User-friendly, High-throughput, and Fully Automated Data Acquisition Software for Single-particle Cryo-electron Microscopy

Summary

Single-particle cryo-electron microscopy demands a suitable software package and user-friendly pipeline for high-throughput automatic data acquisition. Here, we present the application of a fully automated image acquisition software package, Latitude-S, and a practical pipeline for data collection of vitrified biomolecules under low-dose conditions.

Abstract

In the past several years, technological and methodological advancements in single-particle cryo-electron microscopy (cryo-EM) have paved a new avenue for the high-resolution structure determination of biological macromolecules. Despite the remarkable advances in cryo-EM, there is still scope for improvement in various aspects of the single-particle analysis workflow. Single-particle analysis demands a suitable software package for high-throughput automatic data acquisition. Several automatic data acquisition software packages were developed for automatic imaging for single-particle cryo-EM in the last eight years. This paper presents an application of a fully automated image acquisition pipeline for vitrified biomolecules under low-dose conditions.

It demonstrates a software package, which can collect cryo-EM data fully, automatically, and precisely. Additionally, various microscopic parameters are easily controlled by this software package. This protocol demonstrates the potential of this software package in automated imaging of the severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) spike protein with a 200 keV cryo-electron microscope equipped with a direct electron detector (DED). Around 3,000 cryo-EM movie images were acquired in a single session (48 h) of data collection, yielding an atomic-resolution structure of the spike protein of SARS-CoV-2. Furthermore, this structural study indicates that the spike protein adopts two major conformations, 1-RBD (receptor-binding domain) up open and all RBD down closed conformations.

Introduction

Single-particle cryo-EM has become a mainstream structural biology technique for high-resolution structure determination of biological macromolecules¹. Single-particle reconstruction depends on acquiring a vast number of micrographs of vitrified samples to extract two-dimensional (2D) particle images, which are then used to reconstruct a three-dimensional (3D) structure of a biological macromolecule^2,3. Before the development of DEDs, the resolution achieved from single-particle reconstruction ranged between 4 and 30 Å^4,5. Recently, the achievable resolution from single-particle cryo-EM has reached beyond 1.8 Å⁶. DED and automated data acquisition software have been major contributors to this resolution revolution⁷, where human intervention for data collection is minimal. Generally, cryo-EM imaging is performed at low electron dose rates (20-100 e/Ų) to minimize electron beam-induced radiation damage of biological samples, which contributes to the low signal-to-noise ratio (SNR) in the image. This low SNR impedes the characterization of the high-resolution structures of biological macromolecules using single-particle analysis.

The new generation electron detectors are CMOS (complementary metal-oxide-semiconductor)-based detectors, which can overcome these low SNR-related obstacles. These direct detection CMOS cameras allow fast readout of the signal, due to which the camera contributes better point spread function, suitable SNR, and excellent detective quantum efficiency (DQE) for biological macromolecules. Direct detection cameras offer high SNR⁸ and low noise in the recorded images, resulting in a quantitative increase in the detective quantum efficiency (DQE)-a measure of how much noise a detector adds to an image. These cameras also record movies at the speed of hundreds of frames per second, which enables fast data acquisition^9,10. All these characteristics make fast direct detection cameras suitable for low-dose applications.

Motion-corrected stack images are used for data processing to calculate 2D classification and reconstruct a 3D density map of macromolecules by using various software packages such as RELION¹¹, FREALIGN¹², cryoSPARC¹³, cisTEM¹⁴, and EMAN2¹⁵. However, for single-particle analysis, an enormous dataset is required to achieve a high-resolution structure. Therefore, automatic data acquisition tolls are highly essential for data collection. To record large cryo-EM data sets, several software packages have been used over the past decade. Dedicated software packages, such as AutoEM¹⁶, AutoEMation¹⁷, Leginon¹⁸, SerialEM¹⁹, UCSF-Image4²⁰, TOM₂²¹, SAM²², JAMES²³, JADAS²⁴, EM-TOOLS, and EPU, have been developed for automated data acquisition.

These software packages use routine tasks to find hole positions automatically by correlating the low-magnification images to high-magnification images, which assists in identifying holes with vitreous ice of appropriative ice thickness for image acquisition under low-dose conditions. These software packages have reduced the number of repetitive tasks and increased the throughput of the cryo-EM data collection by acquiring a vast amount of good-quality data for several days continuously, without any interruption and the physical presence of the operator. Latitude-S is a similar software package, which is used for automatic data acquisition for single-particle analysis. However, this software package is only suitable for K2/K3 DEDs and is provided with these detectors.

This protocol demonstrates the potential of Latitude-S in the automated image acquisition of SARS-CoV-2 spike protein with a direct electron detector equipped with a 200 keV cryo-EM (see the Table of Materials). Using this data collection tool, 3,000 movie files of SARS-CoV-2 spike protein are automatically acquired, and further data processing is carried out to obtain a 3.9-4.4 Å resolution spike protein structure.

Protocol

NOTE: Three important steps are required for cryo-EM data collection: 1. cryo-EM grid preparation, 2. calibration and alignment of the microscope, 3. automatic data collection (Figure 1). Furthermore, automated data collection is subdivided into a. suitable area selection, b. optimization of Latitude-S, c. start automatic hole selection, and d. start automatic data acquisition (Figure 1).

1. Cryo-EM grid preparation and sample loading for automatic data acquisition

1. Clean the grids using a glow discharger and vary the glow discharge parameters based on experimental requirements (here, 60 s at 20 mA).
2. Add a freshly prepared protein sample (3 µL) to the glow-discharged grid and incubate for 10 s.
3. Blot the grids for 3-5 s at 100% humidity and quickly plunge them into liquid ethane using a cryo-plunger.
4. Clamp the grids manually into a clip ring to form the cartridge using a flexible C-clip ring.
5. Load the frozen cartridge-mounted grids into the autoloader cassette and transfer the cassette by the nano cap to the precooled autoloader of the microscope for data collection.

2. Microscope tuning and basic alignment before automatic data acquisition

1. Beam shift
   1. Click on Beam shift from the Direct Alignment tab.
   2. Reduce the magnification, and center the beam to the optical axis using the Multifunction X and Y knob.
2. Pivot point alignment
   1. Click on the Beam tilt option in Direct Alignment pp X from the Direct Alignment tab.
   2. Condense the beam to a spot and minimize the movement by using the Multifunction X and Y knob.
3. C2 aperture centering
   1. Select the condenser aperture from the Alignment tab.
   2. Condense the beam to a spot, center the beam to the optical axis, and then expand the beam to cover the circle evenly.
   3. Repeat these steps until the Condenser 2 aperture is adjusted.
4. Coma-free alignment
   1. Click on Coma-free Alignment X from the Direct Alignment tab to align the beam to the optical axis.
   2. Use the Multifunction knob to minimize the shape and the movement of the FFT (ensure it is stable).
   3. Repeat the same procedure for Coma - free alignment Y.
5. Set parallel illumination before data collection in the cryo-EM because of the C twin lens.
   1. Insert the objective aperture (70 µm) in diffraction mode.
   2. Focus the objective aperture on the front focal plane of the diffraction lens by controlling the defocus-intensity knob (objective lens and C2 lens current).
   3. Ensure that the crisp edge of the objective aperture is seen after proper defocusing.
   4. Insert the beam stopper and spread the intensity until the gold powder diffraction rings are minimized.
      NOTE: If the beam is spread properly, a clear diffraction ring of the gold powder is visible on the screen, which indicates that the beam is parallel.
   5. Retract the beam stopper after setting the parallel illumination and change the microscope mode to Nano probe.
      ​NOTE: Check the microscope tuning before starting data collection to ensure the optimum performance of the microscope. All these settings would be performed in the microscope Direct Alignment GUI Tab. All the microscope tuning is performed using a test grid before data collection.

3. Data acquisition with Latitude-S

1. Start Latitude-S automated data acquisition software.
   NOTE: Latitude-S installation also requires microscope calibration, which will be performed before data collection, and the settings will be stored permanently. Five different states for data collection are calibrated with four different magnifications (Figure 1 and Figure 2). Atlas state and grid state are in two different magnifications in LM mode (low-magnification ranges). The hole state is in SA mode (high-magnification ranges) but with a moderate magnification. Focus and data state use high-magnification SA mode.
   1. Click on DigitalMicrograph from the Start menu, or double-click on the DigitalMicrograph icon on the desktop.
   2. Select the Technique manager icon from DigitalMicrograph.
      NOTE: This system will show TEM and Latitude-S icons (Figure 2 and Supplemental Figure S1).
   3. Select the Latitude-S icon for single-particle automated data collection.
      NOTE: The K2 camera operates in three modes: linear/integrated, counted, and super-resolution. The user could select any mode in the interface of DigitalMicrograph. Data images could be saved as either dose-fractionated image stacks or as summed images in MRC, TIF, or .dm4 files with different bit depths. Furthermore, data could be saved as motion-corrected images for the K3 camera. On a K2 camera, an unprocessed image stack could be saved as 4-bit MRC, 8-bit TIF, or 8-bit .dm4 files.
2. Create a new session based on the settings from a previous session.
   1. Check the Based on prior session checkbox in the palette.
   2. Select the New button.
   3. Choose the folder containing the session on which the new session's settings are based. Go to the previous session directory to create the new session. Choose the folder to save the new session and associated data.
   4. Select and choose the folder where the new session and associated data will be saved.
      NOTE: Each state and its underlying settings (magnification, illumination conditions, image, or projector) and beam shift and camera parameters (total exposure, single-frame exposure, and binning) will be exported from the existing session to the new session. The path of the folder is shown as a text string at the bottom of the palette. Each of the states and configuration palettes has an asterisk (*) appended to the title to show that it has already been set up and is ready to use.
3. Continue an existing session.
   1. Press the Continue button in the palette to continue an existing session.
      NOTE: The atlas montage cannot be modified.
   2. Choose and navigate to the folder that contains the session that needs to be continued.
4. Start an entirely new session.
   1. Click on New tab in the palette. Choose the folder that contains the session to be continued. Select a folder to save the data.
      NOTE: The default folder name is built by using the date and time.
   2. Click on the Setting icon. In the Manage state explore box that appears, add state, set the TEM condition, the camera condition, and image/stack, and then name the state.
      NOTE: The automated data acquisition workflow uses 5 different states for automated data collection. These states are configured and stored in their respective state palettes. The state summary is given in Table 1.
5. Configure the atlas state.
   1. Click on Atlas state palette.
   2. Configure the atlas state with the following parameters: magnification 115x LM mode in nano probe, illumination conditions-spot size 8 and brightness 934400, binning: 1 and camera exposure time: 1.0 s for imaging at low magnification. Refer to the state summary given in Table 1.
   3. Click on Next to move to the next state.
      NOTE: Atlas state is the lowest magnification state, which provides the survey of the entire grid (Supplemental Figure S2). Generally, this state helps us visualize the entire grids at low magnification and judge the ice thickness, flatness, and broken square of the grids. It is recommended to generate the atlas at different areas of the grid to observe the optimal ice thickness and suboptimal ice thickness of the grids (Supplemental Figure S3). The mentioned parameters could be varied according to the user's needs.
6. Configure the grid state.
   1. Click on the Grid state palette.
   2. Configure the Grid state with the following microscope imaging optics (magnification 380x LM mode in Nano probe), illumination conditions (spot size: 8 and brightness 626,200), binning: 1 and camera exposure time: 1.0 s.
   3. See the Latitude-S state summary provided in Table 1.
   4. Click on Next to move to the next state.
      NOTE: The grid state is set at a magnification higher than the atlas state such that the field of view is one grid square (Figure 2). In this particular magnification, one grid square is observed. Therefore, holes are observed correctly in this magnification, which helps check the ice thickness of the holes (Supplemental Figure S4). A simple bandpass filter is used in the grid state to locate the holes in the patent grid. The mentioned parameters could be varied according to the user's needs.
7. Configure the hole state.
   1. Click on the hole palette.
   2. Configure the Hole state with the following microscope settings: imaging optics (magnification 4500x SM mode in Nano probe), illumination conditions (spot size: 7 and Beam diameter 8.81 µm), binning: 1 and camera exposure time: 1.0 s.
   3. Change the parameters if required based on the grid type. See the state summary provided in Table 1.
   4. Click on Next to move to the next state.
      NOTE: The SA mode indicates a high-magnification range in the electron microscope. The hole state is in the SA magnification range with a field of view of a few micrometers (10-20 µm) (Figure 2 and Supplemental Figure S4). This magnification range is higher than the Atlas or Grid state but much smaller than Focus/Data state. In this magnification, individual holes will be visible. The hole size is appropriate to observe high degrees of contaminations, empty holes, and the proper ice thickness of the holes. The holes for imaging are selected based on these assumptions. Two filters are used in the hole state: one for cross-correlating a hole reference image with a new hole image and another for adjusting the stage height to the eucentric height.
8. Configure the focus state.
   1. Click on Focus palette.
   2. Configure the focus state with the following microscope settings: imaging optics (magnification 45,000x SA mode in nano probe), illumination conditions (spot size: 8 and brightness 934400), binning: 1 and camera exposure time: 1.0 s.
   3. Focus on the amorphous carbon area near the hole. Refer to the Latitude-S state summary provided in Table 1.
   4. Click on Next to move to the next state.
      NOTE: The SA mode indicates a high-magnification range in the electron microscope. The Focus state is the higher SA range magnification. In focus mode, the beam is shifted to a nearby carbon area of the target hole and performs focus automatically to collect the data in the data state. A bandpass filter combined with a hanning or soft rectangular filter is used in the focus state to measure the offset between two focus state images of the same area (Figure 2). The mentioned parameters could be varied according to the user's needs.
9. Configure the data state.
   1. Click on Data palette.
   2. Configure the data state with the following microscope settings: imaging optics (e.g., magnification 28,000x, 45,000x, 54,000x at SA mode in nano probe), illumination conditions (spot size: 8 and brightness 934400), binning: 1 and camera exposure time: 1.0 s.
   3. Refer to the Latitude-S state summary given in Table 1.
   4. Click on Next to move to the next state.
      ​NOTE: Data state is the highest magnification selected based on pixel size requirements and target resolution (Figure 2). Generally, after focusing, the beam is automatically shifted to the target area to collect the data. The above-mentioned parameters could be changed based on the user's requirements.

4. Focus configuration

1. Click on Focus configuration palette. Specify the range of defocus values and the step size in the given tab.
2. Press Next button to move to the next step.
   ​NOTE: Lower defocus values can be used for high-resolution data acquisition. Generally, -0.5 to -3.0 µm defocus values with 0.25 or 0.5 defocus step sizes are used for image acquisition. Users can skip the focus setup step if they only want to screen the sample. Simply press the Next button on the palette to skip the focus configuration step.

5. Fine alignment

1. Focus on some features on the grid (e.g., ice contamination hexagonal ice); see Figure 3.
   NOTE: Features should not be too big or too small. They should be visible at both Atlas state magnification 115x (LA mode) and data magnification.
2. Click on the Capture button. Position the red cross mark on the same feature on each image of different states.
3. Start with focus, data, and hole states because their field of view is much bigger than the atlas and grid states. Zoom on atlas and grid states to position the red cross mark on the same feature in the atlas and grid states.
4. Click on the Calculate button to calculate the positions of five different states, which will calculate the offsets between each of the states and reflect these to the output window.
   ​NOTE: The offset values are integrated into the states for further use (Figure 3). Fine alignment is performed to provide high accuracy of the position of each state (Figure 3). This fine alignment helps pinpoint the exact position in all the five states. Fine Alignment is critical for single-particle data acquisition. Therefore, it is highly recommended to perform Fine Alignment before imaging.

6. Data acquisition procedure using Latitude-S

1. Click on the Capture palette.
   NOTE: Generally, atlas data is collected at low magnification (115x) to visualize most of the grid squares.
2. Choose the size of the atlas to cover the entire grid or part of the grid based on the requirement (e.g., 6 x 6, 8 x 8, 12 x 12, 6 x 8, 8 x 6, 12 x 8, or 8 x 12).
   NOTE: 16 by 16 atlas size covers the entire grid.
3. Click on the Capture button to capture the atlas.
   NOTE: The main Latitude-S navigation window opens and fills the available space in DigitalMicrograph (Supplemental Figure S5). Three picture panes in the main navigation window show images of the system states at three different magnifications. The overall atlas is currently displayed in its current state of acquisition in the leftmost pane. Tiles in the atlas will fill up as each capture occurs.
4. Select the grid square based on the ice thickness by navigating on the atlas (Supplemental Figure S5). Once the desired grid squares are selected, click on the Schedule button and observe the tiles in the grid square fill up as each grid square is captured.
5. Click on the Schedule button once the grid squares are selected.
6. Select a representative hole in the grid square by adding its position. Once a hole image is acquired, define the data and focus positions and save the layout as a template (Supplemental Figure S6).
7. Click on Auto find, give the hole size (e.g., R1.2/1.3), and click on the Find button in the program, which will cause the Find program to automatically find the holes based on the hole diameter. Next, click on the Mark button to add the template (Figure 4) and add red circle marks in all the holes in one grid or partial grid square.
8. Set up the intensity to remove the holes from the grid square and ice contamination (Figure 4).
   NOTE: Finally, the selected holes will be marked in yellow for scheduling the data collection.
9. Click on the Schedule button in Latitude tasks after adding the holes through Auto find.
   ​NOTE: Before scheduling the automated data collection, ensure that the liquid nitrogen tank level is sufficient, the autoloader turbo pump is off, and the RAID drive space is free. Latitude-S task manager shows the number of atlas, grid square, hole, and data states scheduled for data collection (Figure 5). In Latitude-S GUI, various color schemes will be visible, and the meaning of the different color schemes will be displayed: 1. Yellow indicates unscheduled; 2. Green indicates scheduled; 3. Blue indicates Acquired; 4. Red indicates failed.

7. Cryo-EM data processing

NOTE: Cryo-EM image processing of spike protein is described in detail in recent literature²⁵.

1. Perform image processing of spike protein of SARS-CoV2 using RELION 3.0¹¹.
2. Screen the movie images collected using Latitude-S manually, and perform beam-induced motion correction of the individual movies using MotionCor2 software⁹. Perform the initial screening of the motion-corrected micrographs manually with the help of cisTEM software package¹⁴.
   NOTE: Almost 85% of the automatically acquired micrographs were of good quality, and data had signal within 3.7-5.2 Å, which is calculated using cisTEM software¹⁴ (Supplemental Figure S7A,B).
3. Process the data using the RELION 3.0 software package¹¹.
   1. Pick spike particles manually and subject them to 2D class classification (Supplemental Figure S7C). Use the best 2D class as a reference to autopick 3,99,842 single-spike particles from the micrographs using the RELION autopick tool¹¹.
      NOTE: Three rounds of 2D classification were carried out before subjecting the particles to 3D classification (Supplemental Figure S8). Approximately 2,55,982 single particles were selected for 3D classification, and the data set was classified into six classes. The final 3D auto-refinement was performed with the best class; 85,227 spike particles were obtained from the 3D classification.
   2. After auto refinement, perform per particle defocus refinement with proper beam tilt parameters for resolution improvement. Next, subject the particles to Bayesian polishing using the RELION 3.0 software package¹¹. Finally, use the polished particle set for another round of 3D auto-refinement using RELION 3.0¹¹.

Results

In the current pandemic situation, cryo-EM plays a key role in characterizing the structures of various proteins from SARS-CoV-2^26,27,28,29, which may help develop vaccines and drugs against the virus. There is an urgent need for fast-paced research efforts with limited human resources to combat the coronavirus disease of 2019. Data acquisition in single-particle cryo-EM is a time-consuming but...

Discussion

Latitude-S is an intuitive user interface, which provides an environment to automatically set up and collect thousands of high-resolution micrographs or movie files in two days. It provides easy navigation across the grids and maintains the position of the microscope stage while it moves from low magnification to high magnification. Each step of data acquisition with Latitude-S is time-efficient, with features such as a simple user interface, fast streaming of data at up to 4.5 GB/s speed, and simultaneous display of dat...

Materials

Name	Company	Catalog Number	Comments
Blotting paper	Ted Pella, INC.	47000-100	EM specimen preparation item
Capsule	Thermo Fisher Scientific	9432 909 97591	EM specimen preparation unit
Cassette	Thermo Fisher Scientific	1020863	EM specimen preparation unit
C-Clip	Thermo Fisher Scientific	1036171	EM specimen preparation item
C-Clip Insertion Tool	Thermo Fisher Scientific	9432 909 97571	EM specimen preparation tool
C-Clip Ring	Thermo Fisher Scientific	1036173	EM specimen preparation item
EM grid (Quantifoil)	Electron Microscopy Sciences	Q3100AR1.3	R 1.2/1.3 300 Mesh, Gold
Glow discharge Machine	Quorum	N/A	Quorum GlowQube glow discharge machine
K2 DED	Gatan Inc.	N/A	Cryo-EM data collection device (Camera)
Latitude S Software	Gatan Inc.		Imaging software
Loading station	Thermo Fisher Scientific	1130698	EM specimen preparation unit
Talos 200 kV Arctica	Thermo Scientific™	N/A	Cryo-Electron Microscope
Vitrobot Mark IV	Thermo Fisher Scientific	N/A	EM specimen preparation unit