New method could lead to safer medicines
Researchers at the University of Leipzig have developed a method to better understand how drugs work in our cells. Artificial intelligence is helping them in this endeavor.
Many people take medication for high blood pressure or hay fever. These often act on tiny switching points in our cells, known as receptors. They are located on the cell surface, receive signals such as hormones, and relay them into the cell. About one-third of all approved medications work through such receptors.
Researchers at the University of Leipzig have now developed a method that allows them to reconstruct previously hidden forms of these proteins. This is important because a receptor can trigger different signals. An active ingredient can thus produce desired effects, but it can also cause side effects.
“Because every receptor is involved in multiple signaling pathways, we need to understand the structural mechanisms and effects of drug binding in great detail,” explains Matthias Elgeti from the Institute for Drug Development at the Faculty of Medicine.
Three methods for a more accurate picture
Receptors are not rigid. They constantly change their spatial conformation. Some states last only very briefly, while others occur rarely. This makes them difficult to characterize using conventional analytical methods.
To address this, Elgeti and doctoral student Julia Belyaeva combined three approaches. Using a specialized measurement method, they determined distances within the receptor. Artificial intelligence calculated possible spatial structures. Computer simulations then helped verify which models matched the measurement data. “The artificial intelligence generated thousands of possible structural models. The simulations tested their stability, and the measurement data served as a control,” explains Belyaeva.
The two researchers tested the method on the angiotensin II receptor. It plays an important role in regulating blood pressure. In doing so, they investigated how a specific active ingredient influences the shape of the receptor and what bonds form within the molecule. This allowed them to reconstruct structural states that had previously been difficult to access.
Active compounds should act in a more targeted manner
This is of interest for drug development because different forms of the same receptor can favor different signaling pathways. In the future, active ingredients could be developed to trigger desired reactions as precisely as possible while avoiding undesired ones.
However, this is still basic research. The Leipzig scientists have neither developed a new drug nor tested its effects in humans. Their results initially provide more accurate models of the processes at the molecular level.
According to the researchers, the new method requires less computing power than previous approaches. It should also be transferable to other proteins. "The structural models generated are publicly available and serve as a starting point for the computer-aided search for new active compounds. “We’re currently working on that,” says Elgeti. The research group now plans to test its method on additional receptors.
Original publication:
Julia Belyaeva, Matthias Elgeti, “Reconstructing hidden GPCR conformations and ligand binding modes through integrative DEER modeling,” Structure,
2026.