(A) Occurrence of different modes of Mg2+-ATP and Ca2+-ATP interaction in non-redundant high-resolution crystal structures available in the Protein Data Bank. Each mode is indicated by the phosphate groups in ATP that coordinate the cation. Analogous data is shown for complexes of Mg2+ with the non-hydrolyzable analog AMP-PNP. (B) A sample of the simulated trajectories of the P2X3-ATP-Mg2+ complex produced with the solute-tempering protocol (S = 0.39). Trajectories showing spontaneous ATP dissociation from P2X3 (and Mg2+) are those in which the distance between the center of ATP and the center of the binding site in the receptor exceed 12 Å. Additional trajectories are depicted in Figure 2—figure supplement 1. (C) Comparison of P2X3-ATP-Mg2+ and P2X3-Mg2+ following ATP dissociation. For P2X3-ATP-Mg2+, 3D density maps calculated from the simulation data (see Materials and methods) are shown for ATP (orange), the γ-phosphate (gray surface), Mg2+ (magenta) and the surrounding carboxyl groups of acidic residues (cyan). Analogous maps are shown for ion and sidechains (red and blue, respectively), following ATP dissociation. The maps shown are averages of the data for the three protein subunits. The contour levels used for Mg2+ are the same in both cases; likewise for the sidechains. The protein conformation (gray cartoons) is a snapshot of the ATP-bound state. (D) Same as panel C, for Mg2+ and ATP in the ATP-bound state. For the ATP-free state, individual Mg2+ density maps are shown for each binding site in the trimer (red, green, purple), to demonstrate the observed displacement is reproducible. The protein snapshot is that in panel C, now represented as a surface, with adjacent protein subunits in light and dark gray. (E) Same as panel C, for the ATP-bound state, now showing a randomly selected configuration of Mg2+, ATP and the four acidic sidechains that represents the density. Non-polar hydrogen atoms and other components of the simulation system are omitted for clarity. (F) Same as panel E, for the ATP-dissociated state, in a slightly different view, for clarity. (G) Impact of Mg2+ on the nature of the interaction between ATP and P2X3. The plot quantifies the number of ATP dissociation events expected to occur in a 50-ns time-window, deduced from our simulation data, for each value of the scaling factor S, and in the presence or absence of Mg2+. Note that one ATP molecule is bound to each of the three protein subunits, and hence the maximum number of dissociation events is also 3. Each data point is an average of N = 20 independent simulations. Error bars denote the S.E.M. The data was fitted to a sigmoidal function f = 3/(1 + exp [m (S - n)]. The resulting parameters (C.C. > 99%) are: with Mg2+, m = 58.6, n = 0.42; without Mg2+, m = 29.1, n = 0.51.