Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

All five subtypes of muscarinic acetylcholine receptors (mAChRs) are members of the G protein-coupled receptor superfamily and are responsible for mediating the majority of actions in the body of the neurotransmitter acetylcholine (Caulfield, 1993). Despite the widespread tissue distribution of these receptors and their involvement in a variety of physiological processes, the extremely high sequence homology in the classic ligand-binding domain across the mAChRs has made it difficult to design therapeutic agents that can effectively target one receptor subtype to the exclusion of all others. Thus, there is an ongoing need for effective structure-activity studies based on selected mAChRs.

Over the last two decades, one of the most intensively investigated areas for potential treatment with selective mAChR agonists has been Alzheimer's disease (Avery et al., 1997; Fisher et al., 1998; Ladner and Lee, 1998). Research efforts in this field have been spurred by the finding that despite the significant loss in presynaptic cholinergic innervation that occurs in the disorder, postsynaptic M₁ mAChRs, which are involved in learning and memory, remain largely preserved (Whitehouse et al., 1982; Mash et al., 1985; Ferrari-DiLeo et al., 1995). Hence, selective M₁ mAChR agonists would be expected to offer an improvement in cognitive deficits by exploiting the functionality of these postsynaptic M₁ mAChRs. Accordingly, the mAChR agonist xanomeline has been identified as one of the most potent M₁ agonists to date (Shannon et al., 1994), and clinical trials with this agent have shown significant improvement in a number of cognitive and behavioral symptoms associated with Alzheimer's disease (Bodick et al., 1997). Furthermore, we have recently demonstrated that this agonist may possess a novel mode of binding and activation at the M₁ mAChR (Christopoulos et al., 1998, 1999). However, despite the therapeutic promise demonstrated by xanomeline, a number of problems persist. Since this agent displays similar binding affinity at all five mAChRs, its functional selectivity will depend on the degree of receptor reserve associated with each mAChR on which it acts, and this can vary from tissue to tissue. For instance, despite xanomeline's ability to act as a functionally selective M₁ mAChR agonist under a variety of conditions, a recent study by Wood et al. (1999) has shown that this same agent has the potential to yield a robust response at M₃ and M₄ mAChRs in addition to the M₁ mAChR as a result of the presence of receptor reserve for each of these subtypes in CHO cells. Another complication relates to the lipophilic nature of xanomeline and its pronounced biotransformation in humans (Andersen and Hansen, 1997; Bymaster et al., 1997). The need therefore remains for more selective mAChR agonists that also retain oral bioavailability and metabolic stability.

In continued efforts to identify potent and selective mAChR agonists meeting these criteria, we have previously published on the structure-activity relationships leading to the design and synthesis of the mAChR agonist phenylpropargyloxy-1,2,5-thiadiazole-quinuclidine (NNC 11-1314; Fig. 1) (Sauerberg et al., 1998b). This work was further extended by evaluation of M₁/M₂ activity as a function of halogen substitution on the phenylpropargyl part of 1-(1,2,5-thiadiazol-4-yl)-4-azacyclo[2.2.1.0^2,6]heptanes (Jeppesen et al., 1999). Data showed that even the relatively small halogen substituents such as fluorine and chlorine had marked influence on both receptor potency and efficacy of the compounds. The present article expands on the study of another aspect of structure-activity relationships of NNC 11-1314, namely, the effects of its dimerization. This approach was chosen for two reasons: first, because bivalent ligands have previously been shown to possess distinct pharmacological properties, including better selectivity, at several classes of receptors, such as the opioid receptors (Portoghese, 1989), serotonin receptors (Perez et al., 1998), and mAChRs (Melchiorre et al., 1989); second, we recently provided data supporting the use of functionally selective M₁/M₄ mAChR agonists as novel antipsychotic agents, in addition to their potential use in treating cognitive dysfunction (Bymaster et al., 1998; Sauerberg et al., 1998a; Shannon et al., 2000). It was thus reasoned that a bivalent ligand approach may provide another avenue to achieve such target selectivity. In the present study, we evaluated the pharmacology of NNC 11-1607 and NNC 11-1585 (Fig. 1), two dimeric analogs of NNC 11-1314, at the five mAChRs and show that dimerization of mAChR ligands represents a viable approach to the design of potent and selective agonists for these receptors.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Structures of the novel mAChR agonists used in this study.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. N-[³H]Methylscopolamine (84.5 Ci/mmol) was purchased from PerkinElmer Life Sciences (Wilmington, DE); myo-[³H]inositol (107 mCi/mmol), [³H]adenine (26 Ci/mmol) and [adenine-U-¹⁴C]cAMP (244 Ci/mmol) were purchased from Amersham Pharmacia Biotech (Arlington Heights, IL); [¹⁴C]inositol-1-phosphate (300 mCi/mmol) was purchased from American Radiolabeled Chemicals (St. Louis, MO); Dulbecco's modified Eagle's medium was purchased from Invitrogen (Carlsbad, CA); geneticin was purchased from Calbiochem (San Diego, CA); and bovine calf serum was purchased from Hyclone Laboratories (Logan, UT). All other reagents were purchased from Sigma Chemical Co. (St. Louis, MO).

General Chemistry. The dimers were synthesized analogous to the procedure published previously (Sauerberg et al., 1998b) by cross-coupling dibromobenzene with propargyl alcohol using a palladium catalyst and copper(I)iodide to give bis(3-hydroxy-propyn-1-yl)benzenes. Reaction of this bis alcohol with the known (S)-3-(3-butylsulfonyl-1,2,5-thiadiazol-4-yl)-1-azabicyclo[2.2.2]octane under basic conditions gave the desired dimers in moderate yields.

NNC 11-1585. A mixture of 1,3-dibromobenzene (7.1 g, 30.0 mmol), copper(I)iodide (114 mg, 0.6 mmol), and bis(triphenylphophine)palladium(II)chloride (98%) (210 mg, 0.3 mmol) in triethylamine (25 ml) and diisopropylamine (75 ml) was stirred under a nitrogen atmosphere at room temperature for 1 h. 3-Propynol (5.25 ml, 93.8 mmol) was added, and the reaction mixture was heated at 60°C for 16 h. The mixture was filtered, and the filtrate was evaporated to dryness. The product was purified by flash chromatography using toluene/ethyl acetate (9:1) graduated to toluene/ethyl acetate (4:1) as eluent to give 4.1 g (73%) of 1,3-bis(3-hydroxy-propyn-1-yl)benzene as an oil. To an ice-cooled solution of 1,3-bis(3-hydroxy-propyn-1-yl)benzene (372 mg, 2.0 mmol) and (3S)-3-(3-butylsulfonyl-1,2,5-thiadiazol-4-yl]-1-azabicyclo[2.2.2]octane (1.3 g, 4.0 mmol) in tetrahydrofuran (50 ml) was added sodium hydroxide (80%) (360 mg, 12.0 mmol) under a nitrogen atmosphere. The mixture was stirred at 50°C for 4 days. The reaction was quenched with ice water, and 4 N HCl was added to achieve pH 2. The tetrahydrofuran was evaporated, and the residue was washed with ether (three times and then discarded). The solution was made basic with NH₃ to pH 10 to 11, and the product was isolated by extraction with ether (three times). The organic phase was concentrated in vacuo and purified by HPLC on a Source RPC 15 column equilibrated in ethanol/20 mM 1,3-diaminopropane (3:7) and eluted with 30 to 55% acetonitrile (Amersham Pharmacia Biotech AB, Uppsala, Sweden). The pure fractions were pooled, and the acetonitrile was evaporated. NH₃ was added to the residue to pH 10 to 11, and the product was isolated by extraction with ether (9:1) (six times). The organic phases were dried (MgSO₄), filtered, and concentrated in vacuo. The purified product was crystallized with L-(+)-tartaric acid in isopropanol to give the title compound in 350 mg (20%) yield: melting point, 101 to 105°C; ¹H NMR (dimethyl sulfoxide-d₆, 400 MHz)  1.55-1.70 (m, 4H), 1.85-2.05 (m, 4H), 2.30 (bs, 2H), 3.10-3.30 (m, 8H), 3.50-3.65 (m, 6H), 4.0 (s, 4H), 5.40 (s, 4H), and 7.40-7.55 (m, 4H); MS m/z 573 (M + 1).

NNC 11-1607. A mixture of 1,4-dibromobenzene (11.8 g, 5 0.0 mmol), copper(I)iodide (300 mg, 1.6 mmol) and bis(triphenylphophine)palladium(II)chloride (98%) (720 mg, 1.0 mmol) in triethylamine (30 ml) and diisopropylamine (150 ml) was stirred under a nitrogen atmosphere at room temperature for 1 h. 3-Propynol (6.4 ml, 110 mmol) was added, and the reaction mixture was heated at 55°C for 16 h. The mixture was filtered, and the filtrate was evaporated to dryness. The product was purified by flash chromatography using dichloromethane graduated to dichloromethane/methanol (19:1) as eluent to give 2.9 g (32%) of 1,4-bis(3-hydroxy-propyn-1-yl)benzene as an oil. To an ice-cooled solution of 1,4-bis(3-hydroxy-propyn-1-yl)benzene (372 mg, 2.0 mmol) and (3S)-3-(3-butylsulfonyl-1,2,5-thiadiazol-4-yl]-1-azabicyclo[2.2.2]octane 1.3 g, 4.0 mmol) in tetrahydrofuran (50 ml) was added sodium hydroxide (80%) (360 mg, 12.0 mmol) under a nitrogen atmosphere. The mixture was stirred at 50°C for 4 days. The reaction was quenched with ice water, and 4 N HCl was added to pH 2. The tetrahydrofuran was evaporated, and the residue was washed with ether (three times and then discarded). The solution was made basic with NH₃ to pH 10 to 11, and the product was isolated by extraction with dichloromethane (three times). The organic phase was concentrated in vacuo and purified by HPLC on a Source RPC 15 column equilibrated in ethanol/20 mM 1,3-diaminopropane (3:7) and eluted with 5 to 20% acetonitrile. The pure fractions were pooled, and the acetonitrile was evaporated. NH₃ was added to the residue to pH 10 to 11, and the product was isolated by extraction with dichloromethane/isopropanol (9:1) (nine times). The organic phases were dried (MgSO₄), filtered, and concentrated in vacuo. The purified product was crystallized with L-(+)-tartaric acid in isopropanol to give the title compound in 444 mg (26%) yield: melting point, 123 to 126°C; ¹H NMR (dimethyl sulfoxide-d₆, 400 MHz)  1.60 (m, 4H), 1.85-2.0 (m, 4H), 2.30 (bs, 2H), 3.05-3.35 (m, 8H), 3.45-3.65 (m, 6H), 4.02 (s, 4H), 5.40 (s, 4H), and 7.50 (s, 4H); MS m/z 573 (M + 1).

Cell Culture. CHO cells stably expressing the individual human M₁ to M₅ mAChRs (provided by Dr. M. Brann, University of Vermont Medical School, Burlington, VT) were grown for 4 days at 37°C in Dulbecco's modified Eagle's medium, supplemented with 10% bovine calf serum and 50 µg/ml geneticin, in a humidified atmosphere consisting of 5% CO₂ and 95% air. Cells were used 4 days after subculture and were harvested by trypsinization followed by centrifugation (300g, 3 min) and resuspension of the pellet in HEPES buffer A (110 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl₂, 1 mM MgSO₄, 25 mM glucose, 50 mM HEPES, 58 mM sucrose; pH 7.4 ± 0.02; 340 ± 5 mOsm), repeated twice.

Membrane Preparation. CHO cells were harvested as described above and centrifuged (300g, 3 min) before resuspension of the pellet in HEPES buffer B (0.03 M HEPES, 0.5 mM MgCl₂, 0.5 mM EGTA; pH 7.4). The pellet was washed twice and resuspended in HEPES buffer B before homogenization with Ultra-Turrax (Janke & Kunkel, Staufen, Germany) (2 × 10-s bursts, with a 30-s period of cooling on ice between homogenizations) and centrifugation at 1,000g for 10 min. The resulting supernatant was collected and centrifuged at 30,000g for 30 min, after which the pellet (P₂ fraction) was resuspended and immediately used in the binding assays.

Competition Binding Assays. CHO cell (10-50 µg/tube) or rat tissue (50-100 µg/tube) membranes were incubated with 0.2 nM [³H]NMS, in the absence or presence of increasing concentrations of NNC compounds 11-1314, 11-1607, or 11-1585 in a total volume of 1 ml. Incubation was allowed to proceed for 1 h at 37°C before termination by rapid filtration through Whatman GF/B filters (Whatman, Clifton, NJ), positioned on a Brandell cell harvester (Gaithersburg, MD). Filters were washed three times with 4-ml aliquots of ice-cold saline and dried before radioactivity was measured using liquid-scintillation counting. Nonspecific binding was defined using 10 µM atropine.

Phosphoinositide Hydrolysis Assay. CHO M₁, M₃, or M₅ cells were suspended in buffer and loaded with myo-[³H]inositol (8 µCi/ml) for 1 h at 37°C. Labeled cells were then washed with HEPES buffer A (containing 10 mM LiCl), distributed to assay tubes (~5 × 10⁵ cells/tube), and allowed to incubate for 15 min at 37°C. For each individual experiment, concentration-response curves for the stimulation of PI hydrolysis were constructed for oxotremorine-M, NNC 11-1314, NNC 11-1607, and NNC 11-1585. The reactions were allowed to proceed for 30 min after the addition of agonist or vehicle control before being stopped with chloroform/methanol (1:2). Tubes were centrifuged (450g; 15 min), and total inositol phosphates were separated by ion-exchange chromatography on Dowex AG1-X8 resin, with [¹⁴C]inositol-1-phosphate as a recovery standard. The amount of radioactivity (dpm) in each sample was then determined by liquid-scintillation counting.

cAMP Accumulation Assay. CHO cells expressing the M₂ or M₄ mAChR were suspended in buffer and loaded with [³H]adenine (20 µCi/ml) for 1 h at 37°C. Labeled cells were then washed with HEPES buffer A, distributed to assay tubes (~3 × 10⁵ cells/tube), and allowed to incubate for 15 min at 37°C in the presence of 1.5 mM isobutylmethylxanthine. Cells were then exposed to the same agonists as described above for the assays of PI hydrolysis for 2 min before the addition of 10 µM forskolin for an additional 10 min. Reactions were then stopped with 50 µl of 2.5 M HCl, and the contents of the tubes were added to acid alumina columns to isolate [³H]cAMP, with [¹⁴C]cAMP being used as an internal recovery standard. Samples were eluted from the columns with 2 ml of 0.2 M ammonium formate, and radioactivity was subsequently determined by liquid-scintillation counting.

Data Analysis. Competition binding isotherms were analyzed via nonlinear regression using GraphPad PRISM 3.02 (GraphPad Software, San Diego, CA) to derive estimates of n_H (slope factor) and IC₅₀ values (midpoint location/potency parameter). Assuming simple competition, the data were refitted according to both one- and two-site mass-action binding models, and the better model was determined by an extra-sum-of-squares test using PRISM. IC₅₀ values were converted to K_I values (competitor-receptor dissociation equilibrium constant) according to the following equation (Cheng and Prusoff, 1973):

(1)

where [D] and K_D denote the concentration and dissociation constant, respectively, of the radioligand, the latter determined previously in saturation binding assays (not shown).

(2)

(3)

(4)

(5)

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Equilibrium Binding Assays on Cloned Human M₁ to M₅ mAChRs. The binding properties of each of the novel mAChR compounds were initially evaluated by their ability to compete with the antagonist [³H]NMS at each of the cloned human mAChR subtypes expressed in CHO cell membranes (Fig. 2). In all instances, complete inhibition of specific radioligand binding was attained in a manner consistent with simple competition for a single binding site, with Hill slope factors not significantly different from unity (Table 1). The monomeric NNC 11-1314 and the dimeric NNC 11-1607 showed little selectivity across the five subtypes, whereas the other dimer, NNC 11-1585, showed a marked preference for the M₁ and M₂ subtypes relative to all others (Table 1).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Competition between the binding of [3H]NMS (0.2 nM) and NNC 11-1314 (), NNC 11-1607 (), or NNC 11-1585 () in CHO cell membranes stably expressing the human M1 to M5 mAChRs. Incubation was for 1 h at 37°C. Nonspecific binding was defined using 10 µM atropine. Values represent the mean ± S.E.M. of three to eight experiments conducted in triplicate. Error bars not shown lie within the dimension of the symbol.

Equilibrium Binding Assays on Native (Rat) mAChRs. Subsequent experiments were undertaken to determine the binding properties of the novel compounds in native tissue. In these experiments, rat brain and heart membranes were used, the former known to express a mixture of all five mAChRs (Caulfield, 1993; Flynn et al., 1997), whereas the latter is one of the few tissues that expresses the M₂ mAChR almost exclusively (Caulfield, 1993).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Competition between the binding of [3H]NMS (0.2 nM) and NNC 11-1314 (), NNC 11-1607 (), or NNC 11-1585 () in membranes derived from rat whole brain minus cerebellum (A) or rat heart (B). Values represent the mean ± S.E.M. of six to eight experiments conducted in triplicate. All other details are the same as in Fig. 2.

Functional properties at the M₁, M₃, and M₅ mAChRs. To ascertain the functional properties of the novel compounds at the M₁, M₃, and M₅ mAChRs, assays of mAChR-mediated PI hydrolysis were undertaken. This signaling pathway represents one of the best studied second-messenger systems that is preferentially linked to the aforementioned mAChR subtypes (Caulfield, 1993).

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Agonist-mediated PI hydrolysis in CHO cells stably expressing the human M1 (A), M3 (B), or M5 (C) mAChR. Each panel shows the results from a representative experiment demonstrating the effects of oxotremorine-M (), NNC 11-1314 (), NNC 11-1607 (), or NNC 11-1585 (), all normalized to the maximal response of oxotremorine-M in that experiment. Incubation was for 30 min at 37°C. Each point represents the mean ± S.E.M. of triplicate determinations. Error bars not shown lie within the dimensions of the symbol.

Functional Properties at the M₂ and M₄ mAChRs. In contrast to the M₁, M₃, and M₅ mAChR subtypes, the M₂ and M₄ mAChRs signal preferentially through the cAMP pathway (Caulfield, 1993). Hence, the same series of agonists used in the PI assays described above were investigated for their ability to inhibit forskolin-stimulated cAMP accumulation in CHO cells expressing the human M₂ or M₄ mAChR as an index of agonism at these receptor subtypes.

View larger version (28K): [in this window] [in a new window]   Fig. 5.   Agonist-mediated effects on 10 µM forskolin-stimulated [3H]cAMP accumulation in CHO cells stably expressing the human M2 (A and B) or M4 (C and D) mAChR. A, response to oxotremorine-M at the M2 mAChR, normalized to its own maximum inhibitory effect (2442 ± 438 dpm). B, response to NNC 11-1314 (), NNC 11-1607 (), or NNC 11-1585 () at the M2 mAChR, normalized to the maximum inhibitory effect of oxotremorine-M. C, response to oxotremorine-M at the M4 mAChR, normalized to its own maximum inhibitory effect (2819 ± 549 dpm). D, response to NNC 11-1314 (), NNC 11-1607 (), or NNC 11-1585 () at the M4 mAChR, normalized to the maximum inhibitory effect of oxotremorine-M. In all instances, incubation was for 10 min at 37°C. Each point represents the mean ± S.E.M. of four to seven experiments conducted in triplicate. Error bars not shown lie within the dimensions of the symbol.

View larger version (26K): [in this window] [in a new window]   Fig. 6.   Maximum responses for agonist-mediated PI hydrolysis (M1, M3, and M5 mAChRs) or inhibition of forskolin-stimulated cAMP accumulation (M2 and M4 mAChRs) normalized to the maximal response of oxotremorine-M. Data are from Table 3.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

The present study has demonstrated how the dimerization of the potent M₁ mAChR agonist NNC 11-1314 (Sauerberg et al., 1998b) led to an enhancement of binding affinity for selected mAChR subtypes (NNC 11-1585) or to a profound alteration in the functional mAChR selectivity profile (NNC 11-1607). The differential expression of these properties by the dimers was particularly interesting because they differed only in the positioning of the (identical) substituents on a central benzene ring (Fig. 1).

Using the monomeric NNC 11-1314, we previously demonstrated the significance of the phenylpropargyloxy side chain for M₁ mAChR agonism relative to the hexyloxy side chain found in compounds such as xanomeline (Sauerberg et al., 1998b). However, that study was limited because NNC 11-1314 was not investigated under similar conditions at all mAChR subtypes. We have now confirmed the high affinity and potency of this compound at the human M₁ mAChR and extended our observations to include the determination of its pharmacological properties at all the mAChR subtypes. From radioligand binding conducted on CHO cell membranes, it was apparent that NNC 11-1314 did not demonstrate appreciable binding selectivity across the five mAChRs (Table 1). Dimerization of this compound to yield the 1,4-disubstituted NNC 11-1607 led to an enhancement of binding affinity at all five mAChRs. However, because this enhancement in affinity was relatively uniform across the receptors, no substantial selectivity in binding was gained. The most profound differences in binding were noted for the 1,3-disubstituted NNC 11-1585; this compound demonstrated a greater than 300-fold enhancement in M₁ binding affinity and greater than 700-fold enhancement in M₂ binding affinity relative to the monomeric NNC 11-1314.

Subsequent binding experiments used rat brain and heart homogenates to investigate the binding of the novel compounds on native mAChRs. Rat brain is known to possess a mix of all five mAChRs (Levey et al., 1991; Caulfield, 1993), and it is thus not surprising that biphasic binding was evident for NNC 11-1585 but not for NNC 11-1314 or NNC 11-1607 (Table 2). This finding is consistent with the selectivity profile determined from the CHO cell studies, which showed that NNC 11-1585 could discriminate between M₁ and M₂ mAChRs on the one hand and M₃, M₄, and M₅ mAChRs on the other hand, whereas NNC 11-1314 and NNC 11-1607 could not. In contrast, the binding profile of the agonists in the rat heart membranes cannot be reconciled with the predictions of binding selectivity on the basis of the cell-line studies; heart tissue predominantly expresses the M₂ mAChR (Caulfield, 1993), yet biphasic binding was noted for all three agonists (Table 2, Fig. 3B). A common cause of multiphasic agonist competition binding isotherms at individual receptor subtypes is a dispersion of agonist affinity states between coupled and uncoupled receptor-G protein complexes (Kenakin, 1997; Christopoulos and El-Fakahany, 1999). The occurrence of this phenomenon is highly host-system-dependent, and if operative in the current situation, it may thus reflect profound differences in receptor-G protein stoichiometry or coupling efficiency between CHO cells and rat tissue.

A more direct assessment of agonist activity was obtained in the CHO cells using functional assays of PI hydrolysis for the M₁, M₃, and M₅ mAChRs and inhibition of forskolin-stimulated cAMP accumulation for the M₂ and M₄ mAChRs. These second-messenger assays have long been shown to be reliable indicators of agonist activity at G_q/11-preferring receptors (e.g., M₁, M₃, M₅) and G_i/o-preferring receptors (e.g., M₂, M₄), respectively (Harden et al., 1998). The experimental paradigms used in our study were designed to exploit the fact that oxotremorine-M behaves as a full agonist at each of the mAChR subtypes. Thus, the construction of an oxotremorine-M concentration-response curve in parallel to the curves of the novel agents allowed for an independent estimate to be obtained of maximal system responsiveness for each individual experiment. This was advantageous for a number of reasons. First, differences in stimulus-response coupling between cell passages that could lead to variations in the agonist profile of partial-agonist ligands would be better revealed if a reference full-agonist curve were available for each experiment. Second, an important analytical advantage was gained for the assays of PI hydrolysis because the full-agonist curve in these experiments was used to provide an estimate of maximal system responsiveness and transduction efficiency for the application of the operational model of agonism (Black and Leff, 1983) to the novel agonist data. From this analysis, it was concluded that all three novel compounds are partial agonists relative to oxotremorine-M at the M₁, M₃, and M₅ mAChRs, with highest efficacy being observed at the M₁ mAChR. Although this is also evident from a comparison of relative maximal response ranges for all agonists tested (Table 3; Fig. 6), the actual determination of operational efficacy (log) estimates provides a stronger quantitative ground for further structure-activity studies that transcend variations in stimulus-response coupling. In addition, another advantage of the operational modeling was the determination of functional agonist affinity, in which it was found that the high selectivity of NNC 11-1585 for the M₁ mAChR relative to the M₃ and M₅ mAChRs seen in the binding assays was also retained in the functional assays.

The degree of quantitative rigor used in the analysis of the PI data could not be applied to the same extent for the cAMP assays because the agonists displayed complex, bell-shaped, concentration-response profiles. This was most evident for the full agonist, oxotremorine-M, in which the inhibition of forskolin-stimulated cAMP accumulation was followed by an almost complete reversal (Fig. 5, A and C), especially at the M₄ mAChR. Bell-shaped concentration-response curves have previously been attributed to a number of mechanisms, such as convergence of multiple receptor pathways activated by the same agonist (Leff, 1993; Rovati and Nicosia, 1994), agonist-mediated receptor desensitization (Christopoulos and El-Fakahany, 1999), cooperative agonist binding (Järv, 1993), and promiscuous receptor-G protein coupling (Dittman et al., 1994; Vogel et al., 1995). Irrespective of mechanism, the use of an empirical bell-shaped model to derive agonist potency and maximal response range for inhibiting cAMP accumulation still allowed for useful comparisons to be made between oxotremorine-M and the novel agonists. One important aspect noted was the complete lack of agonistic activity of NNC 11-1607 at the M₂ mAChR (Fig. 5B; Table 3) in contrast to the M₄ mAChR. Taken together with its almost negligible efficacy at the M₃ and M₅ mAChRs, this compound may be deemed functionally selective for M₁ and M₄ mAChRs to the practical exclusion of all others (Fig. 6). A second important feature relates to the monomeric NNC 11-1314, which exerted a maximal effect at the M₄ mAChR that was indistinguishable from that of oxotremorine-M. Thus, despite its robust M₁ efficacy, this compound seems to be an even stronger M₄ mAChR agonist. Indeed, when compared with the dimers, NNC 11-1314 demonstrated the lowest propensity toward functional selectivity (Table 3; Fig. 6). A final intriguing finding in the cAMP assays was observed for NNC 11-1585. In contrast to its effects on cAMP accumulation at the M₂ mAChR (Fig. 5B), this compound could only mediate an inhibition of cAMP accumulation at the M₄ mAChR over all the concentration-ranges investigated (Fig. 5D). This may reflect a markedly different coupling profile for the M₄ mAChR when bound to this agonist and may be a provocative indication of the phenomenon of agonist-directed trafficking of stimulus (Kenakin, 1995). Alternatively, our findings may simply reflect the fact that the stimulatory component would only become evident at concentrations higher than those examined, but this would then imply a very shallow concentration-response profile for this particular agonist relative to all others. Either way, these speculations suggest that NNC 11-1585 may prove useful in further studies of M₄ mAChR coupling promiscuity.

Overall, our findings support the general concept of the "bivalent ligand" approach (Portoghese, 1989) as a means of enhancing ligand activity and/or selectivity at a number of receptor systems; dimerization of NNC 11-1314 led to the identification of subtype selective compounds on the basis of binding (NNC 11-1585) or function (NNC 11-1607). Importantly, the relatively subtle change in the structural configuration between NNC 11-1607 and NNC 11-1585 had a profound effect on the functional selectivity profile between the ligands, with the former being functionally M₁/M₄ mAChR-selective. This suggests that the two dimers can adopt profoundly different conformations with respect to their activating properties at the mAChRs, and point to NNC 11-1607 as a useful candidate for an agonist that prefers the central nervous system, a region rich in the M₁ and M₄ mAChR subtypes.

In conclusion, we have demonstrated that dimerization of the potent mAChR agonist NNC 11-1314 results in ligands with greater binding affinities. Furthermore, the resulting dimers displayed greater selectivity either on the basis of binding (NNC 11-1585) or function (NNC 11-1607). The functionally M₁/M₄ mAChR-selective agonist NNC 11-1607, in particular, may prove to be a useful agent for further investigations into central nervous system disorders involving cholinergic mechanisms, such as Alzheimer's disease and schizophrenia.