Improving ABFE Calculations for GPCRs by Addressing Errors in Apo-State Representation

Free energy perturbation (FEP) methods provide a robust computational framework for predicting protein-ligand binding affinities and have seen increased adoption in drug discovery workflows. Among these, absolute binding free energy (ABFE) calculations are especially valuable as they enable affinity prediction for novel compounds without prior experimental data. However, their broader adoption is limited by setup complexity, sampling challenges, and variable performance across systems. These challenges are particularly pronounced in therapeutically relevant complex membrane protein systems such as G protein-coupled receptors (GPCRs). This underscores the need to evaluate ABFE performance in GPCR systems and to develop strategies to mitigate these errors. Here, we evaluate ABFE calculations across three GPCR systems: the adenosine A2A receptor (A2AR), the dopamine D3 receptor (D3R), and the metabotropic glutamate receptor 5 (mGlu5). We focus on three sources of systematic error in these systems: (i) inadequate resolvation of the binding pocket in apo-like alchemical states, (ii) inconsistent treatment of metal ions, and (iii) insufficient sampling of receptor conformational transitions between holo-like and apo-like ensembles. To address these issues, we introduce two methodological improvements: a biased grand-canonical-like water transport scheme to improve binding pocket resolvation, and a replica-exchange umbrella sampling (REUS)-based framework to quantify and correct the free energy associated with receptor relaxation to apo-like conformations. Together, these approaches reduce the systematic overestimation of binding affinities, improving agreement with experimental data while maintaining ranking performance. Overall, our results highlight the importance of accurately representing the apo state in ABFE calculations and provide a practical framework for applying high-accuracy free energy methods to membrane protein targets.