High-speed high-resolution data collection on a 200 keV cryo-TEM

Abstract

Limitations to successful single-particle cryo-electron microscopy (cryo-EM) projects include stable sample generation, production of quality cryo-EM grids with randomly oriented particles embedded in thin vitreous ice and access to microscope time. To address the limitation of microscope time, methodologies to more efficiently collect data on a 200 keV Talos Arctica cryo-transmission electron microscope at speeds as fast as 720 movies per hour (∼17 000 per day) were tested. In this study, key parameters were explored to increase data collection speed including: (1) using the beam-image shift method to acquire multiple images per stage position, (2) employing UltrAufoil TEM grids with R0.6/1 hole spacing, (3) collecting hardware-binned data and (4) adjusting the image shift delay factor in SerialEM. Here, eight EM maps of mouse apoferritin at 1.8-1.9 Å resolution were obtained in the analysis with data collection times for each dataset ranging from 56 min to 2 h. An EM map of mouse apoferritin at 1.78 Å was obtained from an overnight data collection at a speed of 500 movies per hour and subgroup analysis performed, with no significant variation observed in data quality by image shift distance and image shift delay. The findings and operating procedures detailed herein allow for rapid turnover of single-particle cryo-EM structure determination.

Keywords: 200 keV cryo-TEM; beam image shift; cryo-electron microscopy; direct-electron detector.

Figure 1

Data collection setup and SerialEM script. (a) Flowchart outlining steps and the time to set up data collection. A full-grid montage is collected at 62× magnification in lower magnification (LM) mode to identify areas ideal for data collection. Maps of selected grid squares are acquired at 210× magnification in LM mode after finding eucentric height. Multishot record parameters and points are added to each of the grid square maps by microscope operator. (b) Flowchart of the SerialEM script used to collect data. The stage is moved to each of the points (in XYZ) and centered over the hole iteratively using the reference image. Errors in the camera sensor referred to as black strip are checked after hole centering. Autofocus using beam-tilt iterates until target defocus is obtained. Stage drift is measured 20 times, or until drift threshold is obtained. As an option, an additional beam shift is applied to direct the beam away from the center of the hole, we refer to this as touch of carbon and find this useful for particles that cluster close to the edge of the carbon film. Micrographs are acquired in a multishot array, one micrograph is taken at each Quantifoil hole, typically 9–49 holes depending on the hole spacing. (c) Percentage of time spent on each step of the data collection script. Record and delay occur during multishot acquire, with delay including image shift delay, other SerialEM delays (ExtraBeamTime, ShutterDeadTime, ExtraOpenShutterTime), and camera read out and file write time.

Figure 2

Multishot setup for Talos Arctica on an R0.6/1 Quantifoil TEM grid. The carbon foil hole and beam diameter (0.6 µm). (a) Holes and beam diameter (indicated by green circles) drawn to scale. The distance in micrometers from the center hole to adjacent holes is indicated at each position in the multishot. Direction and order of BIS are indicated by arrows.

Figure 3

Data collection speeds using different image shift delay factors. (a) Rate of data collection using image shift delay factors of 0, 0.25, 0.5, 1 and 2 in super-resolution (SR) mode or hardware-binned (HB) mode. (b) Particle contribution of individual multishot positions contributing to the final 3D reconstruction for each of the recording conditions and image shift delay factor tested. The outlier in the hardware-binned image shift delay factor 0 is attributed to beam edge entering the field of the detector during data collection. For further breakdown of particle contributions see Fig S1. (c) Individual panels show visualization of coulombic potentials for amino acids Y29 and L115 for each of the eight apoferritin maps.

Figure 4

EM map of apoferritin collected overnight. (a) Local-resolution heat map for a region of the 1.78 Å overnight apoferritin reconstruction. Density contribution from water molecules is visible (red balls within gray transparent map). (b) Individual panels show the visualization of coulombic potentials for representative examples of each of the 20 amino acids. These data were collected with hardware-binning, using an image shift delay factor of 0.5 and collected at a rate of 500 movies per hour.