Journal of Chemical Information and Modeling Article (a) Phi(degrees) Figure 1. Homology model of S-HTIAR a, Homology model of S-HTIAR in cartoon representation. b, Ramachadran plot calculated for the S-HTIAR For each of the five probes, the grid points were superimposed the ability at a concentration of 10 uM to inhibit the binding of to identify clusters. Corresponding pharmacophore feature was a tritiated radioligand to the corresponding receptor was tested. generated in situ at the geometric center of each cluster with Compounds that inhibited binding by more than 90% were the tolerance volume determined by the radius of gyration of further assayed at six or more concentrations, ranging above corresponding cluster. For the pharmacophore model devel oped in this study, clusters were selected based on the binding and below ICso. The K, values were calculated using the mode of R-8-OH-DPAT in 5-HTR. Clusters of the N1+ following equation: K= ICso/(1+C/Kd). [H]5-HT was used as the standard radioligand for S-HTIA receptor. Duplicate residue D3. 32 were selected; clusters of both O and Ni probes tubes were incubated at 30C for 50 min with increasing within a hydrogen-bonding distance from the conserved SS.42 concentrations(1 nM to 100 uM)of each compound and with were selected; two hydrophobic clusters of Dry probe within a 0.7 nM CHS-HT in a final volume of 200 uL of binding buffer distance of 5 A from conserved F6.51, F6.52, and w7. 40 were containing 50 mM Tris and 4 mM MgCl2 (pH 7.4) elected. Thus,a four-feature pharmacophore model of 5- Nonspecific binding was assessed by parallel incubations with HTIAR was generated. Moreover, to reduce the false positive 10 uM S-HT. The reaction was started by addition of membranes(15 pg/tube) and stopped by rapid Itration position of side chains of D3.32, F6.51, and Y7.43 were added. through a Whatman GF/B glass fber filter and subsequent Dynamic Pharmacoph ore-Based Virtual Screening In washing with cold buffer [50 mM Tris and 5 mM ethy our virtual screening approach, the Ligand Pharmacophore potential molecules from Maybridge and Specs chemical 24-well cell harvester. Scintillation cocktail was added, and the Mapping protocol embedded in DS was employed to retrieve enediaminetetraacetic acid(EDTA)(pH 7.4) using a Brandel databases with the dynamic pharmacophore model as a 3D radioactivity was determined in a Micro Beta liquid scintillation query. The two databases, each containing 61623 and 166778 counter. The ICso and K values were calculated by nonlinear compounds,were filtered by Lipinski's"Rule of Five"36 to regression(PRISM, Graphpad, San Diego, CA)using a create druglike databases in advance using Prepare Ligands sigmoidal function. protocol. For each molecule in the database, a maximum of 1 SH-GTPyS Assays. The [S]GTPyS binding assay was conformers with an energy threshold of 20 kcal/mol were performed at 30C for 30 min with 10 ug of membrane protein generated using FASt algorithm. The Best and Flexible in a final volume of 100 uL with various concentrations of the mapping option was adopted. Compounds were required to compounds. The binding buffer contained 50 mM Tris (pH match at least three features of input pharmacophore(Fit Value 7.5),5 mM MgCl, 1 mM ethylenediaminetetraacetic acid 22.95). Hits were then subjected to GOLD docking. Specially, (EDTA), 100 mM NaCl, 1 mM DL-dithiothreitol (DTT), and e automatic genetic algor h the search efficiency was set at 30% which is recommended for 40 uM guanosine triphosphate(GDP). The reaction was virtual eening. Other parameters were the same as described initiated by the addition of [S]GTPys(final concentration of before in the molecular docking section. Outcome solutions 0.1 nM). Nonspecific binding was measured in the presence of were sorted by GoldScore. Compounds with GoldScore >45 uM S-guanylimidodiphosphate(GpP(NH)P). The re- were retained for further visual inspection action was terminated by the addition of 1 mL of ice-cold Binding Assays. All the selected compounds wer washing buffer(50 mM Tris, 5 mM Mg Cl, I mM EDTA, 100 subjected to competitive binding assays for 5-HTIA receptor, mM Naci)and was rapidly filtered with GE/C glass fber filters sing membrane preparation obtained from stable transfected (Whatman)and washed three times. Radioactivity was CHO cells as previously described by our laboratory. First, determined by liquid scintillation counting dxdoLor/10. 1021/c400481p. Chem. Inf Model. 2013, 53, 3202-3211For each of the five probes, the grid points were superimposed to identify clusters. Corresponding pharmacophore feature was generated in situ at the geometric center of each cluster with the tolerance volume determined by the radius of gyration of corresponding cluster. For the pharmacophore model devel￾oped in this study, clusters were selected based on the binding mode of R-8-OH-DPAT in 5-HT1AR. Clusters of the N1+ probe that were within a distance of 3 Å from the conservative residue D3.32 were selected; clusters of both O and N1 probes within a hydrogen-bonding distance from the conserved S5.42 were selected; two hydrophobic clusters of DRY probe within a distance of 5 Å from conserved F6.51, F6.52, and W7.40 were selected. Thus, a four-feature pharmacophore model of 5- HT1AR was generated. Moreover, to reduce the false positive rate of virtual screening, three excluded volumes right at the position of side chains of D3.32, F6.51, and Y7.43 were added. Dynamic Pharmacophore-Based Virtual Screening. In our virtual screening approach, the Ligand Pharmacophore Mapping protocol embedded in DS was employed to retrieve potential molecules from Maybridge and Specs chemical databases35 with the dynamic pharmacophore model as a 3D query. The two databases, each containing 61623 and 166778 compounds, were filtered by Lipinski’s “Rule of Five” 36 to create druglike databases in advance using Prepare Ligands protocol. For each molecule in the database, a maximum of 100 conformers with an energy threshold of 20 kcal/mol were generated using FAST algorithm. The Best and Flexible mapping option was adopted. Compounds were required to match at least three features of input pharmacophore (Fit Value ≥2.95). Hits were then subjected to GOLD docking. Specially, the automatic genetic algorithm search option was used, and the search efficiency was set at 30% which is recommended for virtual screening. Other parameters were the same as described before in the molecular docking section. Outcome solutions were sorted by GoldScore. Compounds with GoldScore >45 were retained for further visual inspection. Binding Assays. All the selected compounds were subjected to competitive binding assays for 5-HT1A receptor, using membrane preparation obtained from stable transfected CHO cells as previously described by our laboratory.37,38 First, the ability at a concentration of 10 μM to inhibit the binding of a tritiated radioligand to the corresponding receptor was tested. Compounds that inhibited binding by more than 90% were further assayed at six or more concentrations, ranging above and below IC50. The Ki values were calculated using the following equation: Ki = IC50/(1+C/Kd). [3 H]5-HT was used as the standard radioligand for 5-HT1A receptor. Duplicate tubes were incubated at 30 °C for 50 min with increasing concentrations (1 nM to 100 μM) of each compound and with 0.7 nM [3 H]5-HT in a final volume of 200 μL of binding buffer containing 50 mM Tris and 4 mM MgCl2 (pH 7.4). Nonspecific binding was assessed by parallel incubations with 10 μM 5-HT. The reaction was started by addition of membranes (15 μg/tube) and stopped by rapid filtration through a Whatman GF/B glass fiber filter and subsequent washing with cold buffer [50 mM Tris and 5 mM ethyl￾enediaminetetraacetic acid (EDTA) (pH 7.4)] using a Brandel 24-well cell harvester. Scintillation cocktail was added, and the radioactivity was determined in a MicroBeta liquid scintillation counter. The IC50 and Ki values were calculated by nonlinear regression (PRISM, Graphpad, San Diego, CA) using a sigmoidal function. [ 35S]-GTPγS Assays. The [35S]GTPγS binding assay was performed at 30 °C for 30 min with 10 μg of membrane protein in a final volume of 100 μL with various concentrations of the compounds. The binding buffer contained 50 mM Tris (pH 7.5), 5 mM MgCl2, 1 mM ethylenediaminetetraacetic acid (EDTA), 100 mM NaCl, 1 mM DL-dithiothreitol (DTT), and 40 μM guanosine triphosphate (GDP). The reaction was initiated by the addition of [35S]GTPγS (final concentration of 0.1 nM). Nonspecific binding was measured in the presence of 100 μM 5′-guanylimidodiphosphate (Gpp(NH)p). The re￾action was terminated by the addition of 1 mL of ice-cold washing buffer (50 mM Tris, 5 mM MgCl2, 1 mM EDTA, 100 mM NaCl) and was rapidly filtered with GF/C glass fiber filters (Whatman) and washed three times. Radioactivity was determined by liquid scintillation counting. Figure 1. Homology model of 5-HT1AR. a, Homology model of 5-HT1AR in cartoon representation. b, Ramachadran plot calculated for the 5-HT1AR model. Journal of Chemical Information and Modeling Article 3204 dx.doi.org/10.1021/ci400481p | J. Chem. Inf. Model. 2013, 53, 3202−3211