Energies of interaction between portions of rhodopsin averaged over the course of the simulation. For the simulations performed in triplicate (Y7.53F, Y7.53N, F7.60A, and Y7.53F/F7.60A), the results represent averages over all of the relevant simulations. The interaction energies were calculated between the following sets of residues: (A) TM7 (7.49−7.56) and HX8 (7.60−7.70), (B) residues at positions 7.53 and 7.60, (C) TM7 (7.49−7.56) and TM2 (2.40−2.50), (D) residues at positions 7.53 and N2.40, and (E) residues at positions 7.53 and L2.43. (F) The color panel shows a representative snapshot of the cytoplasmic ends of TM7 and TM2 from the WT simulation compared to the same region in the β₂-AR structure (yellow). The relative positions of residues Y7.53, N2.40, and L2.43 (I in the β₂-AR) in the two structures suggest a weakened interaction in the β₂-AR structure. Cα atoms corresponding to positions 7.50−7.57 and 2.40−2.43 were used to fit the structures onto each other.

(A) Average structure of rhodopsin from the WT simulation (0−20 ns) with arrows indicating the positive direction of motion proportional to the mean square fluctuation of each Cα atom along the reference eigenvector (the first principal eigenvector derived from the WT simulation) for the TM7−HX8 segment. (B) TM7−HX8 segments from the photoactivated rhodopsin intermediate (yellow) and the β₂-AR (red) fitted onto the average structure of inactive rhodopsin from the WT simulation (0−20 ns). Cα atoms for the sequence from position 7.50 to 7.57, and from position 2.40 to 2.43, were used to fit the structures onto each other. Note that these segments lie along the positive direction of the reference eigenvector. (C) Projections of the TM7–HX8 segment from each simulation onto the reference eigenvector, which is the first principal eigenvector derived from the WT simulation. The projections were measured and binned starting from t = 5 ns and represent the total distribution from all corresponding trajectories for each construct of rhodopsin (see Table 1). These distributions are plotted as the fold change relative to the mean square fluctuation of the WT simulation along the reference eigenvector.