3D scanning transmission electron microscopy approaches for parasitology
Images produced in scanning transmission electron microscopy and in conventional TEM contain structural information of the inside of the object of interest because they are projection images, as opposed to surface images. Projection images can be combined to reconstruct biological objects by computational approaches thus providing access to the 3D structure. When projection images are acquired by tilting the sample around a single axis the computational approach is called tomography26.
STET principle & applications
Historically, STEM was mostly used in the field of material sciences, even though there were some early uses of the technique to measure the amino acids content of proteins, thanks to the atomic mass sensitivity of the method27,28. In biology, STEM became increasingly popular when the community realised that it could be used to study thick sections of cells embedded in resin4,17,29,30. STET of biological samples was then developed to explore cellular and sub-cellular structures of samples embedded in thick resin sections19,31.
In STET, the sample is tilted at different angles inside the electron microscope and, for each tilt angle an image is collected. In the microscope, the sample cannot be tilted up to ±90° because its apparent thickness would be virtually infinite (at this angle the electrons would have to traverse the whole diameter of the grid). Therefore, the angular collection is incomplete. In the Fourier space, this creates a wedge (called the missing-wedge) which does not contain any information32. A typical tilt range is between -60° to +60°, using 2° or 3° increments, but some studies collect images with tilts up to ±70°. Collecting images at higher tilt angles can be useful to fill in and reduce the missing-wedge. During the experiment, the sample must be tracked so that the same field of view is imaged during the whole data collection and focusing must be performed to avoid collecting out-of-focus images. It is particularly important in STEM to collect images in focus otherwise they suffer a strong blurriness32. This is particularly true if high convergence semi-angles are used (depth of focus limited to a few tens of nm) but can be mitigated by using low convergence semi-angles (depth of focus up to a few microns). After data collection, the 2D projections must be aligned to a common origin, which can be done by using gold beads as fiducial markers or fiducials-free reference mathematical methods. Then, dedicated algorithms are used to reconstruct the 3D volumes using the aligned 2D projections. The details of these steps are out of the scope of this review but can be found in the literature26.
Resin sections up to 1 µm-thick can be studied using STET. Because the field of view is usually greater than 1 µm², the total volume available in a single 3D reconstruction is greater than 1 µm3, which corresponds to about the volume of small prokaryote, or using parasitology standards, the whole flagellar pocket of a T. bruceicell19. In comparison, the sample volume usually captured in conventional TEM ranges between 0.1 and 0.3 µm3 (for 100 and 300 nm-thick resin sections, respectively, and provided that the field-of-view is 1 µm²).
Early STET works in biology were performed using the ADF detector as it was often employed in material science studies4,17. In tomography, the sample is tilted and its apparent thickness increases inversely proportional to the cosine of the tilt angle. This means that at ±60°, the apparent thickness of a sample is twice its nominal thickness. Because of this, multiple inelastic scattering occurs, leading to the undesirable detection of light elements on the ADF or HAADF detector. This, in turn, results in the collection of noisy ADF/HAADF images. For the observation of thick samples, more meaningful images are obtained using the BF detector29,31.
In parasitology, STET has so far only been used on a limited number of organisms19,29. A video article describing the use of STET on T. brucei thick sections is also available33. In this review, we introduce another case, also involving T. brucei . In this example, STET has been performed on a 500 nm-thick resin section of T. brucei , showing structural details next to the mature- and pro-basal bodies, between the flagellar pocket and the kinetoplast of the cell (Figure 6). This is the same region as the one shown in Figure 3D. Each image corresponds to a virtual 10 nm-thick slice within the reconstruction and is separated from the next one by a 20 nm-thick gap. The entire pro-basal body is present in this 500 nm-thick section. The microtubule triplets are clearly visible, with microtubules A, B and C well-delineated. The flagellum is slightly tilted inside the reconstruction as we can see it appearing from the top of the image and disappearing from the bottom. It extends from the mature basal body, which is almost fully contained in the resin section. The penultimate image corresponds to the centre of the flagellum as the central pair is visible above the basal plate. The microtubule quartet is visible on the third and fourth images (Fig. 6C and D). The microtubules of the quartet are then directed towards the anterior part of the cell and are visible on the left side of the flagellar pocket membrane in the fifth and sixth images (Fig. 6E and F). On the bottom part of the images, the large kinetoplast is easily recognised thanks to its specific shape and texture. The disk of the kinetoplast DNA is particularly well defined. Other details could be described in this dataset, such as some thin strings connecting the kinetoplast and the mature basal body or the collarette around the flagellum membrane in the flagellar pocket (Fig. 6F). However, what we want to point out here is the large volume of structural information one can get in using STET on thick resin sections. Using a 200 kV STEM capable electron microscope, 750 nm-thick resin sections can routinely be studied.