1.11 Recommendations and Future Perspectives
Electron microscopes and tomography software have become very user-friendly,
and they often incorporate automated column alignments that assist the acquisition
process to run to completion. Alignment is now comparatively easy thanks to
optimised SNR collection, smoother tracking, and flexible tilt series alignment
software that automates identification and tracking of fiducial markers or
other features in 3D space, perhaps by combining both strategies (treating adsorbed
gold beads as ‘features’). These procedures are non-invasive—gold particles do not
need to be injected destructively into the specimen prior to fixation via a gene gun -
type approach.
Nevertheless, it is just as useful to identify current bottlenecks and impediments
to progress. Most tomography reviews over the past 10–15 years have highlighted
the need for optimised detectors, phase plates, and artefact-free specimen thinning.
Efforts in these areas have led to significant improvements. The ability to make use
of every (elastically scattered) electron incident on the detector surface means that
information yield can be truly optimised. Of course, nothing can be done about the
inelastically scattered electrons, apart from removing their collective contribution to
blurring via an zero-loss energy filter [76]. Depending on the question, STEM
tomography (Chap. 2) provides further flexibility for thicker specimens. The use of
phase plates allows projections to be recorded close to focus, circumventing the
need for CTF correction. Several studies demonstrate the success of this technology
[110, 114–117]. Previously, a defined level of underfocus was required to provide
optimal contrast for features of interest with a given size but at the expense of
high-frequency information [118]. Volta phase plates do not suffer from fringing
artefacts inherent in Zernike-type plates. They are also highly practical, allowing
routine use [99, 119].
These breakthroughs in combination with true advances in sample preparation
mean that single-particle cryo-EM and electron tomography have entered a new era.
Vitrification is now common to both plastic and cryo workflows. Therefore, apart
from the necessity to render infectious agents inactive, the artefacts of chemical
fixation should no longer play a role. It is tempting to conclude that the major
challenges (images acquired in focus, suitable specimen thinning procedures, close
to 100% information yield from calibrated electron dose) have been addressed
satisfactorily, and although there is always room for improvement, perhaps the most
important point to note is that these solutions are costly in terms of instrumentation
and highly-trained, dedicated personnel.
Finally, the success of standard sample collection geometries for numerous
high-impact publications belies the fact that cylindrical samples have numerous
advantages, including constant focus and thickness during tilting. For the last part
of this chapter, I will explain why this is problematic and how this setup could help
electron tomography data collection to be even more efficient, and to realise its true
potential in terms of information yield.
22
A. Leis
Electron microscopes and tomography software have become very user-friendly,
and they often incorporate automated column alignments that assist the acquisition
process to run to completion. Alignment is now comparatively easy thanks to
optimised SNR collection, smoother tracking, and flexible tilt series alignment
software that automates identification and tracking of fiducial markers or
other features in 3D space, perhaps by combining both strategies (treating adsorbed
gold beads as ‘features’). These procedures are non-invasive—gold particles do not
need to be injected destructively into the specimen prior to fixation via a gene gun -
type approach.
Nevertheless, it is just as useful to identify current bottlenecks and impediments
to progress. Most tomography reviews over the past 10–15 years have highlighted
the need for optimised detectors, phase plates, and artefact-free specimen thinning.
Efforts in these areas have led to significant improvements. The ability to make use
of every (elastically scattered) electron incident on the detector surface means that
information yield can be truly optimised. Of course, nothing can be done about the
inelastically scattered electrons, apart from removing their collective contribution to
blurring via an zero-loss energy filter [76]. Depending on the question, STEM
tomography (Chap. 2) provides further flexibility for thicker specimens. The use of
phase plates allows projections to be recorded close to focus, circumventing the
need for CTF correction. Several studies demonstrate the success of this technology
[110, 114–117]. Previously, a defined level of underfocus was required to provide
optimal contrast for features of interest with a given size but at the expense of
high-frequency information [118]. Volta phase plates do not suffer from fringing
artefacts inherent in Zernike-type plates. They are also highly practical, allowing
routine use [99, 119].
These breakthroughs in combination with true advances in sample preparation
mean that single-particle cryo-EM and electron tomography have entered a new era.
Vitrification is now common to both plastic and cryo workflows. Therefore, apart
from the necessity to render infectious agents inactive, the artefacts of chemical
fixation should no longer play a role. It is tempting to conclude that the major
challenges (images acquired in focus, suitable specimen thinning procedures, close
to 100% information yield from calibrated electron dose) have been addressed
satisfactorily, and although there is always room for improvement, perhaps the most
important point to note is that these solutions are costly in terms of instrumentation
and highly-trained, dedicated personnel.
Finally, the success of standard sample collection geometries for numerous
high-impact publications belies the fact that cylindrical samples have numerous
advantages, including constant focus and thickness during tilting. For the last part
of this chapter, I will explain why this is problematic and how this setup could help
electron tomography data collection to be even more efficient, and to realise its true
potential in terms of information yield.
22
A. Leis