Project
Predicting Binding Affinity and Kinetics in GPCR Oligomers: Insights from Enhanced Sampling Simulations
Our work investigates how dopamine D2 receptors, proteins that transmit signals in the brain, change their behavior when they form pairs instead of acting alone. A single receptor is called a monomer, while a pair is called a dimer, and larger assemblies are known as oligomers. These different forms can alter how receptors function and how they interact with drugs. The D2 receptor is central to many brain processes and is strongly linked to conditions such as schizophrenia. While existing medicines already target this receptor, it is not yet clear how drug action is influenced when receptors are in dimers rather than monomers. Understanding this could guide the development of more effective therapies with fewer side effects. Studying these processes experimentally is very challenging, since receptors are embedded in the cell membrane and interact in highly dynamic ways. To address this, we use computer simulations to model receptors and follow how drugs bind at atomic resolution. These simulations must capture hundreds of thousands of atoms and reproduce events spanning very long time scales. Running such large-scale and long-timescale calculations is only possible with High-Performance Computing (HPC), which provides the power needed to complete simulations that would otherwise be impossible on standard computers.
Project Details
Project term
September 4, 2024–September 3, 2025
Affiliations
Forschungszentrum Jülich
Institute
Computational Biomedicine Institute
Principal Investigator
Methods
We used molecular dynamics (MD) simulations to model the dopamine D2 receptor in both its single (monomeric) and paired (dimeric) forms, and to study how the antipsychotic drug risperidone binds to these receptors. In MD, the motion of all atoms in the system is computed by solving Newton’s equations of motion, with interactions between atoms described by an empirical force field. This approach provides an atomistic description of the receptor–drug system in a realistic membrane environment. To explore the binding pathways and calculate the strength of drug binding, we employed well-tempered metadynamics (WT-MetaD). This enhanced sampling method adds a small, history-dependent bias to carefully chosen variables, allowing the system to overcome energy barriers and enabling reconstruction of the free energy landscape. To investigate the kinetics of drug binding and unbinding, we used infrequent metadynamics (I-MetaD). By introducing bias at long time intervals, this method avoids disturbing the transition state region and makes it possible to recover unbiased transition times. From these simulations, we estimated the residence time of the drug and the rate constants associated with binding and unbinding events.
Results
We first investigated how risperidone binds to the dopamine D2 receptor in its monomeric form. Well-Tempered Metadynamics (WT-MetaD) simulations revealed five intermediate states along the binding pathway. Two of these correspond to stable binding poses within the orthosteric site, while the others represent transient intermediates during binding and unbinding. The most stable pose closely matched the experimentally determined crystallographic structure, and the calculated binding free energy (ΔG = −11.06 kcal/mol) agreed well with experimental data (ΔG = −11.78 kcal/mol). In the dimeric receptor, risperidone binding to one subunit (DA) followed a similar multi-step pathway but displayed important differences. The orthosteric pocket in the dimer was more expanded, allowing the drug to adopt a deeper binding pose, stabilized by interactions not present in the monomer. This resulted in a stronger binding affinity, with a calculated ΔG of −13.38 kcal/mol. By contrast, simulations on the second subunit (DB) failed to converge, suggesting weaker or less stable binding. Overall, these results indicate that dimerization enhances risperidone binding in one subunit while destabilizing it in the other, reflecting an asymmetric influence of receptor pairing. To quantify the impact of dimerization on binding kinetics, we used Infrequent Metadynamics (I-MetaD) to estimate risperidone’s residence time. In the monomer, the drug remained bound for an average of ~18 minutes, corresponding to a dissociation rate constant (koff) of 0.05 min⁻¹. While this is shorter than the experimentally reported value of 233 minutes, the result falls within the expected accuracy of the method.In the dimeric receptor, binding proved highly asymmetric. In subunit A, risperidone displayed an exceptionally long residence time of ~4,818 minutes (koff = 0.0002 min⁻¹), consistent with its deeper insertion into the binding pocket. In subunit B, however, binding was much weaker and short-lived, with a residence time of only ~0.83 minutes (koff = 1.20 min⁻¹). This striking difference shows that dimerization can either stabilize or destabilize ligand binding depending on the subunit.
Discussion
Together, our results show that dimerization profoundly alters both the thermodynamic and kinetic features of risperidone binding. In one subunit of the dimer, binding becomes markedly stronger and more stable than in the monomer, whereas in the other subunit, it is weaker and highly transient. This asymmetry reveals how receptor dimers can generate distinct drug-binding environments compared to isolated receptors. Such effects may help explain why drug behavior in cellular contexts often diverges from simplified experimental models and highlight the importance of explicitly accounting for receptor dimers in drug design. Looking forward, extending this approach to other brain receptors will help determine whether similar asymmetries arise upon dimerization. This could provide broader insights into receptor cooperativity in the brain and inform the development of more effective and selective treatments for psychiatric and neurological disorders.
Additional Project Information
DFG classification: 201-07 Bioinformatics and Theoretical Biology
Software: PLUMED, Gromacs, VMD
Cluster: CLAIX