Why X-Ray Protein Crystallography is Hot

A leading application of X-ray protein crystallography is its use to reveal how specific mutations are linked to disease. For example, using X-ray protein crystallography, researchers have been able to explain how the S13F mutation in the chloride channel cystic fibrosis transmembrane regulator (CFTR) disrupts binding between CFTR and filamin to cause cystic fibrosis¹.

X-ray protein crystallography is also widely used during drug development. A popular strategy involves using X-ray protein crystallography to structurally define the active site of an enzyme, allowing for improved drug targeting. More recently, it has become possible to screen whole fragment libraries, accelerating the fragment-to-drug process².

Other applications of X-ray protein crystallography include its use for studying protein-protein interactions, such as those between neutralizing antibodies and the SARS-CoV-2 receptor binding domain, and its implementation for characterizing PROTACs, the advanced protein degraders that have shown encouraging results against seemingly ‘undruggable’ targets ^3,4.

Notably, the first protein to have its structure determined by X-ray crystallography was myoglobin, which saw Max Perutz and John Kendrew awarded the Nobel Prize for Chemistry in 1962. More recently, X-ray crystallography was used to resolve the structure of the ribosome, leading to Ada Yonath, Venki Ramakrishnan and Thomas A. Steitz receiving the Nobel Prize for Chemistry in 2009.

X-ray protein crystallography begins with creating a solution that is supersaturated in the protein of interest, without fundamentally alter the protein’s natural state. This is achieved by adding mild precipitating agents, such as neutral salts or polymers, to the parent solution while simultaneously manipulating parameters such as temperature, pH, ionic strength, and the concentrations of various buffer additives⁵. Often, hundreds of different conditions must be compared to identify those which promote crystal nucleation and growth, making it common to use specialized liquid handling robots and pre-configured screens for these types of studies.

Sitting- or hanging-drop vapor diffusion are widely used crystallization screening methods. In the sitting drop method, a drop comprising a mixture of the parent solution and the precipitating reagent is placed on an elevated platform above a reservoir, which contains the precipitating reagent at a higher concentration than the drop. As the water evaporates from the drop toward the reservoir, the protein concentration is increased. The hanging drop method is similar, but the drop is suspended from a cover slip instead of resting on a platform. Other methods for promoting supersaturation include microbatch crystallization, microdialysis, and free-interface diffusion⁵.

Once protein crystals have been obtained, they are removed from the drop with a loop and plunged into liquid nitrogen, which serves to protect the crystals from radiation damage. X-ray data is then collected, either using a laboratory X-ray source or at a synchrotron, where very high intensity and highly focused X-rays are generated. During this step, each crystal is placed in a narrow beam of X-ray photons that are all traveling in the same direction. By measuring X-ray diffraction, which results from the photons interacting with the crystal, and applying complex mathematical formulae, it is possible to produce a three-dimensional map of the protein. Factors influencing the data collection time include the intensity of the X-ray source, the size of the crystal, and how well it diffracts.

AlphaFold, a state-of-the-art artificial intelligence (AI) system developed by Google DeepMind, promises to revolutionize the modeling of protein structures and interactions⁶. By predicting a protein’s 3D structure from its amino acid sequence, with an accuracy that rivals experimental methods, AlphaFold holds vast potential to accelerate protein research, including by revealing previously unsolvable structures. The latest version of AlphaFold was released in June 2024 and provides even greater accuracy than previous versions⁷.