Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

The -adrenergic receptors (ARs) are members of the seven transmembrane G protein-coupled receptor family and are activated by catecholamine and related molecules. The ligand-binding site of the AR has been extensively characterized by the use of a variety of techniques (Wong et al., 1988; Dohlman et al., 1988; Savarese and Fraser, 1992; Strader et al., 1994; Hockerman et al., 1996). Initial deletion mutagenesis of the hamster ₂AR showed that the hydrophilic loop regions connecting TMs of the receptor are not important for agonist or antagonist binding (Dixon et al., 1987). Point mutations of the hamster ₂AR have revealed a key amino acid residue in TM3 (Asp¹¹³) that is essential for high-affinity binding of both agonists and antagonists, as well as key residues in TM5 (Ser²⁰⁴ and Ser²⁰⁷) that are assumed to interact with two hydroxyl groups of the catechol ring and be critical for agonist activation of the receptor (Strader et al., 1988, 1989). These data suggested that the ligand-binding domain of the AR resided within the hydrophobic TMs. Recently, Wieland et al. (1996) reported that Asn²⁹³ of the ₂AR in TM6 interacts with the -hydroxyl group of AR ligands and is responsible for stereoselectivity.

Binding domains of AR subtype-selective antagonists and an agonist, norepinephrine, were also studied by several groups. Frielle et al. (1988) reported that TM6 and TM7 of AR appear to play an important role in determining binding of the ₁- and ₂-selective antagonists such as betaxolol and ICI118551. They also reported that the selectivity of norepinephrine, which shows about 10 times higher affinity for the ₁AR than for the ₂AR, is largely determined by TM4 of the ₁AR. Dixon et al. (1989) showed that TM4 is responsible for ₁-selective binding of norepinephrine, using chimeras of the hamster ₂AR and the human ₁AR. Marullo et al. (1990) examined the ligand-binding regions for ₁- or ₂-subtype-selective antagonists by analyzing chimeric receptors that replaced several TMs of the ₁AR or the ₂AR with corresponding TMs of the ₂AR or the ₁AR. They concluded that no single TM could be responsible for the selectivity of AR antagonists. Because they exchanged two or more TMs at same time, their methods might not estimate the contribution of a particular single TM to the subtype selective binding. Furthermore, they did not analyze the binding site(s) of highly selective AR agonists.

Thus, the domains and amino acids responsible for the high-affinity binding of ₁- and ₂-selective agonists have not been examined so far. We determined recently that the ₂-selective agonist binding domain was mainly located in TM7 by using ₁/₂-chimeric receptors and ₂-selective agonists such as TA-2005 and salmeterol (Isogaya et al., 1998; Kikkawa et al., 1998). We also determined that Tyr³⁰⁸ in TM7 was a main amino acid that determined the high-affinity binding of ₂-selective agonists by analyzing alanine-substituted mutants. In the present study, we extended the previous finding to other ₂-selective agonists such as formoterol and procaterol. We found that Tyr³⁰⁸ bound ₂-selective agonists via hydrophobic or hydrophilic interactions, which were dependent on the structures of ligands. We also determined the amino acids most important for the ₁AR to bind the subtype selective agonists with high affinity. We built three-dimensional models of AR-subtype-selective agonist complexes based on the predicted structure of rhodopsin and the results of mutagenesis experiments.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. The plasmid constructs pBC-₁ and -₂ encoding for the human ₁- and ₂ARs were kindly provided by Dr. R. J. Lefkowitz (Duke University, Durham, NC). IPS-339 {(t-butyl-amino-3-ol-2-propyl) oximino-9-fluorene-p-hydroxy-benzoate}, salbutamol, salmeterol, procaterol, formoterol, T-0509 {()-(R)-1-(3,4-dihydroxyphenyl)-2-[(3,4-dimethoxyphenethyl)amino] ethanol}, xamoterol, prenalterol, T-1583 {-(3,4,5-trimethoxyphenethylaminomethyl)-[3,4-dihydroxybenzyl-alcohol] hydrochloride}, denopamine, and dobutamine were synthesized at the Lead Optimization Research Laboratory, Tanabe Seiyaku (Saitama, Japan). The structure of these ₁- or ₂AR-selective agonists is shown in Fig. 1. ()Norepinephrine-bitartrate, (±)propranolol, and DEAE-dextran were obtained from Sigma Chemical Co. (St. Louis, MO). Dulbecco's Modified Eagle's medium and gentamicin were from Life Technologies, Inc., (Rockville, MD). Taq and Pfu DNA polymerases were obtained from Takara (Siga, Japan) or Stratagene (La Jolla, CA), respectively. GTP was purchased from Seikagaku (Tokyo, Japan). ¹²⁵I-labelled cyanopindolol (¹²⁵I-CYP) was obtained from Amersham Pharmacia Biotech (Arlington Heights, IL) or New England Nuclear (Boston, MA). Fetal bovine serum was from JRH Biosciences (Lenexa, KS).

Construction of Chimeric ₁/₂ARs and Alanine-Substituted AR Mutants. Chimeric ₁/₂-receptors were constructed by polymerase chain reaction techniques with Taq or Pfu DNA polymerase as described (Higuchi, 1989). The sequences of the amplified regions were confirmed by the dideoxy chain termination method (Sanger et al., 1977). The amplified region was combined with the rest of the ₁- or ₂AR sequences to obtain full-length ₁/₂ chimeras, and chimeric cDNAs were finally inserted into the EcoRI and BamHI or EcoRI and SalI sites of mammalian expression vector pCMV5. The positions of the junction for individual ₁/₂-chimeric receptors are as follows (numbers refer to amino acid positions in the human ₁- and ₂AR sequences): CH-1, ₁ 1-84/₂ 60-413; CH-2, ₂ 1-71/₁ 97-131/₂ 107-413; CH-3, ₂ 1-295/₁ 347-381/₂ 331-413; CH-4, ₂ 1-71/₁ 97-131/₂ 107-295/₁ 347-381/₂ 331-413; CH-5, ₂ 1-59/₁ 85-477; CH-6, ₁ 1-96/₂ 72-106/₁ 132-477; CH-7, ₁ 1-346/₂ 296-330/₁ 382-477; CH-8, ₁ 1-96/₂ 72-106/₁ 132-346/₂ 296-330/₁ 382-477 (Fig. 2). Alanine-substituted mutants of the ₁- and ₂ARs and phenylalanine-substituted mutant of the ₂AR were constructed by PCR using the Quick Change site-directed mutagenesis method as described by Isogaya et al. (1998). After the mutations were confirmed, the fragments containing the substitutions were ligated with other portions of the receptors. The expression vector pCMV5 was used for the alanine-substituted mutants except for the M98A-₁AR. The expression vector pEF/myc/cyto was used for the mutant due to low expression with pCMV5.

Transient Expression of Wild Type (WT) or ₁/₂-Chimeric Receptors in COS-7 Cells. The cDNAs encoding for the human ₂AR in pBC12BI, the human ₁AR, or the ₁/₂-chimeric receptors in pCMV5 were transfected into COS-7 cells by the DEAE-dextran method (Cullen, 1987). Before the day of transfection, COS-7 cells were seeded at 1.0 to 1.5 × 10⁶ cells/100-mm dish. The amount of the WT or ₁/₂-chimeric receptor cDNAs was 5 µg/100-mm dish. All cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and gentamicin (10 µg/ml). Two to three days after the transfection, the cells were harvested for preparation of the crude membrane fraction.

Membrane Preparation. The COS-7 cells were rinsed with 10 ml of ice-cold PBS and mechanically detached in 1 ml of an ice-cold lysis buffer containing 10 mM Tris-HCl (pH 7.4), 5 mM EDTA, 5 mM EGTA, 10 µg/ml benzamidine, 10 µg/ml soybean trypsin inhibitor (TypeII-S), and 5 µg/ml leupeptin. The lysate was centrifuged at 45,000g for 10 min at 4°C. The pellet was resuspended in 1 ml of a lysis buffer with a Potter type homogenizer and stored at 80°C until use. Protein concentration was determined by the method of Lowry et al. (1951).

Radioligand Binding Assay. Radioligand binding studies were carried out in a buffer containing 75 mM Tris-HCl (pH 7.4), 12.5 mM MgCl₂, and 2 mM EDTA in the presence of 100 µM GTP at 37°C for 60 min using 0.2 to 10 µg of membrane protein. Competition binding assays were performed using the indicated concentration of ¹²⁵I-CYP and various concentrations (0-10 mM) of unlabeled ligands in the presence of 100 µM GTP. The binding reaction was terminated by the rapid filtration over Whatman GF/C filters and washed three times with the solution containing 25 mM Tris-HCl (pH 7.4) and 1 mM MgCl₂. Nonspecific binding was determined in the presence of 5 µM (±)propranolol. The radioactivity remaining on the filter was counted by a gamma counter.

Data Analysis. All data shown are mean values ± S.E. for n determinations. Equilibrium dissociation constants were determined from saturation isotherms. Radioligand binding data obtained from competition curves were analyzed by a nonlinear regression analysis to determine EC₅₀ values and K_i values using PRISM software (GraphPad Software Inc., San Diego, CA). Statistical significance was assessed with one-way ANOVA for the multiple comparisons using JMP software (SAS Institute, Cary, NC). ANOVA post hoc comparisons were made with the Dunnett's test.

Computer Modeling of -Selective Agonists-₁- or ₂AR Complexes. The initial coordinates of the backbone and side chain atoms were modeled by assigning the amino acids of the ₁- and ₂ARs to the model of rhodopsin built by Baldwin et al. (1997), using the Biopolymer module of SYBYL software package (TRIPOS Assoc., St. Louis, MO). The side chain conformations were optimized by the dead-end algorithm with the "large-size" rotamer library (Desmet et al., 1992; Tanimura et al., 1994). The selective agonists (procaterol, formoterol, or denopamine) were docked to the ₂- or ₁ARs manually by satisfying the following established interactions: Asp¹¹³ of the ₂AR (Asp¹³⁴ of the ₁AR) with the protonated amine, Ser²⁰⁴ (Ser²²⁹) and Ser²⁰⁷ (Ser²³²) with catechol or equivalent entities, Asn²⁹³ (Asn³⁴⁴) with the hydroxyl group at the  position, and Phe²⁹⁰ (Phe³⁴¹) with the phenyl ring or equivalent groups. After docking procedures, the entire structures were energy minimized with positional restraints on the C atoms in the transmembrane helices by MAXIMIN2 of SYBYL software (TRIPOS Assoc.).

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Affinities of Propranolol for ₁/₂AR Chimeras. We constructed a series of ₁/₂-chimeric receptors. Because a nonselective agonist isoproterenol binds to the AR at at least three sites (i.e., Asp¹¹³, Ser²⁰⁴, and Ser²⁰⁷) and because -selective agonists often have substituents at an amino group, we focused on TM1, TM2, and TM7 of the AR to allow the AR to bind the subtype-selective agonists with high affinity. We used salbutamol, formoterol, and procaterol for ₂-selective agonists (see Fig. 1A for structures) and T-0509, T-1583, xamoterol, denopamine, dobutamine, and norepinephrine for ₁-selective agonists (Fig. 1B). One of TM1, TM2, or TM7, or both TM2 and TM7 of the ₂AR were replaced by the homologous regions of the ₁AR. These chimeric receptors were termed CH-1, CH-2, CH-3, and CH-4. On the other hand, one of TM1, TM2, or TM7 or both TM2 and TM7 of ₁AR were replaced by the homologous regions of the ₂AR. They were termed CH-5, CH-6, CH-7, and CH-8 (Fig. 2). Table 1 shows that the affinities of propranolol for CH-1 to CH-4 were essentially the same as those of the WT ₂AR. Although the affinities of propranolol for the three chimeras (CH-5, CH-7, and CH-8) were significantly changed by the introduction of TMs of the ₂AR into the ₁AR, the changes in the affinities were relatively small compared with those of selective agonists (Table 1). Furthermore, the increases of the affinities for the chimeras did not accompany the decreases of the affinities for the reciprocal chimeras. This suggested that the increases of the affinities for propranolol were not specific for a particular TM.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   A, the structures of 2AR-selective or -nonselective agonists and antagonists. B, the structures of the 1AR-selective agonists.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   The structures of 1/2 chimeric (CH) receptors. The peptide sequences of the 1AR are shown by thin lines and those of the 2AR are indicated by thick lines. The positions of the junctions are described under Experimental Procedures.

View this table: [in this window] [in a new window]   TABLE 1 Effects of replacement of transmembrane regions with corresponding portions of the 1AR on ligand-binding characteristics of the 2AR The binding of ligands to the WT-2AR and 1/2-chimeric receptors were determined by competition with 50pM 125I-CYP. The data were analyzed using the nonlinear least-squares regression computer program as described under Experimental Procedures. The results are shown as the mean ± S.E. from three to four separate experiments.

Affinities of Formoterol and Procaterol for ₁/₂AR Chimeras. Formoterol and procaterol are highly ₂ selective. The ratios of K_i values [K_i(₁) to K_i(₂)] of formoterol or procaterol were 88 or 114, respectively (Table 1). The replacement of TM2 of the ₂AR with that of the ₁AR (CH-2) decreased the affinities of the two agonists about 7- to 17-fold (Table 1). The contribution of TM1 and TM7 to ₂ selectivity of the two agonists was low compared with the contribution of TM2. The affinities of the two agonists were further decreased by the replacement of TM7 together with TM2 of the ₂AR (CH-4), and the affinities for the reciprocal mutant of CH-4 (CH-8) were increased nearly to the same values as for the WT ₂AR (Table 1). This indicated that both TM2 and TM7 determined the high-affinity binding of formoterol and procaterol.

Affinities of Salbutamol for ₁/₂AR Chimeras. Salbutamol was less potent and selective than formoterol and procaterol [ratio of K_i(₁) to K_i(₂) is about 10] (Table 1). The affinity of salbutamol for the ₂AR was 2 to 3 orders of magnitude lower than those of the other ₂-selective agonists. The affinities of salbutamol were significantly decreased by the replacement of TM2 of the ₂AR with that of the ₁AR and were increased by the introduction of TM1 or TM2 of the ₂AR into the ₁AR (Table 1). Although the affinity of salbutamol was decreased in CH-1, the change in the affinity was relatively small compared with that of CH-2 (less than 3-fold versus more than 6-fold). These results suggested that the contribution of TM1 to ₂-selective binding of salbutamol was small. The replacement of TM7 of the ₂AR with that of the ₁AR did not change the affinity of salbutamol for the resulting chimera (CH-3). When both TM2 and TM7 of the ₂AR (or the ₁AR) were replaced with those of the ₁AR (or the ₂AR), the chimeras (CH-4 or CH-8) showed the decreased (or increased) affinities for salbutamol (Table 1). These results suggested that the ₂ selectivity of salbutamol was mainly determined by both TM2 and TM7.

Affinities of the ₂-Selective Antagonist IPS-339 for ₁/₂AR Chimeras. One of the ₂AR-selective antagonists, IPS-339, showed about 40 times higher affinity for the ₂AR than for the ₁AR (Table 1). When TM1, TM2, or TM7 of the ₂AR were replaced by the corresponding regions of the ₁AR (CH-1 to CH-4), the affinities of IPS-339 for these chimeras were decreased by 3.5- to 8.2-fold (Table 1). On the other hand, the transfer of TM1 or TM2 from the ₂AR to the ₁AR (CH-5 and CH-6) did not increase the affinities of IPS-339 for these chimeras (Table 1). The slightly increased affinity of IPS-339 caused by transferring TM7 from the ₂AR to the ₁AR suggested that the major determinant of ₂AR selectivity of IPS-339 was TM7, in spite of structural differences between IPS-339 and the ₂AR-selective agonists.

Affinities of Synthetic ₁-Selective Agonists for ₁/₂AR Chimeras. We examined the AR selectivity of T-0509, denopamine, xamoterol, dobutamine, T-1583, and prenalterol. These are known as ₁-selective agonists when administered to whole animals or to isolated tissues. Dobutamine, T-1583, and prenalterol showed little ₁ selectivity in the binding experiments. The ratios of K_i values [K_i(₁) to K_i(₂)] of these agonists were less than 3-fold (Table 2). We therefore did not study these agonists in detail. Among these agonists, T-0509, xamoterol, and denopamine showed significantly higher affinities for the ₁AR than for the ₂AR. The binding experiments using the recombinant ARs expressed in COS-7 cells showed that the selectivity of these agonists was relatively low, compared with the selectivity of ₂-selective agonists such as procaterol and formoterol (Table 2). Replacement of TM2 of the ₁AR with the homologous region of the ₂AR (CH-6) decreased the affinities of these three agonists, and transfer of TM2 of the ₁AR to the ₂AR (CH-2) increased the affinities of these agonists to nearly same values as for the WT ₁AR (Table 2). The effect of replacement of TM2 together with TM7 on xamoterol binding was essentially the same as the effect of replacement of TM2 alone. Although the affinities of T-0509 and denopamine for CH-6 (the ₁AR with TM2 of the ₂AR) were decreased, those for the CH-8 (the ₁AR with TM2 and TM7 of the ₂AR) were increased, for an unknown reason (Table 2). These data suggested that TM2 of the ₁AR determines the ₁ selectivity, even though it is not a sole determinant of the ₁-selective binding site.

View this table: [in this window] [in a new window]   TABLE 2 Effects of replacement of transmembrane regions with corresponding portions of the 2AR on ligand-binding characteristics of the 1AR The binding of ligands to the WT 2AR and 1/2-chimeric receptors were determined by competition with 50pM 125I-CYP. The data were analyzed using the nonlinear least-squares regression computer program as described under Experimental Procedures. The results are shown as the mean ± S.E. from three to four separate experiments.

Affinities of the Endogenous ₁-Selective Agonist Norepinephrine for ₁/₂AR Chimeras. Norepinephrine is the endogenous ₁-selective agonist. We confirmed the ₁ selectivity of norepinephrine (ratio of K_i(₂) to K_i(₁) is about 9.0) (Table 3). We also examined the affinities of norepinephrine for the various ₁/₂AR chimeras. The replacement of TM7 but not TM2 of the ₁AR with the homologous region of the ₂AR decreased the affinity of norepinephrine. These results indicated that TM7 contributed to ₁-selective binding of norepinephrine, which constrasted with synthetic ₁-selective agonists. The importance of TM7 for ₁-selective binding was further supported by the finding that the introduction of TM7 of norepinephrine of the ₁AR into the ₂AR (CH-3) increased the affinity of norepinephrine. These data suggested the contribution of different TMs to subtype-selective binding of structurally different ₁-selective agonists.

View this table: [in this window] [in a new window]   TABLE 3 Effects of replacement of transmembrane domains of 1AR with corresponding regions of the 2AR on binding characteristics of norepinephrine The binding of norepinephrine to the WT 2AR and 1/2-chimeric receptors were determined by competition with 50-100 pM 125I-CYP. The data were analyzed using the nonlinear least-squares regression computer program as described under Experimental Procedures. The results are shown as the mean ± S.E. from three to five separate experiments.

Effects of Substitution of Amino Acids with Alanine in TM7 of the ₂AR on Binding of Procaterol and Formoterol. We have recently reported that TM7 of the ₂AR played an important role in determining the high-affinity binding of ₂-selective agonists such as TA-2005 and salmeterol, and Tyr³⁰⁸ in TM7 was the most important amino acid for the high-affinity binding (Isogaya et al., 1998; Kikkawa et al., 1998). To examine the role of TM7 for binding of procaterol and formoterol in detail, we expressed alanine-substituted mutants of the ₂AR, in which the amino acids in TM7 of the ₂AR different from those of the ₁AR were individually changed to alanine. Because the effect of exchange of TMs between the ₁AR and ₂AR on the binding characteristics of salbutamol is relatively small compared with formoterol and procaterol, we did not examine the binding characteristics of salbutamol for the alanine-substituted mutants. The K_d values of these muatants for ¹²⁵I-CYP is essentially the same as that of the WT ₂AR, indicating that the substitution did not cause nonspecific alterations of the binding sites (Table 4). Among mutants, the Y308A-₂AR is the only one that showed significantly decreased affinities for formoterol and procaterol. A similar conclusion indicating the importance of Tyr³⁰⁸ was obtained from the previous reports (Isogaya et al., 1998; Kikkawa et al., 1998), using TA-2005 and salmeterol as ₂-selective agonists. Procaterol also showed decreased affinity for the L324A-₂ARs. The amino acid of the ₁AR at the homologous position of Tyr³⁰⁸ is Phe instead of Ala. We made a mutant in which Tyr³⁰⁸ is replaced with Phe, and we examined the binding characteristics of these agonists. Although the affinity of formoterol was not significantly decreased by the replacement, the affinity of procaterol was decreased by the replacement. The K_i value of procaterol for the Y308F-₂AR is essentially the same as that of the Y308A-₂AR. The differential susceptibility of formoterol and procaterol to hydroxyl group at Tyr³⁰⁸ indicated that the high-affinity binding of procaterol but not formoterol required the hydroxyl group of Tyr³⁰⁸. We also found that the affinity of salmeterol did not decrease in the Y308F-₂AR, indicating the importance of hydrophobic interaction for ₂-selective, high-affinity binding of salmeterol (Table 4). The affinity of the nonselective -agonist isoproterenol for the Y308F-₂AR showed essentially the same value as that for the WT ₂AR. The differential susceptibility of isoproterenol and ₂-selective agonists to the removal of the hydroxyl group from Tyr³⁰⁸ indicated that the effect of the removal of the hydroxyl group from Tyr³⁰⁸ is specific for ₂-selective agonists.

View this table: [in this window] [in a new window]   TABLE 4 Effects of replacement of amino acids in TM7 of the 2AR with alanine or phenylalanine on ligand-binding characteristics of the 2AR The binding of ligands to the WT 2AR and alanine-substituted 2AR mutants were determined by competition with 50pM 125I-CYP. The data were analyzed using the nonlinear least-squares regression computer program as described under Experimental Procedures. The results are shown as the mean ± S.E. from three to four separate experiments.

Effects of Substitution of Amino Acids with Alanine in TM2 of ₂AR on Binding of Procaterol and Formoterol. To examine the contribution of individual amino acids in TM2 to the binding of ₂-selective agonists, the amino acids in TM2 of the ₂AR that were different from those of the ₁AR were changed to alanine. It is assumed that alanine makes a hole at the position of the replacement without altering the conformation of the amino acid with the cognate ligand (Clackson and Wells, 1995; Holst et al., 1998). The K_d values of ¹²⁵I-CYP for these mutants were almost same as that of the WT ₂AR, indicating that TM2 did not contribute to the binding of ¹²⁵I-CYP. It is reasonable to assume that TM2 retained the same conformation as in the WT-₂AR (Table 5). Among eight mutants, only H93A-₂AR showed significantly decreased affinity for procaterol. The other mutants did not show significantly decreased affinities for procaterol, formoterol, or salmeterol. Although replacement of TM2 of the ₂AR with that of ₁AR decreased the affinities 32-fold for salmeterol and 7-fold for formoterol (Table 1, and see Table 1 in Isogaya et al., 1998, for salmeterol), we could not identify the specific amino acid(s) that contributed to the high-affinity binding for these agonists. These results suggested that TM2 contributes to selective binding as a whole entity and that a specific amino acid is not important for the high-affinity binding of ₂-selective agonists.

View this table: [in this window] [in a new window]   TABLE 5 Effects of replacement of amino acids in TM2 of the 2AR with alanine on ligand-binding characteristics of the 2AR The binding of ligands to the WT 2AR and alanine-substituted 2AR mutants were determined by competition with 50 pM 125I-CYP. The data were analyzed using the nonlinear least-squares regression computer program as described under Experimental Procedures. The results are shown as the mean ± S.E. from three to four separate experiments.

Effects of Substitution of Amino Acids with Alanine in TM2 of ₁AR on Binding of T-0509, Xamoterol, and Denopamine. The contribution of each amino acid in TM2 of the ₁AR to ₁-selective agonists was examined by expressing and characterizing the alanine-substituted ₁AR mutants. The affinities of T-0509 and denopamine were significantly decreased in the mutants that substituted alanine at Leu¹¹⁰, Thr¹¹⁷, and Val¹²⁰ (Table 6). The structure of denopamine is the same as that of T-0509 except that denopamine lacks a hydroxyl group at the meta position in the catechol ring (Fig. 1B). It is reasonable to assume, therefore, that the mutation of same amino acids decreased the affinities for both agonists. The decreases of the affinities of T-0509 and denopamine were significant but slightly smaller (2- to 6-fold) than the decreases in affinities of these substances for the chimeric receptors (~10-fold). The affinities of xamoterol were not significantly decreased by any substitution of amino acids in TM2.

View this table: [in this window] [in a new window]   TABLE 6 Effects of replacement of amino acids in TM2 of the 1AR with alanine on ligand-binding characteristics of the 1AR The binding of ligands to the WT 1AR and alanine-substituted 1AR mutants were determined by competition with 50 pM 125I-CYP. The data were analyzed using the nonlinear least-squares regression computer program as described under Experimental Procedures. The results are shown as the mean ±S.E. from three to four separate experiments.

Computer Modeling of -selective Agonist-AR Complexes. We built three-dimensional models of -selective agonist-AR complexes to visualize the binding pocket. The structural models of the ₁ and ₂ARs were built on the basis of the predicted structure of rhodopsin simulated by Baldwin et al. (1997) (Fig. 3). General features of the present models are as follows. First, the amino acids of TM2 and TM7 form a binding pocket that can interact with N-substituents of the selective agonists. Second, Tyr³⁰⁸ in TM7 locates at the top of the binding pocket and covers the binding pocket from the upper side. Third, the binding pocket mainly consists of hydrophobic residues. Procaterol, formoterol, and denopamine were well fitted to the binding pocket of the model. The phenyl group of Tyr³⁰⁸ of the ₂AR covers the ₂-selective agonists (procaterol and formoterol) from the top of the pocket and acts like a "barrier" to prevent the ligand from moving freely into extracellular space (Fig. 3, A and B). The N-substituent of formoterol goes a little farther into the binding pocket formed by TM2 and TM7 than does the N-substituent of procaterol (Fig. 3A). Then Tyr³⁰⁸ interacts with the phenyl ring of the N-substituent of formoterol, mainly through hydrophobic interaction. This is consistent with the result that the change of Tyr³⁰⁸ to Phe did not significantly decrease the affinity of formoterol. When Tyr³⁰⁸ is mutated to alanine, alanine cannot inhibit movement of the N-substituents of the agonists to extracellular space. This mutation resulted in the receptors that showed decreased affinities for the ₂-selective agonists.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Three-dimensional models of 2- and 1-selective agonist-AR complexes. Three-dimensional models were built by the predicted structure of rhodopsin by simulating with SYBYL. The possible interaction sites between the AR and selective agonists are shown. A, side view of formoterol-2AR complex. B, side view of procaterol-2AR complex. C, side view of denopamine-1AR complex. The interaction sites of formoterol and procaterol with the 2AR are as follows: Ser204 and Ser207 in TM5 with catechol or an equivalent group of the ligands, Phe290 in TM6 with a phenyl or carbostiryl group of the ligands, Asn293 in TM6 with a hydroxyl group at the  position, Asp113 in TM3 with the protonated amine. Putative interaction sites of denopamine with the 1AR are as follows: Ser232 in TM5 with a hydroxyl group at the para position, Phe341 in TM6 with a phenyl group, Asn344 in TM6 with a hydroxyl group at  position, Asp138 in TM3 with the protonated amine. Phe359 in TM7 of the 1AR is the homologous amino acid of Tyr308 of the 2AR. Other amino acids that were assigned to interact with the ligands are mentioned in the text.

View larger version (67K): [in this window] [in a new window]   Fig. 4.   Space-filling models of TM2 and TM7 of the 1- and 2ARs. The left panel is a model of TM2 and TM7 of the 1AR, and the right panel is a model of TM2 and TM7 of the 2AR. Hydrophobic (aromatic) amino acids consisting of Phe, Trp, and Tyr are shown in green. Hydrophobic (aliphatic) amino acids consisting of Ala, Ole, Leu, Met, Val, and Pro are shown in yellow. Polar or charged amino acids consisting of Asn, Asp, Cys, Gln, Glu, Arg, Lys, Ser, His, and Thr are shown in cyan. The back bone is shown in white.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

We have demonstrated that the affinities of the synthetic ₁-selective agonists such as T-0509, xamoterol, and denopamine were increased or decreased by transferring TM2 of the ₁AR to the ₂AR or TM2 of the ₂AR to the ₁AR. This indicates that TM2 of the ₁AR is a major determinant of the high-affinity binding of the ₁-selective agonists. In contrast with the ₁AR, the replacement of TM2 of the ₂AR with the homologous region of the ₁AR decreased the affinities of ₂-selective agonists, and introduction of TM2 or TM7 into the ₁AR partially restored the high-affinity binding. Furthermore, the affinities for the ₁AR with both TM2 and TM7 of the ₂AR became close to the values for the WT ₂AR. These data on loss-of-function and gain-of-function mutants suggest that TM2 of the ₁AR is a major determinant for the ₁-selective agonist to bind to the receptor with high affinity, and that both TM2 and TM7 of the ₂AR determine the high-affinity binding of the ₂-selective agonists.

There are several reports that the specific amino acids in TM2 and TM7 are close together and are functionally interacting (Zhou et al., 1994; Sealfon et al., 1995; Perlman et al., 1997). In addition to these reports, Ballesteros et al. (1998) recently reported that Arg, which is located at the bottom of TM3 and is well conserved among G protein-coupled receptors, interacts with Asn in TM2 and Asp in TM7 in the gonadotropin-releasing hormone receptor. It is possible that amino acids in TM2 and TM7 form the binding pocket in a cooperative manner and provide the site for high-affinity binding of the -selective agonists.

The structures of T-0509, T-1583, and denopamine are similar to each other (Fig. 1B). Among these agonists, T-1583 did not show ₁AR selectivity. This suggests that the positions of methoxy groups on the phenyl ring extending from the protonated amine are important for the ₁-selective binding, and three methoxy groups are not accommodated by the binding pocket formed by TM2 and TM7. The selectivity of denopamine was lower than that of T-0509, suggesting that the hydroxyl group of the phenyl ring at the meta position also contributes in part to ₁ selectivity, possibly through interaction with Ser²²⁹ in TM5 of the ₁AR or the homologous amino acid Ser²⁰⁴ in TM5 of the ₂AR. Xamoterol has a long side chain extending from the protonated amine and shows relatively high affinity for the ₁AR, compared with other ₁-selective agonists. It indicates that a long side chain (N-substituent) may be necessary to reach TM2, which plays an important role in the ₁ selectivity.

Norepinephrine is the endogenous catecholamine that shows ₁AR selectivity. Two groups of researchers have reported (Frielle et al., 1988; Dixon et al., 1989) that TM4 of the ₁AR is the region that is responsible for the ₁-selective binding of norepinephrine. In contrast with the previous works, the present study showed that TM7 was a primary region that determined the ₁-selective binding of norepinephrine. However, the authors of the previous studies evaluated the ₁ selectivity based on the ratios of K_i values of norepinephrine and epinephrine. When they evaluated the ₁ selectivity with the K_i values of norepinephrine, their results are consistent with our findings. The replacement of TM1 to TM6 of the ₁AR with those of the ₂AR did not confer the high-affinity binding of norepinephrine on the chimera (Frielle et al., 1988).

Figure 5 illustrates the amino acids in TM2 and TM7 that are different between the ₁AR and the ₂AR. Because agonist binding domains are assumed to be located within TMs, and 42% of the amino acids in TM7 of the ₁AR and the ₂AR are different, compared with 29% in TM2, TM7 may be a more favorable target for the subtype-selective agonist.

View larger version (27K): [in this window] [in a new window]   Fig. 5.   The aligned amino acid sequences of TM2 and TM7 of the human 1- and 2ARs. The numbers refer to the first methionine of the 1- or 2AR as first amino acid. The same amino acids between the 1- and 2ARs are shown in the squares.

Space-filling models suggested that the polar amino acids of the binding pockets consisting of TM2 and TM7 of the ₁- and ₂ARs are differentially located and orientated. Although the selectivities of ₁AR-selective agonists are at most 10-fold, the affinities of ₂-selective agonists are high for the ₂AR and low for the ₁AR. This difference may be explained by the differential location of polar amino acids in TM2 and TM7. The ₁AR contains Phe³⁵⁹ at a position homologous to Tyr³⁰⁸ of the ₂AR, an amino acid that is critical for the high-affinity binding of the ₂-selective ligands. However, because there are polar amino acids around Phe³⁵⁹, the N-substituents of the ₁-selective agonists cannot be accommodated by polar amino acids in TM2 and TM7. This also suggests that the design of ₁-selective agonists may be more complex than that of ₂-selective agonists because ligands should contain both hydrophobic and hydrophilic parts in an appropriate position and orientation to interact with polar and hydrophobic amino acids in TM2 and TM7. T-0509 and denopamine, but not T-1583, showed the ₁ selectivity. The only difference between the three agonists is that T-1583 contains three methoxy groups in its N-substituent, compared with the other two agonists, which have two methoxy groups. The interaction between the ₁-selective agonists and TM2 and TM7 may be interfered with by repulsion between polar amino acids in TM2 and TM7 of the ₁AR and the third methoxy group of T-1583.

A three-dimensional model of -selective agonist-AR complexes revealed a unique binding pocket formed by TM2 and TM7, which can explain the binding characteristics of -selective agonists for the mutated ₁- or ₂ARs. We previously reported that Tyr³⁰⁸ in TM7 of the ₂AR played a major role in the binding of ₂-selective agonists such as TA-2005 and salmeterol with high affinity (Isogaya et al., 1998; Kikkawa et al., 1998). We extended the previous observation to other ₂-selective agonists such as procaterol and formoterol and proposed that Tyr³⁰⁸, which is located at the top of TM7, plays two roles in the binding of selective agonists, as determined by mutagenesis and three-dimensional modeling. The first role of Tyr³⁰⁸ is to provide high-affinity binding via hydrophobic or hydrophilic interactions with the ₂-selective agonists. The second role of Tyr³⁰⁸ is to prevent N-substituents of the selective agonists from freely moving into extracellular space. The affinities of salmeterol and formoterol were decreased in Y308A-₂AR but not Y308F-₂AR. However, the affinities of procaterol were decreased in both Y308A- and Y308F-₂ARs. This discrepancy could be explained by the different types of interactions between the ₂-selective agonists and Tyr³⁰⁸, that is hydrophobic and hydrophilic interactions. Alanine substitution cannot complement the interaction and block free movement of the N-substituent of the ligand from the binding pocket. The chimeric receptor, in which TM7 of the ₁AR is introduced into the ₂AR, did not show significantly decreased affinities for procaterol and formoterol. However, the Y308A-₂AR mutant, in which Tyr³⁰⁸ in TM7 is replaced with alanine, did show decreased affinities for both agonists. This apparent discrepancy can be explained by the fact that the amino acid of the ₁AR that is homologous to Tyr³⁰⁸ of the ₂AR is Phe.

It is interesting to try to understand how the amino acids contributing to high-affinity binding participate in the activation steps, because TM7 changes the conformation and contributes to activation of the receptors upon agonist binding (Wess et al., 1993; Abdulaev and Ridge, 1998). It is possible that TM2 and/or TM7 involve not only the high-affinity binding of the selective agonists but also the activation step.

In conclusion, binding domains of AR subtype-selective agonists appeared to be localized in TM2 and TM7. We showed that TM2 was especially important for ₁ selectivity, that both TM2 and TM7 were important for ₂ selectivity, and that interaction of the binding pocket formed by TM2 and TM7 of the ₁- or ₂ARs with N-substituents of the subtype-selective agonists is essential for high-affinity binding. We identified several amino acids that are important for the ₁ or ₂ selectivities. However, our data do not exclude the possibility that other amino acids in TM2 and TM7 participate in subtype-selective binding for the agonists that have different structures from those of the agonists examined in this report.