University of Chicago
As an undergraduate, Laura Watkins got her hands dirty and started down the path that would lead her to computational chemistry.
While a freshman at Washington University in St. Louis, a research course led her to study bacteriophages, viruses that infect bacteria, from soil samples. She isolated DNA from these tiny phages and used bioinformatics to analyze the resulting genetic information. To dive deeper into that project, she took an introductory computer science course. “That’s where I learned that I really liked computers and programming,” says Watkins, now a Department of Energy Computational Science Graduate Fellowship (DOE CSGF) recipient studying at the University of Chicago.
A medicinal chemistry course during her junior year taught Watkins more about the extensive experimental work involved in developing new drugs. She concluded “it would be so much easier if we could do all of this with computers.”
She carried out undergraduate research in computational chemistry with Jay Ponder. Watkins studied peptides, small protein-like snippets made of amino acids, and how flexible charge distributions were necessary to correctly model certain types of interactions in these molecules. To study this flexibility, known as polarizability, she used molecular dynamics (MD) computational methods that approximate atoms as charged particles.
At Chicago, Watkins works with computational chemist Gregory Voth to examine how protein channels and transporters work. These large, complex molecules bridge cellular membranes and shuttle protons – minuscule, positively charged hydrogen ions – from one side to the other. Many important proteins, including M2 in the influenza virus, have this function, but researchers have struggled to understand exactly how these molecules carry out this critical task.
Following the movement of protons within large proteins presents a computational challenge. Most molecular dynamics methods require fixing bonds in place, Watkins says, but following proton movement requires a more flexible approach. She’s using specialized reactive MD methods, which Voth’s group developed.
Reactive MD allows the proton to hop between water molecules, so that Watkins can track its movement. “What is it doing within the protein? What is it interacting with?” she asks. “And how does this relate to its overall function?”
Influenza’s M2 protein helps the virus infect cells. The body’s cells encase the virus in a membrane called an endosome. As protons flow into the virus, the pH of its contents drops, spurring it to escape from the endosome and infect the surrounding cell.
Watkins’ research has examined exactly how the protons interact with the protein and how they alter the hydrogen bonding network of surrounding water molecules. Besides improving scientists’ fundamental understanding of proton transport, her work could eventually help develop new flu drugs. Although a couple of existing medications, amantadine and rimantadine, target M2, some flu virus strains are now resistant, Watkins says. “Any better understanding that we can get could help them come up with something better.”
Watkins plans to use X-ray crystal structures of M2 bound to amantadine and rimantadine from collaborators at the University of California, San Francisco, to model how the drugs affect proton transport. Those insights could help her better understand exactly why drug molecules bind where they do and how that’s related to proton movement in the drug-free protein.
During her 2017 Los Alamos National Laboratory practicum, Watkins also worked on MD simulations with chemist Art Voter. Voter has developed a group of methods called Accelerated MD that have worked well for understanding the dynamics within solids and other hard materials. Watkins focused on adapting these methods to make them useful for softer, more flexible materials such as proteins. She incorporated machine-learning techniques to help the algorithms work on this problem more effectively.
Watkins expects to finish her Ph.D. in 2020 and is considering two career paths: pursuing data science research or working on science’s business side. “I really like understanding and communicating science,” she says.
The CSGF helped Watkins learn the language of many different computational scientists. “Everything that I do is based in statistics,” she says, but chemists avoid talking about their work in those terms. “So getting to take a statistics class and learn the language and learn how they talk about similar problems was cool. It gave me a greater depth of understanding of the research that I do.”
Image caption: Water molecules form hydrogen bonds (shown as green and blue arrows) within a protein channel, like this one from the influenza protein M2 (left). As a proton bound to water (green balls) approaches (center) and moves through the channel (right), the hydrogen bonds rearrange in response. Credit: Laura Watkins.