Abstract
4-n-Butyl-1-[4-(2-methylphenyl)-4-oxo-1-butyl]-piperidine hydrogen chloride (AC-42) is a selective agonist of the muscarinic M1 receptor previously suggested to interact with an “ectopic” site on this receptor. However, the pharmacological properties of this site (i.e., whether it overlaps to any extent with the classic orthosteric site or represents a novel allosteric site) remain undetermined. In the present study, atropine or pirenzepine significantly inhibited the ability of either carbachol or AC-42 to stimulate inositol phosphate accumulation or intracellular calcium mobilization in Chinese hamster ovary (CHO) cells stably expressing the human M1 receptor. However, the interaction between either of these antagonists and AC-42 was characterized by Schild slopes significantly less than unity. Increasing the concentrations of atropine revealed that the Schild regression was curvilinear, consistent with a negative allosteric interaction. More direct evidence for an allosteric mode of action of AC-42 was obtained in [3H]N-methylscopolamine ([3H]NMS) binding studies, in that both AC-42 and the prototypical modulator gallamine failed to fully inhibit specific [3H]NMS binding in a manner that was quantitatively described by an allosteric model applied to both modulator data sets. Furthermore, AC-42 and gallamine significantly retarded the rate of [3H]NMS dissociation from CHO-hM1 cell membranes, conclusively demonstrating their ability to bind to a topographically distinct site to change M1 receptor conformation. These data provide the first direct evidence that AC-42 is an allosteric agonist that activates M1 receptors in the absence of the orthosteric agonist.
Muscarinic acetylcholine receptors (mAChRs) are members of the rhodopsin-like G protein-coupled receptor superfamily (Kristiansen, 2004). There are five known mAChR subtypes (M1–M5). M1, M3, and M5 subtypes couple predominantly through the Gq/11 pathway to stimulate phospholipase C to catalyze the breakdown of phosphatidylinositol bisphosphate to inositol triphosphate and diacylglycerol and the subsequent mobilization of intracellular calcium (Caulfield and Birdsall, 1998). The M2 and M4 subtypes couple predominantly through the Gi/o family of G proteins to inhibit adenylyl cyclase and decrease the intracellular level of cAMP (Caulfield and Birdsall, 1998).
There is a wide array of pharmacological tools with which to study mAChRs. N-Methyl scopolamine (NMS), quinuclidinyl benzilate, pirenzepine, and darifenacin, among mAChR antagonists, and acetylcholine and oxotremorine-M, among mAChR agonists, have been radiolabeled and used extensively to characterize mAChRs. Unfortunately, most of these pharmacological tools show low selectivity between mAChR subtypes (Caulfield and Birdsall, 1998; Ellis, 2002). This reflects the high level of conservation among the residues forming the orthosteric binding site of mAChRs, which makes the design of subtype-selective agonists and antagonists extremely difficult. An alternative route to engendering subtype selectivity among ligands targeted to the mAChRs is through search for allosteric modulators, which act at a site that is topographically distinct from the orthosteric binding site.
Allosteric modulators of mAChRs represent some of the best-characterized examples of allosterism at G protein-coupled receptors. Ligands such as gallamine, alcuronium, heptane-1,7-bis-(dimethyl-3′-phthalimidopropyl)-ammonium bromide (C7/3-phth), brucine, and closely related analogs have all been shown to have varying allosteric effects on mAChRs depending on the interacting orthosteric ligand. The ability of the same allosteric modulator to display different types of cooperative interactions (positive, negative, or neutral) depending on the nature of the ligand occupying the orthosteric site is characteristic of allosteric interactions, a phenomenon previously referred to as “probe dependence” (Christopoulos and Kenakin, 2002).
The best-studied examples of mAChR allosteric modulators, as described above, are thought to interact at a site near, but extracellular to, the orthosteric binding site (Gnagey et al., 1999). Site-directed mutagenesis studies suggest that the interaction with residues at the top of TM7 and in extracellular loops 2 and 3 is responsible for the different mAChR selectivities exhibited by these ligands (Matsui et al., 1995; Gnagey et al., 1999). Spalding et al. (2002) have described a novel compound, AC-42 (Fig. 1), which activated the M1 receptor with an intrinsic activity of approximately 0.7 compared with the traditional orthosteric agonist carbachol but had no significant agonist activity at the M2–M5 mAChRs despite binding to all subtypes with similar affinity (Spalding et al., 2002). Chimeric receptor studies indicated that the N terminus-TM1 and extracellular loop 3-TM7 regions of the M1 mAChR were required to play a role in the selective activation of the receptor by AC-42 (Spalding et al., 2002); regions that are not consistent with AC-42 binding to the orthosteric site of the M1 mAChR deeper in the TM bundle. In an elegant conclusion to their study, Spalding et al. (2002) demonstrated that the potency of AC-42 was not significantly attenuated by either of the point mutations Y381A or N382A in TM6; residues that are believed to form part of the orthosteric site in that their mutation abolishes the agonist activity of carbachol (Ward et al., 1999; Spalding et al., 2002). These findings prompted the authors to categorize AC-42 as an “ectopic” agonist, in the knowledge that its binding site is partially or totally distinct from that for carbachol, but in the absence of definitive pharmacological data proving that it is an allosteric ligand.
To date, the molecular relationship among the binding site used by AC-42, the orthosteric site used by traditional mAChR ligands, and the allosteric site used by prototypical allosteric modulators remains undetermined. If the ectopic site is actually a completely distinct allosteric site, then this would have profound implications for drug discovery, because it would represent the most striking example of a selective allosteric agonist identified to date for any G protein-coupled receptor. In this study, we have used a combination of functional calcium mobilization, inositol phosphate (IP) accumulation, and radioligand inhibition and dissociation kinetic binding assays to characterize the molecular nature of the interaction between AC-42 and the M1 mAChR. We provide, for the first time, direct evidence that AC-42 is an allosteric agonist of this receptor.
Materials and Methods
Materials. All cell culture reagents were from Invitrogen (Paisley, UK). myo-[3H]Inositol and [3H]NMS were from GE Healthcare (Little Chalfont, Buckinghamshire, UK). AC-42 was synthesized in house. All other chemicals were purchased from Sigma Chemical (Poole, Dorset, UK) or BDH (Poole, Dorset, UK).
Cell Culture. CHO-hM1 and CHO-hM3 cells were routinely cultured in α-minimal essential medium with ribonucleosides plus 10% fetal bovine serum. Cells were kept in a humidified atmosphere at 37°C in 5% CO2 and passaged twice weekly by scraping in media, centrifugation (200g; 5 min), and resuspension of the pellet in fresh media.
CHO-hM1 Calcium Mobilization Assays. Intracellular calcium was monitored using the fluorescent wash-free dye Calcium Plus (Molecular Devices, Reading, UK) in a fluorometric imaging plate reader (FLIPR; Molecular Devices). CHO-hM1 cells were seeded into clear-bottomed black-walled 96-well microtiter plates (Costar UK Ltd., High Wycombe, Buckinghamshire, UK) 24 h before use at a density of 50,000 cells/well. Before assay on a FLIPR, medium was removed, and cells were incubated with Calcium Plus according to manufacturer's instructions for 60 min at 37°C in Tyrode's buffer (145 mM NaCl, 2.5 mM KCl, 1.2 mM MgCl2, 10 mM Na-HEPES, 10 mM glucose, and 1.5 mM CaCl2, pH 7.4) containing 2.5 mM probenecid. For agonist studies, compounds were tested for activity in a FLIPR assay by adding 50 μl of test solution into a plate volume of 50 μl. For antagonist studies, cells were preincubated with antagonist or buffer, for 30 min before agonist addition on the FLIPR (50 μl of test solution into a plate volume of 100 μl). Fluorescence was monitored (λex = 488 nm, λEM = 540 nm) before and after the addition of agonist at 1-s intervals for 60 s and at 5 s intervals for a further 60 s. Agonist concentration-response curves were constructed using the peak stimulation (minus basal) for each concentration of agonist.
Inositol Phosphate Accumulation Scintillation Proximity Assay. Inositol phosphate accumulation scintillation proximity assays were carried out as described in Brandish et al. (2003) with minor modifications. CHO-hM1 or CHO-hM3 cells were seeded into clear-bottomed black-walled 96-well microtiter plates (Costar UK Ltd.) at a density of 50,000 cells/well. After 24 h, the cell medium was removed and replaced with 100 μl of inositol-free (IF) DMEM containing 10 μCi/ml myo-[3H]inositol (GE Healthcare). After 16 h, the myo-[3H]inositol was removed, and the cells washed with 2 × 200 μl of IF-DMEM. Cells were then incubated with 100 μl of agonist and 100 μl of antagonist/media in IF-DMEM containing 5 mM LiCl for 1 h at 37°C. The reaction was terminated by the removal of the medium and the addition of 200 μl of 0.1 M formic acid (Sigma Chemical) to extract inositol phosphates. The extraction was allowed to continue for 1 h at room temperature after which 20-μl samples of the extracts were removed from each well and combined with 80 μl of RNA binding yttrium silicate scintillation proximity assay beads (12.5 mg/ml) in opaque-bottomed 96-well picoplates (PerkinElmer Life and Analytical Sciences, Boston, MA). Plates were agitated for 1 h at room temperature, and then the beads were left to settle for 2 h. Total [3H]inositol phosphate accumulation (cpm) for each concentration of agonist was determined on a TopCount plate counter (PerkinElmer Life and Analytical Sciences) and used to construct concentration-response curves.
CHO-hM1 Cell Membrane Preparation. Cell membrane preparation was carried out as described in Lanzafame and Christopoulos (2004); all procedures were carried out at 4°C. CHO-hM1 cells were harvested as described above. The cell pellet was resuspended in 5 ml of ice-cold buffer containing 50 mM Tris-HCl, 3 mM MgCl2, and 0.2 mM EGTA, pH 7.4, and then homogenized using an Ultra-Turrax homogenizer (IKA-Werke GmbH, Staufen, Germany) for 30 s. The homogenate was centrifuged at 1000g for 10 min, the remaining pellet was discarded, and the supernatant re-centrifuged at 39,000g for 30 min. The resulting cell membrane pellet was resuspended in the same buffer at a protein concentration of 2 mg/ml as determined by the method of Bradford (1976). Aliquots were stored at -80°C until further use.
CHO-hM1 [3H]NMS Inhibition Binding Assays. Inhibition binding studies were performed by incubating CHO-hM1 cell membranes (10 μg of protein/well) with [3H]NMS (0.2 or 2 nM) and a range of concentrations of the test compound at 37°C for 1 h in buffer containing 50 mM HEPES, 110 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl2, 1 mM MgSO4, 25 mM glucose, and 58 mM sucrose, pH 7.4. The final assay volume was 0.5 ml. Compounds were serially diluted in HEPES buffer to generate 10 different test concentrations over a 5 log unit range. Nonspecific binding was defined as that remaining in the presence of 1 μM atropine. The reaction was terminated by rapid filtration through GF/B filter mates presoaked in double distilled H2O using a 48-well Brandel cell harvester (Semat International, Ltd., St. Albans, Herts, UK) Filters were washed with 5 × 1 ml of ice-cold 0.9% (w/v) NaCl solution and allowed to dry before bound radioactivity was measured using liquid scintillation counting.
CHO-hM1 [3H]NMS Kinetic Binding Assays. For dissociation studies, CHO-hM1 cell membranes (10 μg of protein/well) were preincubated with [3H]NMS (0.2 nM) and either HEPES buffer or atropine (1 μM final) for 60 min in a final assay volume of 0.55 ml. Triplicate determinations of total and nonspecific binding of [3H]NMS, as defined above, were then measured at 0.5, 1, 2, 5, 10, 20, and 30 min after the addition of atropine (1 μM final) either in the presence or absence of AC-42 (100 μM final) or gallamine (1 mM final). Buffer composition, reaction termination, and radioactivity determination were all as for CHO-hM1 inhibition binding.
Data Analysis. For the functional assays, agonist concentration-response curves, in the absence and presence of antagonist, were globally fitted to the following logistic equation (Motulsky and Christopoulos, 2004) using Prism 4 (GraphPad Software Inc., San Diego, CA): where Top represents the maximal asymptote of the curves, Bottom represents the lowest asymptote (basal response) of the curves, log EC50 represents the logarithm of the agonist EC50 in the absence of antagonist, [A] represents the concentration of the agonist, [B] represents the concentration of the antagonist, nH represents the Hill slope of the agonist curve, s represents the Schild slope for the antagonist, and pA2 represents the negative logarithm of the concentration of antagonist that shifts the agonist EC50 by a factor of 2. In the absence of antagonist ([B] = 0), this equation becomes the standard four-parameter logistic equation for fitting agonist concentration-response data. When fitting agonist concentration-response data obtained in the presence of antagonist, if the estimated Schild slope was not significantly different from unity, it was constrained as such and the estimate of pA2 represented the pKB. For presentation purposes, concentration-ratios were also derived for the concentration-response curves in the absence or presence of antagonist and plotted in the form of a Schild regression (Kenakin, 1997). Where appropriate, concentration-response data were also fitted to the following equation, which is based on a simple allosteric ternary complex model of ligand-receptor interaction (Ehlert, 1988; Lazareno and Birdsall, 1995; Christopoulos and Kenakin, 2002): where Top, Bottom, log EC50, nH, and [A] are as described above for eq. 1, [B] denotes the concentration of allosteric ligand, KB denotes its equilibrium dissociation constant for binding to the allosteric site on the receptor, and α denotes the cooperativity factor, which is a thermodynamic measure of the strength of allosteric interaction between the orthosteric and allosteric sites. Values of α > 1 denote positive cooperativity (allosteric enhancement), whereas values of α < 1 denote negative cooperativity (allosteric inhibition).
Radioligand inhibition binding isotherms for the interaction between two different concentrations of [3H]NMS and either atropine or pirenzepine were globally fitted according to the following one-site orthosteric mass-action equation: where Y denotes the percentage of specific binding; Top and Bottom denote the maximal and minimal asymptotes, respectively; [A] denotes the concentration of [3H]NMS (either 0.2 or 2 nM); [B] denotes the unlabeled antagonist concentration; and KA and KB denote the equilibrium dissociation constants of A and B, respectively. For each series of antagonist curves, the negative logarithm of the KA value for [3H]NMS was fixed at 9.6 (determined separately in saturation binding assays), and the KB parameter was constrained to be shared across both curves.
Radioligand inhibition binding isotherms for the interaction between two different concentrations of [3H]NMS and either AC-42 or gallamine were globally fitted according to the following equation, based on the simple ternary complex model of allosteric interaction: where Y/Ymax denotes fractional receptor occupancy, [A] and KA are as defined for eq. 3 above, and [B], KB, and α are as described for eq. 2 above. For each series of inhibition curves, the negative logarithm of the KA value for [3H]NMS was fixed at 9.6 (as above), and the KB and α parameters were constrained to be shared across both curves.
For the dissociation kinetic experiments, dissociation curves for the radioligand, in the absence and presence of applied modulator, were evaluated by nonlinear regression using the following equation: where Bt denotes the specific binding of radioligand at time t, B0 denotes the specific binding of radioligand at equilibrium (time = 0), and koffobs denotes the observed radioligand dissociation rate constant. For the gallamine/[3H]NMS dissociation kinetic experiments, data sets were globally fitted to eq. 5 with the parameter koffobs defined as follows. where koff denotes the rate constant for dissociation of the radioligand from the free (unoccupied) receptor, koffB denotes the rate constant for the dissociation of the radioligand from the modulator-occupied receptor (fixed to zero for this analysis; see Results), and KB/α denotes the dissociation constant of the modulator for the radioligand-occupied receptor (Lazareno and Birdsall, 1995; Christopoulos et al., 1999). Note that in the absence of modulator, koffobs = koff.
In all cases, potency, affinity, and cooperativity values were estimated as logarithms (Christopoulos, 1998). Data shown are the mean ± S.E.M. Comparisons between mean values were performed by F-test. Unless otherwise stated, values of p < 0.05 were taken as statistically significant.
Results
Functional Analyses of Agonist-Antagonist Interactions. Both carbachol and AC-42 stimulated IP accumulation in CHO-hM1 cells, with pEC50 values of 6.35 ± 0.02 and 6.21 ± 0.10, respectively (Fig. 1). AC-42 was a partial agonist with respect to carbachol (F-test; p < 0.001), with an intrinsic activity of 0.85. Although carbachol also caused a robust stimulation of IP accumulation in CHO-hM3 cells (pEC50 = 6.81 ± 0.04), AC-42 did not show any appreciable agonist activity at this mAChR subtype at concentrations up to 100 μM (Fig. 1).
Subsequent IP accumulation assays investigated the interaction between carbachol or AC-42 and the nonselective mAChR antagonist atropine or the relatively M1-selective antagonist pirenzepine. As shown in Fig. 2, both atropine and pirenzepine produced concentration-dependent rightward shifts in the concentration-response curve to carbachol, with no significant effect on the basal or maximal agonist response, and with Schild slopes not significantly different from unity (Table 1). Because this behavior is consistent with the expectations of simple orthosteric antagonism (Kenakin, 1997), pKB values for each antagonist were also calculated and are shown in Table 1. It is noteworthy that when these experiments were repeated with AC-42 rather than carbachol as the agonist, a different pattern of behavior was noted. In particular, although both atropine and pirenzepine produced concentration-dependent rightward shifts in the concentration-response curve to AC-42 with no significant effect on the basal or maximal agonist response (Fig. 3), the magnitude of these shifts deviated significantly from that expected for orthosteric antagonism. This was manifested as a Schild slope significantly less than 1 for each antagonist (Table 2) and because the AC-42/antagonist IP data did not satisfy the minimum criteria expected for orthosteric antagonism, a pKB value was not determined. However, a pA2 value was still calculated, using eq. 1 under Materials and Methods, as an empirical measure of antagonist potency (Table 2).
To complement the IP accumulations assays, additional experiments were performed using intracellular calcium mobilization as the functional readout. Both carbachol and AC-42 stimulated calcium mobilization in CHO-hM1 cells with pEC50 values of 8.38 ± 0.11 and 7.15 ± 0.12, respectively. Moreover, both ligands were now full agonists in this assay.
Figure 4 illustrates the interaction between carbachol or AC-42 with pirenzepine in the calcium mobilization assay. In agreement with the results of the IP accumulation assay, the interaction between carbachol and pirenzepine (Fig. 4A) was completely consistent with simple orthosteric antagonism; the resulting Schild slope and pKB values are shown in Table 1, where it can also be seen that there was no significant difference between the pKB estimates for pirenzepine from either assay. In contrast, but in agreement with the findings from the IP accumulation assays, the interaction between AC-42 and pirenzepine was characterized by a Schild slope that was significantly less than 1 (Table 2). The estimated pA2 value for this interaction is also shown in Table 2.
At the molecular level, Schild slopes that deviate significantly from unity may be indicative of a negative allosteric interaction between agonist and antagonist (Christopoulos and Kenakin, 2002). In this instance, the reduction in Schild slope reflects the saturability of the negative allosteric interaction, because increasing the concentration of one ligand results in a progressive inability to further antagonize the other ligand. To experimentally confirm that this was the case, a large range of antagonist concentrations was tested in an attempt to reveal saturable antagonism (Christopoulos and Kenakin, 2002). Figure 5 shows the results of calcium mobilization experiments that were specifically designed to investigate whether the interaction between atropine and AC-42 was mechanistically and quantitatively consistent with an allosteric model of interaction. When tested against the classic orthosteric agonist carbachol, atropine inhibited M1 receptor-mediated calcium mobilization in a manner completely consistent with orthosteric antagonism; i.e., the Schild slope was not significantly different from unity over concentrations of antagonist spanning 6 orders of magnitude (Fig. 5A). In contrast, the ability of the highest concentrations of atropine to inhibit the effects of AC-42 clearly approached a limiting value as exemplified by a curvilinear Schild regression (Fig. 5B). Analysis of the data set according to an orthosteric model (eq. 1 under Materials and Methods) yielded a Schild regression with a slope factor of 0.69 ± 0.01, significantly less than unity (Fig. 5B; Table 2). Because the AC-42/atropine data shown in Fig. 5 were consistent with a negative allosteric interaction, the data were fitted to the allosteric ternary complex model (eq. 2 under Materials and Methods), which yielded the parameter values listed in Table 2. From this analysis, the interaction between atropine and AC-42 was characterized by an α value of ca. 0.0002. Statistical comparison of the fit according to the allosteric ternary complex model [eq. 2 and the fit according to eq. 1 in which the slope value was constrained to 1 (indicative of a true competitive interaction)] revealed a clear preference for the former (F-test; p < 0.0001).
[3H]NMS Inhibition Binding Studies. The analysis of the functional data presented above assumes that the molecular mechanism governing the interaction between AC-42 and orthosteric antagonists, such as atropine, is an allosteric modification of ligand binding affinity (quantified by α). To probe and validate this proposed mechanism, radioligand binding assays were performed using [3H]NMS as an orthosteric-site “probe”. Both atropine and pirenzepine fully inhibited specific [3H]NMS binding in membranes from CHO-hM1 cells at radioligand concentrations of either 0.2 nM (ca. KA)or 2 nM (ca. 10× KA; Fig. 6A), as expected for simple orthosteric antagonism. A global analysis of each data set, according to eq. 3 under Materials and Methods, yielded pKB values of 9.37 ± 0.04 for atropine and 7.99 ± 0.05 for pirenzepine. These values were also in excellent agreement with the pKB values determined from the functional assays for these antagonists (Table 1).
Although AC-42 also inhibited [3H]NMS binding in a concentration-dependent manner, it was unable to fully reduce specific binding levels to zero; this was more marked when the concentration of radioligand was increased 10-fold (Fig. 6B); an effect consistent with a saturability of antagonism characteristic of a negatively cooperative allosteric interaction. A global analysis of each pair of inhibition curves, according to the allosteric ternary complex model of interaction (eq. 4), yielded a pKB value of 6.19 ± 0.07 as a measure of the affinity of AC-42 for its allosteric site on the unoccupied M1 receptor, and a log α value of -1.66 ± 0.12 (i.e., α = 0.022). The large difference between the cooperativity between AC-42 and atropine (0.0002) and that between AC-42 and [3H]NMS is a striking example of the probe dependence of allosteric interactions.
For comparison, similar experiments were also performed with the well characterized and prototypical negative mAChR allosteric modulator gallamine. As expected, gallamine also displayed an inability to fully inhibit the specific binding of the radioligand (Fig. 6B). Global analysis of the two curves yielded estimates of pKB = 5.19 ± 0.06 and log α = -1.28 ± 0.06 (α = 0.053).
[3H]NMS Dissociation Binding Studies. Because allosteric modulators of G protein-coupled receptors cause a conformational change in the receptor that is often manifested as an alteration of the dissociation characteristics of a preformed orthosteric ligand-receptor complex (Kostenis and Mohr, 1996), [3H]NMS dissociation kinetic assays were performed. Radioligand dissociation from CHO-hM1 cell membranes under control conditions was rapid and monophasic, with a koff value of 0.19 ± 0.02/min. The presence of gallamine (1 mM) produced a marked retardation of [3H]NMS dissociation, yielding a koffobs of 0.017 ± 0.001/min (Fig. 7; t test; p < 0.01). Because previous studies have shown that gallamine is able to completely prevent [3H]NMS dissociation from mAChRs (Lazareno and Birdsall, 1995), the data were refitted to eq. 6, with the parameter, koffB, constrained to a value of zero, yielding an estimate of the apparent affinity of gallamine for the [3H]NMS-occupied receptor (log KB/α) of -3.86 ± 0.12. This estimate is in excellent agreement with the expected value calculated using the pKB and log α estimates derived from the inhibition binding assays (log KB/α = -3.91), further validating the allosteric ternary complex model as an appropriate mechanistic descriptor of the interactions measured in the current study.
It is noteworthy that AC-42 (100 μM) also significantly slowed the rate of [3H]NMS dissociation, yielding a koffobs value of 0.122 ± 0.005/min (Fig. 7; p < 0.05), although this effect was much less marked than that for gallamine. Because of the modest effect of AC-42 at a concentration that is close to its solubility limits, however, it was not possible to determine the maximum extent of its ability to retard [3H]NMS dissociation. Nonetheless, the ability of AC-42 to significantly alter the dissociation rate of [3H]NMS confirmed that this agonist is also an allosteric modulator.
Discussion
Our findings provide direct evidence that the functionally selective M1 mAChR agonist AC-42 is an allosteric modulator. An investigation of the mode of interaction of this agonist with various orthosteric ligands confirmed that AC-42 exhibited characteristics associated with allosteric ligands, including 1) a saturability to its allosteric effects, 2) a dependence of these effects on the nature of the orthosteric ligand with which it interacted, and 3) the ability to significantly retard the dissociation of [3H]NMS from the M1 mAChR. These results reveal that the M1 mAChR possesses an allosteric site capable of mediating receptor activation in its own right and further reinforce the utility of allosteric binding sites for achieving selectivity at G protein-coupled receptors.
In agreement with previous studies (Spalding et al., 2002; Sur et al., 2003), AC-42 caused a robust stimulation in both IP accumulation and intracellular calcium mobilization in CHO-hM1 cells. The significantly higher potency for both agonists in the calcium assay, relative the IP assay, and the accompanying increase in maximal AC-42 response to match that of carbachol, are consistent with an increase in stimulus-response coupling resulting from signal amplification in the calcium response, which is downstream of the IP pathway. The partial agonist activity of AC-42 in the IP accumulation assay relative to carbachol is also consistent with the study by Spalding et al. (2002). Because partial agonists require maximal receptor occupancy to generate their maximal responses, the pEC50 value of a partial agonist is a useful approximation of its affinity for the receptor (Kenakin, 1997). It is thus noteworthy that the pEC50 in the IP accumulation assay (6.21) is in excellent agreement with the pKB value estimated for the binding of AC-42 according to an allosteric model (6.19).
In functional assays, the best-characterized methods for investigating the mechanism of interaction between agonists and antagonists are the procedures underlying Schild analysis and its variants (Arunlakshana and Schild, 1959; Kenakin, 1997; Motulsky and Christopoulos, 2004). Atropine or pirenzepine antagonized the effects of carbachol in a manner that was consistent with competition for a common (orthosteric) binding site. This was most evident with atropine in the calcium mobilization assay, where a Schild slope of 1 was maintained over antagonist concentrations spanning 6 orders of magnitude (Fig. 5A). The resulting pKB values (Table 1) were consistent with previously reported values in the literature for the binding of atropine or pirenzepine to the M1 mAChR (Ellis, 2002). In contrast, the translocation of the AC-42 concentration-response curve by atropine or pirenzepine in either functional assay format led to Schild slopes significantly less than unity (Table 2), indicating that the antagonism was not consistent with a simple orthosteric interaction.
A clue to the underlying mechanism of action of AC-42 was provided by the atropine antagonism data in the calcium mobilization assay. Negative allosteric interactions have previously been demonstrated in functional assays by a progressive inability of the antagonist to shift the agonist concentration-response curve with increasing antagonist concentrations (Clark and Mitchelson, 1976; Lanzafame et al., 1996), yielding a characteristic curvilinear Schild regression. As shown in Fig. 5B, this was exactly the type of behavior exhibited by atropine when tested against AC-42. However, the conclusion from our experiments is that AC-42 must be the allosteric ligand, because atropine is a well characterized orthosteric antagonist of mAChRs. Subsequent analysis of the data according to an allosteric ternary complex model yielded a pKB estimate for atropine (Table 2) that was in good agreement with its known affinity for the orthosteric site but that also provided an estimate of the cooperativity between atropine and AC-42 (α = 0.0002). Thus, it can be concluded that the low Schild slopes associated with the interaction between atropine or pirenzepine and AC-42 actually reflect the saturability (governed by α in the ternary complex model) of an allosteric interaction between antagonist and agonist. Unfortunately, we could not estimate α values for the interaction between pirenzepine and AC-42 because of the limited number of antagonist concentrations that were able to be studied.
More direct evidence of an allosteric mode of action for AC-42 was obtained using radioligand binding. It is well established that allosteric modulators can have characteristic effects on the binding of orthosteric ligands (Christopoulos and Kenakin, 2002). For example, binding studies using two or more radioligand concentrations can reveal differences in the binding isotherm of the allosteric modulator because the saturability in the allosteric event becomes more evident as the concentration of orthosteric radioligand is increased (Christopoulos and Kenakin, 2002). Therefore, and in contrast to atropine and pirenzepine (Fig. 6A), AC-42 was unable to fully inhibit the specific binding of [3H]NMS at either 0.2 nM, or more strikingly, at the higher radioligand concentration of 2 nM, a property that was shared by the prototypical modulator gallamine (Fig. 6B). The values derived for gallamine via the application of an allosteric model are in good agreement with similar values determined previously (Matsui et al., 1995), whereas our study is the first to apply such a model to AC-42. It is noteworthy that the estimated α value for the interaction between AC-42 and [3H]NMS was significantly different from the value describing the interaction between AC-42 and atropine. This highlights a second characteristic of allosteric interactions (in addition to saturability), namely, that the degree of cooperativity between orthosteric and allosteric sites depends on the chemical nature of the ligands occupying each site. This probe dependence of allosterism can explain why in some studies, an allosteric interaction may be missed altogether if it is characterized by either neutral cooperativity (i.e., seems as if there is no interaction) or by very high negative cooperativity (i.e., seems as if the interaction is competitive).
A further method used to validate an allosteric mechanism of action is to monitor the effects of a putative modulator on the dissociation of a preformed orthosteric ligand-receptor complex. Any alteration in the dissociation of such a complex must be indicative of a conformational change mediated via a topographically distinct site (Kostenis and Mohr, 1996). Indeed, prototypical allosteric modulators of mAChRs, such as C7/3-phth (Christopoulos et al., 1999), strychnine (Lazareno and Birdsall, 1995), and gallamine (Stockton et al., 1983) have all been shown to allosterically retard the rate of [3H]NMS dissociation from mAChRs in a concentration-dependent manner. Thus, the most striking confirmation of the allosteric nature of AC-42 and gallamine was obtained from [3H]NMS dissociation kinetic assays. As illustrated in Fig. 7, both AC-42 and gallamine retarded the dissociation of [3H]NMS from the M1 mAChR. This effect was much more pronounced for gallamine for at least two reasons. First, at the concentrations of each modulator used in the kinetic assays (reflecting solubility maxima), it can be calculated from their relative KB and α values that 1 mM gallamine would occupy ca. 98% of the [3H]NMS-bound receptors, whereas 100 μM AC-42 would only occupy ca. 77% of the [3H]NMS-bound receptors. Second, it is possible that AC-42 simply cannot exert the same maximal effect on radioligand kinetics as gallamine.
As a whole, the data presented here provide unambiguous evidence that AC-42 is appropriately classed as an allosteric, rather than an ectopic, agonist of the muscarinic M1 receptor; i.e., it meets all the criteria for classification as an allosteric ligand in addition to possessing efficacy at the receptor in the absence of other ligands. The prevalence of other confirmed allosteric agonists of G protein-coupled receptors remains relatively limited in the current literature. There is some evidence that the adenosine A1 receptor enhancer PD 81,723 is able to promote a modest degree of receptor G protein-coupling in its own right (Bhattacharya and Linden, 1995). Likewise, the peptide ASLW has been shown to activate the chemokine CXCR4 receptor in a manner that is likely to be allosteric (Sachpatzidis et al., 2003). For the mAChRs, Jakubik et al. (1996) demonstrated that gallamine and alcuronium could promote receptor activation, but these results are not consistently reproduced (Lazareno and Birdsall, 1995; May et al., 2005) and may reflect the requirement of particular stoichiometries of receptor to G protein (Jakubik et al., 1998). Earlier radioligand binding studies suggested that the partial agonist McN-A-343 was also an allosteric modulator of M2 mAChRs (Birdsall et al., 1983), although a subsequent functional study was unable to conclude whether this compound interacted allosterically with very high negative cooperativity or via simple orthosteric competition when tested against carbachol (Christopoulos and Mitchelson, 1997).
In contrast, we have now shown, using functional, equilibrium binding, and dissociation kinetic assays that the mechanism of action of AC-42 at the M1 mAChR is qualitatively and quantitatively consistent with an allosteric model. This finding not only has important ramifications for mAChR-based drug discovery but also raises additional considerations. For example, is the allosteric site recognized by AC-42 the same site as that recognized by gallamine, a second allosteric site recognized by certain staurosporine derivatives (Lazareno et al., 2000), or a third topographically distinct allosteric site? What is it about the nature of the AC-42 binding site that allows almost exclusive activation of the M1 mAChR relative to other mAChR subtypes? Further studies are clearly warranted to elucidate the molecular nature of this allosteric site.
Footnotes
- Received August 8, 2005.
- Accepted October 5, 2005.
A.C. is a Senior Research Fellow of the National Health and Medical Research Council of Australia. Article, publication date, and citation information can be found at http://molpharm.aspetjournals.org.
doi:10.1124/mol.105.017814.
ABBREVIATIONS: mAChR, muscarinic acetylcholine receptor; NMS, N-methyl scopolamine; C7/3-phth, heptane-1,7-bis-(dimethyl-3′-phthalimidopropyl)-ammonium bromide; TM, transmembrane; AC-42, 4-n-butyl-1-[4-(2-methylphenyl)-4-oxo-1-butyl]-piperidine; IP, inositol phosphate; CHO, Chinese hamster ovary; FLIPR, fluorometric imaging plate reader; IF, inositol-free; DMEM, Dulbecco's modified Eagle's medium; PD 81,723, (2-amino-4,5-dimethyl-3-thienyl)(3-(trifluoromethyl)phenyl)-methanone; McN-A-343, 4-(m-chlorophenyl-carbamoyloxy)-2-butynyltrimethylammonium chloride.
- The American Society for Pharmacology and Experimental Therapeutics