatment of experimental effects including finite exposure time, localization error, and blinking of the emitters. Under the assumption of free Brownian motion, these effects can be treated analytically. Accurate estimation becomes more complicated in the case of confined or partially bounding regions. If the boundaries are not considered, the diffusion constant can be severely underestimated. Here, we present a Bayesian method for estimation of the diffusion constant of a membrane protein moving in any arbitrary, but known, landscape of reflecting boundaries. We demonstrated the method on simulated particles undergoing Brownian motion in free regions. Our method improves the diffusion constant estimation but retains a small bias towards underestimation. We evaluated two labeling strategies for super-resolution imaging of actin filaments. We compared Alexa647-phalloidin using a dSTORM approach and Lifeact-Atto655 using a PAINT approach. We found that Lifeact can provide improved super-resolution images at a reduced cost