Improvements and applications of quantitative single molecule localization microscopy

Single molecule localization microscopy (SMLM) covers a group of stochastic photo-switching super-resolution microscopy methods for example PALM, and STORM, in which fluorescent signals are localized using fitting algorithms. The goal is to image a sufficiently low density of non-overlapping, randomly active fluorophores to determine their sub-pixel position and reconstruct a super-resolved image from thousands of images. Localization of sparsely distributed single molecule signals allows SMLM to break the resolution limit and localize emitter centres with sub 20 nm precision.

In the four contributions included in this thesis, I applied SMLM and advanced software tools to analyse the morphology of subcellular structures and the spatial distribution of proteins. In my first contribution, we applied SMLM to observe the actin cytoskeleton of platelets at different morphological states. The three major actin reorganization states of platelets were imaged and the width of filopodia was determined. Improved imaging buffers containing glycerol and an oxygen-scavenger system allowed the localisation of single molecule signals with an accuracy of 12 nm. In the second contribution, we developed software tools for comparative analysis of protein distributions. We studied the distribution of CD41 (integrin α-IIb) and CD62P (P-selectin) in an artificial platelet clot model using 3D SMLM. Hierarchical clustering confirmed that CD41 and CD62P have different distributions at the nanoscale level. In the third contribution, I improved real-time 3D SMLM localization algorithm by combining the least-square approach with template images. Automated control of a reactivation laser based on real-time localizations allowed to counteract photo-bleaching and to ensure optimal molecule density throughout the experiment. In my final contribution, I developed a platform combing microfluidics, simultaneous two-colour 3D SMLM and advanced software tools. This platform was applied to a blood vessel model to study platelet activation on the endothelial monolayer under flow conditions. Our results prove a link between activated platelets and stressed endothelial cells determined via machine learning assisted mitochondrial morphology analysis. Furthermore, laser-treated cells showed no sign of platelet activation. Quantitative analysis of platelet volumes showed that dynamic incubated platelets on an endothelial monolayer have increased volumes compared to static incubation.

In summary, the work presented in this thesis demonstrated that advanced software tools in combination with SMLM enable a more detailed analysis of subcellular processes, cellular structures, and protein distribution at a nanometre level. Automation, real-time analysis, and improved imaging buffers increased the image quality necessary for precise post-processing. Simultaneous two-colour SMLM has proven its utility in linking nanoscale protein distributions with macroscopic subcellular structures in 3D.