Cryogenic (frozen) protein structures are essential for understanding function and developing drugs. Scientists at St. Jude Children’s Research Hospital have created an algorithm to reveal that freezing proteins can create “artifacts,” errors that cause misleading results. The research recently appeared in Angewandte Chemie International Edition and highlighted the importance of water networks in protein-ligand interactions. The results challenge the common view that well-resolved cryogenic water positions are both precise and accurate.

Ligands are molecules that bind to a receptor protein. When a ligand binds to a protein, the conformation (shape) can change, initiating different types of activity in the cell. Protein-ligand binding and the resulting shape changes are crucial considerations during drug development efforts.

“If you only look at cryogenic data, the information used for drug discovery contains artifacts that you wouldn’t know were there,” said corresponding author Marcus Fischer, Ph.D., St. Jude. Departments of Chemical Biology and Therapeutics and Structural Biology. “We have developed a way to disentangle these artifacts. By using pairwise comparisons between cryogenic and ambient temperatures, you can identify which parts of the protein are affected by temperature.

Researchers often use available protein structures by extracting information from a database called the Research Collaboratory for Structural Bioinformatics Protein Data Bank. About 95% of these structures are cryogenically captured and then modeled in the database for ease of use. Drug discoverers rarely look closely at raw experimental data, which comes in the form of an electron density map. Interrogating maps rather than structural models provides an unbiased approach to revealing dynamic features and cryogenic artifacts.

Flipper Algorithm Highlights Important Changes

Fischer and his team have developed an algorithm, called Flipper, that looks at raw experimental data in electron density maps. Flipper identifies map spikes (signals) that would otherwise be invisible. These peaks correspond to the protein portions of specific residues that have temperature sensitive conformations. These residues can change the relative preference of one state over another, or “flip” in their density, moving between conformations, from which the algorithm gets its name.

The researchers used this approach to identify residues that respond to temperature changes and to track the residues in a barcode-like system across the entire protein. This allowed scientists to see how residues inside and outside the ligand binding site react to freezing or warming temperatures.

“With Flipper, we can detect small but important changes in protein structures based on temperature or other factors,” said first author Timothy Stachowski, Ph.D., St. Jude Chemical Biology and Therapeutics. “It is important to correct these details early in the drug discovery process; otherwise, research efforts could be misled.

Since the effects of temperature and water network influence a large number of structures, the results may have a widespread impact on drug development.

A new appreciation for water networks

Using their new approach, the researchers conducted a systematic analysis showing the importance of water networks. Water, one of the most crucial and abundant molecules on Earth, plays an active role in the conformation freezing process. This is especially true at protein-ligand binding sites.

“This is the first time that we have systematically shown the importance of temperature on water networks to modulate the ligand binding interface, where biology occurs,” said Fischer. “Water is often overlooked in the drug discovery process, but we have shown that in addition to having a profound effect on ligand binding, water also influences binding site residues, capturing them in positions that differ depending on the temperature.”

Flipper and the conformational barcoding system that facilitates comparisons of different ligands at different temperatures are freely available to allow other researchers to identify such patterns in their own datasets.