By Ashley Yeager

"Physicists want to do what to me?" (Courtesy SXC.HU)

The spherical cow is a long-standing physics joke, funny to us, but maybe not to her.  (Courtesy Marijn van Braak)

It may be impossible to tip a spherical cow. But by searing one on the computer, scientists may have found the best recipe yet for transforming proteins into crystals.

Crystallizing a protein is not as simple as throwing it in the freezer and letting it chill, the way liquid water turns to ice. “We would like to know what is going on at that fundamental level when we crystallize a protein. But we don’t,” said Patrick Charbonneau, an assistant professor of chemistry, physics and computational biology at Duke.

But by looking at each protein as if it were a spherical cow with sticky patches, Charbonneau said scientists could learn more about the standard set of interactions that transform proteins from a liquid solution into crystals.

The idea of a spherical cow is not new. It comes from an old physics joke where a farmer calls the local university for advice on getting his cows to produce more milk. The university sends a physicist who looks at the barn, the cows and their production and says, “First, assume the cows are spherical and in a vacuum . . . ”

“Physicists idealize everything to the extreme, which could be why their most recent attempts to explain protein crystallization don’t quite work,” Charbonneau said.

Scientists want to know the fundamentals of protein crystallization because proteins are essential for living creatures to survive. Proteins come in many different shapes and have a variety of functions, and their structures explain how they interact with other molecules in the human body — information that often leads to more targeted medications.

To study protein shape, scientists precipitate them from liquid solutions into crystals, which is how DNA’s double helix structure was revealed. Physicists have also tried to model protein shape through computer simulations. In Charbonneau’s new model, he combines physicists’ past simulations based on the spherical cow model with the best protein-crystal-making insights from structural biology.

The new hybrid is what Charbonneau calls the patchy cow model. To develop it, he and his students digitally recreated the structure of the protein rubredoxin as a spherical cow, making it resemble a basic blob-like molecule. The team then applied patchy areas, which acted like the atomic interactions shaping the structure of a protein.

“We found the model actually works, giving us a patchy picture that does predict some of the crystal structure of proteins,” Charbonneau said. It is the first time a simulated model of protein crystallization has matched well with the temperature, salinity and other crystal-forming conditions structural biologists observe in the lab. And “it’s the first time that we can draw clear physical insights from how the model compares with experiments,” he added.

A single protein crystal of lysozyme (Courtesy Wikimedia)

A single protein crystal of lysozyme (Courtesy Wikimedia)

Charbonneau first began working on the protein crystallization problem as a post-doc at Amolf in the Netherlands. When he came to Duke in 2008, he met David and Jane Richardson — leaders in the field of protein crystallization and structural biology. From them, Charbonneau began to realize that physicists and structural biologists don’t know each other’s work. “We don’t cite each other, and we don’t trust each others’ descriptions of protein crystallization,” he said. The goal of his work, he explained, is to convince the disciplines to begin talking to each other.

The biggest problem for both camps is that they know the atomic and molecular interactions that should be involved in crystallizing a protein, but they “cannot make sense of what is going on. It’s figuring out the mechanism, the physical process, that is frustrating,” Charbonneau said. His new model tries to paint the physicists’ spherical cows from the perspective of the observations the Richardsons and others have already made in the crystallography community.

Charbonneau anticipates that the patchy spherical cow model might help structural biologists predict what conditions they need to crystallize any protein of their choice. But first, he said, the team needs to put their model to another test, simulating crystallization in more complex proteins, such as physicists’ favorite, lysozyme.

Citation: “Characterizing protein crystal contacts and their role in crystallization: rubredoxin as a case study,” Diana Fusco, Jeffrey Headd, Alfonso De Simone, Jun Wang, and Patrick Charbonneau. Soft Matter, November 8, 2013. 10.1039/C3SM52175C.