Cryo-Electron Microscope
Take a million blurry photos of identical molecules frozen mid-action — let computation reassemble them into a 3D portrait sharp enough to count atoms.
Determines the 3D structure of proteins, viruses, ribosomes, and membrane receptors at resolutions approaching 1.2 Å — comparable to or better than X-ray crystallography — but without needing crystals. Because each frozen snapshot can be in a different functional state, cryo-EM also captures conformational heterogeneity directly.
A dilute protein sample (~0.1–2 mg/mL) is applied to a holey carbon or gold foil grid; excess liquid is blotted to leave a ~50 nm aqueous film. The grid is plunge-frozen in liquid ethane (~−183 °C) at thousands of degrees per second — fast enough to vitrify the water (form glass-like, non-crystalline ice). A 200–300 kV electron beam (wavelength ~2 pm) passes through the film; electrons scatter off atoms with contrast set by atomic number and density. Because biological molecules are radiation-sensitive, total dose is limited to 20–60 e⁻/Ų and each image is unavoidably noisy. Direct electron detectors capture images as 50-frame movies; motion correction undoes beam-induced drift, and contrast-transfer-function correction undoes lens aberrations. Single-particle analysis software (RELION, cryoSPARC) picks hundreds of thousands of particle images, sorts them by orientation, and iteratively reconstructs a 3D density map.
In plain terms
If you wanted to know what a complicated LEGO set looks like assembled, you could photograph millions of identical sets dumped randomly on a table. A computer figures out which photo is the front, the back, the side, rotates them all to match, and combines them into a perfect 3D model. Cryo-EM does exactly that with proteins. It takes hundreds of thousands of photos of identical protein molecules frozen in a thin layer of ice, sorts them by orientation, and merges them into a single atomic-resolution 3D structure — no crystals, no stains.
Also called: cryo-EM, cryogenic transmission electron microscope, single-particle cryo-EM.
Why it matters
- Cryo-EM structures of SARS-CoV-2 spike, HIV envelope, and GPCR drug targets directly guided vaccines and antivirals — the technique is now mainstream pharma infrastructure.
- Membrane proteins (the targets of ~60% of drugs) rarely crystallize; cryo-EM provides a crystallization-free route to their atomic structure.
- Reconstructions can resolve heterogeneous complexes and capture multiple conformations simultaneously.
- Time-resolved cryo-EM via spray mixing now reaches ~10 ms windows, beginning to make molecular-machine movies possible.
Where it lives
- Thermo Fisher Titan Krios — most widely deployed high-end cryo-EM, ~200+ instruments worldwide
- MRC Laboratory of Molecular Biology, Cambridge UK — Henderson's lab; birthplace of cryo-EM structural methods
- NYSBC, Scripps Research, UCSF, Stanford, MPI Biophysics — major academic cryo-EM centers
- Pfizer, Merck, AstraZeneca — pharmaceutical cryo-EM facilities for drug structure determination
Key numbers
- Electron beam voltage: 200–300 kV (Higher voltage → shorter wavelength → less scattering)
- Best reported resolution (apoferritin): ~1.22 Å (Atomic-scale features visible)
- Routine resolution for well-behaved complexes: 2–4 Å
- Vitreous ice thickness: ~50 nm (Thicker ice degrades contrast)
- Electron dose (total): 20–60 e⁻/Ų (Radiation-damage limit for biological samples)
Surprising facts
- Cryo-EM has resolved apoferritin to ~1.22 Å — fine enough to see individual atoms in the protein backbone.
- Richard Henderson predicted in 1995 that ~10,000 images of a 100 kDa protein should suffice for 3 Å resolution. Reaching that prediction in practice took another 18 years of detector and algorithm development.
- The cryo-EM reconstruction of a whole SARS-CoV-2 virion at ~8 Å covered roughly 50 million atoms — at the time the largest molecular structure ever reconstructed at that resolution.
- The 2017 Nobel Prize in Chemistry went to Henderson, Frank, and Dubochet for the methods that made all this routine.
Open questions
- Preferred-orientation problem: Many proteins adsorb to the air-water interface during grid prep and adopt only a small set of orientations rather than randomly tumbling. The resulting anisotropy degrades 3D reconstruction and produces direction-dependent resolution artifacts. The physical chemistry behind which proteins do this — and why — is not fully understood despite extensive investigation.
A short history
Jacques Dubochet (EMBL, 1984) developed vitrification of biological specimens in liquid ethane. Joachim Frank (HHMI/Columbia) developed single-particle averaging through the 1970s and 1990s. Richard Henderson (MRC-LMB) determined the first membrane-protein structure by electron crystallography (bacteriorhodopsin, 1990). The 'resolution revolution' began around 2012–2013 with direct electron detectors (Gatan K2) enabling routine near-atomic structures. Henderson, Frank, and Dubochet shared the 2017 Nobel Prize in Chemistry.
Sources
- Nobel Prize Chemistry 2017 — advanced background (https://www.nobelprize.org/uploads/2018/06/advanced-chemistryprize2017.pdf)
- Nobel Prize 2017 — popular explanation (https://www.nobelprize.org/prizes/chemistry/2017/popular-information/)
- PNAS (2017) — profile of Nobel laureates (https://www.pnas.org/doi/10.1073/pnas.1718898114)
- Journal of Structural Biology X (2020) — sub-2 Å resolution (https://pmc.ncbi.nlm.nih.gov/articles/PMC7337053/)
- Nature Communications (2025) — foam-film vitrification (https://www.nature.com/articles/s41467-025-61270-7)