UCSF
UC San Francisco Previously Published Works
Title
MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron
microscopy.
Permalink
https://escholarship.org/uc/item/0rx7g4p1
Journal
Nature methods, 14(4)
ISSN
1548-7091
Authors
Zheng, Shawn Q
Palovcak, Eugene
Armache, Jean-Paul
et al.
Publication Date
2017-04-01
DOI
10.1038/nmeth.4193
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
1
Anisotropic Correction of Beam-induced Motion for Improved Single-
particle Electron Cryo-microscopy
Shawn Q. Zheng
1,2
, Eugene Palovcak
1
, Jean-Paul Armache
1
, Yifan Cheng
1,2
and David A.
Agard
1,2
1
Department of Biochemistry and Biophysics and
2
Howard Hughes Medical Institute, University of California, San Francisco, CA 94158
Abstract
Correction of electron beam-induced sample motion is one of the major factors contributing to
the recent resolution breakthroughs in cryo-electron microscopy. Improving the accuracy and
efficiency of motion correction can lead to further resolution improvement. Based on
observations that the electron beam induces doming of the thin vitreous ice layer, we developed
an algorithm to correct anisotropic image motion at the single pixel level across the whole frame,
suitable for both single particle and tomographic images. Iterative, patch-based motion detection
is combined with spatial and temporal constraints and dose weighting. The multi-GPU
accelerated program, MotionCor2, is sufficiently fast to keep up with automated data collection.
The result is an exceptionally robust strategy that can work on a wide range of data sets,
including those very close to focus or with very short integration times, obviating the need for
particle polishing. Application significantly improves Thon ring quality and 3D reconstruction
resolution.
3
In recent years single-particle cryo-electron microscopy (cryo-EM) has made groundbreaking
advancements in obtaining atomic resolution structures of macromolecules that are either
refractory to crystallization or difficult crystallize in specific functional states
1-4
. Central to this
success has been the broad application of direct electron detection cameras, which not only have
significantly improved detective quantum efficiency (DQE) at all frequencies but also have a
high output frame rate, up to 40 frames per second. Nowadays, cryo-EM images of frozen
hydrated biological samples are recorded as dose-fractionated stacks of sub-frames (movies)
5-7
,
which enable correction of beam-induced sample motion that blurs the captured images
6,8
.
Combining high sensitivity imaging, motion correction and advanced image processing and 3D
classification strategies have made it practical to routinely determine three-dimensional (3D)
reconstructions at near atomic resolution by single particle cryo-EM even for the most
challenging samples
9-11
.
Sample illumination with the high-energy electron beam breaks bonds, releases radiolysis
products and builds up charge within the thin frozen hydrated biological samples during image
recording. The result is a combination of physical and optical distortions that can significantly
deteriorate sample high-resolution information through image blurring. This has been one of the
major factors limiting the achievable resolution of single particle cryo-EM
12
. The concept of
recording the image as movie to correct sample motion was proposed long ago
13,14
, but only
became practical once direct electron detection cameras became available
7
. The fast sample
motions can be measured by tracking the movement of images captured in a series of snapshots,
either as the whole frame or as individual particles. Image motion can then be corrected by
registering identical features in the sub-frames to each other, followed by summing the registered
sub-frames to produce a motion-corrected image
6
. From the earliest studies it has been clear that
the apparent sample motions can be non-uniform and idiosyncratic, such that the pattern of local
motion varies across the image. While in principle tracking should be straightforward, the
practical challenge is the extremely low signal-to-noise ratio (SNR) in each individual sub-
frame. Except for very large particles, accurate motion measurement generally requires
correlating the motion between sub-frames over large areas. That said, even sub-optimal motion
correction can significantly restore high-resolution signals and improve the resolution of final 3D
reconstructions
6,8
.
4
We previously developed an algorithm that made use of redundant measurements of
image shifts between all sub-frames to derive a least squares estimate of relative motions
between neighboring sub-frames. This algorithm, implemented in the program MotionCorr,
provided an efficient correction of image motions with sufficient accuracy
6
to enable the
determination of numerous near atomic resolution 3D reconstructions
11,15
. Around the same time
or soon afterwards, a number of different strategies were devised that either assume particles
located nearby have similar motions or assume uniform motion of the entire frame or patches of
the frame. Programs based on the former assumption include RELION that provides a movie-
processing mode but uses a 3D reconstruction to track particle motions
8,16
, Xmipp that
implemented an Optic Flow algorithm
17
, and alignparts_lmbfgs that implemented a regularized
Fourier Space optimization algorithm to track neighboring particles
18
. Programs based on
tracking the full frame or parts of the frame include MotionCorr
6
and iterative whole frame
alignment procedures such as Unblur
19
or those used in electron tomography
20
. All of these
algorithms have demonstrated the ability to recover high-resolution signals to varying degrees
and have improved the resolution of the resultant 3D reconstructions.
Ideally, single particle cryo-EM images should be acquired with the smallest possible
defocus to enhance high-resolution information, and with the shortest sub-frame exposure times
to reduce motion trapped within individual sub-frames, in particular for the first few frames
where the sample has the least radiation damage but moves most rapidly
21
. Additionally, motion
detection should be done on the smallest possible local area to best capture the anisotropic
motion. Unfortunately, minimizing time and or area significantly reduces the SNR in each sub-
frame, ultimately leading to incorrect motion estimates. For example, previous experiments with
MotionCorr revealed that subdividing the images into areas smaller than ~2000 × ~2000 pixels
or going to sub-frame integration times of less than 100 milliseconds worsened resolution due to
increased errors in motion tracking.
Another recent advance, where a model of radiation damage is used to weight the
individual sub-frames in Fourier space, allows collection to very high doses (80-100e
-
/Å
2
)
producing a high contrast image without degradation of the high-resolution signal
19
. While
having a single correctly weighted high contrast image has clear advantages for particle picking,
orientation refinement, and overall computability, taking advantage of this strategy requires that
each of the sub-frames be optimally aligned over the full image area. Therefore, improving