Induced membrane domains as visualized by electron tomography and template matching.

Membranes play a crucial role in many cellular processes, and it is therefore not surprising that many electron tomographic studies in life sciences concern membranous structures. While these tomographic studies provide many new insights into membrane connections and continuities in three dimensions, they are mostly limited to a macro-morphological level. In this paper, we demonstrate that by combining electron tomography and three-dimensional template matching we are able to investigate membrane morphology at a new level: membrane domains in three dimensions. To test this, temperature induced lipid phase separation in the biological model system of the Escherichia coli bacteria was used. We compared the inner (containing phospholipids) and outer (containing lipopolysaccharides) leaflet of the E. coli outer membrane at both 37 and -20 degrees C, and could visualize how these leaflets react differently to temperature shifts. These findings can be explained by the physico-chemical nature of the building blocks and are in line with earlier published data. This study shows that the combination of electron tomography and template matching is robust enough to visualize membrane domains that are beyond the perception of manual annotation.

Template matching as a tool for annotation of tomograms of stained biological structures.

In recent years, electron tomography has improved our three-dimensional (3D) insight in the structural architecture of cells and organelles. For studies that involve the 3D imaging of stained sections, manual annotation of tomographic data has been an important method to help understand the overall 3D morphology of cellular compartments. Here, we postulate that template matching can provide a tool for more objective annotation and contouring of cellular structures. Also, this technique can extract information hitherto unharvested in tomographic studies. To evaluate the performance of template matching on tomograms of stained sections, we generated several templates representing a piece of microtubule or patches of membranes of different staining-thicknesses. These templates were matched to tomograms of stained electron microscopy sections. Both microtubules and ER-Golgi membranes could be detected using this method. By matching cuboids of different thicknesses, we were able to distinguish between coated and non-coated endosomal membrane-domains. Finally, heterogeneity in staining-thickness of endosomes could be observed. Template matching can be a useful addition to existing annotation-methods, and provide additional insights in cellular architecture.