The drug interleukin-2 does a great job of goosing a patient’s immune response to certain kidney and melanoma cancers. In fact, it can crank the immune system so high that it triggers life-threatening toxic conditions.
But now researchers with the University of Washington’s Institute for Protein Design have used AI to engineer a protein that can quickly shut off interleukin-2, as just one example of the technology’s potential applications.
The scientists describe their work in a paper being published in the peer-reviewed journal Nature.
“One way to control a medicine is through dose. We’ve added a second lever by designing molecules that can be switched off rapidly, even after they’ve taken full effect,” said David Baker, senior author of the study and a 2024 Nobel Laureate for his work in computational protein design.
The engineered protein works by inserting itself into a complex of proteins that are bound together, changing their shape and causing them to fall apart. In the example pursued by the UW scientists, the novel protein targets a complex that includes interleukin-2 that’s tightly bound to an immune cell receptor, causing it to quickly release.
With one of these engineered proteins, “you could much more precisely control the duration of [immune response] signaling and also immediately shut down any immune activation, even after it’s already started,” said Adam Broerman, lead author of the Nature paper and a graduate student in Baker’s lab.
The interleukin-2 experiments were performed in human cells in the lab, demonstrating the strategy works. Going forward, the tool could be used to develop therapies for cancer and other diseases. It could also be used in studies that seek to understand and manipulate cellular processes by rapidly turning protein interactions off and on.
In his research, Broerman tried a variety of engineered proteins, tweaking different components to optimize the impact.
Earlier experiments showed modest improvements to the speed at which the proteins broke apart. When Broerman landed on the winning solution, the de novo protein worked so fast he had to recalibrate his measuring devices to clock the disassociation.
“That was probably one of the most fun weeks of my life, when this first started working,” Broerman said.
The study was led by scientists at the Institute for Protein Design and included researchers from the UW’s Department of Chemistry, Osnabrück University, Stanford University and Oregon Health & Science University.
Additional authors of the article titled “Design of facilitated dissociation enables timing of cytokine signalling” include Christoph Pollmann, Yang Zhao, Mauriz A. Lichtenstein, Mark D. Jackson, Maxx H. Tessmer, Won Hee Ryu, Masato Ogishi, Mohamad H. Abedi, Danny D. Sahtoe, Aza Allen, Alex Kang, Joshmyn De La Cruz, Evans Brackenbrough, Banumathi Sankaran, Asim K. Bera, Daniel M. Zuckerman, Stefan Stoll, K. Christopher Garcia, Florian Praetorius and Jacob Piehler.
