FULL PAPER Zhang et al. index. html) was employed in our study for this pur- Table 5 Results of the normal mode analysis of the hSMSI This server makes use of the elastic Network tructure which provides a fast simple tool to compute ze, and analyze low-frequency normal modes of requena biological macromolecules. The structural model of mode 7 0.6182 hSMS I/SM was submitted for the NMA analysis. Key mode 8 107 0.6494 parameters used in computation included: DQMIN mode 9 1.23 0.6552 100, DQMAX=100, DQSTEP=20 and NRBL auto" The default cutoff of eight residues was used to identify elastic interaction ranges. A total of 100 normal mode 11 1.87 0.6801 modes with the lowest frequencies were computed Essential features of the top ten low-frequency normal 0.6152 modes, including its frequency, are summarized in Ta- 0.6192 ble 5. The two lowest-frequency normal modes ( mode 15 2.70 0.5726 modes 7 and 8 in Table 5)are illustrated in Figure 8 0.3389 the 10 normal modes with lowest frequencies are listed here, Collectivity indicates the percentage of residues that are involved in a certain normal mode Figure 8 The two primary normal modes of hsMSI with the lowest frequencies (Left: bending of the extracellular loops to- open state and a closed state, in which the extracellular loops tracellular loops on the top of the taa ight: rotation of the ex- Figure 7 The two major states of the hSMS I structure, i.e. an wards the transmembrane domain adopt a different orientation from the transmembrane domain he initial structure of hSMSI before MD refinement is rendered top-ranked binding poses were output after docking was in pale ribbons, the refined structure after 10 ns molecular dy- finished These binding poses predicted by GOld were namics simulation is rendered in dark ribbons. The arrow on the quite similar to each other. Therefore, we chose the top stands for the hinge axe around which the extracellular loops binding pose with the best binding score, and the corre rotate, which is given by the DynDom program. sponding hSMS 1/D609 complex structure was mini- mized with the same force field parameters mentioned Molecular docking of d609 to hSMSI above in couple with the implicit membrane GB/SA To understand how D609 affects the catalytic activ- model. The final model of the hSMSl/D609 complex is ity of hSMSl, the GOLd program(version 4.1, released illustrated in Figure 9 by CCDC Inc. was employed to perform automated lar docking to explore the possible binding poses Results and discussion between hSMSI and D609. The active site was defined as the residues within 10 A from the reference ligand. Homology modeling of the hSMS1/SM complex i.e. SM. Other key parameters used in docking included structure population size=100, number of GA operations In our study, it was of particular difficulty to select 500000, mutation rate=95%, crossover rate=95% and an appropriate template for homology modeling of the scoring function=Chemscore During docking process, hSMSI structure In fact, no other membrane protein the hSMSI structure was kept fixed. A total of 20 with known three-dimensional structure has a sequence 1572www.cjc.wiley-vch.de02011SioC,Cas,Shanghai&WIley-vcHVerlaggmbh&Co.Kgaa,WeinheimChinj.chem.2011,29,1567-1575FULL PAPER Zhang et al. 1572 www.cjc.wiley-vch.de © 2011 SIOC, CAS, Shanghai, & WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chin. J. Chem. 2011, 29, 1567—1575 index.html)47 was employed in our study for this pur￾pose. This server makes use of the Elastic Network Model, which provides a fast simple tool to compute, visualize, and analyze low-frequency normal modes of biological macromolecules. The structural model of hSMS1/SM was submitted for the NMA analysis. Key parameters used in computation included: DQMIN＝ －100, DQMAX＝100, DQSTEP＝20 and NRBL＝ “auto”. The default cutoff of eight residues was used to identify elastic interaction ranges. A total of 100 normal modes with the lowest frequencies were computed. Essential features of the top ten low-frequency normal modes, including its frequency, are summarized in Ta￾ble 5. The two lowest-frequency normal modes (i.e. modes 7 and 8 in Table 5) are illustrated in Figure 8. Figure 7 The two major states of the hSMS1 structure, i.e. an open state and a closed state, in which the extracellular loops adopt a different orientation from the transmembrane domain. The initial structure of hSMS1 before MD refinement is rendered in pale ribbons; the refined structure after 10 ns molecular dy￾namics simulation is rendered in dark ribbons. The arrow on the top stands for the hinge axe around which the extracellular loops rotate, which is given by the DynDom program. Molecular docking of D609 to hSMS1 To understand how D609 affects the catalytic activ￾ity of hSMS1, the GOLD program (version 4.1, released by CCDC Inc.) was employed to perform automated molecular docking to explore the possible binding poses between hSMS1 and D609. The active site was defined as the residues within 10 Å from the reference ligand, i.e. SM. Other key parameters used in docking included: population size＝100, number of GA operations＝ 500000, mutation rate＝95%, crossover rate＝95% and scoring function＝Chemscore. During docking process, the hSMS1 structure was kept fixed. A total of 20 Table 5 Results of the normal mode analysis of the hSMS1 structure Modea Frequency Collectivityb mode 7 1.00 0.6182 mode 8 1.07 0.6494 mode 9 1.23 0.6552 mode 10 1.85 0.6581 mode 11 1.87 0.6801 mode 12 2.07 0.5377 mode 13 2.16 0.6152 mode 14 2.65 0.6192 mode 15 2.70 0.5726 mode 16 2.75 0.3389 a Only the 10 normal modes with lowest frequencies are listed here; b Collectivity indicates the percentage of residues that are involved in a certain normal mode. Figure 8 The two primary normal modes of hSMS1 with the lowest frequencies (Left: bending of the extracellular loops to￾wards the transmembrane domain. Right: rotation of the ex￾tracellular loops on the top of the transmembrane domain). top-ranked binding poses were output after docking was finished. These binding poses predicted by GOLD were quite similar to each other. Therefore, we chose the binding pose with the best binding score, and the corre￾sponding hSMS1/D609 complex structure was mini￾mized with the same force field parameters mentioned above in couple with the implicit membrane GB/SA model. The final model of the hSMS1/D609 complex is illustrated in Figure 9. Results and discussion Homology modeling of the hSMS1/SM complex structure In our study, it was of particular difficulty to select an appropriate template for homology modeling of the hSMS1 structure. In fact, no other membrane protein with known three-dimensional structure has a sequence