Cryo-electron microscopy (cryo-EM) is a family of techniques that examines specimens at cryogenic temperatures using an electron microscope. In biological applications, rapid freezing embeds proteins, molecular assemblies, or cellular material in glass-like ice, preserving hydrated structures without conventional dehydration or heavy-metal staining. Recorded images support computational reconstruction of three-dimensional structures. Cryo-EM encompasses several approaches, particularly single-particle analysis and cryo-electron tomography, rather than one uniform experimental procedure. (ebi.ac.uk)
Physical principles and specimen preparation
Biological cryo-EM usually employs a transmission electron microscope, in which accelerated electrons pass through a thin specimen. Their interactions with the sample provide structural information, but also cause radiation damage. Cooling limits damage sufficiently for low-dose imaging; it does not make specimens immune to irradiation. Consequently, individual images contain little signal, and recovering fine detail depends on combining many observations. (grigoriefflab.umassmed.edu)
The central preparation process is vitrification: cooling water rapidly enough to prevent ice crystals from forming. Crystalline ice can obscure particles and disrupt specimen organization, whereas vitreous ice traps hydrated material in a noncrystalline solid. A common procedure applies a small volume of sample to a supported metal grid, removes excess liquid, and plunges the grid into liquid ethane cooled by liquid nitrogen. The specimen remains cold during transfer and imaging. (grigoriefflab.umassmed.edu)
Ice thickness, particle concentration, aggregation, and orientation all influence usable image quality. Ice must accommodate the particles while remaining thin enough for effective electron transmission. Although frequently described as “near-native,” a vitrified specimen is an experimentally prepared, frozen sample rather than a living system observed continuously. (grigoriefflab.umassmed.edu)
Single-particle reconstruction
Single-particle analysis reconstructs an assembly from images of many separate copies viewed in different orientations. “Single-particle” therefore describes the objects being analyzed, not reconstruction from just one molecule. Unlike X-ray crystallography, this approach does not require the molecules to form an ordered crystal. Its reconstruction problem includes estimating the initially unknown orientations of noisy particle images. (nobelprize.org)
A typical computational workflow begins by aligning the frames of recorded movies to compensate for specimen motion. Software then identifies particles, extracts their images, and estimates the microscope’s contrast transfer function, which describes how imaging conditions alter spatial information. Two-dimensional classification groups similar views and helps exclude unsuitable images. Initial three-dimensional maps are subsequently refined through repeated estimation of particle orientations and structural parameters. (arxiv.org)
Reconstruction is an inverse problem involving image processing, statistics, and numerical optimization. Some methods use Bayesian inference to handle uncertain orientations and constrain reconstructions. Three-dimensional classification can separate distinct compositions or conformations, although flexible regions and continuous structural variation complicate this separation. A resulting map represents the information retained by its selected particle population and processing model. (arxiv.org)
Tomography and related approaches
Cryo-electron tomography records the same specimen at successive tilt angles and reconstructs a three-dimensional volume, or tomogram. It can examine assemblies within cells, preserving information about their spatial relationships that purification may remove. Repeated complexes can be extracted from tomograms and combined by subtomogram averaging to improve structural detail. (cryoem.wisc.edu)
Thick cellular specimens may require thinning, including cryogenic focused-ion-beam milling. Conventional tilt-series acquisition cannot cover every viewing angle, leaving a “missing wedge” of information that produces direction-dependent reconstruction limitations. Tomography must also distribute its limited electron exposure across multiple images of the same specimen. (journals.iucr.org)
Other approaches include helical reconstruction, which exploits repeating organization in filaments, and electron crystallography of ordered arrays. Their common use of cryogenic imaging does not imply identical acquisition or reconstruction methods. (ebi.ac.uk)
Historical development
Modern cryo-EM emerged from complementary advances in specimen preservation, microscopy, and computation. Joachim Frank developed methods for combining images of isolated particles into three-dimensional reconstructions. Jacques Dubochet and colleagues established practical vitrification of aqueous specimens during the early 1980s. Richard Henderson’s work on bacteriorhodopsin demonstrated that electron microscopy could determine atomic-resolution protein structures; a major result appeared in 1990. (nobelprize.org)
Direct electron detectors, movie-based motion correction, improved instrumentation, and computational methods drove the subsequent “resolution revolution.” The 2017 Nobel Prize in Chemistry was jointly awarded to Dubochet, Frank, and Henderson for developing cryo-EM for high-resolution determination of biomolecular structures. In 2020, an apoferritin reconstruction at 1.25 ångströms demonstrated direct visualization of individual atoms under exceptionally favorable conditions, rather than a resolution routinely achievable for every specimen. (nobelprize.org)
Resolution, validation, and applications
Cryo-EM resolution describes recoverable structural detail, not simply magnification. A common assessment uses Fourier shell correlation to compare two independently refined half-maps across spatial frequencies. The 0.143 threshold is widely used for reporting single-particle resolution. However, a global value can conceal substantial regional variation, and masking or overfitting can affect apparent agreement. Atomic-model validation therefore also examines whether coordinates fit the map and satisfy plausible molecular geometry. (ebi.ac.uk)
Applications range from ribosomes and viruses to membrane complexes, ion channels, and enzymes. Cryo-EM can reveal assembly organization, conformational differences, and binding interactions, but achievable detail depends on specimen quality, structural stability, and imaging conditions. Reconstructed volumes are archived in the Electron Microscopy Data Bank, while associated atomic coordinates are deposited in the Protein Data Bank; maps and fitted models remain distinct experimental and interpretive products. (nobelprize.org)