Opioid Receptor Models — Baiitista, Aslier, and Carpenter 



109 



using the Amber94 force field (Cornell et al. 



1995) . A protein report was generated to de- 

 termine if any angles or dihedrals exceeded 

 biochemical norms (outliers). All outhers were 

 in loop regions, which were ignored and not 

 used in this study. 



Each model was then used for docking of 

 morphine. A model of moiphine was built in 

 MOE and the nitrogen was protonated to rep- 

 resent the structure at physiological pH. Chi- 

 rality of the centers was determined and main- 

 tained in the model. The resulting structure 

 was then minimized in Merck Molecular force 

 field (MMFF94) (Halgren 1996). 



Docking 



Morphine was docked into the mu, kappa, 

 and delta receptors. Docking was accom- 

 plished using the algorithm MOE Dock. The 

 default settings were used with the exception 

 of Tabu search. A Tabu search will disallow a 

 previous docking location for subsequent 

 docking runs to ensure that the entire docking 

 volume is probed. For each position, 700 it- 

 erations were generated to optimize the inter- 

 actions at a location. A database of 25 com- 

 plexes for each receptor with the ligand was 

 generated. MMFF94 force field (Halgren 



1996) was used. Each complex was visually in- 

 spected to determine if the Hgand was in the 

 upper one-third of the receptor and inside the 

 helical bundle. Hydrogen bonds between the 

 ligand and receptor and aromatic residues 

 were determined. Based on these criteria, the 

 best complexes of each were selected and then 

 minimized to an RMSG of 0.1 Kcal/mol-A us- 

 ing the MMFF94 force field. Each complex 

 was checked to determine if hydrogen bond- 

 ing had increased, decreased, or remained the 

 same. Any changes in aromatic interactions 

 were also noted. 



Dynamics 



The best complex for each ligand and re- 

 ceptor was then subjected to unrestrained mo- 

 lecular dynamics simulations. Dynamics were 

 performed on all three complexes generating 

 molecular databases. This calculation used 

 NVT parameters (holding constant moles, vol- 

 ume, and temperature). The simulation was 

 performed for 1 fs time step with 60 ps of 

 heating prior to the 100 ps equilibrium (data 

 collection) phase as in Wang et al. 2001. The 



output databases contained 100 entries col- 

 lected during the equilibrium phase. The sim- 

 ulation calculated potential energy (U in Kcal/ 

 mol), temperature (T in kelvins), pressure (P 

 in Kpa), total energy (E, kinetic and potential 

 in Kcal/mol), and enthalpy (H, E + PV in 

 Kcal/mol). 



RESULTS 



Mu Model 



The mu model was minimized to an RMSG 

 of 0.1 kcal/mol-A. The protein report indicated 

 five cis amide bonds all located in extra- or 

 intra-cellular loops. Since these locations were 

 not used in the results, the amide bonds were 

 not corrected. The helices were between 33 

 and 20 residues in length, with helix 5 being 

 the shortest (Table 3). This helix, in all three 

 receptor models, was terminated early on the 

 intracellular side. This area is indicated in ac- 

 tivation and not in ligand binding and was 

 therefore ignored. 



Kappa Model 



The kappa model was minimized to an 

 RMSG of 0.1 kcal/mol-A. The protein report 

 indicated three cis-amide bonds and all were 

 located in loop regions. Since these locations 

 were not used in any docking results, the am- 

 ide bonds were not corrected. The helices var- 

 ied in length from 20 to 33 residues. Helices 

 4 and 5 were both 20 residues long, unusually 

 short for spanning the lipid bilayer. Helix 4 is 

 terminated by PI 72 located at the intracellular 

 surface. As in the mu receptor, the kappa re- 

 ceptor helix 5 is terminated early at the intra- 

 cellular surface (Table 3). 



Delta Model 



The delta model was minimized to an 

 RMSG of 0.1 kcal/mol-A. The protein report 

 indicated three cis-amide bonds and all were 

 located in loop regions. Since these regions 

 were not used in the docking results, the am- 

 ide bonds were ignored. The helices varied in 

 length from 19 to 29 residues (Table 3). Again, 

 helix 5 was terminated at the intracellular sur- 

 face. Helix 4 was also terminated at the intra- 

 cellular surface by Pro 162. 



