NMR structure of the thromboxane A 2 receptor ligand recognition pocket Ke-He Ruan, Jiaxin Wu, Shui-Ping So, Lori A. Jenkins and Cheng-Huai Ruan Vascular Biology Research Center and Division of Hematology, Department of Internal Medicine, The University of Texas Health Science Center, Houston, TX, USA To overcome the difficulty of characterizing the structures of the extracellular l oops (eLPs) of G protein-coupled receptors (GPCRs) other than rhodopsin, we have explored a strategy to generate a three-dimensional structural model for a GPCR, the thromboxane A 2 receptor. This three-dimen- sional structure was completed by the assembly of the NMR structures of the computation-guided constrained peptides that mimicked the extracellular loops and connected to the conserved seven transmembrane domains. The NMR structure-based model reveals the structural features of the eLPs, in which the s econd extracellular loop (eLP 2 )andthe disulfide bond between the first extracellular loop (eLP 1 )and eLP 2 play a major role in forming t he ligand r ecognition pocket. The eLP 2 conformation is dynamic a nd regulated by the oxidation and reduction of the d isulfide bond, which affects ligand docking in the initial recognition. The reduced form of the thromboxane A 2 receptor experienced a decrease in ligand b inding activity due to the rearrangement of the e LP 2 conformation. The ligand-bound receptor was, however, r esistant to the reduction inactivation because the ligand covered the d isulfide bond and stabilized the e LP 2 conformation. This mole cular mechanism of ligand recog- nition is the first that may be applied to other prostanoid receptors and other GPCRs. Keywords: G protein-coupled receptor; thromboxane A 2 ; thromboxane A 2 receptor; NMR; synthetic peptide. Prostanoids, comprising prostaglandins and thromboxane A 2 , act as local hormones in the vicinity of their production site to regulate hemostasis and smooth muscle function. These hormones are mediated by specific receptors inclu- ding five basic types based on sensitivity to prostaglandin D 2 (PGD 2 ), prostaglandin E 2 (PGE 2 ), prostaglandin F 2 (PGF 2 ), prostaglandin I 2 (PGI 2 ) and thromboxane A 2 (TXA 2 ), termed DP, EP, FP, I P and TP receptors, respectively [1,2]. In addition, EP is subdivided into four subtypes, EP1, EP2, EP3 and EP4 receptors, based on the responses to various agonists and antagonists. Of the prostanoid receptors, human TP was first purified from platelets in 1989 and the cDNA was cloned from the placenta in 1991 [3,4]. All of the known p rostanoid receptors belong to the rhodopsin-type G protein-coupled receptor (GPCR) superfamily, which is one of the largest protein families in nature with seven hydrophobic transmembrane domains [5,6]. Because of the difficulty in crystallizing the membrane proteins of GPCRs, rhodopsin is the only one for which a crystal structure has been determined [7–10]. The crystal structure of rhodopsin has offered a structural template of the conserved transmembrane helices for othe r GPCRs, including prostanoid receptors. For more than a decade, structural and functional studies of the prostanoid receptors have been foc used on the identification of the ligand binding site and specific recognition of ligands. The homology modeling-based mutagenesis for the transmem- brane domains of the prostanoid receptors has suggested that the conserved regions in the third and seventh transmembrane domains are involved in binding the common structures o f the prostanoids, which includes a carboxylic acid, a hydroxyl group at position 15, and two aliphatic side chains [11–13]. To understand the different physiopathological actions of the prostanoids, it is import- ant to know the molecular mechanism of how the prost- anoid receptors recognize ligand molecules selectively on the extracellular side of the receptors, transfer them into the membrane domains, and finally trigger the different G proteins binding on the intracellular side of the receptors. Structural characterization of the extracellular domains of the r eceptors is a key step to revealing the molecular action mechanism at the molecular level. The crystal structure of rhodopsin cannot mimic the extracellular domains of prostanoid receptors, such as the TP receptor, because of the different sizes and lack o f conservation (Fig. 1A) in the segments, which has resul- ted in the inability to model the extracellular domains for functional studies in general for the prostanoid receptors. The possible involvement in ligand recognition of the extracellular loops of prostanoid receptors and other GPCRs has been reported by different research grou ps (Table 1). However, the lack of an experime ntal three- dimensional structural model for any of the receptors has Correspondence to K H. Ruan, Division of Hematology, Department of Internal Medicine, University of Texas Health Science Center at Houston, 6431 Fannin St. Houston, Texas 77030, USA. Fax: + 1 713 500 6810, Tel.: + 1 713 500 6769, E-mail: kruan@uth.tmc.edu Abbreviations: GPCR, G protein-coupled receptor; eLP, extracellular loop; TP, thromboxane A 2 receptor. (Received 1 4 March 2004, revised 13 May 2004, accepted 27 M ay 2004) Eur. J. Biochem. 271, 3006–3016 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04232.x impaired further definition of the ligand recognition pocket on the extracellular side of the receptors. This has become a major obstacle to the further understanding of the molecular mechanism of the specific ligand recognition in the receptors, and to the further develop- ment of specific ligand recognition-based drugs. To overcome this obstacle, we recently developed a strategy to precisely mimic the extracellular l oops of the TP receptor, termed Ôcomputation-guided constrained peptide synthesisÕ for t he solution structural determination using two-dimensional NMR spectroscopy. Three-dimensional structures of eLP 2 [14] and the third extracellular l oop (eLP 3 ) [15] regions of the TP receptor have been successfully determined by this experimental approach individually. In addition, through the use o f the NMR structure of the TP eLP 2 peptide, the unique residues involved in forming the specific ligand recognition site were identified and confirmed by NMR structure-guided mutagenesis [1]. However, to define the specific ligand recognition pocket of the receptor, a three-dimensional structural model f or the t hree extracellular l oops config- ured to the transmembrane domains is required. I n this report, the eLP 1 structure of the TP receptor was determined by two-dimensional N MR spectroscopy, and the information was combined with the defined N MR structures of eLP 2 [14] and eLP 3 [15] domains to construct a solution structure, which includes all three extracellular loops connected to the conserved transmem- brane helices of the T P receptor. The NMR structure- based extracellular loop model is the first experimental three-dimensional structure for the prostanoid receptors and also the first for mammalian GPCRs with the single exception of bovine rhodopsin. As expected, the three- dimensional model provided valuable information reveal- ing the dynamic specific ligand recognition pocket in the extracellular loops of the TP receptor and the ligand/ receptor recognition mechanism, which may also apply to other prostanoid receptors. Materials and methods Peptide synthesis and purification A peptide mimicking the TP eLP 1 (residues 88–104), with homocysteine added at both e nds of the loop was s ynthes- ized using the fluorenylmethoxycarbonyl-polyamide solid phase method. After cleavage with trifluoroacetic acid, the peptide was purified to homogeneity by HPLC [14]. For cyclization of the peptide by the formation of a disulfide bond, the purified peptide was dissolved in 1 mL dimethyl Fig. 1. Sequence alignment of the extracellular loops of TP receptor with rhodopsin (A) and topology model of the TP receptor (B). The heavy line represents t he eLP 1 region studied. The amino acid sequence of the region synthesized is shown in t he op en form and in the con- strained loop form that has a conne ction between the N a nd C termini by a disulfide bond using additional homocysteine (hCys) residu es. eLP, Extracellular loop; iLP, intracellular loop; NT, N-terminal region; CT, C-terminal region. Table 1. Ligand recognition sites localized in extracellular loops. Receptor Method(s) Loop Residues Reference Prostanoid Receptors TP Mutation eLP 2 Cys105, Cys183 [24] EP Mutation eLP 2 Trp199, Pro200, Thr202 [30] TP Affinity labeling eLP 2 Cys183–Asp193 [50] TP Mutation, NMR eLP 2 Val176, Leu185 Thr186, Leu187 [1] Other GPCRs Thyrotropin-releasing hormone receptor Mutation eLP 2 Tyr181 [51] Human P2Y 1 receptor Molecular modeling, mutation eLP 2 , eLP 3 Cys42, Cys296 [52] A1/A3 adenosine receptor Molecular modeling, radioligand binding assays eLP 2 11 residues at the C-terminal of eLP 2 [53] Human A2a adenosine receptor Mutation eLP 2 Glu151, Glu161, Glu169 [54] a1-Adrenergic receptor Mutation eLP 2 Gly196, Val197, Thr198 [55] Ó FEBS 2004 Thromboxane A2 receptor structure and function (Eur. J. Biochem. 271) 3007 sulfoxide,andaddedtoH 2 O at a final concentration of 0.02 mgÆmL )1 . The solution was then adjusted to pH 8.5 using triethylamine, and stirred overnight at room tempera- ture. The cyclic peptide was then lyophilized and purified by HPLC on the C4 column. The sequence of the synthesized TP receptor eLP 1 isshowninFig.1B. 1 H NMR and assignment Proton NMR was carried out on a Brucker 600 spectrometer in the Chemistry Department, University of H ouston (Houston, TX). All t wo-dimensional e xperi- ments (DQF-COSY, TOCSY and NOESY) were performed under the same conditions (298 K, in 20 m M , pH 6.0 sodium phosphate buffer containing 10% D 2 Oto provide a lock signal). The NOESY spectrum was recorded with a mixing time of 200 ms; the TOCSY spectrum was carried out with a MLEV spin-lock sequence with a total mixing time of 5 0 ms. All spectra were composed of 2048 complex points in the F2 and 512 complex points in F1 w ith 64 s cans per t1 in crement. Quadrature detection was achieved in the F1 by the states-time proportional phase increment method. The NMR data were processed using FELIX 2000 (Accelrys Inc., San Diego, CA). All f ree induction decays were zero- filled to 2K-2K before Fourier transformation, and 0° (for DQF-COSY), 70° (for TOCSY), or 90° (for NOESY) shifted sinbell 2 window function was used in both dimensions. The sequence-specific assignment was obtained using the standard method [16] with the help of the FELIX 2000 auto-assignment program. Calculation of structures The overall structure of the peptide was determined through the use of the intraresidue sequential NOEs and dihedral angles of the peptide residues as described previously [14]. The NOE cross-peak volumes in NOESY spectrum were converted into upper bounds of the interproton distances by using the FELIX program. NOE cross-peaks were segmented using a statistical segmentation function and c haracterized as strong, medium and weak, corresponding to upper bound distance range constraints of 2.5, 3 .5 and 6.0 A ˚ , respectively. Dihedral angles were obtained by direct measurement of 3 J NHa values in the DQF-COSY spectrum. Distance geometry calculations were then carried out on an SGI workstation using the DGII program within the INSIGHT II package (Accelrys Inc., San Diego, CA) and initial structures were calculated based on the NOE constraints and dihedral angles. Energy refinement calculations, inclu- ding restrained minimization/dynamics, were carried out in the best distance geometry structures using the DISCOVER program within the INSIGHT II package. Molecular modeling of the transmembrane domains of the TP receptor The three-dimensional structural working model for the seven transmembrane helices of the TP receptor was constructed u sing homology modeling within the INSIGHT II program, based on the bovine rhodopsin crystallographic structure [10]. Results Design and synthesis of peptides mimicking extracellular loops of the human TP receptor Our approach for the structural characterization of the TP extracellular loops is the use of synthetic peptides mimicking the loop domains. Analysis of the TP receptor model, generated from molecular modeling based on the crystal- lographic structure of bovine rhodopsin, indicated that approximately 10–14 A ˚ separates the N and C termini of the extracellular loops. A loop peptide whose termini are constrained to this separation is presumably more likely to mimic the native loop structure than the corresponding loop peptide with unrestricted ends. In our previous studies, a constrained p eptide corresponding to the highly co nserved eLP 2 (residues 173–193) of the TP receptor has been made with the N and C termini connected by a homocysteine disulfide bond. The overall three-dimensional structure of the loop peptide has also been determined through two- dimensional NMR, complete 1 H NMR assignments a nd structural construction [14]. The structure shows b-turns at residues 180 and 185. The distance between the N and C termini of the peptide shown in the NMR structure is 14.2 A ˚ , which corresponds to the distance (14.5 A ˚ ) between the t wo transmembrane helices connecting eLP 2 in the TP receptor model. In addition, the constrained eLP 2 peptide was shown to actively interact with a TP receptor ligand, SQ29 548 which w as identified by fluorescent, CD [14], and NMR studies [1]. The identity of the residues in contact with the ligand, using the peptide, was further confirmed in recombinant protein [1]. These findings suggested that the constrained peptide approach could be used to mimic other extracellular membrane loops of the receptor. Very recently, based on eLP 2 peptide synthesis and structural determin- ation, the NMR structure of a constrained peptide mimicking the TP eLP 3 domain has been determined [15]. To further define the ligand recognition pocket in the extracellular loops for the TP receptor, experimental three- dimensional structure of the eLP 1 region is needed. In t his paper, a synthetic peptide corresponding to the TP eLP 1 (Fig. 1B) was synthesized by the constrained peptide synthesis technique using homocysteines to link t he N and C termini of the peptide, forming a designed distance to connect the corresponding transmembrane domains. After synthesis and cyclization of the constrained eLP 1 peptide, HPLC was used to purify the peptide to homogeneity. The correct molecular mass of the peptide was then confirmed by mass spectrometry [17]. NMR and assignment for TP eLP 1 Two-dimensional 1 H NMR spectra of the constrained eLP 1 peptide were recorded in H 2 O as described in Experimental procedures, and the 1 H NMR assignments were a ccom- plished by the standard sequential a ssignment t echnique [16,18–20] as described [14]. A ll of the assignments were performed i n a procedure including spin system identifica- tion and sequential assignment using a combination of TOCSY (Fig. 2A), DQF-COSY (data not shown), and NOESY (Fig. 2B) spectra. T he complete proton reson- ance assignments for the constrained eLP 1 peptide are 3008 K H. Ruan et al. (Eur. J. Biochem. 271) Ó FEBS 2004 summarized in Table 2. The correct assignment for the peptide was further validated by the FELIX 2000 auto- assignment program. Secondary structure of the constrained eLP 1 peptide analyzed using two-dimensional NMR data Using the assignment information, the secondary structures of the peptides could be predicted by analysis of the chemical shifts, inter-residue NOE connectivities, and the 3 J NHa coupling c onstants. The 3 J NHa coupling c onstants were obtained by the direct measurement of 3 J NHa values and by comparing the intensities of NH-aH cross peaks, which were grouped as strong (J > 6 Hz), medium (J ¼ 4– 6 Hz) and weak (J < 4 Hz ). Weak and strong 3 J NHa coupling constants have been used to identify helical and b-shee t structures, respectively [21,22]. It has been well established that the chemical shifts, which deviate from the Ôrandom coilÕ reference values (conformational shifts), are closely correlated to the type of secondary structure in proteins and peptides [16,23]. In particular, aHandNH conformational shifts have been proposed as markers for characterization of a peptide’s helical structures in solution. Large conformational s hifts (> 0.3 p.p.m. upfield) are a sensitive and powerful sign f or the presence of a helical structure. The medium-range NOE connectivities in the NOESY spectra and the strength of the 3 J NHa coupling from the NH-CaH c ross peaks in the DQF-COSY spectra are summarized in Fig. 3. These data suggests the presence of a b-turn segment within peptide residues 12–15, and an a chelix within peptide residues 2–9. Construction of the three-dimensional structure of the constrained eLP 1 peptide After the complete assignment, the volumes of the identified cross-peaks in the NOESY spectra of the peptide were converted into c onstraints b y the FELIX 2000 program. Fig. 2. Assigned TOCSY and NOESY spectra of the TP eLP 1 . (A) Expanded aH-NH region of the T OCSY spe ctrum (50 ms mixing time) for the T P eLP 1 in H 2 O. The spectrum was recorded at 298 K. (B) Expanded aH-NH region of the N OESY spectrum (200 ms mixing time) for the TP eLP 1 in H 2 O. This spectrum was also r ecorded at 298 K. Ó FEBS 2004 Thromboxane A2 receptor structure and function (Eur. J. Biochem. 271) 3009 A total of 208 constraints were obtained, including 76 intraresidue, 78 s equential and 54 lon g range. The number of the constraints p er residue for the eLP 1 peptide i s s hown in Fig. 4. In addition, nine dihedral angles for eLP 1 were extracted from the DQF-COSY spectrum. On the basis of the NOE constraints and dihedral angles, first-generation structures of the eLP 1 peptide, 100 in total, w ere obtained using the DGII program. To further refine the conformation, energy refinement calculations (minimization/dynamic) were then carried out based on t he best distance geometry structure using DISCOVER , and 12 structures were obtained and superimposed as shown in Fig. 5. The view of the NMR structures clearly shows the b-turn co nformations in the residues 12–15. The distance between the N and C termini of TP eLP 1 is 12.4 A ˚ (Fig. 5 ). The a helical structure was localized between residues 2 and 9. Configuration of the NMR structures of the three extracellular loops in TP receptor model The NMR structures of eLP 2 and e LP 3 were grafted onto the TP receptor working m odel, which was constructed by Table 2. Proton chemical shifts for TP eLP 1 peptide. Residues HN Ha Hb1Hb2 Others hCys1 4.446 2.024 2.456 Gln2 8.691 4.453 1.897 2.283 His3 8.609 4.287 3.114 3.191 7.255, 8.544 Ala4 8.155 4.252 1.280 Ala5 8.202 4.205 1.269 Leu6 7.962 4.211 1.363 1.457 0.738, 0.803, 1.457 Phe7 8.002 4.441 2.896 7.095 Glu8 7.95 4.205 1.764 1.865 2.177 Trp9 8.013 4.435 3.120 7.176, 7.376, 7.495, 10.03 His10 7.878 4.393 2.908 3.049 7.076, 8.473 Ala11 8.025 4.075 1.269 Val12 8.013 3.998 1.976 0.832 Asp13 8.424 4.883 2.638 2.841 Pro14 4.323 2.212 1.941, 3.739 Gly15 8.361 3.794, 3.875 Cys16 7.954 4.411 2.872 Arg17 8.308 4.276 1.704 1.799 1.552, 3.120, 7.132 Leu18 8.178 4.311 1.540 0.797, 0.844, 1.599 hCys19 8.237 4.464 2.005 2.471 Fig. 3. Amino acid sequence of the TP receptor eLP 1 and a survey of sequential and medium range NOEs, HN-Ha coupling constants of the peptide. The values of the 3 J NHa coupling constants a re repo rted in the notation of S ( strong), M ( medium), and W (weak). For the sequential NOE connectivities d aN(i,i+1) ,d aN(i,i+2) ,d aN(i,i+3 ) ,d aN(i,i+4) ,d NN(i,i+1) , d NN(i,i+2) ,andd NN(i,i+3) are indicated by lines starting and ending a t the position of t he intera cting residues. At t he bottom of the figure, the location of the helix struc ture and b-turn are shown. Fig. 4. Number of constraints per residue for TP eLP 1 . Intraresidue (filled bars), sequential (hatched bars), and lo ng range ( open bars). Fig. 5. Superimposition of the best 12 structures of the constrained TP eLP 1 peptide using just the a-carbons obtained from energy refinement calculations. The distance b etween the N and the C termini (residues 2–18) is 12.4 A ˚ . 3010 K H. Ruan et al. (Eur. J. Biochem. 271) Ó FEBS 2004 homology modeling using the crystallographic s tructure of bovine rhodopsin as a template [10,14,15]. The configur- ation was performed by the connection of the individ ual loop structures to the corresponding transmembrane domains, u sing the common trans conformation for t he peptide bond. For eLP 1 , the detailed connections in the three-dimensional structural view are shown in Fig. 6 . After c onnection o f t he NMR structures o f the three extracellular loops to the c orresponding modeled structures of the transmembrane domains (Fig. 7A), the next key factor in setting a correct conformation of the three loops together was to adjust the configuration b etween the loops. A disulfide bond was formed between residue 105 in eLP 1 andresidue183ineLP 2 (Fig. 7B); this bond has been identified by m utagenesis [24] and protein chemical studies [25]. After the formation of the disulfide bond, a 500-step energy-minimization was used to refine the topological arrangement of the three loops (Fig. 7 B). The major conformational change o f the extracellular loops was observed in the region with the sequence WCF in eLP 2 , which is a highly conserved region in the prostanoid receptors (Fig. 7). To test the extent of flexibility for the configured topology of the three loops, a dynamic approach was used, in which the molecular movement was stimulated by changes in the temperature. Limited conformational change (rmsd ¼ 1.2 A ˚ ) was observed in the dynamic studies. These results indicated t hat the configuration of the extracellular loops was in a reasonable format (Fig. 8). Structural characterization of the ligand recognition site of TP receptor Recombinant protein studies, including mutagenesis and chimerical molecules, and molecular modeling based on the crystal structure of rhodopsin have indicated that the nonpeptide ligands are found mainly in deep ligand- bound sites a mong the transmembrane domains [26–28]. However, the conserved residues in the transmembrane Fig. 6. Detailed connectivities for the TP eLP 1 with the transmembrane region before the connection (A), and after connection (B). The distances between the ends of eLP 1 and the two helices (TM1 and TM2) connecting the loop are shown. Fig. 7. Conformation of the assembled extracellular l oops and the residues that form the ligand recognition pocket of the human TP receptor. (A) Prior to formation of disulfide bond, and therefore prior to the formation of the ligand recognition pocket. (B) After the formation of the disulfide bond between Cys105 in eLP 1 and Cys183 in eLP 2 , resulting in the formation of th e ligand recognition pocket. The transmembrane domains are indicated as TMs. Ó FEBS 2004 Thromboxane A2 receptor structure and function (Eur. J. Biochem. 271) 3011 domains could not explain the ligand selectivity in many GPCR subfamilies, such as the prostanoid receptors. To understand the ligand selectivities, it is important to identify the initial contact residues in the extracellular domains of the receptors. R ecently, the identification of ligand recognition sites on the extracellular loops has focused on prostanoid receptors and other GPCR (Table 1). The three-dimensional NMR st ructural model of the extracellular loops of the TP receptor allows a more clear understanding of the molecular mechanism of specific ligand recognition involving the extracellular loops. The NMR experimental structure of the extracellular loops configured to the tramsmembrane domains provided the first three-dimensional structural information about the ligand recognition pocket of the TP receptor. The structural information revealed s everal critical molecular mechanisms of the specific ligand/receptor interaction, which has been studied over a decade. Cyclic oxidation-reduction reactions on the extracellular loops that regulate the ligand binding affinity of the TP receptor were reported in 1990 [29]. In the absence of structural information about the extracellular loops of the receptor no explanation could be offered for t he molecular mechanism. By using our NMR three-dimensional struc- tural model we are able to display, for the first time, the formation of the disulfide bond between eLP 1 and eLP 2 . The conformation of the residues in eLP 2 that form the ligand recognition pocket is very d ynamic in the oxidation and reduction conditions. Fig. 7 shows the conformation of eLP 2 simulated by the presence (Fig. 7A) and absence (Fig. 7B) of dithiothreitol, using a computational approach. The disulfide bond between Cys105 in eLP 1 and Cys183 in eLP 2 was reduced by breaking the bond using BUILDER AND BIOPOLYMER in the INSIGHT II . The conformation changes of the eLPs without the disulfide bond were studied by 500- step energy minimization using DISCOVER . In a ddition, a dynamic calculation according to changes in temperature was also used for monitoring the conformation movement and defining the final conformation of the eLPs. These studies were limited to the three eLP domains. The conformation of the eLPs in the simulated reduced form of the TP receptor is similar to that of the conformation before the oxidation of the disulfide bond (Fig. 7A). The main conformational change w as localized in eLP 2 ,for which the distance of approximately 6.6 A ˚ between eLP 1 Fig. 8. Dynamic study of the configured topology of the three loops extracellular connected to the transmembrane domains (TMs)oftheTPreceptor.Limited conformational change was observed (20 structures, rmsd 1.2 A ˚ ). 3012 K H. Ruan et al. (Eur. J. Biochem. 271) Ó FEBS 2004 and eLP 2 in the oxidized form of the TP receptor was changed to 15.0 A ˚ by reduction in the reduced form; this resulted in a change in the diameter of the ligand recognition pocket (Fig. 1). These structural changes of eLP 2 explain the difference in biological activity of the receptor under oxidation and re duction conditions. T he native receptor with intact binding activity with its ligand should be in the oxidized form. This structural o bservation f urther supports that residues of Val176, Leu185, Thr186, and Leu187 in eLP 2 region are involved in ligand recognition, which has been concluded f rom our very current NMR a nd mutagen- esis studies [1]. Mutation of the conserved sequence WCF in eLP 2 of the EP receptor has been reported t o change the ligand binding recognition of t he receptor [30]. The relationship of t he changes of the conserved residues in eLP 2 that affected the selectivity of the ligand recognition could be a ddressed by structural information about the e xtracellular loops. The possible explanation provided from our structural model is that the changes of the conserved WCF residues in eLP 2 can affect the dynamic conformational cha nges of eLP 2 ,which lead to the changes in the ligand docking affinity. The NMR structural model also suggests that the conserved sequence WCF in the eight prostanoid receptors is essential to the formation of secondary and three-dimensional conforma- tions for all of the ligand docking po ckets of the prostanoid receptors. However, t he residues show no involvement in the selectivity of ligand binding. In the experiment carried out under at pH 7.4 and 30 °C (similar to the physiological conditions) the ligand-dockin g site could be protected from dithiothreitol inactivation of the TP receptor through prior occupation with the ligand [29]. Our NMR three-dimensional structure of the extra- cellular loops revealed the molecular mechanism of the docking of SQ29, 548 into the ligand recognition pocket based on t he contacts between eLP 2 and the lig and as defined by NMR and mutagenesis studies [1]. The initial docking was set up by the three contacts using the constraints between c2H of Val176 with H2 of SQ29,548, c2H of Thr186 with H8 of SQ29 548 and d1H of Lue187 with H7 of SQ29 548 [1]. A 2000-step energy minimization was then used to find the suitable fit of SQ29 548 in the ligand recognition pocket. Figure 9 shows the relationship of the disulfide bond with ligan d docking. In the free form of the TP receptor, the exposed disulfide bond on the surface of the molecule can be easily broken by a reducing reagent (Fig. 9A). In c ontrast, in the ligand-bound form of the receptor, the disulfide bond is completely covered by the ligand, which protects the dithiothreitol reduction (Fig. 9B). This finding confirmed the hypothesis i n which the disulfide bond is near to the ligan d-docking site of the TP receptor predicted by Dorn in 1990 [29] and Tai’s group in 1996 [24]. Identification of the ligand recognition pocket of the TP receptor on the extracellular domain does not conflic t with the ligand-binding pocket identified in TM3 and TM7. The reason for this agreement is the ligand first coming into contact with t he recognition site on the extracellular domain. The second step will then be the deposition of the ligand into the TM pocket causing the conformation change of the receptor and triggering the coupling of the receptor with the G protein in the intracellular domain s. The first step of binding determines the ligand selectively and the second step of binding is required for performance of the receptor function. To test the hypothesis of a two- ligand interaction site, SQ29, 548 was used to dock onto the identified ligand recognition pocket (Fig. 10A), and the ligand was then moved into the transmembrane binding pocket (Fig. 10B) [12,13]. The energy calculation was allowed to move the ligand from the recognition pocket to the transmembrane binding pocket. The distance between the two sites is about 23.0 A ˚ based on the NMR s tructural model. Discussion Synthetic peptides have been widely used to mimic parts of proteins in order to examine the structu re and functions of selected portions of native proteins, particularly for mem- brane-bound proteins, w hich are difficult to be crystallized for X-ray studies [31–33]. Membrane proteins are generally inserted into a bilayer during protein synthesis. Engelman and Steitz [34] proposed that insertion of a helical hairpin loop structure into t he membrane involves the f ormation of a helical dimer through helix–helix interactions. Lin and Addison’s work on the insertion of membrane helices of integral membrane proteins suggested that the connecting peptide forms a loop to stabilize the transmembrane helix dimer in p reparation for membrane insertion [35]. The primary s tructure of the c onnecting loops may contain information su fficient to fold into native turn structures Fig. 9. The role of the disulfide bond in relation to ligand binding of the TP receptor. (A) The ligand bind ing poc ke t i s o pen and the disulfi de bond is exposed. (B) S Q29, 548 is bound to the ligand binding pocket, therefore conceal- ing the disulfide bond. Ó FEBS 2004 Thromboxane A2 receptor structure and function (Eur. J. Biochem. 271) 3013 which have biological activities even in the absence of the flanking transmembrane helices. Both Takemoto et al.[36] and Konig et al. [37] f ound that isolated peptides compri- sing the C-terminal domain, or the second or third loops in the intracellular portions of bovine rhodopsin themselves have biological activity. These results with the prototypical G protein-coupled protein suggest that extramembraneous parts of the GPCRs may independently fold into a native structure. However, the free peptide in solution may not necessarily adopt an ordered conformation, especially for the t erminal residues, which a re particularly important for the configuration of a peptide mimicking a loop structure and connected to other defined structures. This led us to develop an approach of computation-guided constrained peptide synthesis for structural and functional studies of the GPCRs. In most cases, structural studies of the extracellular domains and ligand recognition sites of mammalian GPCRs have been performed b y the homology modeling approach using the crystal structure of rhodopsin as a template [25,38–44]. Due to lower conservation of the extrameme- brane domains between rhodopsin and mammalian GPCRs, especially in the case of prostanoid receptors (Fig. 1), little information about the structural characteris- tics of the extracellular domains of the prostanoid receptors is available. Crystal structures for the prostanoid receptors are unlikely to be obtained in the near future. Assembly of the NMR structures of the extracellular d omains connected to the transmembrane domain described above for the TP receptor offers an alternative way to quickly characterize the structural features of the extracellular loops of mammalian GPCRs. The principle of this strategy includes the steps of the computer-guided constrained peptide synthesis with precise secondary structural configurations, two-dimen- sional NMR structural determinations, and the fragment structural configuration on the transmembrane domains. The homocysteine disulfide bond used to constrain t he secondary structure of the synthetic peptides, mimicking the TP extracellular loops, can be applied to other GPCRs a s it is believed that the seven t ransmembrane domains of most of the mammalian GPCRs share similar conformations and separations between the h elices connecting the loops. In addition, the constrained peptide synthesis approach also provided an opportunity to synthesize peptides that mim- icked the intracellular loops for t he TP and other GPCRs for structural and functional studies. It should be noted that Yeagle et al. determined the synthetic p eptides with free ends that mimicked the intracellular loops of bovine rhodopsin and provided t hree-dimensional s tructures f or the loops in which the peptides adopt a turn conformation [45–49]. But the defined distance between both of the ends of the peptide were not conclusive because the structures of the N- and C-terminal residues of p eptides were varied. Our constrained peptide overcomes this problem because the constrained N- and C-terminal residues are considered as intraresidues and adopted a conformation similar to that of other residues in solution. The NMR structures described above have confirmed this hypothesis in which the c on- strained peptides gave a defined conformation for the terminal residues of the TP extracellular loops. The identified ligand recognition pocket, located mainly between eLP 1 and eLP 2 of the TP receptor, and the residues important for contacting the ligand in e LP 2 have been supported by mutagenesis studies [24,30] and an affinity labeling experiment [50] for the prostanoid receptors reported from different groups. We could not exclude the possibility that the eLP 1 region may also contain residues involved in forming the ligand recognition pocket. How- ever, based on our findings and affinity labeling studies, it can be concluded t hat the ligand is a nchored mainly to the TP eLP 2 region. This information h as offered a structural template to predict the specific ligand recognition pockets in the same location for other prostanoid receptors. This prediction is based o n the following facts of the eight prostanoid receptors: first, all of the receptors share s imilar topological backbones and the eLP 2 s are highly co nserved; second, the cysteine residues making up the disulfide bond between eLP 1 and eLP 2 are also conserved; third, muta- genesis for the residues in the eLP 2 region for t he EP3 receptors showed the effect of the ligand binding selectivities [30]; and lastly, t he affinity-labeling s tudies for t he TP receptor showed that the initial ligand binding site is in eLP 2 [50]. Our finding is also in agreement with the current observation for the human P2Y 1 receptor, a mammalian Fig. 10. Docking of the TP receptor with its ligand. (A) SQ29 548 docking onto the iden- tified ligand recognition pocket. The docking was performed in respect to the contacts be- tween SQ29 5 48 with the residues V176, L185, T186 and L187 in eLP 2 identified previously [1].(B) SQ29 548 at the TM binding pocket . The docking was based on the c ontac ts between SQ29 548 with S201 and R295 as described previously [ 12,13]. The distance SQ29 548 moved from the ligand recognition pocket to the TM pocket was calculated to be  23.0 A ˚ . 3014 K H. Ruan et al. (Eur. J. Biochem. 271) Ó FEBS 2004 GPCR, in which the ligand recognition site (Ômate-binding siteÕ) is localized on the extracellular domain and the Ôprincipal TM binding siteÕ is in the TM domains [25]. By homology alignment of the eLP 2 regions of the eight prostanoid r eceptors, our NMR structural m odel of the ligand recognition pocket in the TP receptor f urther implies that the conserved residues in the eLP 2 s and the disulfide bond configuration maintain general ligand recognition pockets, and that the variable residues within the eLP 2 of the prostanoid receptors play key roles in the specific ligand recognition which determines the affinity between the receptor and ligand. This hypothesis can be used to explain the observation that each prostanoid receptor can cross- react with other prostanoids with the only difference being in the binding affinity. Determination of the three-dimen- sional structural conformation of the extracellular loops and experimental identification of the ligand recognition pockets for other prostanoid receptors will provide evidence to test our predictions. Acknowledgements We thank Dr X. Gao, Chemistry Department, University of Houston, for acc ess to the NMR facility. An acknowledgement is also made to the Robert A. Welch Foundation and the W. M. Keck Center for Computational Biology at the University of Houston for computer resource support. This work was s upported by NIH Grants HL56712 and NS23327. The NMR facility at University of Houston is founded by the W. M . Keck Foundation. References 1. So , S P., Wu, J., H uang, G., Huang, A., Li, D. & Ruan, K H. (2003) Identification of residues important for ligand binding of thromboxane A 2 receptor in the second extracellular loop using the NMR expe riment-guide d mutagenesis approach. J. Biol. Chem. 278, 10922–10927. 2. Coleman, R.A., Kennedy, I., Humphrey, P.P.A., Bunce, K. & Kumley, P. (1990) Prostanoids and Their Receptor in Compre- hensive Medical Chemistry (Hansch,C.,Sammes,P.G.,Taylor, J.B. & Emmett, J.C., eds), Vol. 3, pp. 643–714. Pergamon Press, Oxford. 3. Ushikubi, F., Nakajima, M., Hirata, M., Okuma, M., Fujiwara, M. & Narumiya, S. (1989) Purification of the thromboxane A 2 / prostagiandin H 2 , receptor from human blood platelets. J. Biol. Chem. 264, 16496–16501. 4. Hirata, M., Hayashi, Y., Ushikubi, F., Yokata, Y., Kageyama, R., Nakanishi, S. & Narumiya, S. (1991) Cloning and expression of cDNA for a human thromboxane A 2 receptor. Nature 349, 617–620. 5. Wess, J. (1997) G-protein-coupled receptors: molecular mecha- nisms involved in receptor activation and selectivity of G-protein recognition. FASEB J. 11 , 346–354. 6. Ji, T .H., Grossman, M. & Ji, I. ( 1998) G protein-coupled receptors. I. Diversity of recepto r–ligand interac tions. J. Biol. Chem. 273, 17299–17302. 7. Henderson, R., Baldwin, J.M., Ceska, T.A., Zemlin, F., Beck- mann, E. & D owning, K.H. (1990) Model for the struct ure of bacteriorhodopsin based on high-resolutio n e lectron cryo-micro- scopy. J. Mol. Biol. 213, 899–929. 8. Grigorieff, N., Ceska, T.A., Downing, K.H., Baldwin, J.M. & Henderson, R. (1996) Electron-crystallographic refinement of the struc ture o f bacteriorhodopsin. J. Mol. Biol. 259, 393– 421. 9. Peb ay-Peyroula, E., Rummell, G ., Rosenbusch, J.P. & Landau, E.M. (1997) X-ray structure of bacteriorhodopsin at 2.5 ang- stroms from microcrystals grown in lipidic cubic p hases. Science 277, 1676–1681. 10. Pa lc zewski, K., Kuma saka, T., Hori, T., Behnke, C.A., Moto- shima,H.,Fox,B.A.,LeTrong,I.,Teller,D.C.,Okada,T., Stenkamp, R.E., Yamamoto, M. & Miyano, M. (2000) Crystal structure of r hodopsin : a G p rotein-c oupled r eceptor. Science 289, 739–745. 11. Negishi, M., Sugimoto, Y. & I chikawa, A. (1995) Molecular mechanisms of diverse actions o f prostanoid receptors. Biochim. Biophys. Acta 1259, 109–120. 12. Funk, C.D ., Furci, L ., Moran, N . & F itzgerald, G.A. (1993) P oint mutation in the sev enth hydrophobic dom ain of the hum an thromboxane A 2 receptor allows discrim ination b etween ago nist and antagonist binding sites. Mol. Pharmacol. 44, 934–939. 13. Yamamoto, Y., Kamiya, K. & Terao, S. (1993) Modeling of human thromb oxane A 2 receptor an d a nalysis of the receptor– ligand interaction. J. Med. Chem. 36, 820–825. 14. Ruan, K H., So, S P., Wu, J., Li, D., Huang, A. & Kung, J. (2001) So lution structure of t he second extra cellular loop o f human thromboxane A 2 receptor. Biochemistry 40, 275–280. 15. Wu, J., So, S P. & Ruan, K H. (2003) Solution structure of the third extracellular l oopofhumanthromboxaneA 2 receptor. Arch. Biochem. Biophys. 414, 287–293. 16. Wu ¨ thrich, K. ( 1986) NMR of Proteins and Nucleic Acids. John Wiley & Sons, New York. 17. Moore, W.T. & Caproili, R.M. (1991) Techniques in Protein Chemistry II (Villafranca, J.J., ed.), pp. 511–528. Academic Press, New York. 18. Englander, S.W. & Wand, A.J. (1987) Main-chain-directed strat- egy for the assignment o f 1H NMR spe ctra of proteins. Bio- chemistry 26, 5953–5958. 19. Chazin, W.J., Rance, M. & Wright, P.E. (1988) Complete assignment of the 1H nuclear magnetic resonance spectrum of French bean plastocyanin. Application of an integrated approach to spin system identification in proteins. J. Mol. Biol. 202, 603–622. 20. Basus, V.J. (1989) Proton nuclear magnetic resonance assign- ments. Methods Enzymol. 177, 132–149. 21. Bystrov, V.F. ( 1976) Contribution of N MR spectro scopy to the study of structure-function relations of proteins and peptides. Prog. Nucl. Magn. Reson. Spectros. 10 , 41–82. 22. Pardi, A., Billeter, M., Braun, W. & Thrich, K. (1984) Calibration of the angular dep endenc e of the amide proton-C a lpha proton coupling constants, 3JHN alpha, in a globular protein. Use of 3JHN alpha f or identific ation o f h elic al se co ndary structure. J. Mol. Biol. 180, 741–751. 23. Wishart, D.S., Sykes, B.D. & Richards, F.M. (1992) The chemical shift index: a fast and simple metho d fo r the assignme nt o f p rotein secondary s tructure through NMR spe ct roscopy. Biochemistry 31, 1647–1651. 24. Chiang, N., Kan, W.M. & Tai, H H. (1996) Site-directed muta- genesis of cysteinyl and serine residues of human thromboxane A 2 receptor in insect cells. Arch. Biochem. Biophys. 334, 9–17. 25. Moro, S., Hoffmann, C. & Jacobson, K.A. (1999) Role of the extracellular loops of G pro tein-coupled receptors i n ligand recognition: a molec ular mo deling stu dy of the human P2Y1 receptor. Biochemistry 38 , 3498–3507. 26. Baldwin, J.M. (1993) The p robable arrangement of the helic es in G protein-coupled receptors. EMBO J. 12, 1693–1703. 27. Schwartz, T.W. (1994) Loc ating ligand-binding sites in 7TM receptors by protein engineering. Curr. Opin. Biotechnol. 5, 434– 444. 28. Strader, C.D., Fong, T.M., Graziano, M.P. & Tota, M.R. (1995) The family of G-protein-coupled receptors. FASEB J. 9, 745–754. Ó FEBS 2004 Thromboxane A2 receptor structure and function (Eur. J. Biochem. 271) 3015 [...]... (1997) The first and second cytoplasmic loops of the G-protein receptor, rhodopsin, independently form beta-turns Biochemistry 36, 3864–3869 49 Yeagle, P.L., Alderfer, J.L & Albert, A.D (1997) Three-dimensional structure of the cytoplasmic face of the G protein receptor rhodopsin Biochemistry 36, 9649–9654 50 Turek, J.W., Halmos, T., Sullivan, N.L., Antonakis, K & Le Breton, G.C (2002) Mapping of a ligand- binding... (1999) The role of amino acids in extracellular loops of the human P2Y1 receptor in surface expression and activation processes J Biol Chem 274, 14639–14647 53 Olah, M.E., Jacobson, K.A & Stiles, G.L (1994) Role of the second extracellular loop of adenosine receptors in agonist and antagonist binding Analysis of chimeric A1/A3 adenosine receptors J Biol Chem 269, 24692–24698 54 Kim, J., Jiang, Q., Glashofer,... reactions regulate thromboxane A2/ prostaglandin H2 receptor number and affinity in human platelet membranes J Biol Chem 265, 4240– 4246 30 Breyer, R.M & Audoly, L (1997) The second extracellular loop of the prostaglandin EP3 receptor is an essential determinant of ligand selectivity J Biol Chem 272, 13475–13478 31 Ruan, K.-H., Spurlino, J., Quiocho, F.A & Atassi, M.Z (1990) Acetylcholine receptor alpha–bungarotoxin... (1990) Binding of thyroid hormones to human hemoglobin and localization of the binding site J Protein Chem 9, 349–350 34 Engelman, D.M & Steitz, T.A (1981) The spontaneous insertion of proteins into and across membranes: the helical hairpin hypothesis Cell 23, 411–422 35 Lin, J & Addison, R (1995) A novel integration signal that is composed of two transmembrane segments is required to integrate the Neurospora... Structure of the carboxy-terminal domain of bovine rhodopsin Nat Struct Biol 2, 832–834 46 Yeagle, P.L., Alderfer, J.L & Albert, A.D (1995) Structure of the third cytoplasmic loop of bovine rhodopsin Biochemistry 34, 14621–14625 47 Yeagle, P.L., Alderfer, J.L & Albert, A.D (1996) Structure determination of the fourth cytoplasmic loop and carboxyl terminal domain of bovine rhodopsin Mol Vis 2, 12 48 Yeagle,... receptor alpha–bungarotoxin interactions: determination of the region to region contacts by peptide–peptide interactions and molecular modeling of the receptor cavity Proc Natl Acad Sci USA 87, 6156–6160 32 Ruan, K.-H., Stiles, B.G & Atassi, M.Z (1991) The short-neurotoxin-binding regions on the alpha-chain of human and Torpedo californica acetylcholine receptors Biochem J 274, 849–854 33 Atassi, M.Z., Zhouhair,... N.L., Antonakis, K & Le Breton, G.C (2002) Mapping of a ligand- binding site for the human thromboxane A2 receptor protein J Biol Chem 277, 16791–16797 51 Perlman, J.H., Colson, A.-O., Jain, R., Czyzewski, B., Cohen, L.A., Osman, R & Gershengorn, M.C (1997) Role of the extracellular loops of the thyrotropin-releasing hormone receptor: evidence for an initial interaction with thyrotropin-releasing hormone... Modelling the P2Y purinoceptor using rhodopsin as template Drug Des Disc 13, 133–154 44 Metzger, T.M., Paterlini, M.G., Portoghese, P.S & Ferguson, D.M (1996) Application of the message-address concept to the docking of naltrexone and selective naltrexone-derived opioid antagonists into opioid receptor models Neurochem Res 21, 1287–1294 45 Yeagle, P.L., Alderfer, J.L & Albert, A.D (1995) Structure of the. .. Glashofer, M., Yehle, S., Wess, J & Jacobson, K.A (1996) Glutamate residues in the second extracellular loop of the human A2a adenosine receptor are required for ligand recognition Mol Pharmacol 49, 683–691 55 Zhao, M.M., Hwa, J & Perez, D.M (1996) Identification of critical extracellular loop residues involved in alpha 1-adrenergic receptor subtype-selective antagonist binding Mol Pharmacol 50, 1118–1126... neurotransmitter G-binding proteincoupled receptors Mol Pharmacol 40, 8–15 41 Maloneyhuss, K & Lybrand, T.P (1992) Three-dimensional structure for the beta 2 adrenergic receptor protein based on computer modeling studies J Mol Biol 225, 859–871 42 Teeter, M.M., Froimowitz, M., Stec, B & DuRand, D.J (1994) Homology modeling of the dopamine D2 receptor and its testing by docking of agonists and tricyclic antagonists . be the deposition of the ligand into the TM pocket causing the conformation change of the receptor and triggering the coupling of the receptor with the. 9. Configuration of the NMR structures of the three extracellular loops in TP receptor model The NMR structures of eLP 2 and e LP 3 were grafted onto the TP receptor