G-protein coupled receptors (GPCRs) regulate various aspects of cellular behavior and represent actionable targets for drug discovery. Activation of specific signaling pathways downstream of G-protein coupled receptors (GPCRs) or targeting the receptors at selected cellular locations has the potential to provide therapeutic actions with fewer side effects. However, the understanding of the molecular mechanisms underlying GPCR activation is limited, hampering drug discovery efforts towards selective ligands. Proximity biotin labeling based on an engineered ascorbic acid peroxidase (APEX) combined with quantitative mass spectrometry is a powerful method to delineate these mechanisms given its capacity to simultaneously capture proximal protein interaction networks and the cellular location of the receptor. However, a major challenge is to extract the various information from these complex datasets. Here, we describe a computational framework for proximity labeling datasets which predicts ligand-dependent subcellular location of GPCRs and quantitatively deconvolutes the effect of receptor location and proximal interactors. We applied this approach to the mu-opioid receptor and not only monitored distinct effects of ligands on receptor trafficking, but also discovered two novel regulators, EYA4 and KCTD12, which modulate MOR-driven G protein-dependent signaling.

We stably expressed the mu opioid receptor-APEX fusion construct in HEK293 cells, which retained receptor function with regard to signaling, internalization, and recycling. To perform proximity labeling, the cells were pretreated with biotin-phenol followed by activation of MOR using the DAMGO, morphine, or PZM21 at a concentration of 10µM over a time course of up to 60 minutes. At selected time points after receptor activation, we initiated proximal biotin labeling by addition of hydrogen peroxide (H2O2) for thirty seconds, followed by quenching of the reaction, cell lysis, and enrichment of biotinylated proteins using streptavidin. To quantify relative abundance changes of biotin-labeled proteins after agonist stimulation, we utilized a combination of quantitative proteomics approaches encompassing unbiased, shotgun proteomics as well as targeted proteomics based on selected reaction monitoring (SRM).