the characteristics of the specimens that determine the imaging conditions and
image-processing strategies. For example, beam-sensitive, frozen-hydrated specimens require a dedicated, low-dose acquisition scheme, and the resultant noisy
reconstructions benefit from denoising to aid in surface rendering (‘segmentation’,
see later). Stained plastic sections of membrane-rich cells, on the other hand, are
more amenable to automated segmentation routines; these are suitable for characterising ultrastructure rather than studying molecular arrangements and interactions.
1.3.1 Scenario 1: Plastic sections
The classical TEM protocol developed for biological samples comprises aldehyde
fixation, dehydration in a graded alcohol series, infiltration with resin, heat-induced
polymerisation and sectioning using an ultramicrotome. This procedure was the
mainstay of biological electron microscopy in the second half of the twentieth
century [58], and with mostly subtle variations, it continues to be the most widely
used strategy for studies of cell ultrastructure. Refined versions of this protocol
bypass chemical fixation in favour of rapid freezing and progressive solvent substitution at low temperature, culminating in a UV-polymerised resin block that can
also be sectioned by ultramicrotomy. This improvement avoids the artefacts of
chemical fixation, although it is difficult to study alterations that occur after
freezing. As for any vitrification process, with or without added cryoprotectant, it
depends critically on specimen thickness. Nevertheless, the freeze-substitution
process has been studied systematically from a process perspective, and it was
shown that the processing time can be reduced from days to as little as 3 h [59].
Structural preservation after freeze-substitution is clearly superior to structure
visualised after the use of chemical fixatives. In either case, sections need to be cut
as thin as possible without eliminating the feature of interest. In practical terms, this
is dictated by the accelerating voltage of the microscope. Sections of nominal
thickness 80–90 nm are suitable for electron tomography using a 120 kV TEM,
whilst thicknesses up to 300 nm represent the maximum thickness for use in
300 kV instruments whilst still avoiding the possibility of multiple scattering events
[52] that degrade resolution.
Although freeze substitution is superior to classical fixation and embedding, it is
highly informative to revisit the use of powerful (1 MV) transmission electron
microscopes as projection devices but without the benefits of adequate structural
preservation. This led to the flawed ‘microtrabecular lattice’ concept of cell
structure [60]. Here, removal of solvent and/or fixation with glutaraldehyde caused
the soluble components of the cytosol to be deposited on elements of the
cytoskeleton, portraying mammalian cells erroneously [61] as an irregular network
of cross-linked ‘microtrabeculae’ (Fig. 1.4). Porter interpreted these observations as
the fundamental organisation of the cytoplasm, the physical basis for cytoplasmic
motility, and the determinants of cell shape [61]. In the absence of tomography, he
made use of stereo pairs to demonstrate his results.
1 Electron Tomography: A Primer
7
image-processing strategies. For example, beam-sensitive, frozen-hydrated specimens require a dedicated, low-dose acquisition scheme, and the resultant noisy
reconstructions benefit from denoising to aid in surface rendering (‘segmentation’,
see later). Stained plastic sections of membrane-rich cells, on the other hand, are
more amenable to automated segmentation routines; these are suitable for characterising ultrastructure rather than studying molecular arrangements and interactions.
1.3.1 Scenario 1: Plastic sections
The classical TEM protocol developed for biological samples comprises aldehyde
fixation, dehydration in a graded alcohol series, infiltration with resin, heat-induced
polymerisation and sectioning using an ultramicrotome. This procedure was the
mainstay of biological electron microscopy in the second half of the twentieth
century [58], and with mostly subtle variations, it continues to be the most widely
used strategy for studies of cell ultrastructure. Refined versions of this protocol
bypass chemical fixation in favour of rapid freezing and progressive solvent substitution at low temperature, culminating in a UV-polymerised resin block that can
also be sectioned by ultramicrotomy. This improvement avoids the artefacts of
chemical fixation, although it is difficult to study alterations that occur after
freezing. As for any vitrification process, with or without added cryoprotectant, it
depends critically on specimen thickness. Nevertheless, the freeze-substitution
process has been studied systematically from a process perspective, and it was
shown that the processing time can be reduced from days to as little as 3 h [59].
Structural preservation after freeze-substitution is clearly superior to structure
visualised after the use of chemical fixatives. In either case, sections need to be cut
as thin as possible without eliminating the feature of interest. In practical terms, this
is dictated by the accelerating voltage of the microscope. Sections of nominal
thickness 80–90 nm are suitable for electron tomography using a 120 kV TEM,
whilst thicknesses up to 300 nm represent the maximum thickness for use in
300 kV instruments whilst still avoiding the possibility of multiple scattering events
[52] that degrade resolution.
Although freeze substitution is superior to classical fixation and embedding, it is
highly informative to revisit the use of powerful (1 MV) transmission electron
microscopes as projection devices but without the benefits of adequate structural
preservation. This led to the flawed ‘microtrabecular lattice’ concept of cell
structure [60]. Here, removal of solvent and/or fixation with glutaraldehyde caused
the soluble components of the cytosol to be deposited on elements of the
cytoskeleton, portraying mammalian cells erroneously [61] as an irregular network
of cross-linked ‘microtrabeculae’ (Fig. 1.4). Porter interpreted these observations as
the fundamental organisation of the cytoplasm, the physical basis for cytoplasmic
motility, and the determinants of cell shape [61]. In the absence of tomography, he
made use of stereo pairs to demonstrate his results.
1 Electron Tomography: A Primer
7