Characterization of a Novel, Non-peptidyl Antagonist of the Human Glucagon Receptor*

We have identified a series of potent, orally bioavailable, non-peptidyl, triarylimidazole and triarylpyrrole glucagon receptor antagonists. 2-(4-Pyridyl)-5-(4-chlorophenyl)-3-(5-bromo-2-propyloxyphenyl)pyrrole (L-168,049), a prototypical member of this series, inhibits binding of labeled glucagon to the human glucagon receptor with an IC50 = 3.7 ± 3.4 nm(n = 7) but does not inhibit binding of labeled glucagon-like peptide to the highly homologous human glucagon-like peptide receptor at concentrations up to 10 μm. The binding affinity of L-168,049 for the human glucagon receptor is decreased 24-fold by the inclusion of divalent cations (5 mm). L-168,049 increases the apparent EC50 for glucagon stimulation of adenylyl cyclase in Chinese hamster ovary cells expressing the human glucagon receptor and decreases the maximal glucagon stimulation observed, with aK b (concentration of antagonist that shifts the agonist dose-response 2-fold) of 25 nm. These data suggest that L-168,049 is a noncompetitive antagonist of glucagon action. Inclusion of L-168,049 increases the rate of dissociation of labeled glucagon from the receptor 4-fold, confirming that the compound is a noncompetitive glucagon antagonist. In addition, we have identified two putative transmembrane domain residues, phenylalanine 184 in transmembrane domain 2 and tyrosine 239 in transmembrane domain 3, for which substitution by alanine reduces the affinity of L-168,049 46- and 4.5-fold, respectively. These mutations do not alter the binding of labeled glucagon, suggesting that the binding sites for glucagon and L-168,049 are distinct.

We have identified a series of potent, orally bioavailable, non-peptidyl, triarylimidazole and triarylpyrrole glucagon receptor antagonists. 2-(4-Pyridyl)-5-(4chlorophenyl)-3-(5-bromo-2-propyloxyphenyl)pyrrole (L-168,049), a prototypical member of this series, inhibits binding of labeled glucagon to the human glucagon receptor with an IC 50 ‫؍‬ 3.7 ؎ 3.4 nM (n ‫؍‬ 7) but does not inhibit binding of labeled glucagon-like peptide to the highly homologous human glucagon-like peptide receptor at concentrations up to 10 M. The binding affinity of L-168,049 for the human glucagon receptor is decreased 24-fold by the inclusion of divalent cations (5 mM). L-168,049 increases the apparent EC 50 for glucagon stimulation of adenylyl cyclase in Chinese hamster ovary cells expressing the human glucagon receptor and decreases the maximal glucagon stimulation observed, with a K b (concentration of antagonist that shifts the agonist dose-response 2-fold) of 25 nM. These data suggest that L-168,049 is a noncompetitive antagonist of glucagon action. Inclusion of L-168,049 increases the rate of dissociation of labeled glucagon from the receptor 4-fold, confirming that the compound is a noncompetitive glucagon antagonist. In addition, we have identified two putative transmembrane domain residues, phenylalanine 184 in transmembrane domain 2 and tyrosine 239 in transmembrane domain 3, for which substitution by alanine reduces the affinity of L-168,049 46and 4.5-fold, respectively. These mutations do not alter the binding of labeled glucagon, suggesting that the binding sites for glucagon and L-168,049 are distinct.
Glucagon is a 29-amino acid peptide that is an important counter-regulatory hormone in the control of glucose homeostasis (1). Glucagon secretion from the endocrine pancreas induces an increase in hepatic glycogenolysis and gluconeogenesis, and it attenuates the ability of insulin to inhibit these processes. As such, the overall rates of hepatic glucose synthesis and glycogen metabolism are controlled by the systemic ratio of insulin and glucagon (2,3). Therefore, glucagon antagonists have the potential to improve hepatic insulin sensitivity and to be effective hypoglycemic agents.
Peptidyl glucagon antagonists and their hypoglycemic activity were first described over 15 years ago, and an extensive exploration of the structure/activity relationships of these glu-cagon analogs has been reported (4 -6). The hepatic receptor for glucagon was cloned recently (7,8), confirming that it is a member of the seven-transmembrane domain, G-protein-coupled receptor superfamily. This receptor superfamily has a binding pocket for small-molecule ligands within the transmembrane domain that has made it possible to identify nonpeptidyl antagonists for many receptor families in which the endogenous ligands are small peptides or proteins (9). Thus, we initiated an effort to identify non-peptidyl, orally active antagonists for the human glucagon receptor.
Collins et al. (10) have described a dichloroquinoxaline glucagon antagonist with weak affinity (IC 50 ϭ 4 M) for the rat glucagon receptor. However, there have been no subsequent reports in the patent or scientific literature describing the development of potent antagonists from this series. Our initial screening efforts identified a series of triarylimidazole and triarylpyrrole compounds with significant binding affinity for the human glucagon receptor, and efforts to evaluate the structure-activity relationships of this series have lead to the identification of potent glucagon antagonists (11). In the present article, we describe the identification and characterization of a potent glucagon antagonist from this series.

MATERIALS AND METHODS
Characterization of Binding Affinity and Functional Activity-Stable CHO 1 cell lines or COS cells transiently expressing the human glucagon receptor were prepared as described previously (8,12). Antagonist binding affinity was assessed by measuring inhibition of radiolabeled glucagon binding to CHO cell membranes. Briefly, 125 I-glucagon (58 pM) binding to the membrane preparation was measured in 20 mM Tris, pH 7.4, containing 1 mM dithiothreitol, 5 g/ml leupeptin, 10 g/ml benzamidine, 40 g/ml bacitracin, 5 g/ml soybean trypsin inhibitor, and 3 M o-phenanthroline Ϯ 1 M glucagon for 1 h at room temperature. Bound cpm were recovered by filtration using a Tomtec harvester and quantified in a ␥-scintillation counter.
The ability of compound to inhibit glucagon-stimulated adenylyl cyclase was assessed as described previously (12). Briefly, cells were harvested from monolayers with enzyme-free cell dissociation solution (Specialty Media, Inc.) and were pelleted at 500 ϫ g. The cells were resuspended at 100,000 cells/100 l in 75 mM Tris-HCl, pH 7.4, containing 250 mM sucrose, 12.5 mM MgCl 2 , 1.5 mM EDTA, 0.1 mM Ro-20-1724 (Biomol, Inc.), leupeptin (5 g/ml), benzamidine (10 g/ml), bacitracin (40 g/ml), soybean trypsin inhibitor (5 g/ml), and 0.02% bovine serum albumin. Cells were incubated for 30 min at 22°C with increasing peptide concentrations in the presence or absence of antagonist followed by lysis by boiling. Lysates were analyzed for cAMP content versus a nonacetylated cAMP standard curve using the Amersham cAMP radioimmunoassay scintillation proximity assay kit. Data were analyzed using Packard TopCount with RIASmart and GraphPad Prism software. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Preparation and Assessment of Mutant Glucagon Receptors-
The wild-type glucagon receptor was cloned into pCI-neo (Promega Corp., Madison, WI) at the NheI and XbaI sites. Mutations at amino acid residues 184 (TM2, F184A) and 239 (TM3, Y239A) of the wild-type glucagon receptor were prepared using recombinant polymerase chain reaction. The sense primers for the F184A and the Y239A mutants were 5Ј-TTTGCGTCCGCCGTGCTGAA-3Ј and 5-GTGGCCAACGCCTGCTG-GCTGCT-3Ј, respectively. The resulting mutant fragments were excised with either ApaI/SacII (F184A) or SacII/BclI (Y239A) and ligated into the pCI-neo wild-type glucagon receptor. Mutations were confirmed by automated sequencing. All other mutants were prepared in a similar manner using oligonucleotides specific to the production of an alanine at the indicated position with the wild-type receptor, and mutations were confirmed by sequencing. The restriction sites for the other mutations were as follows: TM1 (Tyr-145 The transfection of COS cells and the preparation of membranes for binding assays were performed as described previously (12). Compound titrations using wild-type receptor transiently expressed in COS cells were performed as controls in all experiments utilizing mutant receptors.

RESULTS
Several triarylimidazole compounds were identified from compound collection screening that inhibit labeled glucagon binding to the human glucagon receptor with potencies ranging from 300 nM to 1 M (11). Optimization of this activity by structural modification of these lead compounds led to the identification of the triarylpyrrole glucagon antagonist, L-168,049 (Fig. 1). This compound inhibits binding of labeled glucagon with an IC 50 of 3.7 Ϯ 3.4 nM (n ϭ 7). Surprisingly, the affinity of L-168,049 is decreased 48-fold by the inclusion of 5 mM MgCl 2 in the assay (Fig. 2). In the presence of 5 mM MgCl 2 , the compound inhibits binding of labeled glucagon with an IC 50 of 179 Ϯ 86 nM (n ϭ 3). An equivalent decrease in affinity is also observed with MnCl 2 and CaCl 2 , but not with NaCl, indicating that the loss in affinity is divalent cation-dependent (data not shown). L-168,049 does not inhibit binding of labeled glucagonlike peptide-1 (GLP-1) to the highly homologous GLP-1 receptor at concentrations as high as 10 M.
Inclusion of increasing doses of L-168,049 increases the apparent EC 50 for glucagon stimulation of adenylyl cyclase and decreases the maximal stimulation observed (Fig. 3). A competitive antagonist should not alter maximal agonist-induced activation of the receptor, suggesting that L-168,049 is a noncompetitive antagonist of glucagon. Schild transformation of these data is linear with a slope of 0.6, which is consistent with a noncompetitive mechanism. The affinity of L-168,049, as measured by the concentration of antagonist that shifts the agonist dose-response 2-fold (K b ), is 25 nM. In the absence of exogenously added glucagon, L-168,049 does not alter intracellular cAMP levels, indicating that it has no agonist activity. 125 I-Glucagon dissociates from the human glucagon receptor in the presence of MgCl 2 with k Ϫ1 ϭ 0.013 Ϯ 0.001 min Ϫ1 (n ϭ 2). Inclusion of L-168,049 (1 M) increases the rate of dissociation 4-fold (Fig. 4). These data confirm that L-168,049 is a noncompetitive antagonist of glucagon. The nonhydrolyzable guanine nucleotide analog, guanosine 5Ј-(␤,␥-imido)-triphosphate, which reduces glucagon affinity by uncoupling the receptor from G-protein, increases the rate of dissociation 29-fold. These data suggest that L-168,049 does not inhibit glucagon binding by this mechanism. Although the data shown are in the presence of divalent cation, L-168,049 also increases the rate of dissociation of 125 I-glucagon 4-fold in the absence of divalent cation. To confirm that L-168,049 binds within the transmembrane domain of the glucagon receptor, 19 residues predicted to be within this region (8) (Table I). Inclusion of divalent cation reduced the affinity of the compound for both the wild-type and mutant receptors, indicating that these residues were not involved in mediating the effect of divalent cations on compound affinity (data not shown).
The triarylpyrrole glucagon antagonists have poor affinity for the rat, guinea pig, and rabbit glucagon receptors (Ͼ1 M). However, L-168,049 inhibits labeled glucagon binding to the murine and canine glucagon receptors with IC 50 ϭ 63 Ϯ 44 nM (n ϭ 3) and 60 nM, respectively. As with the human receptor, these affinities are reduced Ն20-fold in the presence of 5 mM MgCl 2 (IC 50 ϭ 331 Ϯ 122 nM (n ϭ 3)) and Ͼ1 M for the murine and dog receptors, respectively. DISCUSSION We have identified a structural class of non-peptidyl glucagon antagonists. Preliminary data indicate that these compounds are absorbed systemically after oral administration (11), suggesting that they have the potential to be useful, orally active hypoglycemic agents. The prototype of this class, L-168,049, is a high affinity, noncompetitive antagonist of the human glucagon receptor, which inhibits glucagon-stimulated adenylyl cyclase with a K b of 25 nM. This compound has poor affinity for the highly homologous GLP-1 receptor and for a panel of other G-protein-coupled receptors, indicating that it is a selective glucagon antagonist.
The observation that inclusion of L-168,049 alters both the apparent affinity and maximal receptor activation of glucagon suggests that these compounds are noncompetitive glucagon antagonists. The demonstration that L-168,049 increases the dissociation rate of labeled glucagon confirms this hypothesis, because these data suggest that both glucagon and L-168,049 are able to bind to the receptor at the same time.
Whereas mutagenesis data suggest that the binding site for glucagon is within the amino-terminal domain and other extracellular domains of the glucagon receptor (13)(14)(15), our data suggest that the L-168,049 binding site is within the transmembrane domain. The binding of glucagon to the 19 transmembrane domain mutant receptors is unaltered, suggesting that these mutations did not dramatically effect overall receptor conformation. Phenylalanine 184 and tyrosine 239, the two mutations with altered affinity for L-168,049, are predicted to be within the second and third transmembrane domains, respectively. Thus, these compounds appear to bind in the same common binding pocket that has been postulated for other G-protein-coupled receptors (9). These data, and the observation that L-168,049 is a noncompetitive antagonist of glucagon, suggest that the compound allosterically modulates the binding of glucagon to the receptor.
Divalent cations have been shown to allosterically enhance the binding of agonists to many types of G-protein-coupled receptors, including the binding of glucagon to the human glucagon receptor (12), but divalent cation effects on antagonist binding affinity are less well documented. However, inclusion of 5 mM divalent cation decreases the affinity of L-168,049 48-fold. The activity of antagonists in the cellular functional assays that are performed under more physiologically relevant buffer conditions correlates more closely with the binding affinity measured in the presence of divalent cation. More importantly, because circulating levels of MgCl 2 are in the low mM range, the decrease in affinity observed in the presence of divalent cation would be expected to be important for the activity of antagonists in vivo.
We previously showed that the deletion of residues 252-259 in the second intracellular domain of the glucagon receptor attenuates the effects of divalent cation on the binding affinity for glucagon (12). However, this mutation does not significantly alter the affinity for L-168,049, and inclusion of divalent cation reduces affinity of L-168,049 for this mutant protein 30-fold. Thus, the effects of divalent cation on agonist and antagonist binding appear to be distinct.
Because L-168,049 has poor affinity for all of the species orthologs of the glucagon receptor tested to date, it has been difficult to test its in vivo efficacy. However, L-168,049 has the highest affinity for the murine receptor of any compound in this series identified to date, and high levels of compound (3 M) increase the apparent EC 50 for glucagon-stimulated adenylyl cyclase activation in murine liver membranes by Ͼ10-fold.
These data suggest that the compound is an antagonist of glucagon activity in a physiological tissue preparation. However, the compound (50 mg/kg, per os) does not inhibit gluca-  gon-stimulated increases in the blood glucose in mice, suggesting that its activity at the murine receptor is insufficient to be a useful tool in this species. Thus, identification of a suitable animal model is required for the analysis of the effects of this class of orally active glucagon antagonists on glucose homeostasis in vivo. Recently we have generated a murine line in which the native murine glucagon receptor has been replaced with the human receptor by homologous recombination, and preliminary data suggest that the activity of compounds in liver membranes from these mice correlates with their affinity as measured using the cloned human receptor. 2 Our current efforts are focused on expanding this murine line to extend these observations and to determine the effect of this class of compounds on glucose homeostasis.