Fu et al that there are 13 residues in the EL-2 loop of the D2 receptor, whole DI receptor and peak-minimum distance changes of whereas 26 residues exist in the EL-2 loop of the DI K-167-E-302(Fig 4 A)occurred three times during our 10- eptor. The main difference in EL-2 mainly comes from ns MD simulations. Quite similar open-closed conforma the upstream preceding conserved Cys_e2(Fig. S2), which is tional changes were also seen for the free D2(Fig. 3 B) far away from the active site crevice. In addition, there These dynamic properties suggest that there are at least two six residues longer in the EL-3 of the DI receptor than that of distinct conformations for both DI or D2, and the sponta the D2 receptor. Though the lengths of EL-2 and EL-3 are neous oscillation between these two conformational states different, they are flexible and function as the conformational appears to be an intrinsic phenomenon. Our observation is in switch to catch the outcoming ligands(see the discussion in agreement with experimental mutagenesis studies in which it the next paragraph) was postulated that GPCRs exist in an equilibrium between two interchangeable conformational states(40, 41). One of the obvious advantages of this type of dynamical behavior of Open-closed conformational changes of free the binding site entrances(open-closed) is that it enables the D1 and D2 receptors to easily capture different ligands(agonists and antagonists)or different conformations of a ligand, as will be The MD simulations of the free DI and D2 systems showed discussed later that, for all of the systems, the temperature, mass density, Tracing the MD trajectory of both the DI and D2 receptors and volume are relatively stable after 2 ns. About then, the revealed a portion of the atomic-level mechanism of the fuctuation scale became much smaller for both the RMs open-closed motion. A hydrogen-bond(H-bond) network deviations of the Ca atoms and potential energies(Fig. 2)of exists between the charged side chains of K-167 and E-302 the two simulation systems, indicating that the molecular At the beginning of MD simulation, the positively charged ystems were well behaved thereafter. side chain of K-167 forms direct H-bonds and two indirect When analyzing the MD trajectories, a striking"open-(water-bridged) H-bonds with the negatively charged side closed"conformational change is observed from the extracel- chain of E-302(Fig. 5 A). As the simulation proceeds,the lular side for both the DI and D2 receptors, as demonstrated side chain of the E-302 turns away from K-167, breaking the Fig 3. Further examination reveals that such an"open- H-bond network. At-2630 ps, all of the H-bonds between closed'event can be attributed mainly to large conforma- K-167 and E-302 are gone. Meanwhile, K-167 gradually tional changes in the EL-2 and EL-3, coupled with the approaches D-173 of EL-2, forming a new H-bond network connecting TM helices. Typically, EL-2 and EL-3 act as via water bridges(Fig. 5 B)which lasts -I ns. K-167-E-302 lips'"of mouth-like entrances of the binding sites of these switches back to the pairing state similar to that in Fig. 5 A two receptors. This can be demonstrated by the fluctuations and the H-bond network is restored. K-167 changes partners of critical distance between K-167 at EL-2 and E-302 at EL- frequently, either E-302 or D-173, staying with D-173 when 3(K-167-E-302 pairing), which is an objective monitor of the"mouth"is open and with E-302 when the"mouth"is he width of the entrance. The open and closed conforma- closed. Therefore, the H-bond network between K-167 and tions are shown in the dotted lines in Fig. 2. Taking DI as an E-302 seems to act as a"bolt"its position control example, the opening event at -2630 ps for the"mouth whether the"mouth"is open or closed orresponds to changes in the K-167-E-302 pairing distance (Fig 4 A), which reaches one of the "peak" points as shown in Fig. 6. As time proceeds, the"mouth"becomes gradually Binding conformations of SPD in the closed at -4210 ps and the distance between K-167-E-302 D1 and D2 receptors decreases to a local minimum. Such synchronization To identify the most probable binding conformations of SPD between the"open-closed"conformational changes of the to the DI and D2 receptors, 10 potential binding conformation 457000 FIGURE 2 Energy changes along MD mulations. (A)DI simulation system. (B)D2 simulation system. Ten conforma- 45900 Most probably active ons shown in dotted line were identified 243m0 conformation of 2 460000 nd obvious geometrical difference among 346100 different conformations. Among the 4453000 the conformations at 2630 ps in the MD 463000 ajectory of the unliganded DI and that at 310 ps in the MD trajectory of ur 020040060080001000 0 2000 4000 6000 8000 10000 liganded D2 have the lowest binding time(ps) energies toward SPD and were selected as the conformations most probably active. Biophysical Journal 93(5)1431-1441that there are 13 residues in the EL-2 loop of the D2 receptor, whereas 26 residues exist in the EL-2 loop of the D1 receptor. The main difference in EL-2 mainly comes from the upstream preceding conserved Cys_e2 (Fig. S2), which is far away from the active site crevice. In addition, there are six residues longer in the EL-3 of the D1 receptor than that of the D2 receptor. Though the lengths of EL-2 and EL-3 are different, they are flexible and function as the conformational switch to catch the outcoming ligands (see the discussion in the next paragraph). Open-closed conformational changes of free D1 and D2 The MD simulations of the free D1 and D2 systems showed that, for all of the systems, the temperature, mass density, and volume are relatively stable after 2 ns. About then, the fluctuation scale became much smaller for both the RMS deviations of the Ca atoms and potential energies (Fig. 2) of the two simulation systems, indicating that the molecular systems were well behaved thereafter. When analyzing the MD trajectories, a striking ‘‘open￾closed’’ conformational change is observed from the extracel￾lular side for both the D1 and D2 receptors, as demonstrated in Fig. 3. Further examination reveals that such an ‘‘open￾closed’’ event can be attributed mainly to large conforma￾tional changes in the EL-2 and EL-3, coupled with the connecting TM helices. Typically, EL-2 and EL-3 act as ‘‘lips’’ of mouth-like entrances of the binding sites of these two receptors. This can be demonstrated by the fluctuations of critical distance between K-167 at EL-2 and E-302 at EL- 3 (K-167-E-302 pairing), which is an objective monitor of the width of the entrance. The open and closed conforma￾tions are shown in the dotted lines in Fig. 2. Taking D1 as an example, the opening event at ;2630 ps for the ‘‘mouth’’ corresponds to changes in the K-167-E-302 pairing distance (Fig. 4 A), which reaches one of the ‘‘peak’’ points as shown in Fig. 6. As time proceeds, the ‘‘mouth’’ becomes gradually closed at ;4210 ps and the distance between K-167-E-302 decreases to a local minimum. Such synchronization between the ‘‘open-closed’’ conformational changes of the whole D1 receptor and peak-minimum distance changes of K-167-E-302 (Fig. 4 A) occurred three times during our 10- ns MD simulations. Quite similar open-closed conforma￾tional changes were also seen for the free D2 (Fig. 3 B). These dynamic properties suggest that there are at least two distinct conformations for both D1 or D2, and the sponta￾neous oscillation between these two conformational states appears to be an intrinsic phenomenon. Our observation is in agreement with experimental mutagenesis studies in which it was postulated that GPCRs exist in an equilibrium between two interchangeable conformational states (40,41). One of the obvious advantages of this type of dynamical behavior of the binding site entrances (open-closed) is that it enables the receptors to easily capture different ligands (agonists and antagonists) or different conformations of a ligand, as will be discussed later. Tracing the MD trajectory of both the D1 and D2 receptors revealed a portion of the atomic-level mechanism of the open-closed motion. A hydrogen-bond (H-bond) network exists between the charged side chains of K-167 and E-302. At the beginning of MD simulation, the positively charged side chain of K-167 forms direct H-bonds and two indirect (water-bridged) H-bonds with the negatively charged side chain of E-302 (Fig. 5 A). As the simulation proceeds, the side chain of the E-302 turns away from K-167, breaking the H-bond network. At ;2630 ps, all of the H-bonds between K-167 and E-302 are gone. Meanwhile, K-167 gradually approaches D-173 of EL-2, forming a new H-bond network via water bridges (Fig. 5 B) which lasts ;1 ns. K-167-E-302 switches back to the pairing state similar to that in Fig. 5 A, and the H-bond network is restored. K-167 changes partners frequently, either E-302 or D-173, staying with D-173 when the ‘‘mouth’’ is open and with E-302 when the ‘‘mouth’’ is closed. Therefore, the H-bond network between K-167 and E-302 seems to act as a ‘‘bolt’’—its position controls whether the ‘‘mouth’’ is open or closed. Binding conformations of SPD in the D1 and D2 receptors To identify the most probable binding conformations of SPD to the D1 and D2 receptors, 10 potential binding conformation FIGURE 2 Energy changes along MD simulations. (A) D1 simulation system. (B) D2 simulation system. Ten conforma￾tions shown in dotted line were identified with two criteria: lower potential energy and obvious geometricaldifferenceamong different conformations. Among them, the conformations at 2630 ps in the MD trajectory of the unliganded D1 and that at 3310 ps in the MD trajectory of un￾liganded D2 have the lowest binding energies toward SPD and were selected as the conformations most probably active. 1434 Fu et al. Biophysical Journal 93(5) 1431–1441