FULL PAPER Ceramide HaNON Figure 1(A)hSMSI-catalyzed synthesis of sphingomyelin( SM) from phosphatidylcholine(PC).(B) Some other phosphatides related to ceramide, including phosphatidylethanolamine(PE), phosphatidic acid(Pa), phosphatidylserine(PS)and phosphatidylglycerol(PG) The transmembrane domain of hSMSI is composed of six transmembrane helices(TMs). We employed 15 SK different computational methods to predict the locations of these TMs on the hSMSI sequence. Most of them produced consistent predictions (Table 1). In order to select an appropriate template for modeling the TMs of Figure 2 Chemical structures of D609(ICs0=500 umolL-I hSMSl. we retrieved a total of 61 entries with six TMs from the membrane structural proteins database against SMS) and the corresponding dimer d609-dixanthogen (no inhibition activity against SMs). PdbTm(htTp: //pdbtm. enzim. hu). After careful evaluations, which will be explained in the later Results Computational methods and Discussion section, we selected the crystal structure of Escherichia coli GlpG(PDB entry: 2IC8)as the tem Homology modeling of the hSMSl/lipid complex plate(Table 2)35 The Modeler function(as imple- structure mented in the Discovery Studio software suite), was The amino acid sequence of hSMSI used in our employed to generate a total of 30 structural models study, which has 413 residues in total, was retrieved based on this template from PubMed (access ID=NP 671512). The hSMSI As for the extracellular or intracellular loop of has an N-terminal domain, a transmembrane domain, hSMSl, loop 3(residues 235--274)is relatively long and a C-terminal domain. 6 Jiang et al. demonstrated (Figure 3). Therefore, it should be modeled based on an recently that truncation of N-terminal and C-terminal of appropriate template. PDB entry IBWO(residues 144 core structure of hSMSI(M130-Q353)was modeled in 463%quence identity =34. 1%; sequence similarity hSMSI would not abort its activity. Thus, only the 181)(se as selected as the template for this purpose our study Locations of the extracellular or intracellular This structure was selected throughout the entire PDB loop and transmembrane domain on the sequence of database according to the sequence similarity computed hSMSI were predicted by using the PSI-PRED Server by the FASTa algorithm. , No qualified template wa he predicted extracellular or intracellular loops and found for other loops of hSMSI though. These loops transmembrane segments are given in Figure 3. This were constructed from scratch by using the modeler prediction is basically consistent with the results re module in the Discovery Studio software suite. A total ported by Huitema et al. in a previous study of 30 structural models also generated for each 1568www.cjc.wiley-vch.deO2011SioC,Cas,Shanghai&WileY-VchVerlagGmbh&Co.Kgaa,WeinheimChin.j.chem.2011,29,1567-1575FULL PAPER Zhang et al. 1568 www.cjc.wiley-vch.de © 2011 SIOC, CAS, Shanghai, & WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chin. J. Chem. 2011, 29, 1567—1575 HO O O H O O HO H HN O HO PC SM Ceramide O O P O O H O O O O N O P O H HN O HO O O N Diacylglycerol + + SMS + - + - A O O P O O H O O O H3N O O P HO O H O O O O O P O O H O O O O H3N O O O O P O O H O O O OH HO H PE PA PS PG + + - O - - - - O O B Figure 1 (A) hSMS1-catalyzed synthesis of sphingomyelin (SM) from phosphatidylcholine (PC). (B) Some other phosphatides related to ceramide, including phosphatidylethanolamine (PE), phosphatidic acid (PA), phosphatidylserine (PS) and phosphatidylglycerol (PG). O S S K O S SS O S D609 D609 dixanthogen + - Figure 2 Chemical structures of D609 (IC50＝500 µmol•L－1 against SMS) and the corresponding dimer D609-dixanthogen (no inhibition activity against SMS). Computational methods Homology modeling of the hSMS1/lipid complex structure The amino acid sequence of hSMS1 used in our study, which has 413 residues in total, was retrieved from PubMed (access ID＝NP_671512). The hSMS1 has an N-terminal domain, a transmembrane domain, and a C-terminal domain.16 Jiang et al. demonstrated recently that truncation of N-terminal and C-terminal of hSMS1 would not abort its activity.17 Thus, only the core structure of hSMS1 (M130-Q353) was modeled in our study. Locations of the extracellular or intracellular loop and transmembrane domain on the sequence of hSMS1 were predicted by using the PSI-PRED Server.18 The predicted extracellular or intracellular loops and transmembrane segments are given in Figure 3. This prediction is basically consistent with the results re￾ported by Huitema et al. in a previous study.10 The transmembrane domain of hSMS1 is composed of six transmembrane helices (TMs). We employed 15 different computational methods to predict the locations of these TMs on the hSMS1 sequence.19-33 Most of them produced consistent predictions (Table 1). In order to select an appropriate template for modeling the TMs of hSMS1, we retrieved a total of 61 entries with six TMs from the membrane structural proteins database PDBTM (http://pdbtm.enzim.hu).34 After careful evaluations, which will be explained in the later Results and Discussion section, we selected the crystal structure of Escherichia coli GlpG (PDB entry: 2IC8) as the tem￾plate (Table 2).35 The Modeler function (as imple￾mented in the Discovery Studio software suite)36,37 was employed to generate a total of 30 structural models based on this template. As for the extracellular or intracellular loop of hSMS1, loop 3 (residues 235—274) is relatively long (Figure 3). Therefore, it should be modeled based on an appropriate template. PDB entry 1BW0 (residues 144— 181) (sequence identity＝34.1%; sequence similarity＝ 46.3%) was selected as the template for this purpose. This structure was selected throughout the entire PDB database according to the sequence similarity computed by the FASTA algorithm.38,39 No qualified template was found for other loops of hSMS1 though. These loops were constructed from scratch by using the Modeler module in the Discovery Studio software suite. A total of 30 structural models were also generated for each