The G protein-coupled cannabinoid receptor subtypes CB1 and CB2 have been cloned from several species. The CB1 receptor is highly conserved across species, whereas the CB2 receptor shows considerable cross-species variations. The two human receptors share only 44% overall identity, ranging from 35% to 82% in the transmembrane regions. Despite this structural disparity, the most potent cannabinoid agonists currently available are largely undiscriminating and are therefore unsatisfactory tools for investigating the architecture of ligand binding sites. However, the availability of two highly specific antagonists, SR 141716A for the CB1 receptor and SR 144528 for the CB2 receptor, has allowed us to adopt a systematic approach to defining their respective binding sites through the use of chimeric CB1 receptor/CB2 receptor constructs, coupled with site-directed mutagenesis. We identified the region encompassed by the fourth and fifth transmembrane helices as being critical for antagonist specificity. Both the wild type human receptors overexpressed in heterologous systems are autoactivated; SR 141716A and SR 144528 exhibit classical inverse agonist properties with their respective target receptors. In addition, through its interaction with the CB1 receptor SR 141716A blocks the Gi protein-mediated activation of mitogen-activated protein kinase stimulated by insulin or insulin-like growth factor I. An in-depth analysis of this discovery has led to a modified three-state model for the CB1 receptor, one of which implicates the SR 141716A-mediated sequestration of Gi proteins, with the result that the growth factor-stimulated intracellular pathways are effectively impeded.