Single-Particle Cryo-EM Data Collection with Stage Tilt using Leginon

Summary

The present protocol describes a generalized and easy-to-implement scheme for tilted single-particle data collection in cryo-EM experiments. Such a procedure is especially useful for obtaining a high-quality EM map for samples suffering from preferential orientation bias due to adherence to the air-water interface.

Abstract

Single-particle analysis (SPA) by cryo-electron microscopy (cryo-EM) is now a mainstream technique for high-resolution structural biology. Structure determination by SPA relies upon obtaining multiple distinct views of a macromolecular object vitrified within a thin layer of ice. Ideally, a collection of uniformly distributed random projection orientations would amount to all possible views of the object, giving rise to reconstructions characterized by isotropic directional resolution. However, in reality, many samples suffer from preferentially oriented particles adhering to the air-water interface. This leads to non-uniform angular orientation distributions in the dataset and inhomogeneous Fourier-space sampling in the reconstruction, translating into maps characterized by anisotropic resolution. Tilting the specimen stage provides a generalizable solution to overcoming resolution anisotropy by virtue of improving the uniformity of orientation distributions, and thus the isotropy of Fourier space sampling. The present protocol describes a tilted-stage automated data collection strategy using Leginon, a software for automated image acquisition. The procedure is simple to implement, does not require any additional equipment or software, and is compatible with most standard transmission electron microscopes (TEMs) used for imaging biological macromolecules.

Introduction

The advent of direct electron detectors over the past decade^1,2,3 has spurred an exponential increase in the number of high-resolution structures of macromolecules and macromolecular assemblies solved using single-particle cryo-EM^4,5,6. Almost all purified macromolecular species are expected to be amenable to structure determination using cryo-EM, except for the smallest proteins ~10 kDa in size or below⁷. The amount of starting material needed for grid preparation and structure determination is at least an order of magnitude less than other structure determination techniques, such as nuclear magnetic resonance spectroscopy and X-ray crystallography^4,5,6.

However, a principal challenge for structure determination by cryo-EM involves suitable grid preparation for imaging. An extensive study evaluating diverse samples using different vitrification strategies and grids suggested that most approaches for vitrifying samples on cryo-EM grids lead to preferential adherence of macromolecules to the air-water interface⁸. Such adherence can potentially cause four suboptimal outcomes: (1) the macromolecular sample completely denatures, in which case no successful data collection and processing is possible; (2) the sample partially denatures, in which case it may be possible to obtain structural insights from regions of the macromolecule that are not damaged; (3) the sample retains native structure, but only one set of particle orientations relative to the direction of the electron beam are represented in the images; (4) the sample retains native structure, and some but not all possible particle orientations relative to the direction of the electron beam are represented in the images. For cases (3) and (4), tilted data collection will help with minimizing directional resolution anisotropy affecting the reconstructed cryo-EM map and provides a generalizable solution for a wide variety of samples⁹. Technically, tilting can also benefit case (2), since the denaturation presumably occurs at the air-water interface and similarly limits the number of distinct orientations represented within the data. The extent of orientation bias in the dataset can potentially be altered by experimenting with solution additives, but a lack of broad applicability hampers these trial-and-error approaches. Tilting the specimen stage at a single optimized tilt angle is sufficient to improve the distribution of orientations by virtue of altering the geometry of the imaging experiment⁹ (Figure 1). Due to the geometric configuration of the preferentially-oriented sample with respect to the electron beam, for each cluster of preferential orientations, tilting the grid generates a cone of illumination angles with respect to the cluster centroid. Hence, this spreads out the views and consequently improves Fourier space sampling and the isotropy of directional resolution.

There are, in practice, some detriments to tilting the stage. Tilting the specimen stage introduces a focus gradient across the field of view, which may affect the accuracy of contrast transfer function (CTF) estimations. Tilted data collection may also lead to increased beam-induced particle movement caused by increased charging effects when imaging tilted specimens. Grid tilting also leads to an increase in apparent ice thickness, which in turn leads to noisier micrographs and may ultimately impact the resolution of reconstructions^5,9,10. It may be possible to overcome these issues by applying advanced computational data-processing schemes that are briefly described in the protocol and discussion sections. Lastly, tilting can lead to increased particle overlap, hindering the subsequent image processing pipeline. Although this can be mitigated to some extent by optimizing on-grid particle concentration, it is nonetheless an important consideration. Here, a simple-to-implement protocol is described for tilted data collection using the Leginon software suite (an automated image acquisition software), available open access and compatible with a broad range of microscopes^11,12,13,14. The method requires at least version 3.0 or higher, with versions 3.3 onward containing dedicated improvements to enable tilted data collection. No additional software or equipment is necessary for this protocol. Extensive instructions on computational infrastructure and installation guides are provided elsewhere¹⁵.

Protocol

1. Sample preparation

1. Use grids containing gold foil and gold grid support¹⁶ (see Table of Materials) because tilted data collection can accentuate beam-induced motion¹⁷.
   NOTE: For the present study, samples on grids were vitrified using the manual plunging and blotting technique¹⁸ in a humidified (greater than 80%) cold room (~4 °C).
2. Avoid using grids containing copper support and carbon foil or a continuous layer of amorphous carbon unless absolutely necessary, as these grids may lead to greater beam-induced motion¹⁶ when the specimen stage is tilted.
   NOTE: Alternative support layers, such as graphene/graphene oxide, appear to reduce beam-induced movement compared to amorphous carbon^19,20.
3. Pre-screen grids and identify regions characterized by acceptable ice thickness and particle distribution. Grids containing too tightly packed particles will lead to particle overlap during tilted data collection, which may affect downstream data-processing steps.
   NOTE: These steps are subjective since identifying good areas of ice is performed by visually inspecting defocused images for regions where particle contrast is clear. This may not be feasible for all samples since some samples will not distribute efficiently in areas of thin ice, leading to challenges during data collection (described in the Discussion section).
4. Vitrify the grids containing your protein sample. Here, for demonstration purposes, we use DNA Protection during Starvation (DPS) protein (see Table of Materials) at a range from 0.1-0.5 mg/mL with gold foil and gold support grids.
   ​NOTE: The protein was purified as described previously, except no TEV protease cleavage was performed²¹. The protein concentration range for a sample of interest will have to be optimized individually, since it is hard to gauge an ideal range that is universally applicable and will almost certainly vary between different samples.

2. Setting up tilted data collection

1. Align the microscope to ensure parallel illumination of the specimen and minimize coma aberrations²².
   NOTE: The microscope must be well aligned for standard SPA data collection without stage tilt. No special alignments are necessary for tilted data collection, but a good alignment will ensure that targeting and imaging proceed smoothly. A general scheme comparing tilted and untilted data collection is provided in Figure 2.
2. Record a grid atlas without stage tilt to identify squares suitable for data collection or manually inspect squares at the magnification used in Square Acquisition Node. Look for squares where the foil is intact, does not look dehydrated, and has ideal ice thickness.
   NOTE: Square Acquisition Node is the low-magnification node used for multi-scale imaging in Leginon.
   1. For typical untilted automated data collection, record a grid atlas, which provides an overview of the overall grid quality and an initial indication of suitable areas for data collection.
   2. Subsequently, select suitable squares through the atlas and submit them to the queue. Then, either through manual selection of holes or through the automated EM hole finder, queue and submit the hole targets.
   3. Finally, use the automated EM hole finder for submitting high magnification exposure targets.
      NOTE: For tilted data collection, squares may need to be queued manually for consistent results, especially if the optimal tilt angle has not been pre-determined and is likely to be adjusted during data collection. The grid atlas could also be recorded using a pre-defined stage tilt if the tilt angle used for data collection had previously been established.
3. Move the specimen stage to a square of interest.
4. Determine the eucentric height for the stage position using α-wobbler at ± 15° stage tilt. Adjust the Z-height to bring the stage to eucentric height using the keypad panel for the microscope. Ensure the image shift is minimal during the α-wobble routine.
   NOTE: If the eucentric height is not properly identified, a large image shift will be observed upon tilting the stage at the square magnification. This can also happen if there are local deformations on the grid, for example, if the grid is broken or severely bent in the vicinity of the imaged area. Although it is best to avoid such regions for data collection, it is imperative to accurately estimate eucentric height if these represent one of the few promising regions for data collection. Figure 3 shows how targeting without properly identifying eucentric height can cause large image shifts in the square magnification.
5. Find a more accurate Z-height, use the Focuser node, and press Simulate.
   1. Typically, estimate the Z-height in the Focuser node at the magnification used in Square Acquisition Node.
      NOTE: The Focuser node focus sequence can also include a fine Z focus estimation at the Hole Acquisition node (a tool in Leginon software) magnification during data collection.
   2. Adjust the settings for the Focuser node and enable/disable the fine Z focus option during the initial queuing of squares.
      NOTE: It is important to ensure that eucentric height is accurately identified when automated data collection begins, which may require re-enabling fine Z focus (step 2.10).
6. Tilt the specimen stage to the desired tilt angle for data collection at the true eucentric height, and re-center the stage if necessary. 0°, 30°, and 60° tilt angles were used for this study. Press Simulate in the Square Acquisition node to begin queuing targets for Hole Acquisition node exposures.
   NOTE: As indicated in step 2.2.1, the grid atlas can be recorded using a pre-defined stage tilt, which would obviate the need to tilt the stage again at this step. This works well and speeds up the process if the tilt angle used for data collection is pre-defined. The current protocol is written with new specimens in mind, wherein the user may wish to test different tilt angles for data collection.
7. Select a Z focus target and regions with holes suitable for high magnification exposures.
8. Press Submit targets to queue for imaging. Do not press Submit Queued Targets until finished queuing up all squares.
9. Bring the specimen stage back to its untilted state. Move to the next square and repeat steps 2.3-2.8 until an adequate number of hole exposures have been queued.
10. Go to the Hole Targeting Node and press Submit Queued Targets once all squares are queued.
    NOTE: If fine Z focus was disabled previously to save time (step 2.5), it needs to be re-enabled before submitting the queue.
11. Manually inspect targets selected by the high magnification Exposure Acquisition node to test if the automated EM hole finder can accurately identify suitable regions for image acquisition when the specimen stage is tilted.
    1. During this procedure, select 'Allow for user verification of selected targets' in Exposure Acquisition node settings. Once the user is satisfied with targeting accuracy, deselect this option for automated data collection.
       ​NOTE: Targets in high magnification Exposure Acquisition node are typically imaged using a beam-tilt image shift strategy, which works equally well for both tilted and untilted data collection^23,24,25,26. For accurate CTF estimation in downstream data processing steps, the lens-coma aberration calibration must be performed for the beam-tilt image shift data collection strategy.

3. Data Processing

1. Initiate on-the-fly data processing^10,27,28,29 with motion-correction of recordedmovies, CTF estimation, particle selection, and generation of initial reconstructions during data collection.
   NOTE: For the present study, cryoSPARC Live¹⁰ (see Table of Materials) has been utilized for pre-processing. On-the-fly data processing provides an initial cryo-EM reconstruction and an approximation for the angular distribution, which can inform the user about the extent of resolution anisotropy. These can, in turn, be used to guide the user as to whether or not the tilt angle used for data collection is sufficiently high.
2. Visualize the reconstructed map and plot the Euler angle distribution to gauge the extent of preferred particle orientations.
   NOTE: The Euler angle distributions can be converted directly into Fourier space sampling distributions to determine the potential extent of resolution anisotropy. A graphical user interface (GUI) has been developed to assist the user in evaluating the quality of an Euler angle distribution and determining an optimal tilt angle^30,31. The tool can be obtained from the Github repository, https://github.com/LyumkisLab/SamplingGui.
3. If necessary, adjust the stage tilt angle at which data is collected to overcome the effects of preferential orientation. The angle can be increased if the preferential orientation remains a problem, as evidenced by the map and Euler distribution in 3.2. Alternatively, the user may wish to split the data collection into groups and record using several different tilt angles, such as 20°, 30°, and 40°.
   NOTE: Although most TEMs must have the capability of tilting the stage to 70°, common specimen stage tilts (that we have used) range from 20°-40°.

Results

DPS at 0.3 mg/mL was used to demonstrate imaging at 0°, 30°, and 60° tilts. Data from different tilt angles were collected on the same grid at different grid regions. CTF resolution fits for higher angle tilts tend to be poorer, as was the case when comparing the three datasets in this study. Figure 4 demonstrates comparative representative images and 2D classification averages. Although the protein concentration is unchanged across the different tilt angles, a higher tilt angle makes the im...

Discussion

Preferred particle orientation caused by specimen adherence to the air-water interface is one of the last major bottlenecks to routine high-resolution structure determination using cryo-EM SPA^4,5,6. The data collection scheme presented here provides an easy-to-implement strategy for improving the orientation distribution of particles within a dataset. We note that the protocol requires no additional equipment or software and doe...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Cryosparc Live v3.1.0+210216	Structura Biotechnology
DPS protein			Purification adapted from protocol described in K.Naydenova et al IUCrJ. 2019 Nov 1; 6(Pt 6): 1086–1098.
K2 Summit Direct Electron Detector	Gatan
Leginon software suite			C Suloway et al Journal of Structural Biology 151 (1): pp. 41-60.
Manual plunging device			Homemade guillotine-like device for vitrification of EM grids
Talos Arctica	FEI/Thermo Fisher
UltrAufoil R1.2/1.3 300 mesh grids	Quantifoil	N1-A14nAu30-01