4.1 Introduction
Studying and understanding cells in their entirety has long been a major goal of the
cell biology community. Despite advances in light microscopy, this method cannot
deliver the resolution necessary to describe some of the cell’s essential working
parts. Electron microscopes (EMs) solve the resolution problem, yet most cells are
too thick to image directly and must be sectioned into thinner slices to view in the
EM. In addition, most cells are so large that the field of view in a single image,
recorded at useful resolution, is not wide enough to capture some structures of
interest. Therefore, to visualize large areas and volumes of cells in three dimensions
(3D), it is necessary to expand the field of view and to increase the thickness of the
observed volume. This has led to the evolution of new imaging technologies termed
‘Volume-EM’; methods that use either a transmission electron microscope
(TEM) or scanning electron microscope (SEM) to reconstruct large areas and
volumes of cells in 3D [3, 4].
There are several approaches to performing large-scale 3D reconstructions by
EM. The first, serial section TEM (ssTEM) involves the use of serial, thin (50–
80 nm) sections to reconstruct larger volumes of cells. This method requires considerable technical skill to manually cut and collect the serial sections for imaging.
Although large areas can be imaged using montage strategies, the z resolution of
any 3D reconstruction is limited by the thickness of the slices. Nevertheless, ssTEM
has been used for many years to address questions in cell biology, such as the 3D
organization of mitotic spindles [5], and to study structural aspects of neurobiology
[6–8]. An analogous approach, array tomography, uses a SEM to image large
numbers of serial sections collected onto tape, silicon or glass discs [9]. Again,
array tomography is limited in z resolution but has the advantage of imaging very
large areas of material in any given section. This technique is now contributing to
the field of neuroscience where a significant volume of the mouse neocortex has
been reconstructed ([10, 11]; see Chap. 6).
SEM has also been used to image a face on a block of embedded material
(BF-SEM), employing either a microtome mounted in the chamber of an SEM
(SBF-SEM; Gatan 3View) or a focused beam of ions (FIB-SEM) to remove thin
slices of material from the block after each image is recorded (see Chap. 5).
SBF-SEM and FIB-SEM can efficiently gather large quantities of data from a single
specimen with minimal human intervention. SBF-SEM uses surface images at
intervals of 10–50 nm [12], implying a resolution in that direction of at least twice
the section thickness. FIB-SEM is closer to isotropic resolution with a sampling
interval said to be as little as 3 nm under the best conditions [13]. Unlike TEM
based techniques, the majority of the specimen is simply disposed of in both these
methods, but they are very useful for imaging large ultrastructural features, such as
whole plasma membranes and neuron axonal networks. On the other hand, features
whose size is on the same order as the slice interval are poorly reproduced or even
omitted completely if they do lie at an appropriate position relative to the Z slices.
Further details on SBF-SEM and FIB-SEM can be found in Chap. 5.
96
E. O’Toole et al.
Studying and understanding cells in their entirety has long been a major goal of the
cell biology community. Despite advances in light microscopy, this method cannot
deliver the resolution necessary to describe some of the cell’s essential working
parts. Electron microscopes (EMs) solve the resolution problem, yet most cells are
too thick to image directly and must be sectioned into thinner slices to view in the
EM. In addition, most cells are so large that the field of view in a single image,
recorded at useful resolution, is not wide enough to capture some structures of
interest. Therefore, to visualize large areas and volumes of cells in three dimensions
(3D), it is necessary to expand the field of view and to increase the thickness of the
observed volume. This has led to the evolution of new imaging technologies termed
‘Volume-EM’; methods that use either a transmission electron microscope
(TEM) or scanning electron microscope (SEM) to reconstruct large areas and
volumes of cells in 3D [3, 4].
There are several approaches to performing large-scale 3D reconstructions by
EM. The first, serial section TEM (ssTEM) involves the use of serial, thin (50–
80 nm) sections to reconstruct larger volumes of cells. This method requires considerable technical skill to manually cut and collect the serial sections for imaging.
Although large areas can be imaged using montage strategies, the z resolution of
any 3D reconstruction is limited by the thickness of the slices. Nevertheless, ssTEM
has been used for many years to address questions in cell biology, such as the 3D
organization of mitotic spindles [5], and to study structural aspects of neurobiology
[6–8]. An analogous approach, array tomography, uses a SEM to image large
numbers of serial sections collected onto tape, silicon or glass discs [9]. Again,
array tomography is limited in z resolution but has the advantage of imaging very
large areas of material in any given section. This technique is now contributing to
the field of neuroscience where a significant volume of the mouse neocortex has
been reconstructed ([10, 11]; see Chap. 6).
SEM has also been used to image a face on a block of embedded material
(BF-SEM), employing either a microtome mounted in the chamber of an SEM
(SBF-SEM; Gatan 3View) or a focused beam of ions (FIB-SEM) to remove thin
slices of material from the block after each image is recorded (see Chap. 5).
SBF-SEM and FIB-SEM can efficiently gather large quantities of data from a single
specimen with minimal human intervention. SBF-SEM uses surface images at
intervals of 10–50 nm [12], implying a resolution in that direction of at least twice
the section thickness. FIB-SEM is closer to isotropic resolution with a sampling
interval said to be as little as 3 nm under the best conditions [13]. Unlike TEM
based techniques, the majority of the specimen is simply disposed of in both these
methods, but they are very useful for imaging large ultrastructural features, such as
whole plasma membranes and neuron axonal networks. On the other hand, features
whose size is on the same order as the slice interval are poorly reproduced or even
omitted completely if they do lie at an appropriate position relative to the Z slices.
Further details on SBF-SEM and FIB-SEM can be found in Chap. 5.
96
E. O’Toole et al.