Using AI to break the opioid addiction cycle

When Ben Brown, research assistant professor of chemistry, thinks about the opioid epidemic, he views the problem on a molecular level. Painkillers used legitimately in medicine, such as oxycodone, are highly addictive, but better understanding of how their molecules interact with proteins in the body could lead to the formulation of nonaddictive alternatives, he said.

In May, the National Institute on Drug Abuse awarded Brown $1.5 million over five years to further his work in this area. Brown, faculty affiliate of the Vanderbilt Center for Addiction Research and the Center for Applied Artificial Intelligence in Protein Dynamics, is developing artificial intelligence that analyze billions of potential opioid drugs to reveal detailed insights into how they interact with key proteins. The remaining $875,000 of the grant will flow to Vanderbilt for indirect and administrative costs connected to Brown's research.

Brown will focus his research on Mu-opioid receptors, signaling proteins in the central nervous system that bind with opioids. These receptors modulate pain, stress, mood and other functions. Drugs that target these receptors are among the most powerful analgesics, but they also are the most addictive.

The grant, an Avenir Award in Chemistry and Pharmacology of Substance Use Disorders, is awarded by NIDA to early-stage investigators who propose highly innovative studies and represent the future of addiction science.

The energy and enthusiasm Ben brings to his science and scientific collaborations are outstanding, and it is fitting that he be recognized as a young pioneer in his field. Ben is one of the intellectual contributors behind the founding of the Center for Applied AI in Protein Dynamics. I anticipate that Ben will make fundamental advances in multiple core aspects of computer-aided drug design."

Hassane Mchaourab, director of the Center for Applied AI in Protein Dynamics and Louise B. McGavock Chair and professor of molecular physiology and biophysics

Brown's computational platform models drug-protein interactions in a way that accounts for their dynamic physical movements. These movements, called conformational changes, can occur in milliseconds and make a big difference in how a protein behaves and binds or interacts with a small molecule drug.

By computationally modeling this motion, algorithms can more effectively predict how tightly proteins and drugs will interact and the effects of this interplay. This information is used to screen billions of potential drugs-;an unprecedented scale for highly dynamic proteins-;or design new ones with properties that lead to fewer addictive side effects.

Computational platforms that model the structure of proteins and how they interact with drugs already exist, but they largely neglect conformational changes and poorly predict how a new drug will behave. That's due in part to the paucity of data available for training algorithms.

With data-rich material from researchers Craig Lindsley, Heidi Hamm and Vsevolod V. Gurevich from Vanderbilt, Matthias Elgeti of Leipzig University and Wu Beili of Shanghai Institute of Materia Medica, Brown will synthesize, functionally validate and structurally characterize drug molecules and receptors designed by the researchers. Following this component of the grant, Brown will feed the data back into the computational platform so it can be used as a starting point for more rounds of optimization-;a computational-experimental iterative feedback loop.

"You see pediatric patients have a surgery, and they're on an opioid postoperatively, and then they've got a problem after that. It's really sad," Brown said. "So the goal is to provide analgesia without risking addiction. And for those who have addiction, to provide new medications to help with recovery."

In addition to the Center for Applied AI in Protein Dynamics and VCAR, Brown's research affiliations include the Center for Structural Biology and the Vanderbilt Institute of Chemical Biology.