Fig. 1|. Removal of the C tail of arrestin leads arrestin to fluctuate between active and inactive states; the receptor core and R_P tail each independently stabilize the active state.

a, Inactive-state (left; PDB 1CF1) and receptor-bound, active-state arrestin-1 (right; PDB 5W0P). b, Upon activation, the arrestin C domain twists with respect to the N domain. In simulations of arrestin-1 starting from its inactive state, the interdomain twist angle remained close to 0° (grey trace), while in simulations starting from its active state with rhodopsin bound, the twist angle remained close to 20° (blue trace). Thick traces represent a 50-ns sliding mean and thin traces represent unsmoothed values. c, In simulations of arrestin without its C tail and without a receptor, arrestin spontaneously transitioned between inactive and active conformations. Top, simulation of arrestin-1 with C tail removed, starting from inactive structure; bottom, arrestin-1 with C tail removed, starting from active structure. d, Distributions (histograms) of interdomain twist angles under different simulation conditions: grey, arrestin-1 with C tail (starting from inactive structure); blue, arrestin-1 bound to full-length rhodopsin (starting from active structure); green, arrestin-1 with C tail removed (starting from inactive structure); purple, arrestin-1 with C tail removed (starting from active structure); magenta, arrestin-1 bound to rhodopsin R_P tail (starting from active structure); yellow, arrestin-1 bound to rhodopsin core (starting from active structure). Removal of the arrestin C tail leads to an increased range of interdomain twist angles (the standard deviation increases; P < 0.001, two-sided t-test with six simulations per condition, see Methods). Binding of either part of the receptor in simulation substantially increases the fraction of time that arrestin spends in active conformations (histograms, yellow and magenta; P = 0.002 for core binding; P = 0.003 for R_P tail binding). Binding of the entire receptor has an even stronger effect (histogram, blue; P = 0.01 compared to both core-bound and to R_P-tail-bound conditions). Histograms are based on all production simulations. Traces are shown for representative trajectories; all are shown in Extended Data Fig. 1.