3.3.4 Cryo-EM Data
Collection
When it comes to data collection of 2D crystals, there are two broad
strategies. If 2D crystals a well ordered and large (at least micronscale), it is possible to collect electron diffraction data by cryo-EM,
as described in [32]. It is then necessary to obtain phases through
indirect means, such as molecular replacement. On the other hand,
it is always possible to image in real-space, with the added benefit
that both phase and amplitude information is conserved. The later
data collection approach has been common for the large majority of
2D crystals. Real-space cryo-EM imaging protocols will be instrument- and facility-specific, but it is vital to image using low-dose
procedures, which can be easily utilized using modern highthroughput imaging programs, e.g., Leginon [33], SerialEM
[34], or EPU (Thermo Fisher Co.). Under ideal conditions, imaging will be performed using direct electron detection to maximize
achievable resolution.
3.4 Image
Processing
Initial stages of image processing proceed as with other cryo-EM
methods. If a direct electron detector was used during data collection, movies should be aligned to correct for motion blurring
using, e.g., MotionCorr [35], MotionCor2 [36], or Unblur
[37]. CTF estimation is also necessary, for instance, with
CTFFIND4 [38] or gCTF [39].
These initial stages of data processing, as well as later stages, can
be easily accomplished using the user-friendly Focus software package [31]. This package, which can be operated in-line with data
collection, provides access to the full pipeline of image processing
necessary for 2D crystallography, containing the core elements first
developed in the MRC 2D crystallography programs, including
lattice estimation, tilt estimation, unbending, projection map generation, 2D merging, 3D merging, and final data evaluation
[40]. The highest achievable resolution will depend on crystal
quality as well as grid preparation, including sample and carbon
film flatness [27, 28], as well as careful data collection and image
processing [31]. The first important data to provide information on
the membrane protein structure is a projection map, which provides information perpendicular to the plane of the phospholipid
bilayer. At resolutions that are lower/worse than 10 A ˚ , information
on the oligomeric state of largely hydrophobic membrane proteins
can be discerned. For these types of proteins at 6–9 A ˚ resolution in
projection, α-helices, provided that they are not too highly tilted,
can be identified. Projection maps at higher resolutions tend to not
allow clear interpretation of α-helices, which may be overcome by
reducing the resolution in an additional map of the same data set to
the intermediate resolution range of 6–9 A ˚ to reveal projection
density typical of α-helices [42]. For membrane proteins with
large soluble domains, the projection map will not allow for interpretation of secondary structure. The 3D map will provide critical
information perpendicular to the plane of the membrane
240
Matthew C. Johnson et al.
Collection
When it comes to data collection of 2D crystals, there are two broad
strategies. If 2D crystals a well ordered and large (at least micronscale), it is possible to collect electron diffraction data by cryo-EM,
as described in [32]. It is then necessary to obtain phases through
indirect means, such as molecular replacement. On the other hand,
it is always possible to image in real-space, with the added benefit
that both phase and amplitude information is conserved. The later
data collection approach has been common for the large majority of
2D crystals. Real-space cryo-EM imaging protocols will be instrument- and facility-specific, but it is vital to image using low-dose
procedures, which can be easily utilized using modern highthroughput imaging programs, e.g., Leginon [33], SerialEM
[34], or EPU (Thermo Fisher Co.). Under ideal conditions, imaging will be performed using direct electron detection to maximize
achievable resolution.
3.4 Image
Processing
Initial stages of image processing proceed as with other cryo-EM
methods. If a direct electron detector was used during data collection, movies should be aligned to correct for motion blurring
using, e.g., MotionCorr [35], MotionCor2 [36], or Unblur
[37]. CTF estimation is also necessary, for instance, with
CTFFIND4 [38] or gCTF [39].
These initial stages of data processing, as well as later stages, can
be easily accomplished using the user-friendly Focus software package [31]. This package, which can be operated in-line with data
collection, provides access to the full pipeline of image processing
necessary for 2D crystallography, containing the core elements first
developed in the MRC 2D crystallography programs, including
lattice estimation, tilt estimation, unbending, projection map generation, 2D merging, 3D merging, and final data evaluation
[40]. The highest achievable resolution will depend on crystal
quality as well as grid preparation, including sample and carbon
film flatness [27, 28], as well as careful data collection and image
processing [31]. The first important data to provide information on
the membrane protein structure is a projection map, which provides information perpendicular to the plane of the phospholipid
bilayer. At resolutions that are lower/worse than 10 A ˚ , information
on the oligomeric state of largely hydrophobic membrane proteins
can be discerned. For these types of proteins at 6–9 A ˚ resolution in
projection, α-helices, provided that they are not too highly tilted,
can be identified. Projection maps at higher resolutions tend to not
allow clear interpretation of α-helices, which may be overcome by
reducing the resolution in an additional map of the same data set to
the intermediate resolution range of 6–9 A ˚ to reveal projection
density typical of α-helices [42]. For membrane proteins with
large soluble domains, the projection map will not allow for interpretation of secondary structure. The 3D map will provide critical
information perpendicular to the plane of the membrane
240
Matthew C. Johnson et al.