Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

₂-Adrenergic receptors serve as a prototypic member of the large superfamily of G protein-coupled receptors. Apart from the visual pigment rhodopsin, they were the first to be cloned (Dixon et al., 1986) and subjected to site-directed mutagenesis (Strader et al., 1988, 1989). The latter two studies identified Asp113 in helix III and two serines in helix V (Ser204 and 207) of the human ₂-adrenergic receptor as anchor points for the protonated amine function and the catechol group of the endogenous agonist epinephrine, respectively. In the previous two decades, thorough medicinal chemistry had already led to the introduction of an important class of antiasthmatic drugs, all ₂-receptor agonists, exemplified by salbutamol and clenbuterol. The nonselective catecholamine isoproterenol has been and still is the reference compound when the pharmacology and molecular biology of -adrenergic receptors are the subject of investigation. In a recent study, we identified Asn293 in helix VI as a residue that determines the stereoselectivity of catecholamines by virtue of its interaction with the -OH group in the aliphatic side chain of this class of compounds (Wieland et al., 1996). The hypothesis for this interaction was based on the molecular modeling of the ₂-adrenergic receptor for which we used the atomic coordinates of bacteriorhodopsin as a template. Figure 1 is a visualization of this receptor model, defining the positions of the four above-mentioned amino acid residues and the levorotamer of isoproterenol.

View larger version (55K): [in this window] [in a new window]   Fig. 1.   Model visualizing the binding of ()-isoproterenol to the human 2-adrenergic receptor. The receptor -helical backbone (helices I-VII) is viewed from the extracellular side. ()-Isoproterenol and the receptor residues thought to be involved in ligand binding are shown in red and yellow (Asp113, Ser204, Ser207, and Asn293, respectively).

We thought it worthwhile to further characterize the atomic details of this stereoselective interaction. Therefore, we synthesized "mutated" derivatives of -adrenergic receptor agonists in the present study and analyzed their effects on the wild-type and aforementioned mutant receptor in which Asn293 was changed for a leucine residue. We reasoned that such an approach (mutant ligands for mutant receptors) might lead to a "gain of function" in terms of affinity. This would provide further physical evidence for a direct interaction between a specific receptor residue and a functional group in the ligand, indeed allowing an interpretation at almost the atomic rather than the molecular level of the ligand-receptor interaction.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Synthesis of Isoproterenol and Clenbuterol Analogs

General Methods and Materials. Schleicher and Schüll DC Fertigfolien F 1500 LS 254 were used for thin layer chromatography analysis. Compounds were detected under UV light. For structures of compounds, see Fig. 2. Column chromatography was performed on silica gel 60, 230 to 400 mesh (Merck, Darmstadt, Germany). ¹H NMR spectra (300 MHz) were recorded at 25°C with a Bruker WM 300 spectrometer. ¹³C NMR spectra (50 MHz) were recorded with a Jeol JNM-FX 200 spectrometer. ¹H chemical shifts () are given in parts per million relative to that of Me₄Si [CDCl₃, MeOD, or dimethyl sulfoxide (DMSO)]. ¹³C chemical shifts are not included but are available on request from the corresponding author. All mass experiments were performed on a Finnigan MAT900, equipped with a programmable direct insertion probe. Temperature was raised from 30 to 300°C in 5 min. Mass spectra were recorded in electron impact mode with 70 eV. Melting points were uncorrected. All compounds appeared pure on thin layer chromatography, and in ¹H NMR, ¹³C NMR, and mass spectrometry.

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Structural formulae of clenbuterol and isoproterenol and their derivatives used in this study.

4-Amino-3,5-dichlorobenzylcyanide (9). To a stirred solution of commercially available 4-aminobenzylcyanide (0.5 g, 3.8 mmol) in acetic acid (10 ml) was added dropwise N-chlorosuccinimide (1.1 g, 8.4 mmol) in acetic acid (5 ml). After 17 h at room temperature, the reaction mixture was concentrated under reduced pressure. The residue was dissolved in CH₂Cl₂ (25 ml) and extracted with water (10 ml), dried (MgSO₄), and concentrated. The residue was purified by column chromatography on silica gel (CH₂Cl₂) to give pure 9 (0.46 g, 60%). ¹H NMR (MeOD)  3.65 (s, 2H, CH₂CN), 7.15 (s, 2H, H_arom). MS: m/z 200 (M⁺).

4-Amino-3,5-dichloro--methylbenzylcyanide (10). Compound 9 (0.4 g, 2.0 mmol) was dissolved in tert.-BuOH (10 ml). To this solution was added potassium tert-butanolate (0.2 g, 2 mmol) dissolved in tert-BuOH (5 ml), and subsequently MeI (0.3 g, 2.2 mmol). After 4 h at room temperature, the reaction mixture was diluted with CH₂Cl₂ (20 ml) and the organic layer extracted with water (10 ml), dried (MgSO₄), and concentrated. Purification of the remaining oil on silica gel [1:0 to 0:1 light petroleum (bp 40-60°C)/ether] afforded 10 (0.3 g, 67%). ¹H NMR (CDCl₃)  1.54 (d, 3H, CH₃, J = 7.2 Hz), 3.60 (q, 1H, CH), 7.18 (s, 2H, H_arom).

3,4-Dimethoxy--methylbenzylcyanide (11). A solution of potassium tert-butanolate (8.3 g, 73 mmol) in tert-BuOH (100 ml) was added at once to a solution of commercially available 3,4-dimethoxybenzylcyanide (10 g, 57 mmol) and MeI (13.3 g, 57 mmol) in tert-BuOH (100 ml). The temperature was maintained at 35-38°C for 60 min. Then the reaction mixture was concentrated and the residue diluted with CH₂Cl₂ (250 ml). The organic layer was washed with water (125 ml), dried (MgSO₄), and concentrated to give 11 (13 g, 85%). ¹H NMR (CDCl₃)  1.06 (t, 3H, CH₃, J = 7.2 Hz), 3.67 (t, 1H, CH, J = 7.2 Hz), 3.86 (s, 6H, 2× OCH₃), 6.83 (dd, 1H, H_arom, J_o 8.1 Hz, J_m 2.0 Hz), 7.25 (d, 1H, H_arom, J_m 2.0 Hz), 7.28 (d, 1H, J_o 2.0 Hz).

General Procedure for One-Pot Reduction-Transimination-Reduction. To a cooled (70°C) solution of cyanide (5 mmol) in dry ether (40 ml) was added 1 M diisobutylaluminum hydride in cyclohexane (10 ml, 10 mmol). After stirring at 70°C for 3 h, dry MeOH (15 ml) was added. The cooling bath was removed and the amine (25 mmol) added. Stirring was continued for 2 h during while the temperature was allowed to rise to room temperature. The mixture was stirred for another 2 h, and 1 N HCl (50 ml) was added. The organic layer was extracted with an additional 1 N HCl (30 ml). The aqueous layer was made alkaline with 5 N NaOH and then extracted with CH₂Cl₂ (3 × 25 ml). The combined organic layers were dried (K₂CO₃) and evaporated. The residue was purified on silica gel (1:0 to 9:1 CH₂Cl₂/MeOH).

2-(4-Amino-3,5-dichlorophenyl)-1-(tert-butylamino)ethane (2). Prepared as described above, starting from cyanide 9 and with tert-BuNH₂ as the amine in a yield of 45%. ¹H NMR (DMSO)  1.27 (s, 9H, 3× CH₃-tert-Bu), 2.78 (m, 2H, CH₂NH), 3.02 (m, 2H, CH₂CH₂NH), 7.21 (s, 2H, H_arom).

2-(4-Amino-3,5-dichlorophenyl)-1-(tert-butylamino)propane (3). Prepared as described above, starting from cyanide 10 and with tert-BuNH₂ as the amine in a yield of 51%. ¹H NMR (CDCl₃)  1.07 (s, 9H, 3 × CH₃-tBu), 1.21 (d, 3H, CH₃, J = 6.4 Hz), 2.66 (m, 2H, CH₂NH), 2.69 (m, 1H, CHCH₂NH), 7.05 (s, 2H, H_arom).

2-(3,4-Dimethoxyphenyl)-1-(iso-propylamino)ethane (12). Prepared as described above, starting from dimethoxybenzylcyanide (Brussee et al., 1978) and with iPrNH₂ as the amine in a yield of 89%. ¹H NMR (CDCl₃)  1.05 (d, 6H, 2× CH₃-iPr, J = 6.2 Hz), 2.71-2.89 (m, 5H, CH₂CH₂NH, CH₂NH, CH-iPr), 3.86, 3.87 (2× s, 6H, 2× OCH₃), 6.75 (dd, 1H, H_arom, J_o 8.2 Hz, J_m 1.9 Hz), 6.80-6.85 (m, 2H, H_arom).

2-(3,4-Dimethoxyphenyl)-1-(iso-propylamino)propane (13). Prepared as described above, starting from cyanide 11 and with iPrNH₂ as the amine in a yield of 91%. ¹H NMR (CDCl₃)  1.02 (d, 3H, CH₃-iPr, J = 6.2 Hz), 1.05 (d, 3H, CH₃-iPr, J = 6.3 Hz), 1.25 (d, 3H, CH₃, J = 7.2 Hz), 2.42-2.95 (m, 4H, CHCH₂NH, CH₂NH, CH-iPr), 3.85, 3.87 (2 × s, 6H, 2 × OCH₃), 6.78 (dd, 1H, H_arom, J_o 8.1 Hz, J_m 2.1 Hz), 6.70-6.75 (m, 2H, H_arom).

General Procedure for Demethylation. The dimethoxy derivative (2 mmol) was dissolved in 48% HBr (10 ml) and the solution was refluxed for 17 h. Then the solvent was evaporated and the last traces of the solvent were removed with the aid of repeated addition and evaporation of toluene (3 × 25 ml). The residue was purified by column chromatography on silica gel (EtAc/MeOH/NH₄OH, 85/15/1, v/v).

2-(3,4-Dihydroxyphenyl)-1-(iso-propylamino)ethane (6). Prepared as described above, starting from dimethoxy derivative 12 in a yield of 52%. ¹H NMR (DMSO)  1.21 (d, 6H, 2 × CH₃-iPr, J = 6.6 Hz), 2.72 (m, 2H, CH₂NH), 3.02 (m, 2H, CH₂CH₂NH), 3.30 (q, 1H, CH₃-iPr, J = 6.2 Hz), 6.49 (dd, 1H, H_arom, J_o 8.0 Hz, J_m 2.1 Hz), 6.63 (d, 1H, H_arom, J_m 2.1 Hz), 6.67 (d, 1H, H_arom, J_o 8.0 Hz). MS: m/z 195 (M⁺).

2-(3,4-Dihydroxyphenyl)-1-(iso-propylamino)propane (7). Prepared as described above, starting from dimethoxy derivative (compound 13) in a yield of 63%. ¹H NMR (MeOD)  1.28 (d, 3H, CH₃, J = 6.1 Hz), 1.30 (d, 6H, 2× CH₃-iPr, J 6.2 Hz), 2.95 (m, 1H, CHCH₂NH), 3.12 (m, 2H, CH₂NH), 3.32 (q, 1H, CH-iPr, J = 6.2 Hz), 6.62 (dd, 1H, H_arom, J_o 8.1 Hz, J_m 2.2 Hz), 6.72 (d, 1H, J_m 2.2 Hz), 6.76 (d, 1H, J_o 8.1 Hz). MS: m/z 209 (M⁺).

2-(4-Amino-3,5-dichlorophenyl)-2-[(methyl)oxy]-1-(tert-butylamino)ethane (4). Clenbuterol (0.5 g, 1.8 mmol) was suspended in MeOH (10 ml) and hydrogen chloride was passed in with stirring at 10°C until a clear solution resulted. The reaction mixture was concentrated and the hydrochloride salt thus obtained (87% yield) was recrystallized (MeOH-diisopropylether): mp 212-214°C. ¹H NMR (MeOD)  1.34 (s, 9H, 3 × CH₃-tBu), 3.07 (m, 2H, CH₂NH), 3.28 (s, 3H, OCH₃), 4.28 (t, 1H, CHCH₂NH, J = 5.8 Hz), 7.23 (s, 2H, H_arom). MS: m/z 290 (M⁺).

2-(3,4-Dihydroxyphenyl)-2-[(methyl)oxy]-1-(isopropylamino)ethane (8). Prepared as described for compound 4 starting from (±)-isoproterenol in a yield of 85%. The hydrochloride salt thus obtained was >97% pure and was used as such in the test systems. ¹H NMR (MeOD)  1.33 (d, 3H, CH₃-iPr), J = 6.5 Hz), 1.35 (d, 3H, CH₃-iPr, J = 6.1 Hz), 3.05 (dd, 1H, CHCHHNH, J = 12.9 Hz, J = 3.5 Hz), 3.13 (dd, 1H, CHCHHNH, J = 12.9 Hz, J = 10.3 Hz), 3.25 (s, 3H, OCH₃), 3.31 (q, 1H, CH-iPr), 4.35 (dd, 1H, CHCH₂NH, J = 3.5 Hz, J = 10.3 Hz), 6.70 (dd, 1H, H_arom, J_o 8.1 Hz, J_m 1.9 Hz), 6.80 (d, 1H, H_arom, J_m 1.9 Hz), 6.81 (d, 1H, H_arom, J_o 8.1 Hz). MS: m/z 225 (M⁺).

Expression of Wild-Type and Mutant ₂-Adrenoceptors in Stable CHO Cells

CHO cell lines stably expressing either wild-type or mutated ₂-adrenoceptor cDNA (encoding a single point mutation at amino acid position 293 from asparagine to leucine) (Wieland et al., 1996) were grown in monolayers on Dulbeccos's modified Eagle's medium (DMEM F-12) supplemented with 10% fetal calf serum, 2 mM glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin (all purchased at Pan Systems, Berlin, Germany) and 0.5 mg/ml of G418 (Gibco Laboratories, Eggenstein, Germany) in a 7.5% CO₂ incubator at 37°C. Clones with comparable densities (~0.2 pmol/mg membrane protein) were selected for the experiments.

Preparation of Crude Cell Membranes

Cells were washed three times with ice-cold phosphate-buffered saline, scraped in 50 ml of ice-cold lysis buffer (5 mM Tris-HCl, 2 mM EDTA, pH 7.4), homogenized with an Ultra Turrax for 30 s at full speed and centrifuged at 1000g for 10 min (4°C). The supernatants were then centrifuged at 50,000g for 15 min (4°C), and the resulting pellets were resuspended in an appropriate volume of incubation buffer (50 mM Tris-HCl, pH 7.4) to give a final concentration of 2 mg/ml. Protein content was determined by the method of Bradford (1976) with bovine serum albumin (Sigma Chemical Co., Deisenhofen, Germany) as the standard.

Radioligand Binding Assay

For inhibition assays, 0.1 ml of fresh membranes was incubated in triplicates with 50 pM ¹²⁵I-cyanopindolol (¹²⁵I-CYP) and various concentrations of isoproterenol/clenbuterol (Sigma Chemical Co.) and their derivatives (100 nM-1 mM) in the presence of 100 µM Gpp(NH)p (Sigma Chemical Co.) for 1 h at 37°C. The latter addition was used to uncouple the ₂-adrenergic receptors from G_s and thereby generate monophasic competition curves for agonists. Nonspecific binding was determined in the presence of 10 µM ()-propranolol (Sigma Chemical Co.). The incubations were terminated by filtration through Whatman GF/C filters and washing with ice-cold 50 mM Tris-HCl, pH 7.4.

Adenylyl Cyclase Assays

Adenylyl cyclase activity was determined as described in Wieland et al. (1996). Crude cell membranes were prepared freshly immediately before the assay as described above. Incubations contained 40 to 50 µg of protein; 50 mM Tris-HCl, pH 7.4; 500 µM RO 20-1724; 100 µM cAMP; 2 mg/ml BSA; 1 mM MgCl₂; 5 mM creatine phosphate; 0.4 mg/ml creatine kinase, 1 µM GTP; 100 µM [-³²P]ATP (0.2 µCi/tube; Amersham Corp., Braunschweig, Germany); and the desired concentrations of the test compounds in a final volume of 100 µl. Incubations were done for 30 min at 30°C. Adenylyl cyclase activities measured in the presence of 100 µM ()-isoproterenol were set to 100%. Basal activities were not subtracted.

UV-Spectrometry

Isoproterenol and clenbuterol (100 µM in 0.1 M KCl) were analyzed at 25°C, under N₂ atmosphere and in the dark, essentially to prevent oxidation of isoproterenol during measurements. Solutions were adjusted to either pH 5 or pH 11 by addition of minute amounts of HCl or NaOH, respectively. UV spectra were recorded on an Aminco DW-2A UV/VIS spectrophotometer.

Molecular Modeling

The receptor-ligand model presented in Fig. 1 was developed exactly as described in Wieland et al. (1996). The figure was generated with programs MOLSCRIPT and RASTER3D.

Data Analysis

Determination of ligand-binding parameters was performed by nonlinear curve fitting with SCTFIT and plotted with KALEIDAGRAPH, respectively. K_D values of ¹²⁵I-CYP were 7.1 and 7.3 pM for the wild-type and mutant receptor, respectively (Wieland et al., 1996).

Concentration-response curves of adenylyl cyclase experiments were fitted to the operational model developed by Black and Leff (1983) as described in Lohse (1990) to obtain an estimate of the transducer ratio :
with E denoting the effect, E_m the maximum possible effect, A the agonist concentration, and K_A the dissociation constant of the agonist-receptor complex. The transducer ratio  is a parameter describing the signal transduction efficacy of the system and is estimated from the fit of the data.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Chemistry. The clenbuterol derivatives 2-4 were prepared as follows. First, commercially available 4-aminobenzylcyanide was chlorinated with N-chlorosuccinimide to give its 3,5-dichloro derivative 9, which was converted into the desired end product 2 by a one-pot reduction/transimination/hydride reduction sequence in a yield of 45% (Zandbergen et al., 1992). Methylation of derivative 9 with methyl iodide in the presence of potassium tert-butoxide gave methyl derivative 10 in a yield of 67%. Conversion of the latter into clenbuterol analog 3 was achieved by the one-pot reduction/transimination/hydride reduction sequence. The protected isoproterenol analogs 12 and 13 were likewise obtained. Demethylation of analogs 12 and 13 in aqueous HBr gave the unprotected isoproterenol analogs 6 and 7. The methoxy derivatives 4 and 8 were synthesized by HCl/MeOH treatment of clenbuterol and isoproterenol, respectively.

Biology. The parent compounds clenbuterol and isoproterenol and their derivatives modified at the -OH position (Fig. 2) were tested in radioligand binding studies and adenylate cyclase assays. For that purpose the wild-type ₂-adrenergic receptor was expressed in CHO cells and membranes were prepared, as was done for a mutant ₂-adrenergic receptor in which an asparagine residue in the sixth transmembrane domain was changed into a leucine (N293L) (Wieland et al., 1996). All clenbuterol and isoproterenol derivatives were capable of displacing ¹²⁵I-CYP, a radiolabelled antagonist, from both wild-type and mutant receptors (Fig. 3 and Table 1). The K_i values listed in Table 1 show that the methoxy analogs 4 and 8 had highest affinity on the wild-type receptor among the derivatives. All derivatives had lower affinities for the wild-type receptor than their parent compounds isoproterenol and clenbuterol. This latter observation held also on the mutant receptor. Here, the methyl derivative 7 was most potent among the isoproterenol derivatives, whereas both the unsubstituted (2) and methoxy derivative 4 were the most potent clenbuterol analogs.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Stereospecific binding of isoproterenol derivatives to CHO cell membranes containing wild-type and mutant 2-adrenergic receptors. Data are means ± S.E. (n = 3). Top, competition for 125I-CYP binding to the wild-type 2-adrenergic receptor by various isoproterenol derivatives (6, ; 7, ; 8, ). Bottom, competition for 125I-CYP binding to the mutant N293L 2-adrenergic receptor by various isoproterenol derivatives (6, ; 7, ; 8, ).

View this table: [in this window] [in a new window]   TABLE 1 Affinities of clenbuterol/isoproterenol and their derivatives (all racemates, except ()-isoproterenol) for wild-type and N293L mutant 2-adrenergic receptors The affinities were determined in competition experiments with 50 pM 125I-CYP. The data were analyzed by nonlinear curve fitting to obtain estimates for the affinity. Data are means ± S.E., n = 3.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Affinity ratios (Ki, wild-type/Ki, N293L) for all eight compounds, calculated from the data in Table 1. Data are means ± S.E. (n = 3).

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Concentration-response curves (adenylyl cyclase activation) for all compounds on both wild-type and N293L mutant -adrenergic receptors. Maximal stimulation by ()-isoproterenol was set to 100%. A typical experiment is shown, which was repeated at least three more times in duplicate with similar results.

View larger version (20K): [in this window] [in a new window]   Fig. 6.   Transducer ratios  (see Experimental Procedures) for all eight compounds on both wild-type and N293L mutant receptors. Data are means ± S.E. (n = 4-5).

UV-Spectroscopy. The UV spectra of isoproterenol and clenbuterol were recorded at two pH values. At pH 5, the aromatic ring systems of both compounds are neutral, whereas at pH 11 the catechol ring of isoproterenol is negatively charged, and no changes occur in clenbuterol's ring system (IJzerman et al., 1984). As a result, the UV spectra of clenbuterol at both pH values are almost coinciding (Fig. 7A), whereas isoproterenol's spectra are different (Fig. 7B). Interestingly, at pH 11 the spectra of isoproterenol and clenbuterol are virtually identical, whereas at pH 5 the UV spectra of both do not match.

View larger version (12K): [in this window] [in a new window]   Fig. 7.   UV spectra of clenbuterol (A) and isoproterenol (B) at pH 5 and pH 11 (mol: molar extinction; : wavelength).

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

In Fig. 1, a possible binding mode of ()-isoproterenol, the prototypic -adrenergic receptor agonist, to the human ₂-adrenergic receptor is visualized. This representation is entirely based on the receptor model developed by Wieland et al. (1996), but viewed from a different angle. In that previous study, we demonstrated the involvement of Asn293 in the stereospecific agonist recognition and activation of the human ₂-adrenergic receptor, through its interaction with the -OH group in the ligand side chain (see also Fig. 2). In particular, catecholamines such as isoproterenol proved highly sensitive to a mutation from asparagine to leucine on this position. The mutant receptor displayed strongly diminished affinity for the active ()-stereoisomer of isoproterenol, whereas the (+)-isomer was less affected. However, clenbuterol, a structurally dissimilar agonist (Fig. 2), did not appear to suffer from the mutation.

In the present study, we focused on this apparent discrepancy at the agonist ligand binding site in more detail. By "mutating" isoproterenol and clenbuterol, we sought to complement our analysis of the receptor-agonist interaction. Hence, we synthesized a number of isoproterenol and clenbuterol analogs, all varied at the -OH position in the side chain (Fig. 2). Radioligand binding studies and adenylate cyclase assays on both wild-type and mutant receptors were performed with the two parent compounds and their derivatives. We reasoned that the mutation of an asparagine residue into a leucine would cause a change in the characteristics of part of the ligand binding site from essentially hydrophilic with hydrogen-bonding capacity to far more lipophilic. As a consequence, certain analogs of isoproterenol should be able to recognize the mutant receptor better than the wild-type receptor, and we hypothesized that replacement of the -OH group in the aliphatic side chain of the ligand by a more lipophilic group might do in this respect. Such a gain of function also would complement and corroborate the findings in our previous study (Wieland et al., 1996), and, moreover, help us in appreciating atomic rather than molecular mechanisms in receptor recognition.

Indeed, we succeeded in making an isoproterenol derivative (7) that displayed higher affinity for the mutant receptor (Table 1; Fig. 4). Apparently, a methyl substituent (7) instead of a hydrogen atom (6) or a methoxy group (8) is an appropriate point of interaction for the leucine side chain. The affinity of 7 for the N293L mutant receptor was comparable to isoproterenol's affinity, given the fact that 7 is a racemate and isoproterenol was tested as the active enantiomer only. Thus, the ~100-fold affinity difference on the wild-type receptor between isoproterenol and its methyl derivative 7 was fully canceled on the mutant receptor. Interestingly, 7 did not prove superior to isoproterenol on the mutant receptor, suggesting that the receptor environment around the -position in the aliphatic side chain of the ligands remains hydrophilic in nature, despite the presence of the leucine side chain.

It might be argued that such a gain in affinity as observed for 7 could be due to more indirect effects, e.g., changes in overall receptor conformation. However, in this case a direct effect is more probable due to the other observations we made. First, the two ligands clenbuterol and isoproterenol behaved differently. The ratios shown in Fig. 4 demonstrate that hardly any gain or loss of function was found for clenbuterol and its derivatives, suggesting that Asn293 is not relevant for the recognition of this -adrenergic receptor agonist. If the effect observed for isoproterenol were due to general changes in receptor conformation, a similar change would have been expected for clenbuterol and analogs. Second, the data from the adenylate cyclase experiments show that the intrinsic activities on the two receptors are almost similar for each derivative (Fig. 5). Also, the transducer ratios  (Fig. 6) hardly varied between the wild-type and the mutant receptors. Apparently, the transduction mechanism for the wild-type and the mutant receptor had not drastically changed. Such a change would have been more probable if the mutant receptor had acquired a different overall conformation.

The mutated ligands were all less effective than their parent compounds because maximal cyclase activation was not observed (Fig. 5) and their  values were lower (Fig. 6). Apparently, the -OH group is preferred over the other substituents for full receptor activation. This finding corroborates the observations in our previous study (Wieland et al., 1996) that this position is not only important for stereoselective receptor recognition but also for receptor activation. The adenylyl cyclase experiments confirmed the preference of 7 for the mutant receptor. Its EC₅₀ value on the mutant receptor was 20 ± 9 µM compared with 58 ± 5 µM on the wild-type receptor.

The present study also provides evidence that there is not one single agonist binding site on the ₂-adrenergic receptor, based upon the above-mentioned observations that the recognition of clenbuterol and analogs is different from isoproterenol and derivatives. Hence, the critical involvement of Ser204 and Ser207 on helix V in agonist binding (Fig. 1), as elegantly shown by Strader et al. (1989), may be limited to catecholamines. These two amino acids are supposed to interact with the two catechol-OH groups via hydrogen bonding and, as a consequence, the ligand is directed to position its -OH group in the vicinity of Asn293. In binding clenbuterol, Asn293 does not seem to play a significant role, and, thus, Ser204 and Ser207 are hypothesized to be of less importance, too. Preliminary experiments in which we analyzed the binding of a series of agonists to a "double mutant" ₂-adrenergic receptor (Ser204Ala, Ser207Ala) strongly corroborated this notion (data not shown). Isoproterenol was >100-fold more potent on the wild-type receptor, whereas clenbuterol's affinity was only slightly diminished on the double mutant receptor (~3-fold).

Another line of evidence for this hypothesis came from the UV experiments (Fig. 7). In hydrogen bonding to the two serine residues, the two catechol-OH groups in isoproterenol may undergo a weakening of their OH bond. This gradual transition from a fully intact OH group to a situation that resembles a deprotonated form is mimicked in the UV recordings at pH 5 and pH 11, respectively. Under slightly acidic conditions (pH 5) the catechol structure is intact, whereas basic conditions (pH 11) promote the dissociation of a proton from the catechol OH groups (IJzerman et al., 1984). Clenbuterol's UV spectrum, unlike that of isoproterenol, is not pH dependent. The electronic characteristics of the aromatic ring systems in clenbuterol and isoproterenol (as exemplified by their UV spectra) are almost coinciding at pH 11. It has been shown that the electronic features of the aromatic ring system in -adrenergic receptor agonists are strongly correlated to their intrinsic activity (IJzerman et al., 1986). Apparently, clenbuterol's dichloro-anilino ring system may have the appropriate characteristics for productive receptor interaction without the need to undergo some sort of proton abstraction. This would make a direct interaction with the two serine residues unnecessary, and allow a different orientation of clenbuterol in the receptor.

Interestingly, -receptor antagonists having an oxypropanolamine side chain as in propranolol and in the radioligand used in the present study (¹²⁵I-CYP), also show a stereoselective interaction through their side chain OH group with an asparagine residue, but now in helix VII. This has been convincingly demonstrated, in particular on several subtypes of the 5-hydroxytryptamine receptor, because compounds such as propranolol and pindolol display nanomolar affinity for these receptors, too (Guan et al., 1992). In a recent study, we applied a similar strategy as described above to a series of propranolol analogs (Kuipers et al., 1997). Asn386 (helix VII) of the human 5-hydroxytryptamine_1A receptor was mutated into a valine rather than a leucine residue. It was found that a propranolol derivative with methoxy instead of hydroxy in the oxypropanolamine side chain showed a 25-fold affinity gain on the mutant receptor. Valine has a branched side chain with one -CH₂ group less than leucine, which might explain this receptor's preference for the bigger methoxy group compared with the methyl substituent in isoproterenol.

In summary, our study demonstrates that by simultaneous mutations in both ligands and receptors, gain of function can be established. To our knowledge, this is the first study of its kind focusing on agonist-receptor interactions. This approach is helpful to rule out mutations that have indirect effects on this interaction; it is a first step in the unraveling of ligand-receptor interactions at atomic detail. It is expected that future elucidation of the three-dimensional architecture of G protein-coupled receptors combined with the approach outlined in the present study, i.e., a combination of synthetic chemistry with molecular biology, will enhance our insight in the atomic events that govern the ligand-receptor interaction.