PMC:2726924 / 7955-10903 JSONTXT

{"project":"2_test","denotations":[{"id":"19345691-16473369-62520774","span":{"begin":312,"end":314},"obj":"16473369"},{"id":"19345691-16515450-62520775","span":{"begin":560,"end":562},"obj":"16515450"},{"id":"19345691-19038266-62520775","span":{"begin":560,"end":562},"obj":"19038266"},{"id":"19345691-19264960-62520775","span":{"begin":560,"end":562},"obj":"19264960"}],"text":"Three-dimensional reconstructions of amyloid fibrils\nThe intriguing substructure revealed by the images prompted us to undertake a 3D analysis of the fibril architecture. Of particular interest was the appearance of a subunit repeat, reminiscent of an earlier negative stain EM analysis of prion protein fibrils.36 However, this feature contrasts with current models for amyloid based on fibre diffraction, electron paramagnetic resonance, solid-state NMR, 3D crystals of small peptides, and cryo-EM images of fibrils formed from Aβ and other peptides.6,14,37,38 Using estimates of the long-range helical repeat and the subunit repeat determined from the images, 3D maps were calculated from the class average views. Reprojections of the maps compare well with the corresponding input class averages, showing the consistency of the maps with the input data (Fig. 1b and c). From a data set of 2000 repeats, 17 classes of the A-type structures and 14 classes of the B-type were selected, each with clear structural features, yielding a total of 31 individual 3D fibril maps (Fig. 2d and e). Side views, cross-over repeats, and resolution figures are tabulated for a selection of maps in Fig. 3. The resolution was estimated by comparing pairs of similar maps, using the 0.5 Fourier shell correlation criterion (Supplementary Fig. 2). Despite their structural variations, fibrils in the different classes have similar cross sections and bi-lobed globular repeats along the axial columns of density. The twist variation is most pronounced for the A-type classes where, for 85% of the data, the cross-over repeat distance fell into the range 128–152 nm (Fig. 2b and c). Outliers had cross-over repeats up to 185 nm, but their numbers were insufficient for classification and 3D reconstruction. For the B-type classes, the repeat is more constrained at 122–135 nm. A representative 3D map of each type is shown in more detail in Fig. 4, for which the resolutions are estimated as 29 Å (A-type) and 25 Å (B-type). The maps reveal a remarkably elaborate structure composed of two crescent-shaped units. A-type and B-type half-fibrils show a clear structural similarity when overlaid (Fig. 4e), consistent with the notion that the crescent-shaped half-fibrils are independent structural elements that can join back to back in either a parallel fashion (A-type; Fig. 4a and c) or an anti-parallel fashion (B-type; Fig. 4b and d). Each crescent contains three globular regions of density (about 3 nm × 4 nm in cross section), which represent constituent protofilaments (Fig. 4c and d). Most striking, however, is the complexity of the fibril architecture, with different interfaces between protofilaments both within and between the crescents. As shown in Fig. 2d and e, the A-type and B-type reconstructions also revealed considerable flexibility between the protofilaments and between the two crescents, adding to the heterogeneity of the samples obtained."}