Diwakar Shukla, a professor of chemical and biomolecular engineering at the University of Illinois Urbana Champaign, is using large scale molecular simulation to answer a question that has challenged drug designers for years: why certain synthetic cannabinoids produce severe side effects while chemically similar compounds do not. In a recent study, Shukla and his colleagues combined deep learning with advanced simulation methods to examine how new psychoactive substances interact with cannabinoid receptors in the human brain, revealing molecular mechanisms that could guide the design of safer cannabinoid based drugs.
Dutta, S., & Shukla, D. (2025). Characterization of binding kinetics and intracellular signaling of new psychoactive substances targeting cannabinoid receptor using transition-based reweighting method. ELife, 13. https://doi.org/10.7554/eLife.98798.3
Many new psychoactive substances were originally synthesized as potential pain treatments but were abandoned after producing dangerous psychological and physiological effects. Despite this, they continue to appear in illicit markets under names such as Fubinaca, Chimica, and Pinaca. These compounds are difficult to regulate and detect because their chemical formulas change frequently. From a scientific perspective, they also present a challenge because they bind very tightly to cannabinoid receptors and disengage slowly, making their behavior hard to observe in both laboratory experiments and conventional computer simulations.
Diwakar Shukla, a professor of chemical and biomolecular engineering at the University of Illinois Urbana Champaign stated,
“New psychoactive substances bind very strongly to cannabinoid receptors in the brain and are slow to unbind, making them difficult to observe and simulate in standard laboratory or computer experiments. It can take a huge amount of computer time to see these rare binding and unbinding events.”
The Illinois team focused on how these synthetic compounds differ from classical cannabinoids at the molecular level. Cannabinoid receptors can activate multiple signaling pathways inside cells. Classical cannabinoids tend to favor pathways associated with therapeutic effects, while many synthetic cannabinoids preferentially activate a different route known as the beta arrestin pathway. Previous studies have linked this pathway to stronger and often more harmful neurological responses. Understanding what drives this signaling bias is critical for separating desired effects from unwanted ones.
To investigate these differences, graduate researcher Soumajit Dutta applied a simulation framework known as the Transition Based Reweighting Method. This approach is designed to capture rare and slow molecular events, such as the binding and unbinding of drug molecules from receptors, without requiring impractical amounts of computing time. Using this method, the team was able to estimate both the strength of binding and the rate at which different synthetic cannabinoids disengage from the receptor, parameters that are closely tied to their biological effects.
The researchers also relied on the Folding at Home distributed computing platform, which allows simulations to be run in parallel across millions of volunteer computers worldwide. By combining results from many simulations and using algorithms to guide subsequent calculations, the team was able to observe receptor interactions that would normally be inaccessible to a single laboratory or computing cluster. This strategy made it possible to map out how subtle structural features influence receptor behavior and downstream signaling.
The simulations showed that many new psychoactive substances not only bind more strongly than classical cannabinoids but also stabilize receptor conformations that favor beta arrestin signaling. This helps explain why these compounds are associated with more severe side effects, including anxiety, hallucinations, and cardiovascular complications. Importantly, the results also suggest specific design strategies for reducing risk, such as creating molecules that bind less tightly or that unbind more readily from the receptor.
Beyond cannabinoid research, the study highlights how modern computational tools are reshaping drug discovery. By combining machine learning, enhanced sampling methods, and distributed computing, researchers can now explore biological processes that were previously too slow or complex to simulate. For engineers and scientists working at the interface of computation and medicine, this work provides a clear example of how simulation can move from explanation toward prediction.
Shukla and his colleagues emphasize that the goal is not to rehabilitate harmful substances, but to use what is learned from them to inform better drug design. By identifying how molecular structure controls receptor signaling, the study points toward cannabinoid based therapeutics that could retain medical benefits while avoiding pathways linked to adverse effects. As computational power and modeling techniques continue to improve, approaches like this are likely to play an increasingly central role in developing safer and more targeted pharmaceuticals.
Adrian graduated with a Masters Degree (1st Class Honours) in Chemical Engineering from Chester University along with Harris. His master’s research aimed to develop a standardadised clean water oxygenation transfer procedure to test bubble diffusers that are currently used in the wastewater industry commercial market. He has also undergone placments in both US and China primarely focused within the R&D department and is an associate member of the Institute of Chemical Engineers (IChemE).
