Abstract
Although the functional presence of β3-adrenergic receptors (β3-AR) in rodents is well established, its significance in human adipose tissue has been controversial. One of the issues confounding the experimental data has been the lack of potent and selective human β3-AR ligands analogous to the rodent-specific agonist BRL37344. Recently, we described a new class of aryloxypropanolamine β3-AR agonists that potently and selectively activate lipolysis in rhesus isolated adipocytes and stimulate the metabolic rate in rhesus monkeys in vivo. In this article, we describe novel and selective β3-AR antagonists with high affinity for the human receptor. L-748,328 and L-748,337 bind the human cloned β3-AR expressed in Chinese hamster ovary (CHO) cells with an affinity of 3.7 ± 1.4 and 4.0 ± 0.4 nM, respectively. They display an affinity of 467 ± 89 and 390 ± 154 nM for the human β1-AR. Their selectivity for human β3-AR versus β2-AR is greater than 20-fold (99 ± 43 nM) and 45-fold (204 ± 75 nM), respectively. These compounds are competitive antagonists capable of inhibiting the functional activation of agonists in a dose-dependent manner in cells expressing human cloned β3-AR. Moreover, both L-748,328 and L-748,337 inhibit the lipolytic response elicited by the β3-AR agonist L-742,791 in isolated nonhuman primate adipocytes. The aryloxypropanolamine benzenesulfonamide ligands illustrated here and elsewhere demonstrate high-affinity human β3-AR binding. In addition, we describe specific 3′-phenoxy substitutions that transform these compounds from potent agonists into selective antagonists.
β-Adrenergic receptors (ARs) are integral membrane proteins belonging to the class of G protein-coupled receptors. The pharmacological characteristics of β1-AR and β2-AR have been studied and described exhaustively. In tissue preparations, most of the pharmacological properties of the endogenous catecholamine ligands have been easily explained by the presence of predominantly β1-AR or β2-AR (Lands et al., 1967). Nevertheless, in some tissues, most notably, adipose tissue, β1-AR and β2-AR subtype-selective antagonists do not predictably attenuate agonist response, and these data lend support to the notion of the existence of an additional atypical β-AR (Harms et al., 1974). The development of compounds that elicited lipolysis in white adipose tissue and thermogenesis in brown adipose tissue, although displaying minimal activity in tissue preparations enriched for the β1-AR or β2-AR, was important in validating the existence of a third β-AR subtype (Arch et al., 1984; Holloway et al., 1991). When cloned from human genomic (Emorine et al., 1989) and cDNA (Granneman and Lahners, 1994) libraries, the human β3-AR was shown to have 49 and 51% overall homology at the amino acid level to human β2-AR and β1-AR, respectively. The greatest degree of homology was displayed in the putative transmembrane domain, which has been shown to be responsible for the binding of biogenic amine ligands by site-directed mutagenesis (Strader et al., 1995) and biophysical analysis (Tota and Strader, 1991). Other species homologs of β3-AR have been cloned (Granneman and Lahners, 1994; Pietri-Rouxel et al., 1995;Atgie et al., 1996; Strosberg, 1997). These can be classified as species homologs rather than distinct subtypes based on the degree of sequence homology (approximately 80%) and a unique β3-AR molecular signature, the presence of three consecutive serine residues in the fifth putative transmembrane domain.
In rodent white adipose tissue, β3-AR accounts for 90% of the β-ARs on the cell surface, as determined by saturation binding (Feve et al., 1995). In contrast, human β3-AR has only been detected in human adipose tissue, colon, and gallbladder with the sensitive reverse transcription-polymerase chain reaction method (Krief et al., 1993). Determination of the relative amount and importance of β3-AR on human adipocytes has been complicated by the lack of human-selective compounds. Recently, we described a series of agonists that are highly potent and selective at the human β3-AR (Weber et al., 1998). These benzenesulfonamide aryloxypropanolamine derivatives are not only efficacious in vitro at human cloned β3-AR expressed in Chinese hamster ovary (CHO) cells lines but, more important, are capable of eliciting lipolysis and elevation of metabolic rate in rhesus monkeys (Fisher et al., 1998).
Full pharmacological and functional characterization of the role of β3-AR in human and nonhuman primate adipose tissue requires the development of potent and selective antagonists. A class of aryloxypropanolaminotetralin β3-AR antagonists has been developed (Manara et al., 1995). The most potent member of this class, SR 59,230A has been described as β3-AR selective in rat brown adipocytes (Nisoli et al., 1996), rat colonic motility assays (Manara et al., 1996), and human colonic circular smooth muscle relaxation activity assays (De Ponti et al., 1996). In this article, we describe a new class of human β3-AR-selective antagonists developed via heterologously expressed cloned human receptors.
Experimental Procedures
Materials.
Tissue-culture reagents were obtained from Gibco BRL (Gaithersburg, MD) except for hypoxanthine-thymidine, which was purchased from American Type Culture Collection (Rockville, MD), and EFDM, which was purchased from Specialty Media (Lavallete, NJ). [125I]Iodocyanopindolol was obtained from NEN (Boston, MA). Isoproterenol, propranolol, Tris, EDTA, leupeptin, benzamidine, bacitracin, 3-isobutyl-1-methylxanthine sodium metabisulfite, and nadolol were obtained from Sigma Chemical Co. (St. Louis, MO). CGP 20712A was purchased from Research Biochemicals Inc. (Natick, MA). Metoprolol (racemic mixture) was kindly supplied by Ciba-Geigy. Compounds SR 59,230A, (−)-metoprolol, and carvedilol were synthesized in-house. L-748,328, (S)-N-[4-[2-[[3-[3-(aminosulfonyl)phenoxy]-2-hydroxypropyl]amino]ethyl]phenyl]benzenesulfonamide, and L-748,337, (S)-N-[4-[2-[[3-[3-(acetamidomethyl)phenoxy]-2-hydroxypropyl]ami-no]ethyl]phenyl]benzenesulfonamide as well as analogs (a) to (j) were synthesized according to routes previously outlined (Weber et al., 1998).
Cell Culture.
The human β1-, β2-, and β3-AR cDNAs were expressed in CHO dhfr (dihydrofolate reductase)-cells at levels of 100 to 200 fmol/mg of protein for the β1-AR and β2-AR and 40 to 60 fmol/mg of protein for the human β3-AR, as previously described (Candelore et al., 1996). Cells were grown in Iscoves modified Dulbecco’s medium with 25 mM HEPES, 10% fetal bovine serum, hypoxanthine-thymidine, glutamine, penicillin, streptomycin, and 700 μg/l of G418 at 37°C in a humidified incubator.
Binding Assay.
Cell membranes were prepared at either 3 or 4 days after seeding as previously described (Fisher et al., 1998). Ligands were diluted as 100× concentrated stocks in 100% dimethyl sulfoxide. Control experiments showed no adverse effects of up to 2% dimethyl sulfoxide in the binding assay. Equilibrium binding assays were performed in a final volume of 250 μl of TME buffer (75m M Tris, pH 7.4, 12.5 mM Mgcl2, 1.5 mM EDTA, 5 μg/ml leupeptin, 1 μg/ml benzamidine, 5 μg/ml soybean trypsin inhibitor, and 40 μg/ml bacitracin) containing 10 to 30 μg of membrane protein, [125I]iodocyanopindolol (125I-CYP; 40 pM for β1-AR and β2-AR and 250 pM for β3-AR), and serial dilutions of competing ligands. Nonspecific binding was determined in the presence of 100 μM (S)-(−)-propranolol and was 5 to 10% of the total binding. Assays were incubated for 90 min at room temperature and terminated by rapid filtration over GF/C filters presoaked in 0.1% polyethylenimine. Radioactivity was quantified with a Packard gamma counter. Protein determinations were made with the Bio-Rad protein assay kit (Bio-Rad, Richmond, CA) with γ-globulin as the standard.
Cyclic AMP (cAMP) Assay.
cAMP was measured in whole cells with the cAMP scintillation proximity assay kit from Amersham (RPA 538; Arlington Heights, IL) according to the manufacturer’s instructions. Activation with 1 μM isoproterenol was 6.1 ± 1.9 pmol cAMP/105 cells for the cells expressing human β3-AR and 9.0 ± 1.8 and 12.9 ± 3.8 pmol cAMP/105 for the cell lines expressing β1-AR and β2-AR, respectively. The maximal activation for all three cell lines was 4.4- to 5.8-fold over basal. For Schild plots, the ligands were prepared as 12× stocks, and the antagonist was incubated with the dissociated cells for 10 min at room temperature before addition of the agonist. The cells were then incubated for an additional 20 min at room temperature before termination of the assay. The maximal activation (%act at 10 μM) for each compound was determined at a concentration of 10 μM and is expressed relative to the maximal cyclase stimulation obtained for (−)isoproterenol at 10 μM.
Lipolysis Assays
Rhesus monkey adipose tissue was obtained by surgical biopsy, removed, and placed in Krebs-Ringer buffer containing 0.75 mM glucose and 4% fatty acid-free BSA (Fisher et al., 1998). The tissue was minced and preincubated for 10 min at 37°C. After three washes, minced tissue pieces (50–100 mg/assay) were transferred to 24-well plates. For antagonism of functional activation, the antagonist was added 30 min before addition of the agonist ligand. The incubation mixture was shaken gently in an incubator under 5% CO2 atmosphere for 2 h. The infranatant was collected for glycerol determination. The glycerol content of the samples was determined with Sigma kit 337A. Each assay point was done in quadruplicate. All animal handling procedures were reviewed and approved by the Institutional Animal Care and Use Committee.
Statistical Analysis.
IC50 and EC50 values were determined from concentration-response curve experiments measuring either binding or cAMP (inhibition or stimulation) levels as described above from at least three separate experiments done in duplicate with the iterative, nonlinear, least-squares curve-fitting computer program Prism (GraphPAD Software, San Diego, CA). IC50 values measured in competition binding assays were converted toKi values according to the method ofCheng and Prusoff (1973). The data are expressed as means ± S.D.
Results
Ligand-Binding Properties.
Whereas phenoxypropanolamines are typically antagonists of the β1-ARs and β2-ARs, some compounds in this class have also been shown to activate β3-AR (Mejean et al., 1995; Fisher et al., 1998; Weber et al., 1998). Nevertheless, the benzenesulfonamide derivatives L-747,328 and L-748,337 (Fig. 1) are high-affinity antagonists (Figs. 2 and3 and Table1) of all three human β-AR subtypes expressed in CHO cells. They inhibit binding of125I-CYP to human β3-AR, with Ki values of 3.7 ± 1.4 and 4.0 ± 0.4 nM, respectively. In contrast, these compounds have a much lower affinity for the β1-AR and β2-AR subtypes (Table 1). L-748,328 and L-748,337 are greater than 90-fold selective for β3-AR over β1-AR and are 27-fold and 51-fold selective for the β3-AR over the β2-AR. Human and rhesus monkey β3-AR displayed similar affinities for L-748,328 and L-748,337; however, these compounds did not bind well to rat β3-AR (Table2). The antagonist SR 59,230A (Manara et al., 1996) inhibits 125I-CYP binding to β3-AR with high affinity, but it displayed no selectivity for human β3-AR over the other two subtypes (Table 1 and Fig. 2). SR 59,230A showed similar affinity for human, rhesus monkey, and rat β3-AR (Table 2). In our hands, SR 59,230A is more potent at the human and rat cloned receptors than previously reported (Levasseur et al., 1995). Moreover, although having no agonist activity in cells expressing human β3-AR at levels of 40 to 60 fmol/mg, SR 59,230A did have agonist activity when tested with cells expressing 10 times the level of receptors, with an EC50 value of 71 ± 10 nM and a maximal activation of 63 ± 7%, in agreement with a previous report (Strosberg and Pietri-Rouxel, 1997). This is not an unexpected finding because it is well known that receptor expression affects both ligand efficacy and potency (Wilson et al., 1996).
Inhibition of Functional Activation
L-748,328, L-748,337 and SR 59,230A were tested for functional activation at the three cell lines expressing human β-AR, and none had significant agonist activity (Table 1). The nonselective β-AR agonist isoproterenol increases cAMP production in the CHO cells expressing human β3-AR, with an EC50 value of 95 ± 40 nM (Candelore et al., 1996). L-748,328 and L-748,337 inhibited cAMP production in CHO cells in response to 70 nM isoproterenol, with IC50 values of 4.6 ± 2.2 and 6 ± 2.8 nM, respectively (Fig. 3). L-748,328 at concentrations of 12, 30, 60, and 120 nM and L-748,337 at concentrations of 8, 12, 36, 72, and 120 nM increase the apparent EC50 for isoproterenol stimulation of cAMP production (Fig. 4). Schild regression plots of these data are linear, with slopes not significantly different from unity (1.1 ± 0.5 and 0.99 ± 0.20), and calculated pA2 values are 8.5 ± 0.1 for L-748,328 and 8.5 ± 0.1 for L-748,337. Whereas increasing doses of the antagonists shifted the concentration-response curves to the right, the maximal activation achieved was the same; thus, L-748,328 and L-748,337 are competitive antagonists of the nonselective agonist isoproterenol at the human β3-AR.
Rhesus Monkey Adipose Tissue Lipolysis Assays
L-748,328 and L-748,337 inhibit the lipolytic response elicited by a β3-AR-selective agonist in a manner consistent with their in vitro potencies. Figure5A shows the stimulation of glycerol production as a measure of lipolysis in rhesus monkey adipose tissue in response to the human β3-AR-selective agonist (S)-N-[4-[2-[[3-(4-hydroxyphenoxy)-2-hydroxypropyl]amino]ethyl] phenyl]-4-iodobenzenesulfonamide (L-742,791; Weber et al., 1998). Both L-748,328 and L-748,337 are capable of inhibiting glycerol production elicited by 100 nM L-742,791 in a dose-dependent manner, with IC50 values of 13 ± 10 and 15 ± 19 nM, respectively (Fig. 5). L-742,791-induced lipolysis in rhesus monkey adipocytes is not inhibited by propranolol or nadolol, even at concentrations as high as 10 μM (data not shown). Moreover, L-748,328 and L-748,337 inhibit 125I-CYP binding to rhesus monkey atrial and lung membranes, with IC50values consistent with their affinity at the rhesus monkey cloned β1-AR and β2-AR. L-748,328 and L-748,337 bind rhesus monkey cloned β1-AR, withKi values of 653 ± 4 and 176 ± 24 nM, and the rhesus monkey cloned β2-AR, withKi values of 97 ±25 and 142 ± 42 nM, respectively.
Structure-Activity Relationship of β3-AR Antagonism
Several analogs were prepared in the L-748,337 series to gain some insights into the determinants of β3-AR-selective antagonism (Table3). All of the compounds listed were tested for their ability to inhibit isoproterenol and L-742,791-stimulated cAMP production in cells expressing the β3-AR as described above. All were shown to inhibit agonist activity consistent with their affinity for the receptor. The unsubstituted amino derivative (a) is much less potent at β3-AR than amide derivatives (b) to (i). The amide derivatives examined were all quite potent at β3-AR, withKi values ranging from 1.3 nM for trimethylacetamide (f) to 4 nM for L-748,337. These derivatives were generally greater than 90-fold selective for binding to β3-AR over β1-AR, with the exception of the β-branched analogs isobutyramide (e), cyclohexamide (h), and benzamide (i), which were 61-, 72-, and 16-fold selective, respectively. Whereas both L-748,337 and its one carbon homolog (c) were nearly 50-fold selective for binding to β3-AR over β2-AR, increasing lipophilicity further resulted in a decrease in selectivity. For example, butyramide (d) and isovaleramide (g) were both only 12-fold selective over β2-AR. Thus, in this series, acetamide L-748,337 and propionamide (c) were optimal in terms of selectivity, with the latter being slightly more potent. Because all three β-AR subtypes can accept n-alkyl, branched, or cyclic/aromatic substitutions, there does not appear to be specific steric restriction in this part of the molecule. Moreover, these compounds expose the presence of a hydrophobic binding interaction. This contact appears to be more important for β2-AR than for β3-AR, because increasing the hydrophobicity decreases the selectivity by increasing the affinity for β2-AR while the affinity for the β3-AR remains constant.
Table 4 shows the activity of several classic β-AR antagonists at the three cloned β-ARs. These data agree with previous published results (Blin et al., 1993; Levasseur et al., 1995). With the exception of carvedilol, all show poor binding affinity for human β3-AR. Carvedilol is equipotent at all three receptors and therefore not selective. This contrasts with recently published data (Hieble et al., 1998) showing carvedilol binding to human cloned β3-AR with 92- and 400-fold less affinity than to human β1-AR and β2-AR, respectively. The greater human β3-AR selectivity of the compounds described herein appears to derive from the optimization of the aryloxypropanolamine parent structure with the benzenesulfonamide moiety (Weber et al., 1998).
Discussion
Classic receptor classification has been based on the efficacy and rank order of potency of ligands and the use of subtype-selective antagonists to determine specific interactions in isolated tissue preparations. For example, the development of compounds capable of stimulating the rodent atypical β-AR was important in validating its existence (Arch et al., 1984). In experiments designed to elucidate the pharmacological relevance of the β3-AR in humans, progress has been hampered by the lack of potent and selective human-specific agonists and antagonists.
Human β3-AR-selective agonists have been described recently (Fisher et al., 1998). Herein, we describe a series of compounds that are selective antagonists of human and rhesus monkey β3-AR. Among the most selective in the series, L-748,328 and L-748,337 display greater than 90-fold selectivity for human β3-AR versus β1-AR and 20- and 45-fold selectivity versus human β2-AR, respectively. In contrast, a previously described β3-AR antagonist, SR 59,230A, displays higher affinity at human cloned β1-AR and β2-AR, thus appearing to be a potent, nonselective β-AR antagonist.
The data presented here differ from earlier reports on SR 59,230A showing selectivity for β3-AR over the other two β-AR subtypes. In the studies by Manara et al. (1995; 1996), the β3-AR-specific effects were demonstrated by functional assays in rodent models. In vivo, SR 59,230A inhibits SR 58,611A-inhibited rat colonic motility (β3-AR), with an ID50 value of 3.6 mg/kg, which is 20- and 10-fold selective over inhibition of isoproterenol-mediated chronotropic effects (β1-AR) and salbutamol-mediated bronchodilatory action (β2-AR). In the in vitro rat colonic motility assays, the IC50 value for SR 59,230A was 1.7 nM. In the guinea pig atrial strips (β1-AR) and trachia relaxation assays (β2-AR), the IC50 values were 81.3 nM (48-fold selective) and 245 nM (140-fold selective), respectively. In another study with rodents, SR 59,230A was shown to inhibit SR 58,611A- and CGP 12177A-stimulated adenylyl cyclase activation in brown adipose tissue membranes, with IC50 values of 26.7 and 21 nM, respectively. In contrast, the antagonist was a poor inhibitor of isoproterenol-stimulated cAMP production in rat frontal cortex (β1-AR) and cerebellum (β2-AR) membranes, with IC50 values of greater than 10 μM (Nisoli et al., 1996), demonstrating greater than 500-fold selectivity. Thus, the selectivity varies widely depending on the assay.
The reasons for the discrepancy between our assays with cloned receptors and the work described above are unknown. However, several explanations are possible. There is a documented species difference in the binding and activation of β3-AR ligands (Granneman and Lahners, 1994). Endogenous catecholamines show similar activity and rank order of potency for both human and rodent β3-receptors. This is not always the case for synthetic ligands. Because the greatest discrepancy exists between the results with human cloned receptors and previously reported data with rodent models in the affinity of SR 59,230A for β1-AR and β2-AR, it may be possible that there are pharmacological species differences between human and rodent receptors. Nevertheless, extensive comparisons of human and rat cloned β1-AR and β2-AR with various classes of agonists and antagonists (including SR 59,230A) have failed to show any significant differences (our unpublished data). Therefore, the difference in human cloned β1-AR and β2-AR data compared with the rodent work cited herein cannot be attributed to species differences.
In vitro assays with human cloned receptors are, by design, a simpler model in which to study protein-ligand interactions and lack the complexity of organ bath preparations and in vivo models of drug action. It is now recognized that accessory proteins found in native tissue preparations affect ligand binding to endogenous receptors and receptor function (McLatchie et al., 1998; Möhler and Fritschy, 1999). It is not known whether the expression of these accessory proteins and or receptor dimerization may have an effect on the activity of ligands that bind to the β-ARs coexpressed in native tissues.
In addition, it is possible that the actions of the tetralins SR 58,611A and SR 59,230A on isolated rat colon motility may not be due to the presence of β3-AR in this tissue. The presence of an atypical β-AR activity (not β1-AR, β2-AR, or β3-AR) has been documented in guinea pig ileum; rat gastric fundus, cardiac tissue, and airway smooth muscle; and human platelets, among other locations (Arch and Kaumann, 1993). The identification of this receptor as a member of the β-AR family is supported by a lack of interaction with ligands for the histamine, dopamine, muscarinic, α1, α2-, and serotonin receptors. Recently, CGP 12177A was used as a selective ligand in rat atrium (Sarsero et al., 1998), rat and human adipocytes (Galitzky et al., 1997), and β3-AR knockout mice (Kaumann et al., 1998;Preitner et al., 1998) to demonstrate the existence in these tissues of a β4-AR. The use of CGP 12177A may be misleading because this compound has partial agonist activity at both human and rat cloned β1-ARs (our unpublished data), and it has been reported that β1-AR is up-regulated in β3-AR knockout mice (Susulic et al., 1995). Recent work with β3-AR knockout mice and mice in which β3-AR was replaced in brown adipose tissue and white adipose tissue with the adipose-specific aP2 promoter has shown that the effects of β3-AR agonists on gastrointestinal transit time are indirect and mediated solely by β3-AR expressed in adipose tissue (Fletcher et al., 1998).
Thus, the interpretation of the reported activity of SR 59,230A in rat tissues is unclear. However, the activity of specific agonists and antagonists, as measured in the clonal cell lines we have used in this work, has predicted activity and efficacy in the native primate tissue (Fisher et al., 1998)
In conclusion, we describe herein selective human β3-AR competitive antagonists L-748,328 and L-748,337 as useful pharmacological tools for the in vitro study of β3-AR action. The structure-activity relationship established for benzylsulfonamide antagonists of the human β3-AR described herein demonstrates the existence in this receptor of binding interaction that can accommodate small (acetamidomethyl, L-748,337) to larger and bulkier [benzamido, (i)] side chains. At this time, it is not possible to specify the portion of the molecule responsible for this activity. This is the subject for future, site-directed mutagenesis studies.
Acknowledgments
We thank Professor James G. Granneman (Wayne State University) for supplying the cloned human and rat β3-receptors. We also thank Olga Marko and Dr. Pasquale P. Vicario for performing the rhesus lipolysis assays and Paul Cunningham, Dr. Bonnie Friscino, Dr. William Feeney, and Don Hora from Merck’s Laboratory Animal Resources group for invaluable assistance in procuring the rhesus adipose tissue.
Footnotes
-
Send reprint requests to: Mari Rios Candelore, Merck & Co., P.O. Box 2000 Ry80M-213, Rahway, NJ 07065. E-mail:mari-candelore{at}merck.com
- Abbreviations:
- AR
- adrenergic receptor
- 125I-CYP
- [125I]iodocyanopindolol
- Received December 29, 1998.
- Accepted April 5, 1999.
- The American Society for Pharmacology and Experimental Therapeutics