Functional Selectivity of Muscarinic Receptor Antagonists for Inhibition of M₃-Mediated Phosphoinositide Responses in Guinea Pig Urinary Bladder and Submandibular Salivary Gland

The cholinergic nervous system is the major pathway by which bladder contraction is initiated in humans (Andersson, 1993). Molecular techniques have identified five muscarinic acetylcholine (mACh) receptor subtypes (m1-m5), whereas pharmacological data can distinguish only four subtypes, denoted as M₁ to M₄ (Caulfield and Birdsall, 1998). Both M₂ and M₃ mACh receptors have been identified in bladder (detrusor) smooth muscle at the level of mRNA (Yamaguchi et al., 1996) and protein, with quantitative immunoprecipitation demonstrating a 75 to 90% predominance of the M₂ subtype in all species studied (Wang et al., 1995). Roles for the majority M₂ receptor population in the contraction of various smooth muscle types (including bladder detrusor) have been proposed, based upon observations in tissues treated with N-2-chloroethyl-4-piperidinyl diphenylacetate selectively to inactivate the M₃ receptor population. Under these conditions M₂ receptor activation may potentiate M₃-stimulated contraction (via activation of nonselective cation currents or inhibition of Ca²⁺-dependent K⁺ efflux), in addition to reversing the relaxation mediated by agents that increase cytosolic cyclic AMP (Ehlert, 2003). However, pharmacological characterization of the mACh receptors mediating contraction in native detrusor smooth muscle indicate the predominant involvement of the minority M₃ receptor population in a variety of species including rat (Longhurst et al., 1995), guinea pig (Wang et al., 1995), and human (Chess-Williams et al., 2001). Compounds with high affinity for the M₃ receptor have therefore been used in the management of overactive bladder (OAB) (Wallis and Napier, 1999). Indeed, the major current therapies for OAB, oxybutynin and tolterodine, both display high-affinity antagonism at the M₃ receptor and potently inhibit mACh receptor agonist-induced urinary bladder contractions in vitro and in vivo (Nilvebrant et al., 1997; Gillberg et al., 1998). However, the clinical utility of mACh receptor antagonists has been limited by lack of selectivity, which leads to classical antimuscarinic side-effects, such as dry mouth, tachycardia, and blurred vision. This has led to the search for mACh receptor antagonists with greater selectivity for the M₃ receptor and, in particular, for mACh receptor agonist-mediated bladder contraction.

The mACh receptor antagonist darifenacin displays both high affinity (pK_i = 9.12) and selectivity for the human M₃ receptor (Wallis and Napier, 1999). Intriguingly, darifenacin has been reported to inhibit urinary bladder contraction at concentrations lower than those required to influence salivary secretion in vivo in both rat (Wallis and Napier, 1999) and dog (Gupta et al., 2002), despite both responses being predominantly or exclusively M₃-mediated. In contrast, atropine inhibited both responses equipotently, whereas oxybutynin has been shown to be either nonselective (Newgreen and Naylor, 1996; Ikeda et al., 2002) or slightly selective for the salivation response (Nilvebrant et al., 1997).

Similar tissue-specific selectivity had earlier been reported between different smooth muscle types for other muscarinic receptor antagonists, including zamifenacin (Watson et al., 1995) and p-fluorohexahydrosiladifenidol (Eglen et al., 1990). However, tissue-dependent selectivity may not only be a property of subtype-selective antagonists. Thus, tolterodine, a compound widely used in the clinical management of OAB, has been reported to display a greater functional affinity for urinary bladder than for salivary gland, despite failing to exhibit selectivity between mACh receptor subtypes (Nilvebrant et al., 1997). It has been proposed that such “functional selectivity” could lead to a superior clinical side effect profile in the management of OAB and may, at least in part, explain why tolterodine is equally as effective as oxybutynin for improving the symptoms of OAB, yet is better tolerated by patients (Malone-Lee et al., 2001).

The aim of this study was to establish whether a range of subtype-selective and nonselective mACh receptor antagonists display in vitro selectivity for M₃-mediated responses between urinary bladder and salivary gland, within the same species, at the level of second messenger generation. Darifenacin, along with the current OAB therapies oxybutynin and tolterodine, and the classical nonselective muscarinic antagonist atropine were investigated for their ability to inhibit mACh receptor agonist-mediated phosphoinositide turnover in guinea pig urinary bladder and submandibular salivary gland.

Materials and Methods

Materials. Radiolabeled compounds were obtained from Amer-sham Biosciences UK, Ltd. (Little Chalfont, Buckinghamshire, UK), with the exception of [³H]Ins(1,4,5)P₃, which was obtained from PerkinElmer Life and Analytical Sciences (Zaventem, Belgium). Darifenacin and tolterodine were synthesized in the laboratories of Pfizer Global Research and Development (Sandwich, UK). MT-7 was purchased from Peptides International Inc. (Louisville, KY). All other reagents were purchased from Sigma Chemical (Poole, Dorset, UK) or Fisher Scientific (Loughborough, UK).

Cell Culture. Chinese hamster ovary lines stably expressing cloned human m2 or m3 receptors (CHO-m2 and CHO-m3, respectively) were grown in minimum essential medium-α supplemented with 10% newborn calf serum, 100 IU ml^-1 penicillin, 100 μg ml^-1 streptomycin, and 2.5 μg ml^-1 amphotericin B. Cells were maintained at 37°C in a humidified atmosphere of O₂/CO₂ (19:1) and were routinely split 1:5 every 3 to 4 days, using trypsin-EDTA.

Animals and Tissue Slice Preparation. Male Dunkin Hartley guinea pigs (David Hall Guinea Pigs, Burton-on-Trent, UK), 300 to 500 g in weight, were euthanized by injection of 1 g kg^-1 pentobarbitone. For each experiment, urinary bladder and submandibular glands were dissected from four animals. The tissue was cleared of connective tissue and immediately transferred into ice-cold Krebs-Henseleit buffer (KHB; composition: 118 mM NaCl, 4.7 mM KCl, 1.3 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 25 mM NaHCO₃, 5 mM HEPES, 10 mM glucose, pH 7.4) and kept on ice before chopping. Each tissue was cross-chopped (300 μm × 300 μm) with a McIlwain tissue chopper. The resulting slices were transferred to 20 ml of KHB, gassed with O₂/CO₂ (19:1) at 37°C. Slices were shaken vigorously in a water bath at 37°C with three to four changes of KHB over a 30-min period.

Membrane Preparation from Guinea Pig Tissues. Fresh or frozen tissue samples were finely chopped with scissors in 20 mM HEPES, 1 mM EDTA (pH 7.4) buffer at 4°C and any excess fat was decanted off. Tissue was then homogenized using a Polytron homogenizer (14,000 rpm, 20 s). Homogenates were centrifuged at 5000g (10 min, 4°C) and the supernatant fraction was pooled following passage through Miracloth (Calbiochem, San Diego, CA). The filtrate was then centrifuged at 40,000g (20 min, 4°C). The resulting pellet was re-homogenized in buffer before a second high-speed spin as before. This final pellet was resuspended in 20 mM HEPES, 0.1 mM EDTA, pH7.4 buffer, and aliquots were frozen down at -80°C at a concentration of 2 to 3 mg protein ml^-1.

CHO Cell Membrane Preparation. Confluent monolayers of CHO cells were washed with Hanks' balanced salt solution-EDTA (10 mM HEPES, 0.9% NaCl, 0.2% EDTA, pH 7.4) and cells were lifted from the flask by the addition of HEPES-buffered saline-EDTA for approximately 15 min. A cell pellet was recovered by centrifugation at 1700 rpm for 5 min. The cell pellet was homogenized, and a membrane fraction was recovered and stored exactly as described for the tissue membrane preparations above.

Radioligand Binding to Membranes. Saturation binding was performed using a range of concentrations of [³H]NMS (0.03-6.0 nM; 81 Ci mmol^-1) in the absence and presence of atropine (10 μM) to define nonspecific binding. Binding assays were performed in a final volume of 400 μl (assay buffer: 20 mM HEPES, pH 7.4) containing 25 to 100 μg of membrane protein and were incubated for 120 min at 25°C. Bound radioligand was separated from free by rapid vacuum filtration through Whatman GF/B filters on a 24-well Brandel cell harvester (Brandel Inc., Gaithersburg, MD), and radioactivity was quantified by liquid scintillation counting. Competition binding experiments were performed using a single concentration of [³H]NMS (0.3-0.5 nM) in the absence and presence of a range of antagonist concentrations. Bound radioligand was separated from free by rapid vacuum filtration through Whatman GF/B Unifilter Plates (Whatman, Clifton, NJ) on a 96-well Unifilter Harvester (PerkinElmer Life and Analytical Sciences, Boston, MA). All other conditions were as stated above for saturation binding assays.

Radioligand Binding to Intact Cells. Intact cell binding assays were performed on cell monolayers on 24-well plates, in KHB. The medium was aspirated from each well and the cells washed three times with KHB at 37°C. Where applicable, antagonists were preincubated for 30 min before the addition of [³H]NMS, in a total assay volume of 500 μl. After 60-min incubation at 37°C, cells were washed three times with warm, gassed KHB before solubilizing cells with NaOH (0.1 M) and subsequent liquid scintillation counting.

[³H]Inositol Phosphate Accumulation in Bladder Slices. After the 30-min washing period, bladder slices were transferred to a 50-cm² culture flask containing minimum essential medium-α (10 ml), supplemented with 100 IU^-1 penicillin and 100 μg ml^-1 streptomycin. myo-[³H]Inositol (25 μCi; specific activity 88 Ci mmol^-1) was added to the flask and incubated for 24 h at 37°C in CO₂/air (1:19). Following the labeling period, slices were transferred to KHB at 37°C and incubated at 37°C with vigorous shaking for 20 min. During this period, the buffer was changed three to four times to ensure complete removal of medium. Gravity-packed slices (50 μl) were transferred to insert vials containing KHB. Vials were then gassed with O₂/CO₂ (19:1), capped, and incubated at 37°C. At this point, antagonists were added where appropriate. Lithium chloride (LiCl) (5 mM final concentration) was then added to each vial. Incubations were initiated, 15 min later, by the addition of carbachol (CCh), to give a final assay volume of 300 μl. Incubations were terminated after 30 min by the addition of ice-cold 1 M trichloroacetic acid (TCA; 300 μl). Samples were left on ice to extract for 20 min. At the end of this period, vials were centrifuged for 15 min at 5000 rpm. Five hundred-microliter samples of the resulting supernatants were transferred to microcentrifuge tubes containing 10 mM EDTA (100 μl), and the [³H]IP_x fraction was separated (Batty et al., 1997).

Ins(1,4,5)P₃ Mass Determination in Cross-Chopped Submandibular Gland. Gravity-packed submandibular gland slices (50 μl) were transferred to insert vials containing KHB. Vials were gassed with O₂/CO₂ (19:1), capped, and incubated at 37°C. At this point, antagonists were added where appropriate. Incubations were initiated, 15 min later, by the addition of CCh, to make a final assay volume of 300 μl. Incubations were terminated after 5 min by the addition of ice-cold 1 M TCA (300 μl). Samples were left on ice to extract for 20 min. At the end of this period, vials were centrifuged for 15 min at 5000 rpm. A sample (500 μl) of the resulting supernatant was transferred to a microcentrifuge tube containing 10 mM EDTA (100 μl). Samples were neutralized using 1,1,2-trichlorotrifluoroethane/tri-n-octylamine (1:1 v/v; 500 μl) and assayed for Ins(1,4,5)P₃ mass content as described previously (Batty et al., 1997). Protein concentrations were determined according to the method of Lowry et al. (1951) and data were corrected for these values.

Cyclic AMP Accumulation in CHO Cells. After antagonist additions for 20 min, CHO-m2 cell monolayers were incubated with methacholine (MCh; 1 μM) for 10 min before addition of forskolin (10 μM). After further incubation for 10 min, assays were stopped by aspiration and addition of ice-cold 0.5 M TCA (400 μl). Samples were neutralized as described above and cyclic AMP was determined using the method of Brown et al. (1971).

[³H]Inositol Phosphate Accumulation in CHO Cells. CHO-m3 cell monolayers were labeled with myo-[³H]inositol for 24 h before each experiment. After a 30-min preincubation with antagonist, cells were incubated for 15 min with LiCl (10 mM) before addition of MCh (3 μM). After further incubation for 10 min, assays were stopped by aspiration and addition of ice-cold 0.5 M TCA (400 μl). Samples were neutralized and the [³H]IP_x fraction was separated (Batty et al., 1997).

Data and Statistical Analysis. Data are shown as mean ± standard error of the mean (S.E.M.) for the indicated number of experiments. Saturation binding data were analyzed by nonlinear regression using GraphPad Prism 3.0 (GraphPad Software Inc., San Diego, CA). The equilibrium binding dissociation constant (K_D) and the maximum number of binding sites (B_max) were derived from the Langmuir equation RL = R_tL/(K_D+ L), where L is the concentration of free ligand, RL is the concentration of receptor-bound ligand at equilibrium, and R_t is the total receptor concentration. Antagonist displacement curves were fitted to the “four-parameter logistic equation” using GraphPad Prism 3.0, and the best fit between a variable Hill coefficient and a Hill coefficient fixed to unity was determined using an F test. IC₅₀ values generated by these inhibition curves were corrected to give binding constant (K_i) values for each test compound, using the Cheng-Prusoff equation (Craig, 1993). Sigmoidal concentration-response curves for carbachol and methacholine, as well as all antagonist inhibition curves, were fitted to the four-parameter logistic equation using GraphPad Prism 3.0. IC₅₀ values generated by these inhibition curves were corrected to give binding constant (K_B) values for each test compound, using the functional equivalent of the Cheng-Prusoff equation (Craig, 1993; Lazareno and Birdsall, 1993): K_B = IC₅₀/[1 + (A/EC₅₀)], where A = [agonist] and EC₅₀ = agonist EC₅₀ (derived from agonist concentration-response curve generated each day as part of the same experiment). The statistical significance of differences between data were determined using Student's t test, with K_i and K_B values first being converted to the respective normally distributed negative logarithm (pK_i or pK_B).

Results

Radioligand Binding in Intact CHO Cell Monolayers. The relative selectivity for M₂ and M₃ mACh receptor subtypes of atropine, darifenacin, oxybutynin, and tolterodine was first determined at the level of receptor binding affinity, in CHO-m2 (M₂B_max ≈1 pmol mg^-1 protein) or CHO-m3 (M₃B_max ≈1.5 pmol mg^-1 protein) cells. Saturation binding analysis yielded affinity binding constant (K_D) values for [³H]NMS at CHO-m2 and CHO-m3 cells of 0.24 ± 0.07 and 0.37 ± 0.07 nM, respectively (data not shown). The affinity (pK_i) and associated Hill slopes for atropine, darifenacin, oxybutynin, and tolterodine for the M₂ and M₃ receptor subtypes are summarized in Table 1. Darifenacin displayed marked selectivity (14.8-fold; p < 0.01) for the M₃ over the M₂ mACh receptor. In contrast, oxybutynin was nonselective, whereas both atropine (5.1-fold; p < 0.05) and tolterodine (6.6-fold; p < 0.05) displayed modest, though significant M₂ selectivity (Table 1).

Functional Affinity Estimates in CHO Cells.Figure 1a illustrates the protocol adopted to examine the “functional” affinity of mACh receptor antagonists at the M₂ receptor, showing a representative curve for the inhibition of forskolin-stimulated cAMP accumulation by MCh in CHO-m2 cells (EC₅₀ = 0.27 ± 0.03 μM, n = 7). This allowed the estimation of an approximate EC₇₀ concentration (1 μM) (Fig. 1a) from which antagonist inhibition curves were constructed (Fig. 1b). pK_B values were derived from these curves as described earlier (see Materials and Methods) and are summarized in Table 2.

Figure 2a shows a concentration-response curve for MCh-stimulated accumulation of [³H]IP_x (in the presence of Li⁺)in CHO-m3 cells (EC₅₀, 1.26 ± 0.14 μM, n = 6). An approximate EC₇₀ concentration (3 μM) was estimated from this curve and used to construct antagonist inhibition curves (Fig. 2b). Affinity estimates were derived as described above and are summarized as pK_b values in Table 2.

Comparing the functional affinity estimates at M₂ and M₃ mACh receptors, darifenacin was again found to be the most M₃-selective antagonist (32.4-fold; p < 0.01), followed by oxybutynin, which was modestly, but significantly M₃-selective (3.4-fold, p < 0.05). Atropine (2.1-fold) and tolterodine (2.8-fold) were not significantly subtype-selective in functional assays (Table 2).

Characterization of Responses in Guinea Pig Tissue Slices. In cross-chopped, myo-[³H]inositol-loaded guinea pig bladder, CCh caused a time- and concentration-dependent accumulation of [³H]IP_x in the presence of LiCl (5 mM). Figure 3a illustrates the time course of the response to CCh (300 μM), which was linear over a 30-min time course. The latter time point was chosen for all subsequent experiments. Figure 3b shows a concentration-response curve for the accumulation of [³H]IP_x; analysis of each curve yielded a mean EC₅₀ value of 8.9 ± 1.8 μM (n = 9).

Figure 4a illustrates the time course of the Ins(1,4,5)P₃ response to CCh (300 μM) in cross-chopped guinea pig submandibular gland. The response increased linearly up to 60 s, after which a plateau was reached and maintained throughout the 10-min time course investigated (Fig. 4a). Figure 4b shows the concentration-dependent nature of Ins(1,4,5)P₃ accumulation in response to 5 min of CCh stimulation. The EC₅₀ value calculated for this response was 25.3 ± 4.8 μM (n = 9). Since differences in agonist potency are accounted for within the functional equivalent of the Cheng-Prusoff relationship, comparison of pharmacological values derived from the two tissue responses is valid, despite the slight differences in carbachol potency.

Functional Affinities of mACh Receptor Antagonists for Inhibition of Agonist-Stimulated Phosphoinositide Turnover in Guinea Pig Tissues.Figure 5 shows inhibition curves for each of the test compounds, expressed as a percentage of the maximal response (i.e., that evoked by 50 μM CCh in the absence of antagonist) in both cross-chopped guinea pig bladder and submandibular gland. Oxybutynin (9.3-fold; p < 0.01), darifenacin (7.9-fold; p < 0.05), and tolterodine (7.4-fold; p < 0.05) each displayed significant selectivity for inhibition of the functional response in the bladder over that seen in the submandibular gland. In contrast, atropine displayed similar affinities for inhibition of phosphoinositide responses seen in each tissue (Fig. 5a; Table 2). This provides further support for the validity of the comparison of pharmacological values derived from the two different tissue responses.

Determination of mACh Receptor Subtypes Present in Guinea Pig Bladder and Submandibular Gland. Saturation binding analysis with [³H]NMS established the total mACh receptor populations in bladder and submandibular gland as 335 ± 23 and 120 ± 22 fmol mg^-1 protein, respectively (data not shown).

To determine the mACh receptor subtypes comprising these populations, a range of subtype-selective mACh receptor antagonists was utilized in competition radioligand binding assays against [³H]NMS. The compounds tested were as follows: atropine (nonselective), pirenzepine (M₁-selective), 4-diphenylacetoxy-N-methylpiperidine (M₁/M₃-selective), methoctramine (M₂-selective), darifenacin (M₃-selective), and MT-7 (a highly M₁-selective toxin; see Adem and Karlsson, 1997). Figure 6 illustrates the resulting inhibition curves generated for the six compounds against membranes prepared from guinea pig bladder (Fig. 6a) or submandibular gland (Fig. 6b). Affinity binding constant (pK_i) values were derived from these curves and are summarized, along with the corresponding slope factors, in Table 3.

The binding profile in membranes prepared from bladder tissue was consistent with a mixed M₂/M₃ mACh receptor population. Thus, the displacement curve for the M₂-selective antagonist was best fitted by two-site analysis. The high-affinity site, accounting for approximately 84% of the total mACh receptor population, was consistent with the reported affinity of methoctramine at the M₂ receptor (Eglen et al., 1996), whereas its affinity at the remaining 16% of bladder mACh receptors was consistent with the reported affinity of methoctramine at the M₃ receptor (Eglen et al., 1996). The low affinity of pirenzepine and the absence of any displacement by the highly M₁-selective toxin MT-7 was inconsistent with a detectable M₁ mACh receptor population in the guinea pig bladder (Table 3).

In submandibular gland membranes, pirenzepine displayed low affinity and MT-7 caused no significant displacement of [³H]NMS binding (Fig. 6b; Table 3). Affinities of the other compounds were consistent with a homogeneous population of M₃ mACh receptors; thus, these data indicate that no significant population of M₁ mACh receptors is expressed in guinea pig submandibular gland. Consistent with the radioligand binding data, MT-7 (100 nM) had no significant effect on the Ins(1,4,5)P₃ response to CCh (1 mM) in cross-chopped submandibular gland (data not shown).

To confirm that MT-7 does displace [³H]NMS binding at the (guinea pig) M₁ receptor, competition binding assays were performed using membranes derived from CHO-m1 cells and membranes prepared from guinea pig cerebral cortex. Figure 7 shows that MT-7 binds with high affinity to both human and guinea pig M₁ mACh receptors, displacing the specific [³H]NMS binding component by 94 ± 1% and 46 ± 3% in CHO-m1 and guinea pig cortex membranes, respectively. MT-7 displayed high affinity for the human M₁ mACh receptor expressed in CHO cells (pK_i, 9.48 ± 0.01; n = 3) and a similar affinity for the 46% of mACh receptors that bound MT-7 with high affinity in guinea pig cortex (pK_i, 9.86 ± 0.04; n = 3).

Discussion

Muscarinic receptor antagonists are a mainstay in the pharmacological management of OAB (Wyndaele, 2001). However, a lack of receptor subtype selectivity can result in a significant side effect profile (e.g., dry mouth), which is associated with a high degree of patient noncompliance (Wallis and Napier, 1999). This has prompted a search for anti-muscarinic compounds with greater selectivity. The observation that some, but not all, mACh receptor antagonists display in vivo functional selectivity for bladder versus salivary gland (see the Introduction) represents a significant step toward more selective OAB therapies.

The main aim of this study was to investigate the potential for mACh receptor antagonists to exhibit tissue-dependent functional selectivity at the level of second messenger generation. Initially, the binding and functional affinities of the four compounds chosen for investigation were determined in CHO cells stably expressing either the human M₂ or M₃ mACh receptor. Darifenacin was shown to be M₃-selective, at the level of both receptor binding (15-fold) and inhibition of agonist-induced second messenger responses (32-fold), in agreement with earlier studies (Eglen et al., 1996; Wallis and Napier, 1999). Oxybutynin was found to be nonselective in competition radioligand binding assays and only weakly (3- to 4-fold) M₃-selective in functional assays. This is in contrast to previous reports, which found oxybutynin to be approximately 10-fold M₃- over M₂-selective in radioligand binding studies in membrane homogenates (Nilvebrant et al., 1997). Atropine and tolterodine were weakly M₂-selective in both binding and functional assays. This differs somewhat from other work that reports these compounds to be nonselective (Nilvebrant et al., 1997; Caulfield and Birdsall, 1998). These small discrepancies may be explained by differences in binding affinity and selectivity of mACh receptor antagonists between membrane preparations, which are widely used to characterize ligand binding profiles, and intact cell assays that were used in this study (Nelson et al., 2002).

Using indices of mACh receptor-stimulated phospholipase C activity, we have been able to show that darifenacin, oxybutynin, and tolterodine each display higher affinity for the inhibition of responses in urinary bladder slices compared with submandibular gland slices. Data presented here therefore support the notion that certain mACh receptor antagonists show differential intertissue effects on cellular responses reputedly mediated by the same mACh receptor subtype in the same species. We have demonstrated that, in the case of darifenacin, oxybutynin, and tolterodine, this tissue-specific selectivity is manifested in vitro at the level of inhibition of agonist-induced second messenger turnover, in line with earlier reports of the in vivo selectivity of these compounds (Nilvebrant et al., 1997; Wallis and Napier, 1999; Gupta et al., 2002).

Our finding that oxybutynin displays selectivity (9.3-fold) for bladder versus salivary gland is at odds with previous reports in the literature. Newgreen and Naylor (1996) found that oxybutynin was nonselective in vitro in the guinea pig (by comparing inhibition of bladder contractions and ⁸⁶Rb efflux from submandibular glands). Ikeda et al. (2002) also found that oxybutynin inhibited bladder contraction and salivation with similar potency in vivo, in anesthetized rat. In contrast, Nilvebrant et al. (1997) found oxybutynin to exhibit selectivity for inhibition of salivation over bladder contraction in anesthetized cat. Oxybutynin is subject to metabolism in vivo, producing both active and inactive metabolites (Waldeck et al., 1997), and has also been reported to have significant “nonmuscarinic” effects, such as blockade of Ca²⁺ channels (Wada et al., 1995). Thus, the selectivity we observe in vitro may be reduced or lost in vivo. In contrast, the bladder selectivity demonstrated by darifenacin and tolterodine is in agreement with previous studies. Thus, Gupta et al. (2002) reported that darifenacin displays a 10-fold selectivity for inhibition of pelvic nerve-stimulated bladder contractions versus salivary secretion in dog, whereas tolterodine was 4- to 5-fold selective in the same study (Gupta et al., 2002). It should also be emphasized that in the vast majority of studies, including this one, atropine fails to discriminate bladder and salivary gland responses. This cogently argues against an affinity “frame-shift” between preparations and strongly supports the conclusion that observed tissue selectivities are particular to a subset of mACh receptor antagonists.

Although it is generally accepted that acetylcholine-induced phosphoinositide hydrolysis and contraction are mediated almost exclusively by the M₃ receptor in bladder (Harriss et al., 1995; Chess-Williams et al., 2001), some controversy remains over the potential involvement of “non-M₃” mACh receptors in cholinergic responses of salivary glands (Laniyonu et al., 1990; Watson et al., 1996). We therefore sought clarification of whether the presence of non-M₃ (in particular, M₁) muscarinic receptors could be involved in the phosphoinositide response in submandibular gland, using competition radioligand binding assays with a range of the most selective muscarinic ligands available. It was found that whereas the binding profile was consistent with a mixed M₂ and M₃ receptor population in bladder, affinities derived for submandibular gland membranes were indicative of a homogeneous population of M₃ receptors. The striking absence of any significant displacement of [³H]NMS binding by the highly M₁-selective toxin MT-7 indicates that the guinea pig submandibular gland does not express a significant population of M₁ receptor. This was confirmed by the lack of effect of MT-7 (at 100 nM) on carbachol-stimulated phosphoinositide responses in submandibular gland slices (data not shown). In contrast, MT-7 did cause significant inhibition of specific [³H]NMS binding to membranes prepared from CHO-m1 cells and guinea pig cortex, confirming that this toxin binds to both human and guinea pig M₁ mACh receptors with high affinity.

Although our findings are in agreement with some other studies (Laniyonu et al., 1990; Moriya et al., 1999) it is likely that the makeup of the mACh receptor population in submandibular gland varies from species to species. Thus, a recent study, using methods and range of muscarinic antagonists similar to those used here, reported a significant M₁ mACh receptor subpopulation in canine submandibular gland (Clarke et al., 2003). It should also be noted that controversy regarding non-M₃ mACh receptor expression in submandibular gland is not limited to the potential involvement of the M₁ subtype, and there is some evidence for the presence of M₅ mACh receptors (e.g., Meloy et al., 2001). Recent studies of transgenic mice lacking the individual mACh receptor subtypes implicate the M₅ receptor in slow salivary secretion, since mice lacking the M₅ receptor gene showed reduced salivation in response to pilocarpine only at a longer time (15-60 min) after injection (Yeomans et al., 2001). In contrast, Bymaster et al. (2003) reported that salivation measured at 15 min postinjection was not altered in M₅ knockout mice. Additionally, we found that the binding affinities of a range of the most M₃/M₅-selective agents available (darifenacin, p-fluorohexahydrosiladifenidol, and AF DX 384) in submandibular gland membranes did not differ significantly from affinity estimates in CHO-m3 membranes (data not shown). Our data are therefore consistent with the expression of a homogeneous M₃ mACh receptor population in the guinea pig submandibular gland.

Thus, if agonist-stimulated phosphoinositide turnover in both bladder and submandibular gland is mediated via activation of M₃ mACh receptors, the question remains as to why some, but not all mACh receptor antagonists inhibit these responses with different affinities? In the absence of any evidence for genetic heterogeneity of the M₃ receptor, the influence of the cellular environment on M₃ receptor pharmacology must be considered. In this context it could be speculated that GPCRs express “phenotypic” profiles specific to the host cell environment, such that the same gene can exhibit distinct pharmacologies when expressed in different cells (Kenakin, 2003).

Interactions with other integral membrane proteins expressed by the cell can influence receptor phenotype, as demonstrated by the influence of receptor activity-modifying proteins upon GPCR pharmacology. Coexpression of receptor activity-modifying protein 3 with the calcitonin receptor has been shown to influence not only agonist potency ranking orders, but also antagonist affinity estimates (Armour et al., 1999). Numerous recent studies have also highlighted the potential for GPCRs to form dimer or higher order oligomer assemblies (see Angers et al., 2002). Such receptor-receptor interactions, especially between nonidentical monomers (i.e., heterodimerization), can generate novel properties in the resulting complex (e.g., Jordan and Devi, 1999). Evidence has been provided for dimerization of mACh receptor subtypes (Wreggett and Wells, 1995), and it is therefore tempting to speculate that heterodimerization can occur between M₂ and M₃ mACh receptors to influence the pharmacology in a tissue-dependent manner.

In conclusion, the main finding of this study is that tolterodine, darifenacin, and oxybutynin each display selectivity for the inhibition of phosphoinositide turnover in guinea pig urinary bladder versus submandibular salivary gland in the same species. In contrast, atropine is nonselective between these two tissues. We have also shown that the lower functional affinities of tolterodine, darifenacin and oxybutynin in the submandibular gland are unlikely to be due to the involvement of an M₁ mACh receptor population in the functional response of this tissue. Our data therefore support the notion that certain mACh receptor antagonists are able to display a functional selectivity for the inhibition of responses in bladder versus salivary gland. The molecular mechanism(s) underlying this phenomenon remains to be resolved.