Electron tomography is an electron microscopic technique for obtaining a 3-D image from any electron microscopy specimen, whether ordered or not. Electron tomography can be applied to almost any kind of specimen, provided that it is not too thick. There is a thickness, radiation damage and signal to noise dependency. However, it is generally agreed that ~ 4 nm resolution can be obtained from most specimens that are not too thick. There are a number of reviews that can get the reader up to speed on the latest developments (Baumeister et al., 1999; Baumeister et al., 2002; Grunewald et al., 2003; McEwen and Marko, 2001; O'Toole et al., 1999; O'Toole et al., 2002; Steven and Aebi, 2003). A monograph on electron tomography is also available (Frank, 1992).
Data collection for electron tomography involves collecting images while tilting the specimen around a single axis. Usually data can be collected between 70o if the specimen is not too thick. Complete data collection would require tilting through 90o but because at really high tilt angles, the specimen becomes quite thick this is not possible; at 60o tilt the specimen is 2x as thick as at 0o and at 70o it is nearly 3x as thick.
There are effectively two ways to collect the images: fixed tilt increments or graduated tilt increments where the tilt increment is proportional to the cosine of the tilt angle (Saxton et al., 1984). This method is often referred to as cosine rule or Saxton rule tilting. The advantage of the fixed tilt increment lies in the simplicity of the weighting that is applied to the reconstruction. However, general schemes for weighting are available (Radermacher, 1992) that work quite well. The disadvantage of fixed increment tilting is that it may either underweight the high tilt data, which give the depth information, or if allowance for the increasing tilt angle is made, the scheme will over weight the low tilt increments. Only the cosine rule gives correct balance to the high tilt and the low tilt data. In our tomographic work, we use exclusively cosine rule tilting. For those interested, Radermacher's chapter in Frank's book is an excellent description of the generalized weighting function used for computing the 3-D map.
A key feature of electron tomography is the alignment of the tilt series. There are two ways to do this. One of these is called fiducial, or marker, alignment and the other is marker free alignment. The vast amount of tomographic data alignment is done using marker alignment (Penczek et al., 1995; Ress et al., 1999). Excellent software for this is available that is incorporated into a popular segmentation tool (Kremer et al., 1996). Another method of image alignment is the use of crosscorrelation functions. In our tomography, we make exclusive use of crosscorrelation functions for image alignment (Brandt et al., 2001; Liu et al., 1995; Owen and Landis, 1996; Taylor et al., 1997), otherwise known as marker free alignment. An additional problem in electron tomography is the classification of large heterogeneous data sets. We are investigating this problem using an ideal system for such work, the flight muscle of the large waterbug Lethocerus sp. We have been producing electron tomograms from this system for a number of years (Liu et al., 2004; Schmitz et al., 1997; Schmitz et al., 1996; Taylor et al., 1999). A characteristic of this specimen is that it has a regular arrangement of actin and myosin filaments, but a disordered arrangement of myosin crossbridges. We have developed methods for alignment and classification (Pascual-Montano et al., 2002; Winkler and Taylor, 1999), refinement of atomic models into the reconstructions (Chen et al., 2001; Chen et al., 2002) and for correction of specimen distortions in 3-D (Winkler and Taylor, 1996).
Our labs efforts are focused at obtaining molecular resolution of complex assemblies and therefore we are pushing the resolution and signal to noise envelope. This also means that our efforts emphasize the use of thin specimens because only with such specimens can the resolution be pushed. Problems that are being addressed with the goal of extending the resolution include improvements in our marker-free alignment, improved alignment and averaging of structural motifs within tomograms, correction of image distortions that will affect image alignment and correction of the focus gradient in tilted images.
Click on link in article title for more information.
Winkler, H. and K.A. Taylor, Focus gradient correction applied to tilt series image data used in electron tomography. J. Struct. Biol.143(1), 24-32 (2003). (Pubmed Link)Winkler, Hanspeter, Taylor, Kenneth A. Accurate marker-free alignment with simultaneous geometry determination and reconstruction of tilt series in electron tomography. Ultramicroscopy106, 240-254 (2006).
Winkler, Hanspeter. 3D reconstruction and processing of volumetric data in cryo-electron tomography. J. Struct. Biol 157(1), 126-167 (2007).
Tomography Software Site.