0%

Experimental methods

Experimental models are built directly from physical data collected from the macromolecule itself using sophisticated experimental techniques. These techniques provide raw data that is then processed computationally to produce a 3D model. Macromolecular models deposited in the Protein Data Bank (wwPDB) originate from diverse methodologies, each with its own characteristics, strengths, and limitations.

The three major experimental techniques used to determine macromolecular structures are:

X-ray crystallography

This technique requires growing highly ordered crystals of the macromolecule. When X-rays are shone on the crystals, they scatter (diffract) off the electrons of the atoms in a specific pattern, known as a diffraction pattern. By analysing this diffraction pattern using complex computational methods, researchers can reconstruct an electron density map of the molecule within the crystal. This 3D map shows where electrons are concentrated, allowing researchers to build a 3D model by placing atoms (representing amino acids or nucleotides) into the density. The process is iterative: a preliminary model helps generate a map, which is then used to improve the model, and so on, until no further significant improvements can be made.

X-ray crystallography can often provide very high-resolution models, showing atomic details, but successfully obtaining well-diffracting crystals can be challenging for some macromolecules.

Electron density map for a tyrosine residue (PDB ID: 3HYD)

The blue mesh represents the 2mFo-DFc electron density, which ideally surrounds all atoms of the fitted model, confirming their placement based on experimental data.

Cryo-Electron Microscopy (cryo-EM)

In cryo-EM, samples are rapidly frozen in a thin layer of ice and imaged using an electron microscope. 2D images of many individual particles or regions are captured and computationally processed to reconstruct a Coulomb potential map (often referred to simply as an “EM map”). This map is then used to build the atomic model. This method is particularly powerful for determining the structures of large protein complexes, viruses, and membrane proteins that are difficult to crystallise (Quentin, D., Raunser, S., 2018).

There are subtypes of cryo-EM:

  • Averaging techniques: These include Subtomogram Averaging (STA), Helical Reconstruction (HR), and, finally, the most commonly known, Single Particle Analysis (SPA), which focuses on obtaining high-resolution structures of purified, isolated macromolecules or complexes by imaging and averaging many thousands of identical particles (Wu M, Lander GC., 2020).
  • Cryo-Electron Tomography (cryo-ET) is a technique used to visualise large fields of view in three dimensions (Francis J. O’Reilly et al., 2020). It is particularly useful for studying membrane proteins and their interactions in their normal environment (in situ), providing insights into their function and organisation within cells.

Cryo-EM sample preparation. Grid setup. In cryo-electron microscopy, the protein sample is applied as a thin layer onto a specialised grid, typically made of copper or gold, with a perforated carbon film. This grid, shown here with the protein sample distributed across its holes, is then rapidly plunge-cooled into liquid ethane. This vitrification process ensures that the water surrounding the macromolecules turns into amorphous ice, preventing the formation of ice crystals that would damage the sample and obscure the structural details. The individual protein particles are then imaged in these vitrified ice layers.

Through improvements to sample preparation, instrumentation and software solutions, the field saw a “resolution revolution” which was awarded the Nobel Prize for Chemistry in 2017. The technique now routinely solves structures to resolutions that allow for confident building of model coordinates. 

Nuclear Magnetic Resonance (NMR) Spectroscopy

This technique uses strong magnetic fields and radio waves to probe the local environment of atomic nuclei within the macromolecule. By analysing the NMR spectra, structural information, such as distances and rotation angles between atoms, can be determined. These measured distances and angles serve as restraints that guide the calculation of a set (an ensemble) of possible 3D structure models that satisfy these restraints.

NMR is particularly useful for smaller proteins (25-30 kDa) and provides information about dynamics and flexibility. Structures can be determined using both solution NMR (molecules freely moving in liquid) and solid-state NMR (molecules in a solid or semi-solid state).

An NMR structural ensemble (PDB ID: 5KGQ)

Unlike X-ray crystallography or cryo-EM, which typically yield a single static model, Nuclear Magnetic Resonance (NMR) spectroscopy often produces an ensemble of possible 3D structure models, reflecting the inherent flexibility and dynamic nature of molecules in solution. In this Mol* viewer representation, you can click through the “Model 1 / 20” navigation in the top left corner to visualise the slightly varying conformations within the calculated ensemble.