Electron tomography is an imaging technique which can be applied to a wide range of biological specimens. It does not rely on internal symmetries of the object nor does it require periodic objects. However, since tomographic reconstructions yield a density map from only one single object, image-averaging techniques, which are used implicitely (crystals) or explicitly (single-particle analysis) for noise reduction, cannot be applied. A limiting factor is the radiation sensitivity of biological samples, which is addressed by the recent development of automated low-dose data-acquisition schemes. Together with cryo-sample preparation techniques, electron tomography has the potential to allow imaging of macromolecular complexes and even cells, without exposing them to harming chemical procedures like fixation and dehydration. The next step was to improve the resolution of tomographic reconstructions to bring it into the realm of molecular resolution, enabling the recognition of larger individual macromolecules by their size and shape. To increase the resolution of reconstructed cells, an electron microscope with an accelerating voltage of 300kV, with an energy filter and a 2048x2048 Pixel CCD camera was installed and equipped with a tomography-software package. Several tilt- series of archaea of the species Pyrodictium abyssi were acquired and reconstructed. The analysis of protein crystals, found in the tomograms, revealed ar esolution of down to 3.5nm. A new cryo-holder allowing to rotate the specimen in-place by 100° was designed and manufactured in order to reduce reconstruction-artifacts resulting from the limited angular range of single-axis tilting. With this new device it is possible to acquire double-tilt series with an intermediate rotation of the specimen around the vertical axis. The missing information, forming a missing-wedge in Fourier space is thus reduced to a missing-pyramid. As a consequence, the quality of reconstructed cells was dramatically improved. In an application, a network of archaea could be visualized in 3D, which allowed the tracing of a connecting Cannula inside the quasi-periplasmic space of a cell. The low contrast and the low signal-to-noise ratio, due to the limited dose, both adversely affecting the cell reconstructions, led to the development of pattern recognition methods for molecular structures. Based on correlation techniques, two algorithms were implemented that are able to find and identify macromolecules inside cells. To reduce the computation time the programs were parallelized and can be used on any massive parallel computer system. The application of these algorithms to the tomograms of Thermoplasma acidophilum cells resulted in a first identification of macromolecular complexes.
«
Electron tomography is an imaging technique which can be applied to a wide range of biological specimens. It does not rely on internal symmetries of the object nor does it require periodic objects. However, since tomographic reconstructions yield a density map from only one single object, image-averaging techniques, which are used implicitely (crystals) or explicitly (single-particle analysis) for noise reduction, cannot be applied. A limiting factor is the radiation sensitivity of biological sa...
»