Chapter 1 – Cryo EM 101

Chapter 1: Sample purification

Overview and learning objectives

Cryo-EM has a distinct advantage over other structural biology methods such as X-ray crystallography and NMR in that only a relatively small amount of sample is required. Although cryo-EM can tolerate low purification yields and certain degrees of impurity, the highest quality 3D reconstructions are generally derived from samples that have gone through extensive biochemical optimization.

In this chapter, we will cover basic principles during sample purification that can make or break a cryo-EM project. Special considerations need to be taken into account based on the nature of the particle, such as its size and structural heterogeneity. The chapter introduces common considerations and strategies that are used to optimize particles for structure determination. The chapter then concludes with a negative stain section that includes protocols of how to prepare negative stain grids and how to evaluate their images. By the end of this chapter, users should understand general guidelines to assess particle quality via negative stain TEM in preparation for cryo-EM.

A successful cryo-EM project is dependent on the ability to reconstruct 2D projections of a particle into its 3D structure. The particles should meet certain criteria even before delving into cryo-EM. Most importantly, the particles should be monodisperse in solution, meaning that the sample solution consists of molecules that are roughly the same mass, shape, and size. The following section describes a few strategies that are used to improve particle behavior for single particle analysis.

Single particle analysis requires particles to be “average-able” and mono-disperse. This chapter describes some common approaches that are used to improve particle behavior. Sample suitability is often assessed by negative stain analysis and 2D class averaging. Click here to download the video.

Methods to improve particle behavior

An ideal monodispersed and homogeneous sample is shown on the left. A less ideal sample showing structural heterogeneity is shown on the right.

1. Size exclusion chromatography: Gel filtration can be a powerful tool to predict structural heterogeneity. Whenever possible, gel filtration should be incorporated to the final steps of a purification protocol. Elution profiles can reveal the degree of monodispersity, including whether particles have tendencies to aggregate (i.e., large peaks in void volume) or whether multiple protein peaks are observed during elution. The shape of the peak is also an important indicator of monodispersity and ideally the sample produces a symmetric profile free of shoulders or tails.

An example of an optimal gel filtration profile is shown on the left. The protein complex is enriched and abundant, and well separated from other peaks (e.g., excess substrate). On the right, a less optimal gel filtration profile for the same protein complex is shown (prepared in the absence of its substrate). Note that there are multiple peaks that are not well resolved, and the major complex species is less abundant.

2. Protein or protein complex stabilization: If negative stain or cryo-EM analysis reveals significant structural heterogeneity, then the user may consider additional methods to stabilize a structural target in a defined conformation. These approaches have proven to be effective in trapping conformationally variable proteins and complexes in energetic minima that allow high-resolution structure determination. Some examples are shown in the table below and include using small molecule inhibitors, non-hydrolyzable nucleotides (if applicable), Fab antibody fragments, or genetic perturbations (e.g., mutating catalytic residues).

ParticleResolution (Å)Stabilization methodEMDB Accession No.
β-galactosidase1.9Small molecule inhibitor (PETG)EMD-7770
Vps43.2Non-hydrolyzable nucleotide analog (ADP·BeFx)EMD-8887
Ribosome Quality Control Complex8.2Catalytic inactive mutantEMD-6170
Insulin degrading enzyme3.7Fab antibody fragmentEMD-7062
ABCG23.1Catalytic inactive mutantEMD-0190
IGPD23.1Small molecule inhibitor (C384)EMD-3999
CLC-K channel4.0Fab antibody fragmentEMD-8454
MERS ectodomain trimer3.6Fab antibody fragmentEMD-8784
HIV-1 envelope trimer3.8Fab antibody fragmentEMD-9294
YME13.4Catalytic inactive mutantEMD-7023
Isocitrate dehydrogenase3.8Small molecule inhibitor (ML309)EMD-8193
Lactate dehydrogenase2.8Small molecule inhibitor (GSK 2837808A)EMD-8191
Dicer-2 bound to dsRNA6.8Non-hydrolyzable nucleotide analog (ATPγS)EMD-7290
Some proteins, like the IDE enzyme shown here, are inherently flexible and have mobile domains. The conformations of these molecules can be locked in place using specific ligands. The use of monoclonal antibodies or antibody fragments increases the apparent size of the complex, which allows particles to be aligned more accurately. (Animation based on Zhang et al. eLife 2018). Click here to download the video.

3. Chemical crosslinking: One common way to counteract structural heterogeneity is to chemically crosslink the purified sample, such as with glutaraldehyde or BS3. However, crosslinked samples run a risk of producing structural artifacts and this approach should therefore be considered only after a control, non-crosslinked sample has been evaluated.

To crosslink samples in a controlled manner, Holger Stark’s group developed the GraFix (Gradient Fixation) approach that reduces problems associated with particle heterogeneity. GraFix is performed using ultracentrifugation of samples through a density gradient that contains increasing concentrations of crosslinking reagent. Heterogeneous particles will separate through the gradient based on their masses and become stabilized through crosslinking. Crosslinkers should be used at low concentrations to minimize non-specific intermolecular crosslinks (e.g., 0.01-0.1% v/v glutaraldehyde may be a good starting point). Following ultracentrifugation, the tube is fractionated and samples can be assessed by SDS-PAGE and negative stain TEM. Before preparing cryo-EM grids, the gradient material (e.g., glycerol or sucrose) may interfere with vitrification or decrease contrast in cryo-EM images and should thus be removed via buffer exchange.

The GraFix method can be used to stabilize protein complexes. As the sample migrates through the ultracentrifugation gradient, particles will separate based on their size (see Stark, H. Methods Enzymol. 2010). Click here to download the video.

Membrane proteins

Membrane proteins are an increasingly popular class of macromolecules for single particle cryo-EM analysis. These types of proteins are important targets of pharmacological agents but are challenging to crystallize and study by other structural biology methods. There are now many examples demonstrating that cryo-EM is ideally suited for resolving high-resolution structures of membrane proteins. In order to visualize membrane proteins, they are either solubilized in the presence of detergents or detergent-like amphipols, or reconstituted in lipid nanodiscs.

Series of 3D illustrations showing common biochemical approaches to preparing membrane protein complexes for single particle cryo-EM, including stabilization of membrane proteins in detergents (left), amphipols (middle), or lipid nanodiscs (right).

Detergents are required to solubilize membrane proteins, but the molecules may also destabilize their structures. The choice of detergent is therefore a critical part of the project workflow and multiple types may need to be evaluated before one that strikes the right balance between solubility and structure preservation is determined. The concentration of the detergent is also a variable that affects particle quality. Most detergent-stabilized membrane protein structures are solved using detergent concentrations at or below the critical micelle concentration (CMC), and this range may be a good starting point for single particle EM projects. However, keep in mind that detergents may complicate the cryo-EM grid preparation process (by affecting the surface tension of the remaining solution on the blotted grid) and detergent molecules may lower the contrast of cryo-EM images.

An attractive alternative to detergents is to use amphipols to stabilize membrane protein structures. Amphipols are long, amphipathic polymers (~8-9 kDa) that can be used to replace detergents in buffer. The hydrophobic chains wrap irreversibly around transmembrane regions, forming a tight belt that shields the hydrophobic domains from the aqueous environment. The detergents are then removed from solution via Bio-beads or dialysis. Most of the current high-resolution membrane protein cryo-EM structures have been solved using amphipols.

One emerging tool to study membrane proteins via single particle EM is the use of lipid nanodiscs that stabilize the proteins in a lipid bilayer environment. These nanodiscs consist of a patch of lipid bilayer that is surrounded by a belt of membrane scaffolding proteins. The major advantage to using lipid nanodiscs is that the proteins are retained in more a native-like environment. Thus, lipids that directly interact with membrane proteins are also retained and resolvable by cryo-EM. Although nanodiscs are the most desirable choice to study membrane protein structure, there is currently no ‘one-size-fits-all’ system that allows the nanodiscs to form. The size of the membrane scaffolding proteins often needs to be optimized to form the lipid belt properly and their successful assembly can be validated by size exclusion chromatography.

The following table provides examples of several membrane protein structures that have been solved using one of these three different methods (detergents, amphipols, or nanodiscs). Keep in mind that the right choice for your protein-of-interest will probably require extensive trial-and-error and may be the rate-limiting step of your project.

Membrane ProteinResolution (Å)Stabilization methodEMDB Accession No.
TFP4.2Detergent (DDM)EMD-6945
Orco3.5Detergent (digitonin)EMD-7352
Class B GPCR-G-protein complex4.1Detergent (LMNG/CHS)EMD-8623
RyR13.6Detergent (CHAPS)EMD-8342
CFTR3.9Detergent (digitonin)EMD-8516

Small particles

Particles must be visible under the electron microscope in order to be properly analyzed. This constraint poses a lower molecular weight limit of what can be a viable cryo-EM project. If the particle is too small, it may be entirely invisible in the cryo-EM micrograph. Even if the particle is visible, its small size may lead to inaccurate orientation determination during downstream image processing steps and produce unreliable structures. For practical purposes, a general guideline is that particles should be at least 100 kDa in size, although the true lower limit is unknown.

Negative stain TEM

How do you know that your purified sample is good enough for single particle cryo-EM? Virtually every cryo-EM project begins with negative stain analysis to evaluate particle quality. This simple and quick method provides a first-order estimation of the particle’s monodispersity, concentration, and averageability.

A general protocol for preparing negative stain grids. The procedure begins with glow discharging continuous carbon grids. A few microliters of sample are applied and adsorbed onto the surface of the hydrophilic grids. Excess sample is then removed with filter paper followed by a wash step and finally a few microliters of negative stain are deposited onto the grid. The excess stain is removed with filter paper and the grid is allowed to dry. The grid is then ready for examination via TEM (see next video). Click here to download the video.


  1. Glow discharge grids with a continuous carbon coat.
  2. Apply 3-4μl of sample to the grid and wait for ~1 minute.
  3. Remove excess sample by blotting with filter paper.
  4. Wash the grid over a water droplet and dry the grid with filter paper.
  5. Apply 3.5μl of heavy metal stain to the grid and wait for ~20-30 seconds.
  6. Remove excess stain with filter paper and allow the grid to dry in air for several minutes.
  7. Grid is stored and is ready to be imaged by TEM. (Negative stain grids can be stored for years if stored properly, i.e. under vacuum and away from light)

Table of common reagents used for negative stain TEM

StainWorking concentrationComments
Uranyl acetate1-2%Most common stain used for single particle analysis. Light sensitive. Store at 4°C. Filter periodically to remove precipitates.
Uranyl formate0.5-1%Light sensitive. Difficult to dissolve. Store single use aliquots at -80°C. The fine granular size of this stain is useful for visualizing smaller particles (<100 kDa).
Phosphotungstic Acid1-3%Prepared at neutral pH and may be preferred if particles are sensitive to acidic stains. Provides less contrast than uranyl-based stains.
Ammonium molybdate1-5%Prepared at neutral pH and may be preferred if particles are sensitive to acidic stains. Provides less contrast than uranyl-based stains.

Visualizing the sample at the TEM

At the TEM. This video shows the negatively-stained sample being viewed using a transmission electron microscope. Click here to download the video.

Negative stain images should be recorded at a few microns underfocus. Optimally stained areas of the grid should show fine details within individual particles. The following interactive module shows a few negatively stained bacteriophage particles. Use the module to adjust the focus parameter of the images. Note how the images become blurry when you’re far from focus. For negative stain particles, 1 to 2 microns of underfocus generally provides a good balance of detail and contrast in the final images.

Focus Level

Ideally, particles should be well dispersed, not aggregated. There should be enough particles to saturate the field of view because cryo-EM samples generally show at least 5-10x fewer particles than negative stain. Particles that are well stained will show textured features. Samples that are poorly stained or overstained will appear as dark blobs with no identifiable features.

Examples of good (top row) and poor (bottom row) negative staining.

Negative stain images can be used to provide a sense of how well the particles could be averaged. In this procedure, individual particle images are extracted from the micrographs and then sorted into separate classes based on their similar features. A dataset of a few thousand particles is usually sufficient for this exercise. Attaining well-averaged 2D classes is a good indicator that the particle is ready for cryo-EM and is a minimal requirement expected by many cryo-EM facilities. A more detailed overview of 2D class averaging is available in Chapter 5 (Data Processing).

Limitations of negative stain analysis

Note that negative stain analysis is not always an accurate predictor of whether the particle will lead to a high-resolution cryo-EM structure. Proteins could be destabilized by different stains, for example due to the acidic pH of some stain solutions. Small particles may be harder to visualize by negative stain than by cryo-EM because of poor interactions between the stain and particle. Nucleic acids can also be challenging to visualize by certain types of negative stain. Other issues with single particle cryo-EM, such as preferred orientations and particles failing to enter holes, cannot be addressed at the negative stain level. Nevertheless, it is always worth performing negative stain analysis because of the quick feedback it provides.

Assessing particle quality with mass photometry

Overview of mass photometry. Animation courtesy of Refeyn, Inc.

Mass photometry is an emerging method that provides information about particle mass and monodispersity of biomolecules in solution.  A small volume of sample is deposited onto a glass surface, which is then illuminated by a laser. The light is reflected off the glass surface and is also scattered by the particles on the surface. The scattered and reflected light interfere with each other, which is captured by an objective. The amount of interference from each biomolecule scales linearly with its mass, independent of shape or volume. Mass photometry can measure the masses of biomolecules (protein, nucleic acid or other biomolecule of interest) ranging between 30 kDa to 5 MDa in mass.

A number of features of mass photometry make it suitable for a cryo-EM workflow. Mass photometry is compatible with most aqueous buffers and detergents, so it is possible to analyze biomolecules, including membrane proteins and macromolecular complexes, in their native states. The method offers a fast readout time and has low sample requirements, thus making it a desirable technique to screen different buffer conditions before proceeding to negative-stain EM and cryo-EM.

Mass photometry is a single-particle technique, so it can also detect rare species in a sample. These attributes allow mass photometry to provide better resolution over heterogeneous samples than techniques which rely on averaging particle masses, like dynamic light scattering (DLS). The sample concentration needed for mass photometry experiments is very low (10uL of sample at nanomolar concentration), which conserves valuable sample during early optimization steps; however, this can also be a limitation if the complex requires higher concentrations for stability or grid preparation. Sometimes even a higher concentration of the biomolecule is not enough to withstand the effects of the air-water interface or freezing. In these cases, additional stabilization from crosslinkers or fixatives may be necessary, and mass photometry can be used to optimize structure stability under these conditions.

Overall, mass photometry offers a rapid sample optimization and analysis method for cryo-EM projects and is complementary to other quality control methods, such as negative stain screening. This technique can provide information that is not otherwise accessible with other instruments and can streamline the sample characterization process before committing additional resources to cryo-EM.