Cryo-electron microscopy (cryo-EM) determines the structures of biological macromolecules by imaging individual particles flash-frozen in vitreous (non-crystalline) ice using an electron microscope. Unlike X-ray crystallography, cryo-EM does not require crystals — purified protein in solution is applied to a grid, blotted to a thin film, and rapidly plunged into liquid ethane to trap molecules in their native, hydrated state. The "resolution revolution" (enabled by direct electron detectors and improved image processing algorithms since ~2013) has transformed cryo-EM from a low-resolution technique into a method capable of near-atomic resolution (2-4 Angstroms) for many biological complexes, earning Jacques Dubochet, Joachim Frank, and Richard Henderson the 2017 Nobel Prize.
For most of structural biology's history, determining a protein structure meant growing crystals. Cryo-EM has changed this fundamental constraint. By imaging individual protein molecules frozen in a thin layer of vitreous ice, cryo-EM determines structures without crystals — directly from purified protein in solution. This eliminates the crystallization bottleneck that has frustrated structural biologists for decades and opens the door to structures of flexible complexes, heterogeneous samples, and membrane proteins in lipid environments that resist crystallization.
The specimen preparation is conceptually simple but technically demanding. A few microliters of purified protein (at 0.5-5 mg/mL) are applied to a thin carbon or gold grid with tiny holes. Most of the solution is blotted away, leaving a thin film (~30-100 nm) of protein solution spanning the holes. The grid is then plunge-frozen into liquid ethane, cooling it so rapidly that water vitrifies (forms amorphous glass) rather than crystallizing. The frozen grid is kept at liquid nitrogen temperature throughout imaging to prevent ice crystallization and to protect the radiation-sensitive specimen.
In the electron microscope, the frozen specimen is imaged at low electron dose (to minimize radiation damage) using a 200-300 kV electron beam. Each image captures thousands of individual protein particles in random orientations, frozen mid-tumble. The images are noisy — each particle is imaged with very few electrons to limit damage — but computational processing extracts the signal. Image processing (the intellectual contribution of Joachim Frank) involves: identifying and extracting individual particle images from the micrographs, classifying them by orientation and conformation (2D classification), determining the 3D orientation of each particle (orientation determination), and averaging many particles in the same orientation to produce a high-signal 3D reconstruction. Hundreds of thousands to millions of particle images are typically needed for a near-atomic resolution reconstruction.
The resolution revolution since ~2013 transformed cryo-EM from a niche technique producing blobby shapes into a mainstream structural method producing maps at 2-4 Angstrom resolution — sometimes rivaling crystallography. Three technological advances drove this: direct electron detectors (higher sensitivity, faster readout enabling motion correction), improved algorithms (maximum-likelihood approaches, GPU-accelerated processing), and better specimen preparation (thinner ice, better grids). Cryo-EM now accounts for the majority of new high-resolution structures of large complexes deposited in the PDB. Its advantages over crystallography — no crystals needed, ability to handle conformational heterogeneity (sorting particles into different conformational classes), and visualization of complexes in near-native conditions — make it complementary to crystallography and, for many targets, the method of first choice.