Identification of macromolecular complexes in cryoelectron tomograms of phantom cells

Abstract

Electron tomograms of intact frozen-hydrated cells are essentially three-dimensional images of the entire proteome of the cell, and they depict the whole network of macromolecular interactions. However, this information is not easily accessible because of the poor signal-to-noise ratio of the tomograms and the crowded nature of the cytoplasm. Here, we describe a template matching algorithm that is capable of detecting and identifying macromolecules in tomographic volumes in a fully automated manner. The algorithm is based on nonlinear cross correlation and incorporates elements of multivariate statistical analysis. Phantom cells, i.e., lipid vesicles filled with macromolecules, provide a realistic experimental scenario for an assessment of the fidelity of this approach. At the current resolution of Ϸ4 nm, macromolecules in the size range of 0.5-1 MDa can be identified with good fidelity.

Classification and 3D averaging with missing wedge correction in biological electron tomography

“…Previous studies have acknowledged this problem and various solutions have been proposed but a rigorous analysis has not yet been reported. Alignment of volumes affected by the missing wedge to an external reference that has no missing information (i.e., it is not altered by the presence of a missing wedge) is addressed in Frangakis et al (2002) by constraining the calculation of the crosscorrelation score to regions of reciprocal space where measurements are available. The more general problem of aligning two volumes affected by the missing wedge was considered in Schmid et al (2006) where the cross-correlation score was scaled with the reciprocal of the number of non-zero terms in the complex multiplication, i.e., the size of the region in Fourier space where information from both volumes is available.…”

“…The more general problem of aligning two volumes affected by the missing wedge was considered in Schmid et al (2006) where the cross-correlation score was scaled with the reciprocal of the number of non-zero terms in the complex multiplication, i.e., the size of the region in Fourier space where information from both volumes is available. More recently, Förster et al (2008) have also studied this case but only in the context of classification (not for alignment), extending the constrained cross-correlation initially proposed in Frangakis et al (2002) to account for the missing wedge of both volumes. The approaches taken in Frangakis et al (2002), Förster et al (2005), Schmid et al (2006) and Förster et al (2008), all rely almost exclusively on exhaustive search of the rotational space to determine the best matching orientation between the two volumes, precluding the analysis of very large data sets.…”

“…More recently, Förster et al (2008) have also studied this case but only in the context of classification (not for alignment), extending the constrained cross-correlation initially proposed in Frangakis et al (2002) to account for the missing wedge of both volumes. The approaches taken in Frangakis et al (2002), Förster et al (2005), Schmid et al (2006) and Förster et al (2008), all rely almost exclusively on exhaustive search of the rotational space to determine the best matching orientation between the two volumes, precluding the analysis of very large data sets. Although in some cases the search may be restricted by exploiting the geometry of the problem, e.g., by only searching for in-plane rotations (Winkler and Taylor, 1999;Förster et al, 2005), this is still a very computationally demanding procedure.…”

“…Although this does not hold if the signals are not aligned, one would still expect true corresponding matches to yield better scores than those accounted for by scaling differences alone. We have then chosen to apply this type of normalization in our approach, which is equivalent to that implied in the use of constrained cross-correlation presented in Frangakis et al (2002) and Förster et al (2005Förster et al ( ,2008.…”

Single particle cryoelectron tomography characterization of the structure and structural variability of poliovirus–receptor–membrane complex at 30 Å resolution

“…The ultimate goal is to be able to interpret tomographic reconstructions of crowded cell interiors in terms of their smaller components by matching a library of known higher-resolution structures to features in the low-resolution landscape. Several recent papers have demonstrated substantial progress in this direction, showing that is possible to map individual particles in cells (Bohm et al, 2000;Frangakis et al, 2002;Garvalov et al, 2006;Medalia et al, 2002;Murphy et al, 2006;Ortiz et al, 2006), and in favorable cases to derive reconstructions of macromolecular complexes such as the ribosome, to a few nanometers resolution (Ortiz et al, 2006).…”