Figure 3. Most probable merge between the atomic resolution G-actin structure [Kabsch et al 1990] and the EM-based 3-D reconstruction such as shown in Figure 1 (lower right inset). The monomers are packed with their F-actin filament axis running parallel with the tube surface and roughly along the tube axis. Only one of the two possible antiparallel dimers is shown. Subdomains 1 and 2 are at the outer surface, whereas subdomains 3 and 4 reside at the inner surface of the crystalline actin tubes. Subdomain 2 is marked in blue. The hydrophobic loop of subdomain 4 is colored red. Yellow segments of the ribbon structure mark areas of the actin molecule which protrude from the EM-based envelope. Subdomain 2 was the only part of the actin molecule which could not be clearly localized within the EM-based 3-D reconstruction. The intersubunit contacts along the 56 Å lattice direction (i.e. those in the direction close to the tube axis) resemble those along the two long-pitch helical strands of the atomic model of the F-actin filament [Holmes et al 1990 and Lorenz et al 1993]. Lateral contacts are between subdomains 1 and 2 and subdomains 3 and 4. The hydrophobic loop might play an important role in the intermolecular contact between adjacent dimers along the 65 Å lattice direction. All Figures are presented as stereo pairs. Methods: Merging the EM with the X-ray data was performed using the program O [Jones et al 1991].

Figure 4. Comparison of surface relief data (Figure 2) with the atomic model (Figure 3) of the crystalline actin tubes. The modeled crystalline actin tube reveals the distinct surface relief data of the inner and outer tube surface with remarkable accuracy. On the outer surface the S-shaped ridge is formed by two consecutive subdomains 1, whereas the alternating smaller protrusions represent subdomains 2. The location and relative arrangement (i.e. represented by bars) of distinct surface peaks is faithfully reproduced. The same is true for the inner tube surface. The combined approach reflects the sensitivity of the shadowing procedure which is able to detect grooves or depressions which are less than 5 Å deep. A and B mark the dimer configuration as shown in Figure 3. Scale bars represent 10 Å. Methods: The crystalline packing of the actin monomers was modeled using O [Jones et al 1991] according to the EM-based 3-D reconstruction (Figure 3). These data were exported to RASMOL [Sayle and Milner-White 1995] and visualized as a space-filling van der Waals volume representation.

Figure 2. TEM micrograph of a freeze-dried, high resolution rotary metal-shadowed (i.e. with Ta/W) ruptured crystalline actin tube yielding distinct surface reliefs corresponding to the inner and outer crystalline tube surface. The outer surface data were obtained from the area in the lower half of the electron micrograph. An averaged surface picture is presented in the lower left inset. White areas correspond to protein elevations or surface protrusions, whereas dark areas represent grooves or depressions residing in-between elevations. According to the radial correlation function [Saxton and Baumeister 1982] and the spectral signal-to-noise ratio (SSNR) criterion [Unser et al 1987] a resolution of 15 Å was achieved. The outer surface reveals an S-shaped ridge running approximately perpendicular to the tube axis and with alternating small protrusions at the two ends. The inner tube surface was obtained from the area in the upper half of the electron micrograph, and its correlation-averaged image (upper right inset) resolves two elongated lobes per dimer. Atomic interpretation (see Figure 3 and Figure 4) of the crystalline actin tube data enabled us to estimate the sensitivity of this surface metal shadowing approach. The smallest grooves detected here reside between the two closer packed lobes on the inner tube surface. According to the space filling model of Figure 4 these grooves are only about 4 Å deep (see Figure 4 for a more detailed interpretation). Scale bars represent 500 Å (overview micrograph), and 50 Å (insets). Methods: Samples were freeze-dried at −80°C under high-vacuum conditions (5 × 10⁻⁸ mbar) for about two hours in a custom-built freeze drying/metal shadowing unit which is directly hooked up to a Philips CM12 TEM (Midilab; [Gross et al 1990]). This system allows us to transfer the specimen from the freeze drying/metal shadowing unit to the microscope under high-vacuum and cryo-conditions. Rotary metal shadowing was performed with Ta/W (at an elevation angle of 45°) depositing an 8 Å thick metal film onto the specimen surface. Samples were recorded as described in Figure 1. 2-D image averaging was performed with the software package MILAN [Fuchs et al 1995].