Single-Particles Electron Cryo-Microscopy

Single-particle techniques can be separated into techniques requiring tilts of the specimen holder in the EM and "zero-tilt" techniques. The classical tomography approach requires many pictures of the same particles tilted into different orientations. The technique, which traces back to - among others - the group of Professor Walther Hoppe (Hoppe et al. 1974), has been rejuvenated by the introduction of automatic tomography where the computer controls the data collection in the cryo-EM. Another tilting approach requiring only two exposures of the sample is the Random Conical Tilt (RCT) approach pioneered by Michael Radermacher and co-workers. The (macroscopic) tilts required for these approaches intrinsically limit the achievable resolution level : the huge differences in defocus existing within the tilted image are very difficult to tackle computationally. The RCT approach includes explicit tilting of the specimen and may thus provide the absolute handedness of an object in cases where that information is not available.

Zero-tilt reconstructions exploit the random orientations that particles assume in the vitrified solvent to extract 3D information. Because no explicit macroscopic tilting of the sample holder is required, zero-tilt single-particle approaches can intrinsically reach very high resolution levels. The two main zero-tilt techniques are "projection matching" and "angular reconstitution".

Projection matching is based on finding the projection directions of the individual molecular images by correlating the images to "re-projections" of an earlier 3D reconstruction. It was shown (van Heel, 1984; Harauz & van Heel, 1986) that it is even feasible to first assign Euler angles using a random number generator and then to iteratively refine the Euler-angles assignments by projection matching. Various forms of projection matching have been applied to icosahedral structures and to asymmetric structures. More general and often useful is the angular reconstitution approach, which is based on finding common line projections between 2D projection images of the 3D structure. Noise reduction by multivariate statistical classification of the raw EM projections is an integral part of the approach. The structure analysis of single particles by cryo-EM involves a large number of techniques and algorithms that are all essential for achieving quasi-atomic levels of resolution.

Single-particle analysis

Single-particle techniques can be separated into techniques requiring tilts of the specimen holder in the EM and "zero-tilt" techniques. The classical tomography approach requires many pictures of the same particles tilted into different orientations. The technique, which traces back to - among others - the group of Professor Walther Hoppe (Hoppe et al. 1974), has been rejuvenated by the introduction of automatic tomography where the computer controls the data collection in the cryo-EM. Another tilting approach requiring only two exposures of the sample is the Random Conical Tilt (RCT) approach pioneered by Michael Radermacher and co-workers. The (macroscopic) tilts required for these approaches intrinsically limit the achievable resolution level : the huge differences in defocus existing within the tilted image are very difficult to tackle computationally. The RCT approach includes explicit tilting of the specimen and may thus provide the absolute handedness of an object in cases where that information is not available. Zero-tilt reconstructions exploit the random orientations that particles assume in the vitrified solvent to extract 3D information. Because no explicit macroscopic tilting of the sample holder is required, zero-tilt single-particle approaches can intrinsically reach very high resolution levels. The two main zero-tilt techniques are "projection matching" and "angular reconstitution". Projection matching is based on finding the projection directions of the individual molecular images by correlating the images to "re-projections" of an earlier 3D reconstruction. It was shown (van Heel, 1984; Harauz & van Heel, 1986) that it is even feasible to first assign Euler angles using a random number generator and then to iteratively refine the Euler-angles assignments by projection matching. Various forms of projection matching have been applied to icosahedral structures and to asymmetric structures. More general and often useful is the angular reconstitution approach, which is based on finding common line projections between 2D projection images of the 3D structure. Noise reduction by multivariate statistical classification of the raw EM projections is an integral part of the approach. The structure analysis of single particles by cryo-EM involves a large number of techniques and algorithms that are all essential for achieving quasi-atomic levels of resolution.


From: van Heel et al. (2000). Q. Rev. Biophys. 33

Last update: Tue, January 10, 2006	Back to my home page	Comments ?