Towards the automation of 3D correlative light and electron microscopy: streamlining substrate preparation and fiducial-free registration

Abstract Imaging molecular identity within intact subcellular architecture is essential for understanding structure–function relationships in cell biology and for diagnostic applications. Three-dimensional correlative light and electron microscopy (3D CLEM) uniquely combines the molecular specificity of fluorescence microscopy with the ultrastructural context provided by electron microscopy, but widespread application is hindered by low throughput and high technical complexity. Here, we present an integrated, practical workflow to expedite both cryogenic and room-temperature 3D CLEM: In the first part, we aim at (i) engineering and quantitatively evaluating CLEM substrates through surface treatments and micropatterning strategies to improve cell adhesion on standard EM substrates (EM grids) for cryogenic CLEM approaches. In the second part, we propose (ii) the implementation of a fiducial-free, automated 3D registration pipeline that exploits endogenous lipid droplets as intrinsic landmarks to align confocal fluorescence volumes with focused-ion-beam scanning EM (FIBSEM) datasets (acquired at room temperature) with minimal manual input. We compare common coatings for standard EM grids and plasma-cleaning regimes and demonstrate reproducible approaches to confine cells at known coordinates using a cost-effective, documented UV photopatterning prototype. For image correlation, segmentation of lipid droplets in both modalities yields robust 3D transforms without exogenous beads or training datasets with an accuracy of below 1 µm in 3D across entire cell volumes. The computational tools are provided as open-source image-processing scripts. Together, substrate engineering and lipid droplet-based registration increase the probability of obtaining useful 3D CLEM datasets per specimen, reduce manual intervention, and lower the barrier for adoption of volumetric CLEM workflows in standard laboratories. These methods advance the scalability and accessibility of quantitative 3D CLEM for studying heterogeneous biological samples.