Plant Cytokinesis – Insights Gained from Electron Tomography Studies 253
mium tetroxide, are multifunctional reagents that preserve cell ultrastructure
by crosslinking subsets of cellular molecules, thereby preserving their spatial
organization. However, because chemical fixation takes seconds to minutes
to immobilize cellular processes, and because different cellular components
are fixed at different rates, chemical fixatives are not ideal tools for preserv-
ing cell ultrastucture (reviewed in Gilkey and Staehelin (1986)). For structural
studies of cytokinesis, the most limiting aspect of chemical fixatives is their
inability to preserve short-lived structural intermediates of cell plate assem-
bly and phragmoplast MT dynamics in a reliable manner. Furthermore, due
to the selective crosslinking activity of chemical fixatives, which leaves many
types of molecules in an unfixed state, many cellular structures can continue
to undergo structural changes long after the sample has been “fixed”.
Most of these limitations of chemical fixatives can be overcome by em-
ploying ultra-rapid freezing methods such as high pressure freezing (HPF)
to stabilize the cells, due to their ability to immobilize all cellular molecules
within milliseconds (Gilkey and Staehelin 1986). Samples preserved in this
manner can then be freeze-substituted (FS) at –80 to 90 ◦C, which increases
the probability of viewing even the most labile cellular structures.
The second limitation of classical electron microscopy, even when cry-
ofixed and freeze-substituted samples are examined, is that conclusions have
to be drawn from essentially 2-dimensional (2-D) samples (most frequently,
thin sections). Since the thin sections are typically 60 – 80 nm thick, this
means that whereas the resolution in thexandyaxes of the sections is
∼ 4 nm, the resolution in thez-axisisonly 100 – 150 nm, which greatly limits
the ability of researchers to determine the complete 3-D architecture of most
cellular organelles and cytoskeletal systems at high resolution. The solution
to this latter problem is the use of dual-axis electron tomography (McIntosh
et al. 2005).
Electron tomography is analogous to the various tomographic techniques
used in modern medicine, which rely on X-rays, magnetic resonance or ul-
trasound. By taking multiple images of a specimen (typically a 150 – 300 nm
thick plastic section) as it is systematically tilted from + 60 ◦to – 60 ◦along
two orthogonal axes, it is possible to record information from thez-axis of
a sample, and by computing the back-projection of these images, it is then
possible to produce a tomogram, which is a 3-D block of data that is repre-
sented as an array of volume elements (voxels). This 3-D data block, in turn,
can be used to display 2-D slice images of differentz-axis levels of the sam-
ple. These images resemble thin section images (compare Fig. 1A and B), but
because they are only∼ 2 nm thick, versus 60 – 80 nm for thin sections, many
more details of cellular architecture can be resolved. Furthermore, by tracing
the observed structures in serial slices in a process called hand-segmentation
or modeling, it is possible to produce reconstructions with a 3-D resolution of
5to 8 nm (Fig. 1C). The overall size of the reconstructions can be further in-
creased by producing tomograms and models of serial sections (Fig. 1D), and