Report

Rapid Identification of Subtype-Selective Agonists of the Somatostatin Receptor Through Combinatorial Chemistry

See allHide authors and affiliations

Science  23 Oct 1998:
Vol. 282, Issue 5389, pp. 737-740
DOI: 10.1126/science.282.5389.737

This article has a correction. Please see:

Abstract

Nonpeptide agonists of each of the five somatostatin receptors were identified in combinatorial libraries constructed on the basis of molecular modeling of known peptide agonists. In vitro experiments using these selective compounds demonstrated the role of the somatostatin subtype-2 receptor in inhibition of glucagon release from mouse pancreatic alpha cells and the somatostatin subtype-5 receptor as a mediator of insulin secretion from pancreatic beta cells. Both receptors regulated growth hormone release from the rat anterior pituitary gland. The availability of high-affinity, subtype-selective agonists for each of the somatostatin receptors provides a direct approach to defining their physiological functions.

Somatostatin is distributed throughout the endocrine system and has multiple physiological functions including inhibition of secretion of growth hormone (1), glucagon (2), insulin (3), gastrin, and other hormones secreted by the pituitary and gastrointestinal tract (4). It also acts as a neuromodulatory peptide in the central nervous system (5) and has been implicated as an inhibitor of tumor cell growth acting through somatostatin receptors coupled to tyrosine and serine-threonine phosphatases (6).

Somatostatin occurs naturally in two major forms: a tetradecapeptide (ss-14) and a 28–amino acid form (ss-28), both of which bind to somatostatin receptors that are coupled to heterotrimeric guanine nucleotide–binding proteins. Five distinct somatostatin receptors (sst1 through sst5) have been cloned from human tissues (7). Antibody probes have been developed for the sst2 and sst5 receptor subtypes and have been useful for establishing their potential functions (8–10). The sst2 receptor is localized with glucagon in rat pancreatic α cells (8), and the sst5 receptor localizes with insulin in the pancreatic β cells (9). In the pituitary gland, somatotrophs have sst2 and sst5 receptors, indicating that both receptor subtypes may have a role in regulation of growth hormone secretion (10).

In receptor-ligand binding assays, short peptide analogs of somatostatin, including MK-678 and octreotide, display selectivity for the sst2 receptor (11). These probes have been useful in assigning specific functions such as inhibition of secretion of growth hormone and glucagon to this particular receptor subtype (12,13). However, the lack of somatostatin analogs with selectivity for the sst1, sst3, sst4, and sst5 receptors has impeded progress in understanding functions associated with these receptors.

Our objective was to identify nonpeptide agonists selective for each sst receptor subtype. We used an integrated approach of combinatorial chemistry and high-throughput receptor-binding techniques to rapidly identify subtype-selective compounds. A cyclic hexapeptide somatostatin agonist (L-363,377) (Fig. 1) was used as a probe in a search of the Merck chemical collection, which consists of approximately 200,000 randomly assembled compounds. The side chains of residues Tyr-Trp-Lys in the cyclic peptide were given priority in the search on the basis of their similarity to the pharmacophore created by residues Trp8-Lys9 in the somatostatin-14 peptide (Ala1-Gly2-Cys3-Lys4-Asn5-Phe6-Phe7-Trp8-Lys9-Thr10-Phe11-Thr12-Ser13-Cys14). Seventy-five compounds were selected based on their similarity to the pharmacophore for evaluation in binding assays. The most potent of these compounds (L-264,930) had an apparent inhibition constant of 100 nM for the human sst2 receptor. L-264,930 is tripartite in structure with an aromatic moiety, a tryptophan moiety, and a diamine moiety, making it quite amenable to a combinatorial chemistry approach.

Figure 1

Molecular modeling of the somatostatin pharmacophore. The three-dimensional model of L-363,377 was used as a probe in a search of the Merck sample collection database. The coloration of the molecules depicted in two dimensions schematically highlights the superposition between the residue side chains with the correspondingly colored moieties of L-264,930. Substitutions of the aromatic group are shown in blue, of the tryptophan in green, and of the diamine in red. The search strategy is described in greater detail elsewhere (11).

Combinatorial library #1 was synthesized with 79 different substituents representing the aromatic moiety. Twenty different substituents for the Trp-amino acid module and 20 different substitutions for the diamine moiety yielded a library with 79 × 20 × 20 possible compounds, each occurring as multiple stereoisomers for a total of approximately 130,000 compounds. This library was screened against all five receptors in ligand-binding assays adapted to a 96-well format. One of the most active and sst2-selective groups of compounds contained a family of 1330 spiroindane analogs with the original lead structure (L-264,930) among them. However, this group of compounds was not analyzed further, because traditional medicinal chemistry efforts already identified potent sst2-selective leads from this class (14). Another potent and sst2-selective group that consisted of 1330 compounds, related by virtue of a benzimidazolone substitution at the aromatic position rather than a spiroindole substitution, was chosen for further analysis.

In the second round of screening, 20 samples were examined, each of which contained a family of compounds with the same amino acid substitution linked to the benzimidazolone and to all possible diamine substitutions. The family of compounds bearing β-methyl tryptophan had greater potency than those with other substitutions for the Trp.

In the final round of screening, a restricted diamine moiety was identified as the most favorable for sst2 potency and selectivity. The compound isolated from this series of experiments has an inhibition constant (K i) of 0.05 nM and greater than 6000-fold selectivity for the hsst2 receptor (Table 1). This compound (L-779,976) (Fig. 2) is a potent agonist that inhibited forskolin-stimulated accumulation of cyclic adenosine 3′,5′-monophosphate (cAMP) in Chinese hamster ovary cells (CHO-K1) expressing the sst2 receptor (Table 2). L-779,976 is also a potent inhibitor of growth hormone secretion from rat pituitary cells with a median inhibitory concentration (IC50) of 0.025 nM and comparable efficacy to that of somatostatin-14 (Table 2). Arginine-stimulated secretion of glucagon from mouse pancreatic islet cells was inhibited with an IC50 of 0.1 nM, similar to the binding affinity of L-779,976 for the sst2 receptor. Inhibition of insulin secretion from mouse pancreatic islets required a 1000-fold higher concentration of L-779,976 (Table 2).

Figure 2

Subtype-selective nonpeptide agonists of the somatostatin receptor. Color coding of sst receptor–selective compounds illustrates the relationship of various parts of each molecule to the original lead structure L-264,930. Colors are as inFig. 1. L-797,591, L-779,976, L-796,778, L-803,087, and L-817,818 are selective for the sst1, sst2, sst3, sst4, and sst5 receptors, respectively.

Table 1

Ligand binding assays. Receptor-ligand binding assays were done with membranes isolated from CHO-K1 cells expressing each of the cloned human sst receptors. Results shown are expressed asK i values (in nanomoles). The ligand 3-[125I]iodotyrosyl25–somatostatin-28(Leu8, D-Trp22, Tyr25) was obtained from Amersham and was used at a final concentration of 0.1 nM. Assays were done in 96-well polypropylene plates in a final volume of 200 μl. The assay buffer consisted of 50 mM tris-HCl (pH 7.8), 1 mM EGTA, 5 mM MgCl2, leupeptin (10 μg/ml), pepstatin (10 μg/ml), bacitracin (200 μg/ml), and aprotinin (0.5 μg/ml). Test compounds were examined over a range of concentrations from 0.01 to 10,000 nM. CHO-K1 cell membranes, ligand, and test compound were incubated at room temperature for 45 min and then collected on Packard 96-well glass fiber filter plates treated with 0.1% polyethyleneimine. Plates were washed with cold 50 mM tris-HCl (pH 7.8), then dried before counting. K i values were calculated with the Cheng-Prussof equation (17).K d values for 3-[125I]iodotyrosyl25–somatostatin-28(Leu8,D-Trp22, Tyr25) with each of the five receptors were determined from Scatchard plots of saturation binding curves. The apparent dissociation constants were 1.5, 0.1, 0.4, 4.2, and 0.7 nM for receptors sst1 through sst5, respectively.

View this table:
Table 2

Functional activity of the sst receptor–selective somatostatin analogs. cAMP accumulation in cells expressing the sst receptors (18): Each receptor subtype was evaluated in at least three separate experiments. CRE–β galactosidase assay: The assay protocol used for measuring sst1-mediated inhibition of cAMP accumulation has been previously described (11). Somatostatin-14 had an IC50 of 0.2 nM in this assay. The sst1-selective compound was evaluated in duplicate in two separate experiments. Growth hormone (GH) release from primary cultures of rat pituitary cells: Primary cultures of rat anterior pituitary cells were maintained for 3 to 4 days at 37°C in 5% CO2 and 95% air. Cells were treated with test compounds for two hours at 37°C. Secreted growth hormone was detemined by radioimmunoassay (11). Compounds were evaluated in two separate experiments. Mouse pancreatic islet preparations: The protocol for pancreatic islet preparation and measurement of secreted glucagon and insulin has been described previously (11). The sst1-selective compound was examined once in triplicate. All other compounds were evaluated in triplicate in two separate experiments. CHO K1, CHO K1 cells; RPC, rat pituitary cells; MI, mouse islets; ss-14, somatostatin-14.

View this table:

Combinatorial library #2 consisted of 350,000 different compounds. It was constructed according to the same principle as combinatorial library #1, but was expanded in all three dimensions (147 × 22 × 21). Of the 147 pools of compounds screened in the receptor-binding assays, two were chosen for further analysis, which resulted in identification of compounds selective for the sst1 and sst3 receptors (Fig. 2). In CHO-K1 cells expressing the hsst3 receptor, L-796,778 was a partial agonist with an IC50value of 18 nM for inhibition of forskolin-stimulated production of cAMP. Efficient coupling of the hsst1 receptor to adenylate cyclase was not achieved in the CHO-K1 cell line (15). Therefore, an alternative cell line was chosen for evaluation of the sst1-selective compound. In the mouse L-cell line (16) transfected with the hsst1 receptor and a cAMP response element (CRE)–β-galactosidase reporter, L-797,591 displayed agonist activity with an IC50 of 3 nM. Neither of these compounds inhibited release of growth hormone, glucagon, or insulin (Table 2).

An sst4 receptor–selective compound was isolated from combinatorial library #3, an aryl indole library of limited complexity. The compound, L-803,087, has a diamine moiety that maps to lysine on the pharmacophore (Fig. 2), but relation of this molecule to the aromatic and the Trp substituents of the pharmacophore is not obvious. In binding and functional assays, L-803,087 was an hsst4 receptor agonist. L-803,087 did not inhibit secretion of growth hormone, insulin, or glucagon (Table 2).

An sst5-selective ligand was identified in combinatorial library #4, another extension of library #1. L-817,818 was 130-fold more selective for sst5 as compared to sst2 and eightfold more selective for sst1 (Table 1). The compound inhibited growth hormone secretion from rat primary pituitary cell cultures, with an IC50 of 3 nM, comparable to its binding affinity for sst5. Because a role for sst1 in growth hormone secretion was ruled out with the sst1-selective compound L-797,591, sst5 apparently participates in control of growth hormone release. L-817,818 also inhibited insulin secretion from pancreatic islets but did not inhibit glucagon secretion (Table 2).

Combinatorial chemistry proved useful for rapid refinement of new lead compounds. By adapting the sst receptor ligand binding assays to a 96-well format and screening complex mixtures, we were able to examine hundreds of thousands of new chemical entities in five different somatostatin receptor assays in a short period of time and to identify sst receptor–selective compounds that may be useful for dissecting functions of the individual receptors. These nonpeptide, small-molecule analogs of somatostatin may be useful in development of orally active chemotherapeutic agents capable of crossing the blood-brain barrier.

  • * To whom correspondence should be addressed. E-mail: susan_rohrer{at}merck.com

REFERENCES AND NOTES

View Abstract

Navigate This Article