The Journal of Physical Chemistry B Article 77A iew(right)of 1-SPD-D3R complex embedded in a hydrated POPC lipid bilayer. nd stick representation(middle). Water molecules are shown as lines, and lipid molecules are re )and phosphorus(orange sphere)atoms in van der Waals representation. Na and CI ions are also All nonpolar hydrogen atoms are omitted for clarity b8, SPDD|R—SPDD2R Time (ns) gure 2. Root-mean-square deviation(rmsd)of the backbone atoms(CA, N, c)of all six simulation systems as a function of simulation time: (a) backbone rmsd for unliganded DRs;( b)backbone rmsd for I-SPD-DRs. membrane was in agreement with the suggestions from the water, ions, and lipid headgroups held fixed to allow the melting Orientations of Proteins in Membranes(OPM) database. 5 of lipid tails. Subsequently, 1000 cycles of minimization Lipids whose phosphorus atom lies within the region of the followed by a 0.5 ns equilibration was effected with harmoni protein and those within 0.6 A of the protein were removed. constraint on protein so as to guide the system to the nearest The protein-membrane complex was further solvated using local energy minimum. During this stage, forces were applied to the VMD Solvate plug-in. Water molecules added inside the water molecules entering the highly hydrophobic membrane ipid bilayer and the proteins were eliminated. The VMD protein interface to prevent this region from hydration. Aft Autoionize plug-in was then employed to create an ionic minimization and equilibration with the protein constrained, concentration of 150 mM NaCl by transmuting random water the harmonic constraint was released to permit the whole brief summary of the configuration of each final system is listed finally conducted for 50 ns for all systeme duction runs were molecules into Nat and Cl and to neutralize the system. a system to equilibrate for 0.5 ns further. Pr in Table Sl, and a snapshot of one of the constructe simulation system(I-SPD bound D3R system) is shown in RESULTS AND DISCUSSION Molecular Dynamics Simulation Parameters. All MD 3D Structures of D1R, D2R, and D3R. Sequence simulations were carried out using the parallel molecular gnment indicated that the sequence identity in trans- dynamics program NAMD 2.6 and the CHARMM27 force membrane helices was 42. 4% between DiR and B2AR and field7with CMAP terms8for all protein molecules, POPC 72.9% between D2R and D3R, respectively(Figure S1).The lipid molecules, and ions along with the TIP3P model for 3D models of dir and D2R were constructed using the newly water molecules. A cutoff of 12 A(switching function starting at resolved agonist-state crystal structure of PAR and antagonist 10 A)for van der Waals interactions was imposed. The particle- state X-ray structure of DR as templates, respectively. The mesh Ewald techni ique was employed to calculate long-range PRoCHECK statistics calculations showed that 100% of the electrostatic forces without cutoff. Periodic boundary con- residues in our DIR and D2R models were either in the most ditions were imposed on all simulations with an integration favored or in the additionally allowed regions of the time step of 2 fs to allow a multiple time stepping algorithms to Ramachandran plot (Figure S2), suggesting that the overall be employed. The simulations were equilibrated as an NPT main-ch d side-chain conformations are much more semble, using the Langevin Nose-Hoover method* to reliable as compared with our previous models. The previous maintain the pressure at 1 atm, and the temperature was kept at models of DiR and d2r were built by taking bovine rhodopsin 310 K by using Langevin dynamics with a very weak friction as the template, which was the only crystallized structure in coefficient GPCR family at that time and had ca. 25% sequence identity in Molecular Dynamics Simulation Protocol. All TM region with DIR and D2R. The overall architecture of our lation systems were first subjected to 1000 cycles of ste present DiR and D2R models is similar to that of our descent and 1000 cycles of conjugate gradient DIR and D2R minimization followed by a 0.5 ns equilibration with models, three intramolecular hydrophobic interaction clusters 81 dx. dolora/o.021/p30492351 Phys. Chem. B2012116,8121-813membrane was in agreement with the suggestions from the Orientations of Proteins in Membranes (OPM) database.35 Lipids whose phosphorus atom lies within the region of the protein and those within 0.6 Å of the protein were removed. The protein−membrane complex was further solvated using the VMD Solvate plug-in. Water molecules added inside the lipid bilayer and the proteins were eliminated. The VMD Autoionize plug-in was then employed to create an ionic concentration of 150 mM NaCl by transmuting random water molecules into Na+ and Cl− and to neutralize the system. A brief summary of the configuration of each final system is listed in Table S1, and a snapshot of one of the constructed simulation system (l-SPD bound D3R system) is shown in Figure 1. Molecular Dynamics Simulation Parameters. All MD simulations were carried out using the parallel molecular dynamics program NAMD 2.636 and the CHARMM27 force field37 with CMAP terms38 for all protein molecules, POPC lipid molecules, and ions along with the TIP3P model39 for water molecules. A cutoff of 12 Å (switching function starting at 10 Å) for van der Waals interactions was imposed. The particle￾mesh Ewald technique40 was employed to calculate long-range electrostatic forces without cutoff. Periodic boundary con￾ditions were imposed on all simulations with an integration time step of 2 fs to allow a multiple time stepping algorithms to be employed. The simulations were equilibrated as an NPT ensemble, using the Langevin Nose−́ Hoover method41 to maintain the pressure at 1 atm, and the temperature was kept at 310 K by using Langevin dynamics with a very weak friction coefficient. Molecular Dynamics Simulation Protocol. All simu￾lation systems were first subjected to 1000 cycles of steepest descent and 1000 cycles of conjugate gradient energy minimization followed by a 0.5 ns equilibration with protein, water, ions, and lipid headgroups held fixed to allow the melting of lipid tails. Subsequently, 1000 cycles of minimization followed by a 0.5 ns equilibration was effected with harmonic constraint on protein so as to guide the system to the nearest local energy minimum. During this stage, forces were applied to water molecules entering the highly hydrophobic membrane− protein interface to prevent this region from hydration. After minimization and equilibration with the protein constrained, the harmonic constraint was released to permit the whole system to equilibrate for 0.5 ns further. Production runs were finally conducted for 50 ns for all systems. ■ RESULTS AND DISCUSSION 3D Structures of D1R, D2R, and D3R. Sequence alignment indicated that the sequence identity in trans￾membrane helices was 42.4% between D1R and β2AR and 72.9% between D2R and D3R, respectively (Figure S1). The 3D models of D1R and D2R were constructed using the newly resolved agonist-state crystal structure of β2AR and antagonist￾state X-ray structure of D3R as templates, respectively. The PROCHECK statistics calculations showed that 100% of the residues in our D1R and D2R models were either in the most favored or in the additionally allowed regions of the Ramachandran plot (Figure S2), suggesting that the overall main-chain and side-chain conformations are much more reliable as compared with our previous models. The previous models of D1R and D2R were built by taking bovine rhodopsin as the template, which was the only crystallized structure in GPCR family at that time and had ca. 25% sequence identity in TM region with D1R and D2R. The overall architecture of our present D1R and D2R models is similar to that of our previously reported DRs. Similar to previous D1R and D2R models, three intramolecular hydrophobic interaction clusters Figure 1. Side view (left) and top view (right) of l-SPD-D3R complex embedded in a hydrated POPC lipid bilayer. Protein is shown in cartoon representation, with l-SPD in ball and stick representation (middle). Water molecules are shown as lines, and lipid molecules are represented by sticks with their nitrogen (blue sphere) and phosphorus (orange sphere) atoms in van der Waals representation. Na+ and Cl− ions are also shown as gray and green spheres, respectively. All nonpolar hydrogen atoms are omitted for clarity. Figure 2. Root-mean-square deviation (rmsd) of the backbone atoms (CA, N, C) of all six simulation systems as a function of simulation time: (a) backbone rmsd for unliganded DRs; (b) backbone rmsd for l-SPD-DRs. The Journal of Physical Chemistry B Article 8123 dx.doi.org/10.1021/jp3049235 | J. Phys. Chem. B 2012, 116, 8121−8130