One of the most promising and exciting developments in electron tomography is its application to ice-embedded cells and cellular organelles. The primary goal of cellular electron tomography is not to obtain a high-resolution structure of a particular protein complex, but to identify a macromolecule by virtue of its structural signature and to locate it in the context of the native cellular environment . We present a computationally feasible approach, which demonstrates that an objective identification of large macromolecular complexes in cells is possible, given that a highresolution reconstruction and the structure of the molecular assembly under scrutiny are available. Our aim is to quantify the identification results using a realistic test specimen, open for the modification of experimental parameters like thickness, protein density or protein composition. in this study, the reliability of the algorithm applied to electron tomograms of ‘phantom cells’ - mimicking a real cell - is shown.

We prepared phospholipid vesicles that were filled with different macromolecular complexes. First, a phospholipid solution (SOPC) was sonicated and mixed with His-tagged macromolecular complexes, such as thermosomes, 20S proteasomes or a mixture of both. The lipid-protein mixture was subjected to several freeze-thaw cycles and extruded to produce uni-laminar ‘phantom cells’ of various sizes.